1 NCEA-C-1765
2 November 2006
3
4
5
6 Toxicological Reviews of Cyanobacterial
7 Toxins:
Microcystins LR, RR, YR and LA
10
11
12
13
14
15
16
17
18
19
20
21 National Center for Environmental Assessment
22 Office of Research and Development
23 U.S. Environmental Protection Agency
24 Cincinnati, OH 45268
25
-------
1 DISCLAIMER
2
o
J
4 This report is an external draft for review purposes only and does not constitute Agency
5 policy. Mention of trade names or commercial products does not constitute endorsement or
6 recommendation for use.
DRAFT - DO NOT CITE OR QUOTE
-------
1 TABLE OF CONTENTS
2
o
3
4 LIST OF TABLES vi
5 LIST OF FIGURES vii
6 LIST OF ACRONYMS viii
7 FOREWORD x
8 AUTHORS, CONTRIBUTORS, AND REVIEWERS xi
9
10 1. INTRODUCTION 1
11
12 2. CHEMICAL AND PHYSICAL INFORMATION 3
13
14 3. TOXICOKINETICS 5
15 3.1. ABSORPTION 5
16 3.2. DISTRIBUTION 6
17 3.2.1. Organ Distribution 6
18 3.2.2. Cellular Uptake 8
19 3.2.3. Subcellular Localization of Cystolic Protein Binding 9
20 3.3. METABOLISM 10
21 3.4. ELIMINATION 13
22 3.5. PHYSIOLOGICALLY-BASED TOXICOKINETIC MODELS 14
23
24 4. HAZARD IDENTIFICATION 15
25 4.1. STUDIES IN HUMANS - EPIDEMIOLOGY, CASE REPORTS, CLINICAL
26 CONTROLS 15
27 4.1.1. Oral Exposure 15
28 4.1.1.1. Short-Term Studies and Case Reports 15
29 4.1.1.2. Long-Term Studies and Epidemiological Studies 19
30 4.1.2. Inhalation Exposure 24
31 4.2. ACUTE, SHORT-TERM, LONG-TERM AND CHRONIC STUDIES AND
32 CANCER BIOASSAYS IN ANIMALS - ORAL AND INHALATION 24
33 4.2.1. Oral Exposure 25
34 4.2.1.1. Acute Studies 25
35 4.2.1.2. Short-Term Studies 28
36 4.2.1.3. Subchronic Studies 31
37 4.2.1.4. Chronic Studies 34
38 4.2.1.5. Initiation/Promotion Studies - Cyanobacterial Extracts 36
39 4.2.2. Inhalation Exposure 36
40 4.2.2.1. Acute Studies 36
41 4.2.2.2. Short-Term Studies 37
42 4.2.2.3. Subchronic and Chronic Studies 37
iii DRAFT - DO NOT CITE OR QUOTE
-------
1 TABLE OF CONTENTS cont.
2
o
3
4 4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES - ORAL AND INHALATION...39
5 4.3.1. Oral Exposure 39
6 4.3.1.1. Purified Microcystins 39
7 4.3.1.2. Cyanobacterial Extracts 39
8 4.3.2. Inhalation Exposure 39
9 4.4. OTHER STUDIES 40
10 4.4.1. Neurological Effects 40
11 4.4.2. Immunological Effects 40
12 4.4.3. Hematological Effects 41
13 4.4.4. Effects by Dermal Exposure 41
14 4.4.5. Effects by Parenteral Exposure 42
15 4.4.5.1. Effects in Humans after Parenteral Exposure 42
16 4.4.5.2. Effects in Animals after Parenteral Exposure 43
17 4.4.6. Effects by Intratracheal or Intranasal Instillation 54
18 4.4.7. Mechanistic Studies 54
19 4.4.7.1. Target Organ/Cell Type Specificity 55
20 4.4.7.2. Characterization of Subcellular Effects in the Liver 56
21 4.4.7.3. Molecular Target: Inhibition of Type 1 and 2A Protein
22 Phosphatases 58
23 4.4.7.4. Cytoskeletal Effects 62
24 4.4.7.5. Apoptosis 63
25 4.4.7.6. Lipid Peroxidation 64
26 4.4.7.7. Prevention of Liver Toxicity and Lethality 65
27 4.4.7.8. Extra-Hepatic Effects of Mi crocystins 66
28 4.4.8. Genotoxicity and Cell Proliferation 67
29 4.4.9. Structure-Activity Relationships 69
30 4.5. MODE OF ACTION - NONCANCER AND CANCER 70
31 4.5.1. Target Organ Specificity 70
32 4.5.2. Key Events in the Mode of Action for Liver Toxicity and Hemorrhage 71
33 4.5.2.1. Molecular Events 71
34 4.5.2.2. Cellular Effects 73
35 4.5.3. Conclusion 73
36 4.6. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS 74
37 4.6.1. Oral 75
38 4.6.2. Inhalation 78
39 4.7. WEIGHT-OF-EVIDENCE EVALUATION AND CANCER
40 CHARACTERIZATION 79
41 4.7.1. Summary of Overall Weight of Evidence 79
42 4.7.2. Synthesis of Human, Animal and Other Supporting Evidence 79
43 4.8. SUSCEPTIBLE POPULATIONS AND LIFE STAGES 81
44
iv DRAFT - DO NOT CITE OR QUOTE
-------
1 TABLE OF CONTENTS cont.
2
o
3
4 5. DOSE-RESPONSE ASSESSMENTS 82
5 5.1. NARRATIVE DESCRIPTION OF THE EXTENT OF THE DATABASE 82
6 5.2. ORAL REFERENCE DOSE 83
7 5.2.1. Acute Oral RfD 84
8 5.2.2. Short-Term Oral RfD 85
9 5.2.2.1. Choice of Principal Study and Critical Effect 85
10 5.2.2.2. Methods of Analysis 85
11 5.2.2.3. RfD Derivation 89
12 5.2.3. Subchronic Oral RfD 90
13 5.2.3.1. Choice of Principal Study and Critical Effect 90
14 5.2.3.2. Methods of Analysis 91
15 5.2.3.3. RfD Derivation 92
16 5.2.4. Chronic Oral RfD 92
17 5.2.4.1. Choice of Principal Study and Critical Effect 92
18 5.2.4.2. RfD Derivation 93
19 5.3. INHALATION REFERENCE CONCENTRATION 93
20 5.4. CANCER ASSESSMENT 95
21
22 6. MAJOR CONCLUSIONS IN THE CHARACTERIZATIONS OF HAZARD AND
23 DOSE RESPONSE 96
24 6.1. HUMAN HAZARD POTENTIAL 96
25 6.2. DOSE RESPONSE 97
26
27 7. REFERENCES 98
28
29 APPENDIX A A-l
v DRAFT - DO NOT CITE OR QUOTE
-------
1 LIST OF TABLES
2
3
4 No. Title
5
6 4-1 Relative Risk of Colorectal Cancer and Microcystin Concentration by Drinking
7 Water Source 20
8
9 4-2 Incidence of Liver Lesions in Mice and Rats Treated with a Single Dose of
10 MCLR 26
11
12 4-3 Serum Enzyme Levels and Relative Liver Weights (Mean + Standard Deviation
13 in Rats Ingesting MCLR in Drinking Water 29
14
15 4-4 Incidence of Liver Lesions in Rats Ingesting MCLR in Drinking Water for
16 28 Days 30
17
18 4-5 Blood Chemistry Results (Mean + Standard Deviation) for Mice Treated
19 with MCLR for 13 Weeks 32
20
21 4-6 Incidence of Liver Histopathology in Mice Treated with MCLR for
22 13 Weeks 33
23
24 4-7 Incidence and Severity of Nasal Cavity Lesions in Mice Inhaling Microcystin
25 Aerosol for 7 Days 38
26
27 4-8 LDso Values of Purified Microcystin Congeners by Intraperitoneal Administration 44
28
29 4-9 Intraperitoneal LD50 Values for Bloom Extracts 49
30
31 4-10 Studies Comparing Protein Phosphatase Inhibition Activity of Microcystin
32 Congeners 59
o o
JJ
34 4-11 Summary Noncancer Results in All Animal Studies of Oral Exposure to Purified
35 Microcystin-LR 76
36
37 5-1 Available Dose-Response Information for Oral Exposure to Purified MCLR 84
38
39 5-2 Incidence of Liver Lesions Used for BMD Modeling 86
40
41 5-3 BMD Modeling Results for Heinze (1999) Liver Lesion Data 87
42
43 5-4 BMD Modeling Results for Fawell et al. (1999) Chronic Liver Inflammation Data 91
44
45 5-5 BMD Modeling Results for Fawell etal. (1999) ALT Data in Male Mice 92
46
47 5-6 Summary of Reference Dose Values 94
vi DRAFT - DO NOT CITE OR QUOTE
-------
1 LIST OF FIGURES
2
o
J
4 2-1 General Structure of Microcystins 3
5
6 3-1 Structures of GSH and Cys Conjugates of Microcystins LR, YR, andRR 12
7
8 4-1 Relationship between Colorectal Cancer and Microcystin Concentration in
9 River and Pond Water in Haining City, China (Zhou et al., 2002) 22
10
11 4-2 Schematic Representation of Interactions between Microcystin-LR and the
12 Catalytic Site of Protein Phosphatase 1 61
13
14 5-1 Exposure Response Array for Oral Exposure: All Studies of Purified
15 Microcystin-LR 83
16
17 5-2 Probit Model Fit to Liver Lesion Incidence Data from Heinze (1999) 88
vii DRAFT - DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
AFB1
ALP
ALT
AST
BMD
BMDL
BMDS
BMR
BUN
CI
DEN
ELISA
EPA
ESR
FETAX
GD
GFR
GGT
GST-P
HPLC
i.p.
i.v.
LCso
LD50
LDH
LOAEL
MAPK
MCAR
LIST OF ACRONYMS
AflatoxinBl
Alkaline phosphatase
Alanine aminotransferase
Aspartate aminotransferase
Benchmark dose
Benchmark dose, lower confidence limit
Benchmark dose software
Benchmark response
Blood urea nitrogen
Confidence interval
Diethylnitrosamine
Enzyme-linked immunosorbent assay
Environmental Protection Agency
Electron spin resonance
Frog Embryo Teratogenicity Assay-Xenopus
Gestation day
Glomerular filtration rate
y-Glutamyltransferase
Glutathione S-transferase (placental form)
High pressure liquid chromatography
Intraperitoneal
Intravenous
Concentration lethal to 50% of population
Dose lethal to 50% of population
Lactate dehydrogenase
Lowest-observed-adverse-effect level
Mitogen-activated protein kinase
Microcystin-AR
Vlll
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
o
6
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
MCLA
MCLR
MCRR
MCYM
MCYR
MPT
MW
NMR
NOAEL
OATP
PAS
POD
PP1
PP2A
RfC
RfD
ROS
RVR
SDH
SRR
TDI
TEF
TUNEL
UF
VAF
WHO
Microcystin-LA
Microcystin-LR
Microcystin-RR
Mi crocy stin-YM
Microcystin-YR
Mitochondrial permeability transition
Molecular weight
Nuclear magnetic resonance
No-observed-adverse-effect level
Organic acid transport protein
Periodic acid-Schiff
Point of departure
Protein phosphatase 1
Protein phosphatase 2A
Reference concentration
Reference dose
Reactive oxygen species
Renal vascular rate
Sorbitol dehydrogenase
Standardized rate ratio
Tolerable Daily Intake
Toxicity equivalency factor
Terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick end-labeling
Uncertainty factor
Virus antibody free
World Health Organization
IX
DRAFT - DO NOT CITE OR QUOTE
-------
1 PREFACE
2
3
4 The Safe Drinking Water Act (SDWA), as amended in 1996, requires the Environmental
5 Protection Agency (EPA) to publish a list of contaminants which, at the time of publication, are
6 not subject to any proposed or promulgated national primary drinking water regulations, that are
7 known or anticipated to occur in public water systems, and which may require regulations under
8 SDWA. This list, known as the Contaminant Candidate List (CCL), was first published in 1998
9 and then again in 2005. The 1998 and 2005 CCLs include "cyanobacteria (blue-green algae),
10 other freshwater algae, and their toxins" as microbial contaminants.
11
12 In 2001, a meeting was held among EPA, researchers from the drinking water industry,
13 academia and government agencies with expertise in the area of fresh water algae and their
14 toxins. The goal of this meeting was to convene a panel of scientists to assist in identifying a
15 target list of algal toxins that are likely to pose a health risk in source and finished waters of the
16 drinking water utilities in the U.S. Toxin selection was based on four criteria: health effects,
17 occurrence in the United States, susceptibility to drinking water treatment and toxin stability.
18 Microcystins were identified at this meeting as being toxins of high priority based on those
19 criteria.
20
21 The National Center for Environmental Assessment has prepared this Toxicological
22 Review of Cyanobacterial Toxins: Microcystins (LR, RR, YR and LA) as one in a series of dose-
23 response assessments to support the health assessment of unregulated contaminants on the CCL.
24 The purpose of this document is to compile and evaluate the available data regarding
25 microcystin toxicity to aid the Office of Water in regulatory decision making. It is not intended
26 to be a comprehensive treatise on the chemical or toxicological nature of microcystins.
27
28 In Section 6, Major Conclusions in the Characterization of Hazard and Dose Response,
29 EPA has characterized its overall confidence in the quantitative and qualitative aspects of the
30 hazard and dose response by addressing knowledge gaps, uncertainties, quality of data and
31 scientific controversies. The discussion is intended to convey the limitations of the assessment
32 and to aid and guide the Office of Water in the ensuing steps of the human health risk assessment
33 of microcystins.
DRAFT - DO NOT CITE OR QUOTE
-------
1 AUTHORS, CONTRIBUTORS AND REVIEWERS
2
o
3
4 AUTHORS
5
6 Belinda Hawkins, Ph.D. (Chemical Manager)
7 National Center for Environmental Assessment
8 Office of Research and Development
9 U.S. Environmental Protection Agency
10 Cincinnati, OH
11
12 Heather Carlson-Lynch
13 Syracuse Research Corporation
14 Syracuse, NY
15
16 Marc Odin
17 Syracuse Research Corporation
18 Syracuse, NY
19
20 Julie Stickney
21 Syracuse Research Corporation
22 Syracuse, NY
23
24
25 REVIEWERS
26
27 INTERNAL EPA REVIEWERS
28
29 Joyce Donohue, Ph.D.
30 Office of Water
31 Washington, DC
32
33 Elizabeth Hilborn, D. V.M., M.P.H.
34 National Health and Environmental Effects Research Laboratory
35 Office of Research and Development
36 Research Triangle Park, NC
37
38 James Sinclair, Ph.D.
39 Office of Water
40 Cincinnati, OH
41
42 Jeff Swartout
43 National Center for Environmental Assessment
44 Office of Research and Development
45 Cincinnati, OH
xi DRAFT - DO NOT CITE OR QUOTE
-------
1 1. INTRODUCTION
2
o
J
4 This toxicological review presents background and justification for hazard and dose-
5 response assessments of microcystins LR, RR, YR and LA. U.S. Environmental Protection
6 Agency (EPA) toxicological reviews may include oral reference dose (RfD) and inhalation
7 reference concentration (RfC) values for chronic and less-than-lifetime exposure durations and a
8 carcinogenicity assessment.
9
10 The RfD and RfC provide quantitative information for use in risk assessments for health
11 effects known or assumed to be produced through a nonlinear (possibly threshold) mode of
12 action. These reference values are defined as an estimate of an exposure, designated by duration
13 and route, to the human population (including susceptible subgroups), that is likely to be without
14 an appreciable risk of adverse effects. Reference values may be derived for acute (<24 hours),
15 short-term (up to 30 days), subchronic (up to 10% of average lifespan) and chronic (up to
16 lifetime) exposures, all considered to be continuous exposures throughout the duration specified.
17 A reference value is derived from a BMDL (a statistical lower confidence limit on the
18 benchmark dose), a no-observed-adverse effect level (NOAEL), lowest-observed-adverse-effect
19 level (LOAEL) or other suitable point of departure with uncertainty/variability factors applied to
20 reflect limitations of the data used. The RfD is expressed in units of mg/kg-day, and the RfC in
21 units of mg/m3.
22
23 The carcinogenicity assessment provides information on the carcinogenic hazard
24 potential of the substance in question and quantitative estimates of risk from oral exposure and
25 inhalation exposure. The information includes a weight-of-evidence judgment of the likelihood
26 that the agent is a human carcinogen and the conditions under which the carcinogenic effects
27 may be expressed. Quantitative risk estimates are presented in three ways. The slope factor is
28 the result of application of a low-dose extrapolation procedure and is presented as the risk per
29 mg/kg-day. The unit risk is the quantitative estimate in terms of either risk per |J,g/L drinking
30 water or risk per |J,g/m3 air breathed. Another form in which risk is presented is a drinking water
31 or air concentration providing cancer risks of 1 in 10,000; 1 in 100,000; or 1 in 1,000,000.
32
33 Development of these hazard identification and dose-response assessments for
34 microcystins has followed the general guidelines for risk assessment as set forth by the National
35 Research Council (1983). EPA guidelines that were used in the development of this assessment
36 include the following: Guidelines for the Health Risk Assessment of Chemical Mixtures (U.S.
37 EPA, 1986a), Guidelines for Mutagenicity Risk Assessment (U.S. EPA, 1986b), Guidelines for
38 Developmental Toxicity Risk Assessment (U.S. EPA, 1991), Guidelines for Reproductive Toxicity
39 Risk Assessment (U. S. EPA, 1996), Guidelines for Neurotoxicity Risk Assessment (U. S. EPA,
40 1998a), Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), Supplemental Guidance
41 for Assessing Susceptibility from Early-Life Exposure to Carcinogens (U. S. EPA, 2005b),
42 Recommendations for and Documentation of Biological Values for Use in Risk Assessment (U.S.
43 EPA, 1988), (proposed) Interim Policy for Particle Size and Limit Concentration Issues in
44 Inhalation Toxicity (U.S. EPA, 1994a), Methods for Derivation of Inhalation Reference
45 Concentrations and Application of Inhalation Dosimetry (U.S. EPA, 1994b), Use of the
46 Benchmark Dose Approach in Health Risk Assessment (U.S. EPA, 1995), Science Policy Council
DRAFT - DO NOT CITE OR QUOTE
-------
1 Handbook: Peer Review (U.S. EPA, 1998b, 2000a, 2005c), Science Policy Council Handbook:
2 Risk Characterization (U.S. EPA, 2000b), Benchmark Dose Technical Guidance Document
3 (U.S. EPA, 2000c), Supplemental Guidance for Conducting Health Risk Assessment of Chemical
4 Mixtures (U.S. EPA, 2000d) and A Review of the Reference Dose and Reference Concentration
5 Processes (U. S. EPA, 2002).
6
7 Microcystins are a group of at least 80 naturally occurring hepatotoxins produced by
8 freshwater cyanobacteria (blue-green algae) including Microcystis, Anabaena, Nodularia,
9 Nostoc and Oscillatoria (Duy et al., 2000). Microcystins were first isolated from cyanobacterial
10 extracts in the 1980s (WHO, 1999).
11
12 Much of the toxicological research on microcystins has focused on a single congener,
13 microcystin-LR (MCLR). In addition to MCLR, this report focuses on three other major
14 microcystin congeners: microcystin-YR, microcystin-RR and microcystin-LA (abbreviated as
15 MCYR, MCRR and MCLA throughout this document). Literature searches were conducted for
16 studies relevant to the derivation of toxicity and carcinogenicity values for these four
17 microcystin congeners. The following databases were searched: MEDLINE (PubMed),
18 TOXLINE, BIOSIS, CANCERLIT, TSCATS, CCRIS, DART/ETIC, EMIC, GENETOX, HSDB
19 andRTECS. The relevant literature was reviewed through May 2006.
DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
2. CHEMICAL AND PHYSICAL INFORMATION
Microcystins are monocyclic heptapeptide toxins produced by a number of
cyanobacterial species, including members ofMicrocystis, Anabaena, Nodularia, Nostoc and
Oscillatoria (Duy et al., 2000). At least 80 microcystin congeners have been identified. A
general structure for microcystins is shown in Figure 2-1.
Me
Me OMe
H I II
C=C—C=C—C—C—C—Ph
H
H H H
8
10
Figure 2-1. General Structure of Microcystins
Microcystins are monocyclic heptapeptides consisting of D-alanine (Ala); two variable
amino acids (at positions X and Z in Figure 2-1); D-p-methylaspartic acid (MeAsp);
(2S,3S,8S,9S)-3-amino-9-methoxy-2,6,8-trimethyl-10-phenyldeca-4,6-dienoic acid (Adda); iso-
linked D-glutamic acid (Glu) and TV-methyl dehydroalanine (MDha). Structural variations occur
in all seven of the amino acid peptides, but most commonly in the L-amino acids at positions X
and Z in Figure 2-1 (shown as "variable" amino acids in the figure). The most common L-amino
acids at position X are leucine (L), arginine (R) and tyrosine (Y), while those at position Z are
arginine (R) and alanine (A). The congeners take their names from the L-amino acids in these
positions. For example, the microcystin congener with leucine in the X position and arginine in
the Z position is microcystin-LR.
Little information on the chemical and physical properties of microcystins was located.
Duy et al. (2000) provided the most thorough (albeit general) review of the properties of
microcystins. The microcystins identified to date have molecular weights (MWs) ranging from
900 to 1200. Microcystins are nonvolatile and generally quite hydrophilic, although a few have
lipophilic properties. Microcystins are soluble in water, ethanol and methanol, and insoluble in
DRAFT - DO NOT CITE OR QUOTE
-------
1 acetone, ether, chloroform and benzene. Laboratory studies show microcystins to be stable
2 under changes in temperature and pH. Microcystins are stable in sunlight; however, in the
3 presence of pigments (type unspecified) and sunlight, microcystins can be decomposed or
4 isomerized (Duy et al., 2000). Limited information suggests that microcystins can
5 bioaccumulate in aquatic organisms; these toxins have been measured in a number offish and
6 aquatic invertebrates.
DRAFT - DO NOT CITE OR QUOTE
-------
1 3. TOXICOKINETICS
2
3
4 The available information on toxicokinetics of microcystins is primarily focused on
5 exposure via injection routes, either intravenous or intraperitoneal. Few data are available on the
6 oral and inhalation exposure routes. No data on the absorption, metabolism or elimination of
7 microcystins after in vivo oral or inhalation exposure in humans or animals were located in the
8 literature. Toxicokinetic data available from parenteral routes of exposure are of uncertain
9 relevance to oral and inhalation exposure routes. The database on microcystins does not contain
10 any toxicokinetic models for microcystins.
11
12 A number of studies on the toxicokinetics of microcystins have used 3H-dihydro-MCLR
13 (usually produced by reduction of the MDha moiety in MCLR with tritiated sodium
14 borohydride) as a test material. While there are similarities between the organ distribution,
15 hepatocellular uptake and clinical syndrome after exposure to dihydroMCLR and MCLR
16 (Meriluoto et al., 1990), there are differences in the binding of these compounds to molecular
17 targets. Craig et al. (1996) showed that, while dihydroMCLA was capable of inactivating
18 protein phosphatase 2Ac through a rapid binding mechanism, it did not subsequently form a
19 covalent bond with PP2Ac, while MCLA did. Further information on the structural requirements
20 for microcystin toxicity is provided in Section 4.4.9. Potential differences in the binding of
21 dihydro-microcystin analogs dictates that caution be exercised in generalizing toxicokinetic
22 information derived using these compounds to microcystins as a group. In particular,
23 information on the subcellular localization of dihydromicrocystins may not be applicable to
24 microcystins containing an intact MDha residue.
25
26 3.1. ABSORPTION
27
28 Pulmonary absorption of MCLR (purified from a bloom sample) was rapid following
29 intratracheal instillation in mice (Ito et al., 2001). Immunostaining of the lung occurred within 5
30 minutes, followed by a lag period of 60 minutes before staining was observed in the liver. Based
31 on the positive immunostaining of alveoli, it was concluded that absorption occurred at the
32 alveoli. The lethal dose level for intratracheal injection was similar to a lethal dose for
33 intraperitoneal (i.p.) injection (i.e., both about 100 |j,g/kg). The authors reported that the lungs
34 were not affected by microcystin administration, but it is unclear whether a detailed
35 histopathological evaluation of the lungs was conducted in addition to the immunostaining.
36
37 The occurrence of hepatotoxicity and lethality following oral exposure to microcystins
38 (see Section 4.2) is evidence of oral absorption of the toxin. However, quantitative assessments
39 of oral absorption were not located. Ito et al. (1997a) qualitatively studied the oral absorption
40 and distribution of MCLR (purified from a bloom sample) following gavage dosing in mice (500
41 |j,g/kg). Immunostaining techniques indicated that MCLR was absorbed primarily in the small
42 intestine, although some absorption did occur in the stomach. Erosion was observed in the
43 surface epithelial cells and in the lamina propria of the small intestine villi. Erosion may
44 facilitate uptake of the toxin into the bloodstream.
45
DRAFT - DO NOT CITE OR QUOTE
-------
1 The oral bioavailability of MCLR was indirectly studied in in situ experiments using
2 isolated intestinal loops of rats (Dahlem et al., 1989). Rats given an infusion of MCLR (>95%
3 pure by high pressure liquid chromatography (HPLC) with UV detection) into the ileum showed
4 clinical signs (i.e., labored breathing and circulatory shock) and evidence of liver toxicity within
5 6 hours of a single 5 mg/kg dose. Liver toxicity was assessed as an increase in the liver-to-body
6 weight ratio and the presence of gross and histopathological liver lesions characteristic of
7 microcystin toxicity (i.e., enlarged livers, hepatocyte rounding and disassociation, hemorrhage).
8 Infusion of a similar dose into a jejunal loop produced a lower degree of liver toxicity, as
9 compared to the ileal loop infusion. These results suggest that there could be site-specificity in
10 intestinal absorption of MCLR; however, differences in absorptive surface area were not taken
11 into account in the experiment. In vitro experiments reported in this publication indicated that
12 cholestyramine, a bile acid sequestrant, bound MCLR, and the presence of cholestyramine in the
13 ileal loop infusion significantly reduced MCLR liver toxicity (Dahlem et al., 1989).
14
15 Oral absorption of 3H-dihydromicrocystin was also demonstrated using ileal loop
16 exposure in swine (Stotts et al., 1997a,b). In the exposed swine, the maximum blood
17 concentration of the toxin occurred 90 minutes after dosing.
18
19 3.2. DISTRIBUTION
20
21 The distribution of microcystins is limited due to the poor ability of these toxins to cross
22 cell membranes. Microcystins are primarily taken up into the liver by the multispecific active
23 transport system for bile acids. Once inside the cell, these toxins bind covalently to cytosolic
24 proteins, resulting in retention in the liver. The cytosolic proteins bound by microcystins have
25 been identified as the protein phosphatase enzymes (PP1 and PP2A). It should be noted that
26 dihydromicrocystin analogs do not appear to form covalent bonds with PP1 and PP2A, although
27 they are able to rapidly bind and inactivate the enzymes (Craig et al., 1996). Binding to and
28 inhibition of these phosphatase enzymes are directly related to the mechanism of toxicity for
29 microcystins and are further discussed in Section 4.4.7.3. This section will describe the overall
30 organ distribution, cellular uptake, subcellular localization and protein binding of microcystins.
31
32 3.2.1. Organ Distribution
33
34 The organ distribution of a 125I-labelled heptapeptide toxin (MW 1019) isolated from
35 Microcystis aeruginosa (while not identified by the study authors as such, probably because the
36 toxin had not yet been named, this is assumed to be a microcystin) was investigated in female
37 rats following intravenous (i.v.) administration (Falconer et al., 1986; Runnegar et al., 1986).
38 The heptapeptide toxin was purified by HPLC prior to reaction with 125I in the presence of Nal
39 and lactoperoxidase. Labelling of the toxin was confirmed by HPLC and mouse bioassay. The
40 highest tissue concentrations of microcystins were detected in the liver and kidney. After 30
41 minutes, 21.7% of the administered dose was present in the liver, 5.6% was present in the
42 kidneys, 7% remained in the gut contents, and 0.9% was cleared in the urine (Falconer et al.,
43 1986). The balance of the administered dose was not reported; however, the authors reported
44 that no significant accumulation was observed in other organs or tissues.
45
DRAFT - DO NOT CITE OR QUOTE
-------
1 Brooks and Codd (1987) reported extensive liver uptake following i.p. injection of 125
2 Hg/kg of a 14C-labelled toxin extracted from M. aeruginosa strain 7820 (assumed to be a
3 microcystin) in mice. Seventy percent of the radiolabel was found in the liver after 1 minute,
4 increasing to almost 90% after 3 hours. Radiolabel was also found in the lungs, kidneys, heart,
5 large intestine, ileum and spleen.
6
7 The distribution of 3H-MCLR (>95% pure) was evaluated following i.p. injection of a
8 sublethal (45 ng/kg) or lethal (101 ng/kg) dose in mice (Robinson et al., 1989). The tissue
9 distribution of radiolabel was similar after injection of either a lethal or a sublethal dose. Liver
10 accumulation reached a maximal value of 60% by 60 minutes. For the 101 |J,g/kg dose, the liver,
11 intestine and kidney contained 56, 7 and 0.9% of the radiolabel, respectively. Heart, spleen, lung
12 and skeletal muscle each contained less than 1% of the radiolabel.
13
14 Immunostaining methods were used to evaluate the organ distribution following
15 intratracheal instillation of MCLR purified from a bloom sample (Ito et al., 2001). Following
16 instillation of a lethal dose (100 ng/kg), the lung, liver, small intestine and kidney were
17 positively stained for MCLR. Intense staining was observed in the lung by 5 minutes post-
18 instillation, followed by the kidney (10 minutes), the small intestine (45 minutes) and the liver
19 (60 minutes). After approximately 90 minutes, bleeding began around the hepatic central vein.
20 The authors described the pathological changes in the liver as essentially the same as those seen
21 following oral or i.p. injection exposure routes. Intratracheal instillation of a sublethal dose (50
22 Hg/kg) resulted in immunostaining of the lung, liver, kidney, cecum and large intestine (Ito et al.,
23 2001). No discernable pathological changes were observed at this dose level. Ito et al. (2002)
24 synthesized glutathione and cysteine conjugates of microsystin-LR and administered them by
25 intratracheal instillation in mice. These conjugates are, according to the authors, known
26 metabolites of microcystins. The metabolites were demonstrated to be less toxic than the parent
27 compound (lethal doses were about 12-fold higher than the MCLR lethal dose) and were
28 distributed primarily to the kidney and intestine, as opposed to the liver.
29
30 The distribution of MCLR (purified from a bloom sample) following oral gavage
31 administration to mice (500 ng/kg) was investigated using immunostaining methods (Ito et al.,
32 1997a). MCLR was detected in large amounts in the villi of the small intestine. Erosion of the
33 villi was observed, which may have enhanced absorption of the toxin into the bloodstream.
34 MCLR was also present in the blood plasma, liver, lungs, kidneys and heart.
35
36 The distribution of 3H-dihydroMCLR in mice was shown to differ for the oral and i.p.
37 injection routes of exposure (Nishiwaki et al., 1994). Intraperitoneal injection of 3H-
38 dihydroMCLR resulted in rapid and continuous uptake in the liver, with approximately 72% of
39 the administered dose present in the liver after 1 hour. The 3H-dihydroMCLR was synthesized
40 by reduction of N-methyldehydroalanine from microcystin-LR. Small amounts of radiolabel
41 were found in the small intestine (1.4%), kidney and gallbladder (0.5%), lungs (0.4%) and
42 stomach (0.3%) following i.p. injection. Oral administration of 3H-dihydroMCLR resulted in
43 much lower concentrations in the liver, with less than 1% of the administered dose found in the
44 liver at either 6 hours or 6 days post administration. 3H-DihydroMCLR is rapidly distributed to
45 the liver of swine following i.v. injection or ileal loop infusion (Stotts et al., 1997a,b). Smaller
46 amounts were distributed to the kidneys, lungs, heart, ileum and spleen.
7 DRAFT - DO NOT CITE OR QUOTE
-------
1
2 MCLR was not found in the milk of dairy cattle that were exposed toM aeruginosa cells
3 via drinking water (Orr et al., 2001) or ingestion of gelatin capsule containing the cells (Feitz et
4 al., 2002).
5
6 3.2.2. Cellular Uptake
7
8 The cellular uptake of 3H-dihydroMCLR was evaluated using primary rat hepatocytes in
9 suspension and in isolated perfused rat liver (Eriksson et al., 1990a; Hooser et al., 1991a). The
10 uptake (as measured by scintillation counting of washed cells) of a mixture of unlabelled MCLR
11 and 3H-dihydroMCLR was shown to be specific for freshly isolated rat hepatocytes (Eriksson et
12 al., 1990a). Uptake was negligible in human hepatocarcinoma cells (Hep G2), mouse fibroblast
13 (NIH-3T3) and human neuroblastoma cells (SH-SY5Y). The uptake of 3H-dihydroMCLR was
14 also shown to be inhibited by bile acid transport inhibitors such as antamanide,
15 sulfobromophthalein and rifampicin, and by the bile salts cholate and taurocholate (by
16 competing for the bile acid transporter).
17
18 The uptake of 3H-dihydroMCLR was demonstrated to be rapid for the first 5-10 minutes,
19 followed by a plateau, in both rat hepatocyte suspensions and the isolated perfused rat liver
20 (Hooser et al., 1991a). Uptake was measured as radioactivity in fractionated cells versus
21 radioactivity in medium. The uptake of 3H-dihydroMCLR was inhibited by incubation of
22 suspended rat hepatocytes at 0°C, suggesting the involvement of an energy-dependent process
23 (Hooser et al., 1991a). Uptake was also inhibited by preincubation of hepatocytes with
24 rifampicin, presumably via competitive inhibition of the bile acid transporter (Hooser et al.,
25 199 la).
26
27 Many studies have demonstrated that inhibition of microcystin uptake at the bile acid
28 transporter reduces or eliminates the liver toxicity observed following in vitro or in vivo
29 exposures (Runnegar et al., 1981, 1995a; Runnegar and Falconer, 1982; Hermansky et al.,
30 1990a,b; Thompson and Pace, 1992). The human organic acid transport protein (OATP) was
31 shown to mediate the transport of 3H-microcystin (type not specified) in Xenopus laevis oocytes,
32 and this uptake was inhibited by sulfobromophthalein and taurocholate. This transport protein is
33 found in the human brain and may be related to the acute neurotoxicity seen in hemodialysis
34 patients exposed to microcystins (see Section 4.4.5.1).
35
36 Runnegar et al. (1991) studied the influence of dose level and exposure time on the
37 uptake of 125I-microcystin-YM in isolated rat hepatocytes (measured as radioactivity in
38 centrifuged cell pellet). Hepatocyte uptake was initially rapid with a plateau in the uptake rate
39 observed after 10 minutes. The initial uptake rate (in the first minute of exposure) increased
40 with increasing concentration, but cumulative uptake ceased at a dose that resulted in plasma
41 membrane blebbing.
42
43 Microcystin-YM uptake by isolated rat hepatocytes was temperature-dependent and was
44 inhibited 70-80% by the addition of sodium deoxycholate or sulfobromophthalein (Runnegar et
45 al., 1995b). This provides evidence to indicate that microcystin uptake occurs by carrier
46 mediated transport, most likely via the bile acid transporter. Pretreatment of mice with bile acid
DRAFT - DO NOT CITE OR QUOTE
-------
1 transporter inhibitors (cyclosporine A, rifamycin, trypan blue and trypan red) abolished
2 microcystin toxicity, suggesting limited or no uptake of microcystins (Runnegar et al., 1995b).
3 Further, in vitro preincubation of hepatocytes with bile acids or bile acid transport inhibitors
4 (taurocholate, trypan blue, cholate, sulfobromophthalein, cyclosporine A, trypan red and
5 rifamycin) each decreased the uptake of microcystin-YM, as measured by assays for protein
6 phosphatase inhibition in cell lysates (Runnegar et al., 1995b). Pretreatment with protein
7 phosphatase inhibitors (i.e., microcystin-YM and calyculin A) also resulted in the inhibition of
8 both microcystin-YM uptake and protein phosphatase inhibition, suggesting that the bile acid
9 transporter is itself regulated by serine/threonine phosphorylation.
10
11 Many cell types and established cell lines, including both rodent and some human cells,
12 have been evaluated for potential susceptibility to microcystin uptake and toxicity. Primary
13 isolated hepatocytes have been shown to be the most sensitive to cytotoxicity, due to the
14 presence of the organic ion/bile acid transport system (Eriksson et al., 1990b). In addition,
15 primary cultures of liver cells cease to express these bile acid transport proteins after 2-3 days of
16 being maintained in culture. Therefore, established liver cell lines are generally not useful for
17 evaluating microcystin toxicity (Eriksson and Golman, 1993; Heinze et al., 2001). Chong et al.
18 (2000) evaluated microcystin toxicity in eight permanent cell lines (including rodent, primate
19 and human cell lines), only two of which (human oral epidermoid carcinoma KB cells and rat
20 Reuber H35 hepatoma H-4-II-E cells) showed cytotoxicity following MCLR exposure. The
21 toxic response in these cells was most evident if MCLR was added when the cells were seeded.
22 Established monolayers were more resistant to microcystin toxicity. Mechanistic studies that
23 evaluate organ and cell type specificity for microcystins are further discussed in Section 4.4.7.1.
24
25 3.2.3. Subcellular Localization and Cytosolic Protein Binding
26
27 Tissue distribution was evaluated in mice given i.v. injection of a sublethal dose of
28 3H-MCLR (Robinson et al., 1991a). The liver contained approximately 67% of the radiolabel by
29 60 minutes, and the amount of hepatic radioactivity did not change throughout the 6-day study
30 period, despite urinary and fecal elimination of 24% of the administered dose. The subcellular
31 distribution of radioactivity in the liver demonstrated that approximately 70% of the hepatic
32 radiolabel was present in the cytosol. In vitro experiments showed that radiolabeled microcystin
33 in the liver was bound to high molecular weight cytosolic proteins (Robinson et al., 1991b). The
34 nature of the binding was demonstrated to be covalent, saturable and specific for a protein with a
35 molecular weight of approximately 40,000. Binding was inhibited by okadaic acid (a potent
36 inhibitor of serine/threonine phosphatases [1 and 2A]), suggesting that the target protein is
37 protein phosphatase 1 or 2A. A discussion of protein phosphatase binding and inhibition by
38 microcystins is provided under mechanistic studies in Section 4.4.7.3, below. Binding proteins
39 for MCLR were found in cytosol derived from several different organs, suggesting that liver
40 specificity is not due to limited distribution of target proteins. Covalent binding to hepatic
41 proteins may be responsible for the long retention of microcystins in the liver. Lin and Chu
42 (1994) evaluated the kinetics of MCLR distribution in serum and liver cytosol derived from
43 mice. Uptake of pure MCLR, as analyzed by direct competitive enzyme-linked immunosorbent
44 assay (ELISA), into the serum was shown to be rapid following an i.p. injection of 35 ng/kg
45 (sublethal dose). The toxin reached a maximum concentration in the serum by 2 hours and in
46 liver cytosol by 12 hours post-injection. MCLR was shown to be bound to liver cytosolic
DRAFT - DO NOT CITE OR QUOTE
-------
1 proteins and the kinetics of binding were correlated with inhibition of protein phosphatase 2 A
2 activity. The maximum decrease in enzyme activity was observed 6-12 hours following
3 injection.
4
5 Pace et al. (1991) demonstrated significant accumulation of 3H-MCLR in isolated
6 perfused liver despite a low overall extraction ratio (16% in liver, 79% in perfusate). In the
7 liver, radiolabel corresponding to MCLR (15%) and a more polar metabolite (85%) was
8 primarily found in the cytosolic fraction.
9
10 The subcellular distribution of 3H-dihydroMCLR was evaluated using primary rat
11 hepatocytes in suspension and the isolated perfused rat liver (Hooser et al., 1991a).
12 3H-dihydroMCLR was primarily localized in the cytosolic fraction in both the hepatocytes and
13 liver. In the hepatocytes, precipitation with trichloroacetic acid indicated that approximately
14 50% of the 3H-dihydroMCLR was found as free toxin, while the remaining 50% was bound to
15 cytosolic proteins. Since little of the radiolabel was in the insoluble pellet containing insoluble
16 actin and other elements, the authors suggested that 3H-dihydroMCLR did not bind significantly
17 to actin or other cytoskeletal proteins (Hooser et al., 1991a).
18
19 The subcellular protein binding of 3H-dihydroMCLR was evaluated in rat liver
20 homogenates (Toivola et al., 1994). Most of the radiolabeled toxin (80%) was bound to
21 cytosolic proteins. 3H-dihydroMCLR was shown to bind both protein phosphatase 1 (PP1) and
22 protein phosphatase 2A (PP2A); however, PP2A was detected primarily in the cytosol, while
23 PP1 was found in the mitochondrial and post-mitochondrial particulate fraction (membrane
24 proteins). The binding of microcystins to PP1 and PP2A and the inhibition of protein
25 phosphatase activity is further discussed in Section 4.4.7.3 (Mechanistic Studies).
26
27 Limited information in humans exposed to microcystins intravenously via dialysate
28 indicates that a large proportion of microcystins in the blood are bound to proteins. Hilborn et
29 al. (2005) compared two techniques for measuring microcystin in the serum of six patients. Use
30 of ELISA, which detects free microcystins, resulted in serum microcystin concentrations ranging
31 from 8 to 51% of the concentrations obtained using gas chromatography/mass spectrometry
32 (GC/MS) detection of 2-methyl-3-methoxy-4-phenylbutyric acid (MMPB, which is derived from
33 both free and protein-bound microcystins by chemical oxidation). These results indicate that
34 microcystins are bound to proteins in human blood, and that analysis for microcystins using
35 ELISA techniques may underestimate total blood concentrations.
36
37 3.3. METABOLISM
38
39 Urinary and fecal metabolites of MCLR were analyzed in samples collected 6 and 12
40 hours following i.v. injection of a sublethal dose of 3H-MCLR in mice (Robinson et al., 1991a).
41 Approximately 60% of the radiolabel in both the urine and the feces was associated with the
42 parent compound. MCLR was metabolized in liver cytosol preparations to a product that binds
43 to a high molecular weight cytosolic protein (Robinson et al., 1991b). The parent compound
44 also binds to this protein, which has been suggested to be the catalytic subunit of protein
45 phosphatase 2A. In isolated perfused rat liver, binding of both the parent toxin (3H-MCLR) and
46 a more polar metabolite to cytosolic proteins was also demonstrated (Pace et al., 1991). Polar
10 DRAFT - DO NOT CITE OR QUOTE
-------
1 metabolites accounted for 65-85% of the hepatic cytosol radiolabel. Metabolites of MCLR were
2 not further characterized in these studies.
3
4 3H-Dihydromicrocystin is not extensively metabolized in swine liver after i.v. injection
5 or ileal loop exposure, and is primarily present in hepatic tissues as the parent compound (Stotts
6 etal., 1997a,b).
7
8 Administration of 125 ng/kg of Microcystis toxin 7820 to mice resulted in decreased
9 levels of cytochrome b5 and cytochrome P450 (Brooks and Codd, 1987). Pretreatment of mice
10 with microsomal enzyme (mixed function oxidase) inducers (p-naphthoflavone,
11 3-methylcholanthrene and phenobarbital) was shown to eliminate this effect on hepatic enzymes
12 and to extend survival and reduce liver toxicity (i.e., changes in liver weight). In an in vitro
13 study using mouse liver microsomes, cytochrome P450 associated enzyme activity (i.e.,
14 metabolism of aminopyrene and p-nitrophenol) was not altered by an unidentified toxin isolated
15 fromM aeruginosa (assumed to be a microcystin; Cote et al., 1986).
16
17 The hepatic metabolism of MCRR and MCLR (purified from blooms) was studied
18 following i.p. injection in mice and rats (Kondo et al., 1996). Glutathione and cysteine
19 conjugates were identified at 3 and 24 hours in both mouse and rat livers. Structural
20 modification of the 3-amino-9-methoxy-2,6,8,-trimethyl-10-phenyldeca-4,6-dienoic acid (Adda)
21 and methyldehydroalanine (MDha) moieties of the toxins was indicated. Figure 3-1 shows the
22 glutathione and cysteine conjugates of microcystins.
23
24 Kondo et al. (1992) demonstrated that glutathione and cysteine conjugates of MCLR and
25 MCYR were less toxic than the parent compounds based on LD50 estimates, but were still
26 significantly toxic (LD50 values ranged from 217 to 630 ng/kg in mice). Glutathione and
27 cysteine conjugates of MCLR were shown to inhibit protein phosphatases 1 and 2A in vitro to
28 the same degree as MCLR; however, these metabolites were primarily distributed to the kidney
29 and intestine following intratracheal instillation in mice (Ito et al., 2002). This result suggests
30 that the lower toxicity of glutathione and cysteine conjugates may be related to distribution to
31 excretory organs and elimination of metabolites in vivo. Metcalf et al. (2000) also demonstrated
32 that microcystin conjugates with glutathione, cysteine-glycine and cysteine were less toxic in the
33 mouse bioassay; however, these conjugates were also shown to be weaker inhibitors of protein
34 phosphatases 1 and 2A in vitro. Takenaka (2001) illustrated that glutathione conjugates of
35 MCLR are formed by glutathione S-transferase enzymes found in both rat liver cytosol and
36 microsomes.
37
11 DRAFT - DO NOT CITE OR QUOTE
-------
H,,, >C°2H
2
o
J
4
5
6
7
X Y
Microcystin LR Leu Arg
Microcystin YR Tyr Arg
Microcystin RR Arg Arg
GSH conjugate Z =
H2N
HOOC H
Cys conjugate Z =
9 H
hLN COOH
,COOH
9
10
11
Figure 3-1. Structures of GSH and Cys Conjugates of Microcystins LR, YR and RR
(Kondoetal., 1992)
12 DRAFT - DO NOT CITE OR QUOTE
-------
1 Several studies have demonstrated an increase in hepatic glutathione levels following
2 exposure to microcystins (Ding et al., 2000a; Boua'icha and Maatouk, 2004; Gehringer et al.,
3 2004). MCLR was shown to induce the de novo synthesis of glutathione in mice exposed to a
4 toxic sublethal dose (75% of the LD50) (Gehringer et al., 2004). Increased transcription of
5 glutathione-S-transferase was also demonstrated in this study.
6
7 3.4. ELIMINATION
8
9 Limited information on the elimination of microcystins from the human body is available
10 from follow-up of dialysis patients exposed to microcystins intravenously (see Section 4.4.5.1
11 for further detail). In two separate incidents in Brazil (one in Caruaru, one in Rio de Janeiro),
12 microcystins were detected in patients' serum more than 50 days after documented exposure
13 (Hilborn et al., 2005; Scares et al., 2006).
14
15 The excretion of microcystins was investigated in female rats (Falconer et al., 1986).
16 The blood half-life was measured following i.v. administration of a 125I-labelled heptapeptide
17 toxin extracted firomM aeruginosa (MW 1019, assumed to be a microcystin). A biphasic blood
18 elimination curve was demonstrated, with the first component having a half-life of 2.1 minutes
19 and the second component having a half-life of 42 minutes. After 120 minutes, 9.4% of the
20 administered dose was present in the intestinal contents and 2.9% was present in the urine,
21 suggesting that biliary excretion plays a significant role in elimination of microcystins. Biliary
22 excretion was also demonstrated in isolated perfused rat liver (Pace et al., 1991). In the bile
23 collected 10-20 minutes after toxin exposure, 78% of the radiolabel was associated with the
24 parent toxin, while the remaining radiolabel was associated with more polar metabolites.
25
26 MCLR excretion was also evaluated in mice (Robinson et al., 1991a). A biexponential
27 plasma elimination curve was observed following i.v. injection of a sublethal dose of 3H-MCLR.
28 Plasma half-lives of 0.8 and 6.9 minutes were reported for the first and second phase of
29 elimination, respectively. Approximately 24% of the administered dose was eliminated in the
30 urine (9%) and feces (15%) throughout the 6-day study period. Approximately 60% of the
31 excreted microcystin, measured at 6 and 12 hours following injection, was present as the parent
32 compound.
o o
34 Ito et al. (1997a) demonstrated that MCLR is secreted in the mucous of goblet cells from
35 both the small and large intestine of mice following administration by oral gavage (500 |j,g/kg).
36 MCLR was not detected in urine in this study.
37
38 Stotts et al. (1997a,b) evaluated the toxicokinetics of 3H-dihydroMCLR in swine
39 following i.v. injection and ileal loop exposure. Elimination of 3H-dihydromicrocystin was rapid
40 and followed a biphasic pattern, suggesting that the liver rapidly removes the toxin from the
41 blood. Clearance from the blood is slower at higher dose levels, presumably due to the liver
42 toxicity and circulatory shock observed at high doses. 3H-Dihydromicrocystin was detected in
43 the bile as early as 30 minutes after i.v. injection. Following ileal loop exposure, the
44 concentration of toxin was consistently higher in the portal venous blood as compared to
45 peripheral blood. This suggests that first pass metabolism may play a role in the clearance of
46 dihydroMCLR.
13 DRAFT - DO NOT CITE OR QUOTE
-------
1
2 3.5. PHYSIOLOGICALLY-BASED TOXICOKINETIC MODELS
3
4 No physiologically based toxicokinetic models have been developed for microcystins.
14 DRAFT - DO NOT CITE OR QUOTE
-------
1 4. HAZARD IDENTIFICATION
2
3
4 4.1. STUDIES IN HUMANS - EPIDEMIOLOGY, CASE REPORTS, CLINICAL
5 CONTROLS
6
7 A number of case reports and epidemiological investigations have examined the
8 relationship between human exposure to cyanobacteria and various health endpoints. In all of
9 these studies, humans were exposed to blooms of cyanobacteria in environmental settings. As a
10 result, the potential for co-exposure to multiple cyanobacterial toxins and/or other
11 microorganisms or compounds to contribute to observed toxicity cannot be ruled out.
12
13 4.1.1. Oral Exposure
14
15 4.1.1.1. Short-Term Studies and Case Reports
16
17 Dillenberg and Dehnel (1960) reported on a series of animal poisonings and human
18 exposures to cyanobacterial blooms in various lakes of Saskatchewan, Canada during the
19 previous year. Several cases of individual or group human exposures during recreational
20 activities were reported. In general, the symptoms were gastrointestinal in nature, including
21 nausea, stomach pain and diarrhea; headache and muscle weakness were also reported. Stool
22 samples from three of the victims showed evidence of cyanobacteria (Microcystis and
23 Anabaend). In addition, water samples from the lakes in which the victims had been swimming
24 showed cyanobacteria. At the time of this report, cyanobacterial toxins had not been fully
25 characterized. Thus, no data on the nature or quantity of toxins in the affected waters were
26 provided.
27
28 Billings (1981) reported a series of outbreaks of human illness potentially associated with
29 exposure to cyanobacteria in two Pennsylvania lakes. Swimmers in both lakes reported
30 symptoms, including headache, abdominal cramping, nausea, vomiting, diarrhea, hay fever-like
31 symptoms, ear aches, eye irritation, sore throat, sneezing, runny nose and swollen lips within a
32 few hours of swimming in the waters. Investigation by the state Departments of Environmental
33 Resources and Health served to rule out common bacterial, protozoal and viral agents in the
34 outbreaks. In the second lake, a bloom ofAnabaena was identified. Indirect evidence (rapid
35 onset of symptoms, absence of other potential causative agents and consistency with previous
36 reports of health effects after exposure to cyanobacteria) led the investigators to postulate a role
37 for exposure to Anabaena in the health outcomes.
38
39 Turner et al. (1990) reported a similar type of outbreak among army recruits who had
40 consumed reservoir water during canoe exercises. Detailed case reports were presented for two
41 recruits. Both 16-year-old recruits presented with several days' history of malaise, sore throat,
42 blistering around the mouth, dry cough, pleuritic pain and abdominal pain. One also had
43 experienced vomiting and diarrhea. Physical examination revealed fever, left basal pulmonary
44 consolidation (pneumonia) and abdominal tenderness in both patients. Blood tests revealed low
45 platelet counts in both patients. Both were tested for a variety of pathogens, including
46 Leptospim, Legionella, Chlamydia, Coxiella, Mycoplasma and influenza and adenovirus, all
15 DRAFT - DO NOT CITE OR QUOTE
-------
1 with negative result. The authors reported similar symptoms (sore throat, headache, abdominal
2 pain, dry cough, diarrhea, vomiting and blistered mouth) in 16 additional soldiers who had taken
3 part in the canoe exercises. The reservoir contained a bloom of cyanobacteria, primarily M
4 aeruginosa. Further, a sample of the bloom taken the day after the patients were admitted into
5 the hospital showed microcystins to be present, including MCLR. High levels of Escherichia
6 coli were also found in reservoir water sampled 2 weeks later. The authors suggested that
7 microcystin exposure may have had a role in the pulmonary consolidation and low platelet count
8 of the two patients, citing evidence from studies in mice (the authors cited Falconer et al., 1981
9 and Slatkin et al., 1983).
10
11 Teixeira et al. (1993) characterized an epidemic of gastroenteritis, primarily among
12 children, in the vicinity of the Itaparica Dam in Bahia, Brazil. The authors collected data on the
13 incidence of treatments for diarrhea between February and May of 1988. Timing of 1118 cases
14 of diarrhea in local health units was recorded, showing a spike in the incidence of gastroenteritis
15 coinciding with the flooding of the Itaparica Dam reservoir. Most (about 70%) of the cases
16 involved children under the age of 5 years. Additional data were collected on the age, residence,
17 symptoms, foods consumed, source of drinking water and travel for 76 patients given outpatient
18 treatment for diarrhea. Fecal, blood and urine samples were collected from these patients and
19 analyzed for chemical and biological contaminants (i.e., bactedologic, virologic, cholinesterase,
20 heavy metals). In addition, water samples were analyzed for chemical and biological
21 contaminants (i.e., organophosphates, carbamates, heavy metals, fecal coliform). Analysis of
22 biological samples showed no contaminants. Untreated water samples showed high levels of
23 fecal coliform, but samples of treated water did not contain significant levels. Untreated water
24 samples also revealed high counts (1104-9755 units per mL) ofAnabaena andMicrocystis cells,
25 4-32 times the World Health Organization (WHO) maximum acceptable cell count for untreated
26 water (300 units/mL at the time). No data were provided on levels of cyanobacteria in the
27 treated water. It is not clear from the publication whether affected persons were exposed to
28 treated or untreated water. This study does not provide information on health effects of
29 microcystin exposure, but provides some circumstantial evidence for gastrointestinal effects
30 from exposure to cyanobacteria.
31
32 A case control study investigated the incidence of gastrointestinal and dermatological
33 symptoms among persons exposed to Murray River water (Australia) (el Saadi and Cameron,
34 1993; el Saadi et al., 1995). Physicians in 8 of 11 towns along the Murray River participated in
35 the study, recruiting 102 gastrointestinal and 86 dermatological cases between January and
36 March, 1992. Gastrointestinal cases were patients with abdominal pain, vomiting or diarrhea;
37 dermatological cases had rash, itching or blistering of the mouth. Control patients (132) were
38 selected as the next patient entering the office after each case, when possible. For each study
39 participant, age, sex, primary source of drinking and domestic water (rain/spring, untreated river
40 water or chlorinated river water from a town supply) and recreational water contact (none, river
41 or lake contact, or other, such as pool contact) during the previous week were recorded. River
42 water samples were collected and cyanobacteria identified and quantified. Anabaena,
43 Aphanizomenon and Planktothrix were the most common cyanobacteria identified in the
44 samples; small numbers of M aeruginosa were infrequently identified. Both univariate and
45 multivariate analysis of the data showed the odds of having gastrointestinal symptoms to be
46 raised in persons drinking chlorinated river water or using untreated river water for domestic
16 DRAFT - DO NOT CITE OR QUOTE
-------
1 purposes. Likewise, both types of statistical analyses showed increased odds of having
2 dermatological symptoms for persons less than 20 years of age and for persons using untreated
3 river water for domestic purposes. The proportion of patients with gastrointestinal symptoms
4 and the proportion with dermatological symptoms both correlated with mean weekly log
5 cyanobacterial cell count in the river, although the correlation was statistically significant only
6 for gastrointestinal symptoms. However, when upper and lower reaches of the river were
7 analyzed separately, nonsignificant correlations were observed.
8
9 No data on cyanobacterial toxins in the river water were provided. The symptoms
10 reported in this study cannot be readily attributed to any particular toxin (if they are indeed
11 associated with toxin exposure rather than exposure to the living cyanobacterial cells) due to the
12 absence of toxin analyses, as well as the identification of genera with potential to produce
13 multiple toxins. In addition, the potential for other microbial or chemical contaminants in the
14 untreated river water was not evaluated in this study.
15
16 Pilotto et al. (1997) conducted a prospective study of gastrointestinal and dermatological
17 symptoms among people exposed to cyanobacteria at water recreation sites in Australia. Study
18 participants were individuals 6 years of age and older who were present at one of several water
19 bodies that were both used for recreational purposes and expected to have algal blooms during
20 the summer. Interviewers visited these sites on several Sundays and invited all individuals to
21 participate. Participants completed a questionnaire to evaluate health status and the nature and
22 duration of water-contact activities. In addition, subjects were asked whether they had
23 symptoms or recreational water contact in the 5 days prior to study initiation, in order to control
24 for the effects of prior health conditions and prior exposure on study findings. Five hundred and
25 fourteen persons had either pre-existing symptoms or water contact in the days prior to initial
26 interview. Participants were contacted by telephone 2 and 7 days later, at which time the
27 occurrence of diarrhea, flu-like symptoms, skin rashes, mouth ulcers, fevers or eye or ear
28 infections in the intervening time was recorded. Water samples for cyanobacterial cell count and
29 toxin analysis were collected at 10 a.m. and 2 p.m. on the day of initial interview. Toxin
30 presence was assessed by mouse bioassay (i.p. injection of 500 mg freeze-dried cells/kg body
31 weight).
32
33 Of 1029 persons invited to participate, 921 persons participated in the study (Pilotto et
34 al., 1997). Interviewers were able to contact 845 of these persons by telephone 2 days after
35 initial interview, and 852 persons 7 days later. No differences in the reporting of gastrointestinal
36 and dermatological symptoms were found between those who had water contact and those
37 without water contact (on the day of the initial interview) when contacted 2 days later (Pilotto et
38 al., 1997). However, when subjects with water contact or symptoms prior to initial interview
39 were excluded, a significant trend to increasing symptom occurrence with duration of exposure
40 was observed among persons contacted 7 days after initial interview. Cyanobacterial cell count
41 showed some correlation with symptom occurrence, but presence/absence of hepatotoxins did
42 not. The authors postulated that any association between symptoms and exposure resulted from
43 the allergenic nature of the cells rather than exposure to toxins.
44 Pilotto et al. (1999) evaluated the relationship between cyanobacterial exposure and
45 perinatal outcomes in an ecological study conducted in Australia. Cyanobacterial monitoring
17 DRAFT - DO NOT CITE OR QUOTE
-------
1 data (cell counts) were collected from raw drinking water supplies in 156 towns. Perinatal
2 outcome data were obtained from several registries (for calendar years 1992-1994) and the
3 following variables assessed: premature birth (<36 weeks), low birth weight (<2.5 kg), very low
4 birth weight (<1.5 kg) and congenital defects (at least one). Maternal residence at birth was used
5 to assess exposure based on cyanobacterial cell counts. Exposure was assessed at various
6 gestational periods, either as the proportion of time with cyanobacterial exposure (proportion of
7 weeks with non-zero levels) or average alert level (alert level 1 = <2000 cells/mL; alert level 2 =
8 2000-15,000 cells/mL; alert level 3 = <15,000 cells/mL). Data on 32,700 births were collected,
9 although the numbers with exposure data in different gestational periods varied. A significant
10 difference in the incidence of low birth weight and very low birth weight babies was observed
11 between mothers with and without cyanobacterial exposure during the first trimester. Very low
12 birth weight incidences increased with increasing exposure to cyanobacteria (as measured by the
13 proportion of first trimester with non-zero cyanobacterial cell counts). At the highest exposure
14 category (exposure to cyanobacteria during 100% of first trimester), the odds ratio (OR) was
15 1.42 (95% confidence interval [CI] = 1.00-2.02). When exposure was assessed as the average
16 alert level (cell concentration interval), there was a significant increase in congenital defects at
17 the highest average alert level of 2.5-3.0 (OR = 2.03, 95% CI = 1.37-3.01), but the trend was not
18 significant. This study had a number of limitations, including a lack of individual exposure data
19 and lack of data on cyanobacteria or toxins in the finished water (after various treatment
20 processes). Further, because the measure of exposure was cyanobacteria rather than toxin, it is
21 difficult to interpret this study in the context of microcystin effects.
22
23 Falconer et al. (1983) compared the hepatic enzyme levels in patients served by a public
24 water supply contaminated with a bloom of M. aeruginosa with levels in patients living in areas
25 served by other water supplies. Enzymes assessed in the study were y-glutamyltransferase
26 (GGT), aspartate aminotransferase (AST), alanine aminotransferase (ALT) and alkaline
27 phosphatase (ALP). The study population consisted of all patients served by a single hospital
28 laboratory and referred for liver function tests before, during and after a bloom of M aeruginosa
29 in the Malpas Dam reservoir of Australia. Patients were classified either as residents of the city
30 of Armidale, which uses the reservoir for drinking water supply, or residents of neighboring
31 towns with independent water supplies. Liver function test results within each comparison group
32 were further sorted by date into three categories: testing during the 5 weeks before the first signs
33 of the bloom appeared, testing during the 3-week bloom or the 2 weeks following copper sulfate
34 treatment of the bloom (identified as the high-risk time interval due to the cell lysis and
35 subsequent toxin release) or testing during the 5 weeks that followed.
36
37 Results of plasma enzyme analyses were compared before, during and after the bloom
38 among residents of Armidale and surrounding areas (Falconer et al., 1983). Analysis of variance
39 was used to assess differences in enzyme levels between comparison groups and between times
40 within comparison groups. Results of the statistical analysis indicated a significant rise in GGT
41 levels in residents of Armidale during the bloom period. ALT levels in Armidale residents
42 increased during the bloom period, but the change was not statistically significant.
43
44 The authors noted substantial variability in enzyme levels, attributing this finding to the
45 imprecise method of selecting study participants (Falconer et al., 1983). It should be noted that
46 several of the enzyme measurements for the referent population were associated with one
18 DRAFT - DO NOT CITE OR QUOTE
-------
1 individual requiring repeat analysis for chronic kidney disease. Alcoholism, which can increase
2 GGT levels, was reported to occur at about the same proportion (7-10%) in the groups assessed
3 before and during the bloom, although it was substantially lower in the post-bloom group of
4 Armidale residents. The authors concluded that the change in GGT among Armidale residents
5 before and during the bloom period might potentially be associated with exposure to M
6 aeruginosa.
1
8 4.1.1.2. Long-Term Studies and Epidemiological Studies
9
10 Zhou et al. (2002) conducted a retrospective cohort analysis of colorectal cancer and
11 exposure to microcystins in drinking water in a Chinese province in which an association had
12 been reported previously (Jiao et al., 1985; Chen et al., 1994). Cases of primary colorectal
13 adenocarcinoma between 1977 and 1996 from eight randomly selected towns within Haining
14 City of Zhejiang Province were used as the study population. Cases were identified using the
15 local cancer registry and independently verified by two pathologists. Drinking water source used
16 longest during the lifetime was used as a surrogate measure of exposure to microcystins.
17 Information on drinking water source was obtained by interview of patients or family members
18 of deceased cases. In each of the eight towns, 10 water sources (3 rivers, 3 ponds, 2 wells and 2
19 taps) were randomly selected and sampled for microcystins twice in each of the months of June
20 through September (total of eight samples from each source). Water samples were analyzed for
21 microcystins by ELISA; the authors did not specify the targeted congeners. The authors do not
22 specify the nature of the "tap" water sources, but the text implies that the tap water derives from
23 one or more treatment plants.
24
25 The average incidence rate of colorectal cancer across all of the study areas was
26 8.37/100,000 per year. The incidence rate was compared among the four different water sources,
27 with well water users serving as the referent population. Compared with the incidence among
28 well water users, the colorectal cancer incidence rates among users of the other water sources
29 were significantly increased. Tap water use was associated with a relative risk of 1.88, while
30 river and pond water use were both associated with a relative risk over 7.0. There was no
31 difference in colorectal cancer incidence between river and pond water users. The authors
32 suggested that exposure to trihalomethane compounds might account for the increase in
33 incidence among tap water users. Table 4-1 shows the incidence rate, relative risk and 95% CIs
34 for these exposure comparisons.
35
36 Microcystins were detected at concentrations exceeding 50 pg/mL (considered by the
37 authors to be the limit for positive detection) only in river and pond water, and the average
38 concentrations in these sources were substantially higher (30- to 50-fold) than well or tap water.
39 A similar proportion (about 25%) of the residents in each of the eight towns used river and pond
40 water for drinking water, allowing an analysis comparing the average microcystin concentration
41 in river and pond water in each town with the incidence rate by town. This analysis showed a
42 strong correlation between incidence rate and concentration of microcystin (Spearman
43 correlation
19 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-1 . Relative Risk of Colorectal Cancer and Microcystin Concentration by Drinking
Water Source (Zhou et al., 2002)
Water
Source
Well
water
Tap
water
River
water
Pond
water
Colorectal
Cancer
Incidence
Rate per
100,000
3.61
6.77
28.5
27.76
Relative
Risk of
Colorectal
Cancer
_
1.88
7.94
7.7
95% CI
_
1.39-2.54
6.11-10.31
5.75-10.30
Number of
Microcystin
Samples >50
pg/mL
0/12
0/17
25/69
6/35
Mean
Microcystin
Concentration
(pg/mL)
0.73
4.85
141.08
106.19
Maximum
Microcystin
Concentration
(pg/mL)
9.13
11.34
1083.43
1937.94
20 DRAFT - DO NOT CITE OR QUOTE
-------
1 coefficient = 0.88, p<0.01). Figure 4-1 shows the relationship between colorectal cancer
2 incidence and average microcystin concentration.
3
4 This study provides suggestive evidence for an association between microcystin exposure
5 and colorectal cancer. It is also consistent with earlier reports of an association between
6 drinking river or pond water and incidence of colorectal cancer in the Zhejiang Province of
7 China (Jiao et al., 1985; Chen et al., 1994; studies published in Chinese and not translated for
8 this review). However, because demographic information was not provided in the report, it is
9 not clear whether dietary, genetic and lifestyle factors associated with colorectal cancer have
10 been adequately controlled in the analysis. Further, other potential biological and chemical
11 contaminants in the river and pond water were not considered.
12
13 Several epidemiological studies have examined the relationship between drinking water
14 source (well, river, pond or ditch) and liver cancer in Haimen City, Jian-Su Province, China, an
15 area with an elevated hepatocarcinoma incidence (Yu, 1989; Yu et al., 1989). These studies,
16 published in Chinese and not translated for this review, showed an increased risk of primary liver
17 cancer associated with consumption of pond or ditch water (Ueno et al., 1996). According to
18 Health Canada (2002), Yu (1989) showed that consumption of pond or ditch water was
19 associated with an 8-fold increase in liver cancer incidence when compared with well water
20 consumption. Health Canada (2002) reported that a larger study of 65 counties in China, also
21 published in Chinese (Chen et al., 1991), showed the opposite; that consumption of deep well
22 water was associated with an increased risk of liver cancer.
23
24 Ueno et al. (1996) conducted a survey of microcystin content in drinking water supplies
25 in Haimen City to test the hypothesis that microcystins in surface drinking water supplies could
26 contribute to the higher incidence of liver cancer. Microcystins were measured by ELISA in
27 shallow and deep wells, as well as in ponds/ditches and river waters. The authors did not
28 indicate which congeners were targeted by the ELISA. Occurrence of microcystins was higher
29 in pond/ditch water (17% reported as positive with concentration >50 pg/mL) and river water
30 (32% positive) samples than in shallow wells (4% positive) or deep wells (no detections >50
31 pg/mL). Further, microcystin concentrations averaged across the drinking water types were
32 different, averaging 101, 160 and 68 pg/mL in pond/ditch, river and shallow well samples,
33 respectively. These data, while suggestive, do not directly associate exposure to microcystins
34 and liver cancer, since individual exposures were not measured or estimated, and other
35 biological or chemical contaminants in the surface waters have not been ruled out.
36
37 In a case-control study of liver cancer in Haimen City, conducted by Yu et al. (2002), a
38 variety of liver cancer risk factors were evaluated, including hepatitis B and C virus infection,
39 aflatoxin Bl or microcystin exposure, smoking, drinking, diet and genetic polymorphisms. From
40 a pool of 248 patients with hepatocellular carcinoma and 248 age-, sex- and residence-matched
41 controls, 134 paired cases and controls assented to blood samples for virus infection and ALDH2
42 and CYP2E1 gene polymorphism analyses. Data from these analyses were combined with
43 questionnaire information on possible lifestyle and dietary risk factors for liver cancer.
44 Microcystin exposure was assessed categorically based on drinking water supply (tap, deep or
21 DRAFT - DO NOT CITE OR QUOTE
-------
2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
47
48
49
50
51
52
53
54
Incidence of colorectal cancer vs. microcystin concentration (Zhou et al., 2002)
14
o
o
o 10
o 12
10 -
o
o
6 -
2 -
y = 0.0267x + 3.0079
R2 = 0.8277
50 100 150 200 250
Microcystin concentration (pg/mL) in river and pond
300
350
Figure 4-1. Relationship between Colorectal Cancer and Microcystin Concentration in River
and Pond Water in Haining City, China (Zhou et al., 2002)
22
DRAFT - DO NOT CITE OR QUOTE
-------
1 shallow well, river, ditch or pond), as in earlier studies (Yu et al., 1989). Neither univariate nor
2 multivariate analysis of the data indicated an association between consumption of river, pond or
3 ditch water and hepatocellular carcinoma. Hepatitis B virus infection was strongly associated
4 with primary liver cancer, and history of i.v. injection was also identified as a risk factor (Yu et
5 al., 2002).
6
7 Fleming et al. (2002) conducted an ecological epidemiological investigation of the
8 relationship between drinking water source and incidence of primary liver cancer in Florida.
9 The study was prompted by data showing cyanobacteria and toxins, especially microcystins, in
10 surface drinking water sources in Florida. The study population consisted of all cases of primary
11 hepatocellular carcinoma reported to the Florida state cancer registry between 1981 and 1988.
12 The study population was divided into comparison groups consisting of those served by surface
13 drinking water supply and those using other sources. Residence at the time of diagnosis was
14 used to place cases into the various comparison groups. Surface water treatment plants and their
15 service areas were geocoded, as were deep groundwater treatment plants. Several comparisons
16 were made. First, incident cases residing in the service area of a surface water treatment plant
17 were compared with those residing in the service area of a deep groundwater treatment plant.
18 Within this comparison, there were several referent groups; one randomly sampled from the
19 available groundwater service areas, one matched on median income and rent, one matched on
20 ethnic makeup and one matched on income, rent and ethnicity. Second, incident cases in the
21 surface water service area were compared with equally-sized buffer areas surrounding the
22 surface water service area, but not served by the treatment plant. Finally, incident cases were
23 compared with the incidence in the general Florida population.
24
25 Evaluation of the individual incidence rates in the 18 surface water service areas with the
26 groundwater service areas did not reveal any statistically significant differences among the
27 individual incidence rates. When the service areas were pooled, residence in a surface water
28 service area was associated with a statistically significant reduced risk of hepatocellular
29 carcinoma compared with either groundwater service areas (standardized rate ratios [SRR]
30 ranged from 0.8 to 0.98 for the four groundwater comparison groups) or the general Florida
31 population (SRR of 0.8). It should be noted that the measure of exposure, residence within a
32 surface water service area, was estimated as the average size plus two standard deviations of the
33 service area for this comparison.
34
35 When comparisons were made between residence in the actual (i.e., not estimated as
36 above) surface water service areas and residence in the buffer areas surrounding the service
37 areas, a statistically significant increase in the incidence of hepatocellular carcinoma was
38 observed for those residing within the surface water service area (SRR = 1.39, CI = 1.38-1.4).
39 Analyses of 1990 census data suggested that the ethnic and socioeconomic backgrounds of the
40 service areas and buffer areas were similar, although the authors did not report these data.
41 Interestingly, the incidence of hepatocellular carcinoma in the buffer areas was significantly
42 lower than that in the general Florida population (SRR = 0.59).
43
44 An ecological study such as this is useful for generating hypotheses, but not for
45 establishing an exposure-response relationship due to the lack of exposure data on individuals.
46 In this case in particular, there is strong potential for misclassification of exposure. Residence in
23 DRAFT - DO NOT CITE OR QUOTE
-------
1 a surface water service area at the time of diagnosis of hepatocellular carcinoma is a poor
2 measure of potential exposure to cyanobacterial toxins, especially given residential mobility and
3 likely latency time for cancer development. Further, the initial comparisons with groundwater
4 service areas used GIS-generated estimates of surface water service areas rather than actual
5 service areas, leading to greater potential for misclassification.
6
7 Fleming et al. (2004) also conducted an ecological study assessing the relationship
8 between incident colorectal cancer and proximity to a surface drinking water treatment plant,
9 with the latter representing a surrogate for exposure to cyanobacteria. Methods for this study
10 were identical to those described above for Fleming et al. (2002) except that colorectal cancer
11 data were abstracted from the Florida Cancer Data System from 1981-1999. As with Fleming et
12 al. (2002), comparisons were made between the colorectal cancer incidence rates in the 18
13 surface water treatment service areas with several referent groups (a random group of
14 groundwater treatment service areas, a group of groundwater treatment service areas matched on
15 median income and rent, a group of groundwater treatment service areas matched on ethnic
16 makeup, a group of groundwater treatment service areas matched on both median income and
17 ethnicity, groups residing in an equally-sized buffer areas surrounding the surface water service
18 area and, finally, the general Florida population). Mann Whitney rank sum tests of all
19 comparisons did not suggest an association between colorectal cancer and residence at time of
20 diagnosis in a surface water treatment area (details not provided). This ecological study is
21 subject to the same limitations as described above for Fleming et al. (2002).
22
23 4.1.2. Inhalation Exposure
24
25 No studies of human exposure to microcystins via inhalation were identified in the
26 materials reviewed for this document.
27
28 4.2. ACUTE, SHORT-TERM, SUBCHRONIC AND CHRONIC STUDIES AND
29 CANCER BIOASSAYS IN ANIMALS - ORAL AND INHALATION
30
31 Early research on cyanobacterial toxins examined the effects of exposure to
32 cyanobacterial organisms rather than on the toxins now known to produce many of the
33 toxicological effects. In the case of microcystins, the isolation and characterization of important
34 toxin(s) did not occur until the 1980s (WHO, 1999). As a result, many studies have used various
35 extracts of cyanobacterial blooms as test substances in toxicological experiments. These studies
36 contribute to the hazard identification for cyanobacterial toxins, but, as discussed below, are not
37 useful for dose-response assessment.
38
39 The quantity of an individual cyanobacterial toxin in different bloom samples and
40 extracts varies widely, being influenced by a number of different factors. Some toxins are
41 produced by more than one genus of cyanobacteria. For example, microcystins have been shown
42 to be produced by Microcystis, Anabaena, Planktothrix, Nostoc and others (WHO, 1999). Some
43 species (e.g., Anabaend) can produce more than one cyanotoxin (WHO, 1999). Even within a
44 species, different strains produce varying levels of toxin; some produce little or no toxin at all.
45 Growth conditions can also contribute to the level of toxin produced by a given species and
46 strain (WHO, 1999). Finally, any given sample may contain multiple genera, species or strains
24 DRAFT - DO NOT CITE OR QUOTE
-------
1 of cyanobacteria, as well as other contaminating organisms. Some of these variables will also
2 apply to cyanobacterial cells cultured in a laboratory, although clonal cultures may be
3 characterized as to toxin content. In general, there is no clear means of predicting the toxin
4 content in a given bloom sample or cell extract.
5
6 In addition to the variations in toxin production within bloom and/or culture samples,
7 there are variations in toxin concentration depending on the method used for producing a
8 material for toxicological administration. Among the studies reviewed, the administered
9 material included lyophilized bloom samples in solution, cell-free extracts, cell-free lysates,
10 partially purified toxins, purified toxins and others. As endotoxins, microcystins exist primarily
11 within the cyanobacterial cell, and are released when cells are lysed. As a result, studies of
12 extracts obtained by removing intact cells may not contain much, if any microcystin. Purified or
13 partially-purified toxins were used in a number of studies; however, the specific microcystin
14 congener or congeners may not have been identified. As a result, data using bloom samples, cell
15 extracts or partially purified toxins cannot be considered reliable information in relating
16 exposure to a given toxin with toxicological effect.
17
18 Giving due consideration to the limitations of algal extract studies, a distinction is made
19 between toxicological data obtained using purified microcystins and toxicological data obtained
20 using a bloom sample or extract. These data are discussed separately in this report, with the
21 latter data being considered supplemental due to the confounding factors outlined above. As a
22 result, studies of cyanobacterial extracts are not reviewed in depth.
23
24 4.2.1. Oral Exposure
25
26 4.2.1.1. Acute Studies
27
28 4.2.1.1.1. Purified Microcystins
29
30 Fawell et al. (1999) conducted acute, subchronic and developmental toxicity studies of
31 MCLR in mice and/or rats. In the acute portion of the study, single oral gavage doses of MCLR
32 (purity not specified) in aqueous solution were given to male and female CR1 :CD-
33 1(ICR)BR(VAF plus) mice and CR1 :CD(SD)BR(VAF plus) rats (five per sex per species).
34 Doses of 500, 1580 and 5000 ng/kg body weight were administered. Untreated control groups
35 were not included. The animals were observed for up to 14 days prior to sacrifice and necropsy.
36 Microscopic examinations of the lung and liver were conducted. LD50 values were calculated.
37
38 Oral LD50 values were estimated to be about 5000 ng/kg for mice and over 5000 ng/kg
39 for rats. Animals that died showed clinical signs, including hypoactivity and piloerection;
40 however, clinical signs were absent in survivors. Body weights among surviving animals were
41 not affected during the 14-day follow-up. Necropsy of the animals that died showed darkly
42 discolored and distended livers, as well as pallid kidneys, spleen and adrenals. Livers of all
43 animals that died had moderate or marked centrilobular hemorrhage. The incidence and severity
44 of liver lesions increased in a dose-dependent fashion, as shown in Table 4-2.
45
25 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-2. Incidence of Liver Lesions in Mice and Rats Treated with a Single Dose of MCLR
(Fawell et al., 1999)
Dose
Og/kg)
Number
Animals
Treated
Mortality
Diffuse
Hemorrhage
Moderate
Centrilobular
Hemorrhage
Marked
Centrilobular
Hemorrhage
Centrilobular
Necrosis
Cytoplasmic
Vacuolation
Mice
500
1580
5000
10
10
10
0
1
5
2
2
1
0
2
7
0
1
0
0
0
2
0
0
0
Rats
500
1580
5000
10
10
10
0
0
1
8
7
8
0
0
1
0
0
1
0
0
1
0
0
1
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
Diffuse hemorrhage in the liver was seen in rats and mice of all dose groups, but the
incidence was not clearly related to dose. Because an untreated control group was not included
in this study, it is not possible to say whether the liver effects were treatment-related; thus,
neither a NOAEL nor a LOAEL could be determined. The single mouse death at 1580 ng/kg
indicates that this is a frank effect level (PEL) in this species; the PEL in rats was 5000 ng/kg
with a single rat death at this dose.
Yoshida et al. (1997) assessed the acute oral toxicity of purified MCLR (>95% pure by
HPLC) in female BALB/c mice. Preliminary experiments using doses of 16.8 and 20 mg/kg
resulted in death within 160 minutes in two mice; therefore, doses of 8.0, 10.0 and 12.5 were
chosen for LDso determination. MCLR in saline solution was administered via gavage to a total
of five 6-week-old mice. Two control mice received saline via gavage. Mortality was observed
over a 24-hour period, and dead animals, including those in the preliminary study, were
immediately necropsied. One surviving mouse was sacrificed and necropsied 24 hours after
treatment; the remainder were sacrificed and necropsied after a week. The liver, kidneys and
lung were sectioned and examined by light microscopy. Electron microscopy was used to
identify apoptotic cells in the livers of treated mice. The remaining tissues were subjected to
histopathological analysis.
Mortality within 24 hours was 0/1 at 8 mg/kg, 0/2 at 10 mg/kg and 2/2 at 12.5 mg/kg.
The oral LD50 was calculated to be 10.9 mg/kg. No effects on the stomach, intestine, skin or
26
DRAFT - DO NOT CITE OR QUOTE
-------
1 organs other than the liver and kidneys were observed. Liver effects were pronounced in
2 animals that died, including centrilobular hemorrhage and hepatocyte degeneration, as well as
3 free hepatocytes in the veins of mice receiving doses in excess of 12.5 mg/kg (in the preliminary
4 experiments). Effects on the kidneys included proteinaceous eosinophilic materials in the
5 Bowman's spaces of mice receiving high doses (>12.5 mg/kg). In a single mouse treated with
6 10.0 mg/kg and sacrificed after 24 hours, evidence of hepatocellular necrosis was observed in the
7 centrilobular and midzonal regions, and single cell death (possibly apoptotic) was reported in the
8 centrilobular region, as well as surrounding necrotic areas. In the other mouse treated with 10
9 mg/kg and the two mice treated with 8.0 mg/kg (all sacrificed 1 week after treatment), the livers
10 contained hypertrophic hepatocytes in the centrilobular region and fibrosis in the centrilobular
11 and midzonal regions. A few apoptotic cells were observed in these animals. No kidney effects
12 were reported in animals that survived treatment for at least 24 hours. No control group was
13 included, so neither a NOAEL nor a LOAEL could be determined from this study. The deaths of
14 both mice given 12.5 mg/kg MCLR indicate that this is an FEL.
15
16 Ito et al. (1997a) compared the acute effects of MCLR on the livers and gastrointestinal
17 tracts of young and aged mice. Single doses of 500 ng/kg MCLR (purity not specified)
18 dissolved in ethanol and diluted in saline were administered via oral gavage to aged (29 mice
19 aged 32 weeks) and young (12 mice aged 5 weeks) male ICR mice. Three aged and three young
20 untreated mice served as controls. Twenty-two aged mice were sacrificed at 2 hours, five mice
21 at 5 hours, and two mice at 19 hours after treatment; four young mice were sacrificed at each
22 time point. Liver damage and gastrointestinal erosion were evaluated.
23
24 The results showed marked differences between young and aged mice in both liver
25 damage and gastrointestinal effects. In young mice, no liver pathology or gastrointestinal
26 changes were reported. In contrast, 18 of 29 aged mice treated with the same dose developed
27 pathological changes of the liver. Among the aged mice, 8 of 29 had liver injury of the highest
28 severity, characterized as bleeding, disappearance of many hepatocytes in the whole liver and
29 friable tissue (severity rating of+4). Five of 29 mice had liver changes characterized by
30 bleeding and disappearance of hepatocytes in centrilobular region (severity rating of+3).
31 Necrosis in the centrilobular region was observed in 4 of 29 mice (severity rating of+2) and
32 eosinophilic changes in the centrilobular region were indicated for one mouse (severity rating of
33 +1).
34
35 Gastrointestinal effects observed in the treated aged mice included necrosis to one-third
36 depth of the mucosa and severe duodenal damage (including decreased villi density, separation
37 of epithelial cells from lamina propria and edema of both the submucosa and villi). Details of
38 the incidence of these effects were not reported; however, the authors indicated that the degree of
39 liver injury was related to the severity of gastrointestinal effects. Regeneration of intestinal
40 tissues was evident in some of the mice sacrificed at later time points (5 and 19 hours after
41 treatment). Among untreated aged mice, serum enzyme levels (AST and ALT) were not
42 different, but gastrointestinal condition was somewhat compromised in the aged mice. Aged
43 mice had thinning of surface epithelial cells with consequent exposure of lamina propria and
44 glands in some areas. The authors postulated that the oral uptake of MCLR was dependent on
45 gastrointestinal tract erosion and the loss of permeability in capillaries of the villi. This study
27 DRAFT - DO NOT CITE OR QUOTE
-------
1 identified a freestanding LOAEL of 500 ng/kg (in aged mice only) for liver and gastrointestinal
2 effects.
3
4 In a poorly described study, Fitzgeorge et al. (1994) administered MCLR via gavage to
5 newly weaned CB A/BALBc mice. The commercially-obtained compound was described only as
6 "suitably purified." The LDso was estimated to be 3000 ng/kg, and increases in liver (43%) and
7 kidney (5.9%) weights were reported. The authors reported that there was no change in lung or
8 spleen weight; other endpoints were either not examined or not reported.
9
10 Rogers et al. (2005) evaluated the potential synergism between MCLR and anatoxin-a
11 administered by gavage to CD-I mice (sex not specified). A total of 60 fasted mice were given
12 gavage doses of 0, 500 or 1000 ng/kg MCLR (purity >98%) followed 50 minutes later with
13 gavage doses of 0, 500, 1000 or 2500 |J,g/kg anatoxin-a (purity >95%). Controls were given
14 distilled water by gavage. While not specified, group sizes are presumed to have been five
15 animals per treatment. The animals were observed for clinical signs of toxicity, loss of appetite
16 and mortality; body weight was measured before treatment and 3 hours later. The duration of
17 observation was not reported. No deaths, clinical signs of toxicity or differences in body weight
18 were observed. Effect levels cannot be identified from this study due to inadequate reporting
19 and because few toxicological endpoints were evaluated.
20
21 4.2.1.1.2. Cyanobacterial Extracts
22
23 Rao et al. (2005) compared the acute oral effects of microcystin extract in aged (36
24 weeks old) and young (6 weeks old) Swiss albino mice. A single LD50 dose of extract was
25 administered to male mice; mortality occurred after 4-5 hours. Both groups of mice had
26 increased relative liver weight and DNA fragmentation compared to control, but there was no
27 difference between the age groups. In contrast, glutathione depletion and lipid peroxidation
28 were significantly greater in the aged mice when compared with young mice. Further, while
29 most serum enzymes were increased over controls in both groups, GGT was increased to a
30 greater extent in aged mice than in young mice.
31
32 4.2.1.2. Short-Term Studies
33
34 4.2.1.2.1. Purified Microcystins
35
36 Heinze (1999) evaluated the effects of MCLR in drinking water on 11-week-old male
37 hybrid rats (Fl generation of female WELS/Fohm x male BDIX). Groups of 10 rats were given
38 doses of 0, 50 or 150 ng/kg body weight for 28 days in drinking water. Water consumption was
39 measured daily and rats were weighed at weekly intervals. Dose estimates provided by the
40 authors were not adjusted to account for incomplete drinking water consumption (3-7% of
41 supplied water was not consumed over the 28-day period). The test material was obtained
42 commercially, but the authors did not report a measure of purity. After 28 days of exposure, rats
43 were sacrificed by exsanguination under ether anesthesia. Organ weights (liver, kidneys,
44 adrenals, thymus and spleen) were recorded and hematology, serum biochemistry and
45 histopathology of liver and kidneys were evaluated.
46
28 DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
Hematological evaluation demonstrated an increase in the number of leukocytes in rats in
the highest dose group (38% increase). Serum biochemistry showed significantly increased
mean levels of ALP and lactate dehydrogenase (LDH) in both treatment groups (84 and 100%
increase in LDH, 34 and 33% increase in ALP in low and high doses, respectively) and no
changes in mean levels of ALT or AST. A dose-dependent increase in relative liver weights was
observed (17 and 26% at the low and high doses, respectively). Table 4-3 shows the mean
enzyme levels and relative liver weights.
Table 4-3. Serum Enzyme Levels and Relative Liver Weights (Mean + Standard Deviation) in
Rats Ingesting MCLR in Drinking Water (Heinze, 1999)
Parameter
Relative liver weight (g/100 g body weight)
Lactate dehydrogenase (microkatals/L)
Alkaline phosphatase (microkatals/L)
Control
n=10
2.75 + 0.29
16.64 + 4.48
9.67 + 2.20
50 ng/kg
n=10
3.22 + 0.34*
30.64 + 5.05*
13.00 + 3.81*
150ng/kg
n=10
3.47 + 0.49*
33.58+1.16*
12.86+1.85*
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
* p<0.05 when compared with control
The authors also reported a dose-dependent increase in absolute liver weights, although
the data were not provided. No statistically significant changes in other organ weights or body
weights were observed. In treated animals, histopathological alterations in the liver were
classified as toxic hepatosis. The incidence of liver lesions is summarized in Table 4-4. Lesions
were spread diffusely throughout the parenchyma and included increased cell volume, increased
mitochondria, cell necrosis, activation of Kupffer cells and increased amounts of periodic acid-
Schiff (PAS)-positive substances, indicating cell damage. Liver lesions were observed in both
treatment groups, but the severity of the damage was increased in the 150 ng/kg dose group. No
effects on the kidneys were observed. A NOAEL could not be determined from this study. The
lowest dose tested (50 ng/kg) represents a LOAEL based on liver lesions, increased relative liver
weights and changes in liver enzymes (ALP and LDH).
4.2.1.2.2. Cyanobacterial Extracts
Davidson (1959) treated groups of three mice with drinking water or feed mixed with
different extracts (crude, aqueous and filtered) from aNostoc rivulare water bloom for 21 days.
The only effects reported were ruffled hair and nervousness in the mice treated with crude
extract. Kalbe (1984) observed no change in body weight among juvenile mice and rats given
filtrates from two different water blooms of M. aeruginosa for 2-8 weeks.
29
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-4. Incidence of Liver Lesions in Rats Ingesting MCLR in Drinking Water for 28
Days (Heinze, 1999)
Liver Histopathology
Control
n=10
50 |ig/kg
n=10
150 ng/kg
n=10
Degenerative and Necrotic Hepatocytes with Hemorrhage
Slight
Moderate
Intensive damage
0
0
0
4
6
0
0
6
3
Degenerative and Necrotic Hepatocytes without Hemorrhage
Slight
Moderate
Intensive damage
0
0
0
0
0
0
0
1
0
PAS-positive Material
Slight
Moderate
Intensive damage
1
0
0
5
5
0
0
8
2
Activation of Kupffer Cells
Slight
Moderate
Intensive damage
0
0
0
0
10
0
0
10
0
Lipid Granules and Droplets
Slight
Moderate
Intensive damage
0
1
0
4
2
0
0
1
0
30
DRAFT - DO NOT CITE OR QUOTE
-------
1 Orr et al. (2003) exposed yearling beef cattle to live cultures of M aeruginosa in
2 drinking water in an effort to evaluate whether microcystins accumulated in the liver or blood of
3 the animals. Four steers were treated for 28 days, and four untreated steers served as controls.
4 No effects on body weight, weight gain, food or water consumption or plasma enzymes (GOT,
5 glyceraldehydes dehydrogenase, AST or bilirubin) were observed. The authors reported no
6 detectable microcystins (by HPLC and gas chromatography-mass spectrometry) in either plasma
7 samples collected throughout treatment or in samples of the liver collected upon sacrifice at the
8 termination of exposure. Analysis of the liver samples by ELISA showed measurable
9 microcystins; however, the authors indicated that these results likely represented cross-reaction
10 with something besides microcystins, given the failure of the more sensitive HPLC analysis to
11 detect microcystins.
12
13 Schaeffer et al. (1999) reported the results of an unpublished 1984 study in which
14 Aphanizomenon flos-aquae, a cyanobacterium consumed as a food supplement, was fed to mice
15 in the diet. The authors used recent analysis of the A. flos-aquae, which often coexists with
16 Microcystis species, to estimate the microcystin content in the material consumed by the mice.
17 Analysis of i]\Q A. flos-aquae samples used in the feeding study showed an average concentration
18 of 20+5 |j,g MCLR per gram of A. flos-aquae. The authors estimated the daily exposure of
19 MCLR in the exposed mice to range from 43.3 ng/kg body weight per day to 333.3 |j,g/kg-day.
20 No clinical signs of toxicity were reported, and no effects on mortality, body weight, organ
21 weights or histology were observed in the treated mice. In addition, no effects on reproductive
22 parameters were reported in five treated mice (highest dose group) allowed to breed.
23
24 4.2.1.3. Subchronic Studies
25
26 4.2.1.3.1. Purified Microcystins
27
28 Fawell et al. (1999) conducted acute, subchronic and developmental toxicity studies of
29 MCLR given via gavage to Crl :CD-1(ICR)BR (VAF plus) mice (age not specified). MCLR was
30 obtained commercially and administered in distilled water. The concentration in the dosing
31 solution was verified by HPLC with UV detection. Daily oral gavage doses of 0, 40, 200 or
32 1000 |J,g/kg body weight were given to groups of 15 male and 15 female mice for 13 weeks.
33 Daily clinical observations were made, body weight and food consumption were recorded
34 weekly, and eye examinations were conducted prior to and at the conclusion of treatment.
35 Hematology and serum biochemistry were evaluated for seven mice of each treatment group
36 during the final week of treatment. Upon sacrifice after 13 weeks, gross examination of organs
37 and microscopic evaluation of tissues were performed. All tissues were examined in the control
38 and high dose animals, while only lungs, liver and kidney were examined in the other treated
39 animals.
40
41 Mean body weight gain was decreased approximately 15% in all treated male groups.
42 Mean terminal body weights differed from controls by about 7% in these groups. No dose-
43 related trends were evident for body weight gain or body weight in males. The only body weight
44 change observed in females was an increase in body weight gain in the 200 |j,g/kg-day group.
45 Hematological evaluation showed slight (10-12%) decreases in mean hemoglobin concentration,
46 red blood cell count and packed cell volume among females receiving 1000 ng/kg body weight.
31 DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
ALP, ALT and AST levels were significantly elevated (2- to 6-fold higher) in the high-dose
males, and ALP and ALT were likewise elevated (2- and 6-fold higher, respectively) in high
dose females. ALT and AST were also elevated (2-fold) in the mid-dose males. GGT was
slightly decreased in some treatment groups. Serum albumin and protein were reduced (13%) in
males of the mid- and high-dose groups. Table 4-5 shows the blood chemistry results.
Table 4-5. Blood Chemistry Results (Mean + Standard Deviation) for Mice Treated with MCLR for 13 Weeks
(Fawelletal., 1999)
MCLR Dose
(|ag/kg-day)
Alkaline
Phosphatase
(ALP) (U/l)
Alanine
Aminotransferase
(ALT) (U/l)
Aspartate
Aminotransferase
(AST) (U/l)
Gamma
Glutamyl
Transaminase
(GGT) (U/l)
Total
Protein
(g%)
Albumin
(g%)
Male
Control
40
200
1000
91+22.2
95 + 29.2
94 + 32.3
232b+ 103.2
27 + 8.0
37+17.2
59a+28.0
159c+75
68 + 27.7
64+12.2
121b+43.7
121b+26.3
6+1.0
4 + 0.7
3C + 0.4
4 + 0.4
5.5 + 0.32
5.1+0.26
4.8b + 0.29
4.8C + 0.21
3.2 + 0.19
3.0 + 0.13
2.8C + 0.13
2.8C + 0.11
Female
Control
40
200
1000
167 + 24.6
187 + 76.2
156 + 33.4
339b+ 123.7
32+11.3
25 + 7.8
27 + 9.4
220b+ 149.1
101+38.3
74+13.2
74 + 22.1
144 + 71.7
4+1.0
3+0.5
3+0.0
3+0.4
5.1+0.30
5.2 + 0.28
5.3+0.31
5.1+0.22
3.1+0.14
3.2 + 0.16
3.4a+0.14
3.1+0.18
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
a Significantly different from controls at p<0.05
b Significantly different from controls at pO.Ol
0 Significantly different from controls at p<0.001
Histopathological changes in the liver were reported in the males and females of the mid-
and high-dose groups, with a dose-related increase in incidence and intensity. The liver lesions
were multifocal and observed throughout the liver lobule. Table 4-6 summarizes the incidence
of liver histopathological changes. Sex-related differences in liver pathology were not apparent.
No lesions were found in other tissues.
The authors characterized the 40 ng/kg body weight dose as a clear NOAEL and
indicated that histopathological changes observed in the 200 ng/kg dose group were not severe.
The mid dose (200 |j,g/kg-day) represents a LOAEL based on the liver histopathological changes
and statistically significant blood chemistry changes. The WHO (1999) used the NOAEL value
of 40 |j,g/kg-day from this study as the basis for its provisional Tolerable Daily Intake for
MCLR.
32
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-6. Incidence of Liver Histopathology in Mice Treated with MCLR for 13 Weeks
(Fawell et al., 1999)
Liver Histopathology
Male
Acute inflammation
Chronic inflammation
Congestion
Hepatocyte vacuolation
Hemosiderin deposits
Hepatocyte degeneration
Female
Autolysis
Chronic inflammation
Congestion
Hepatocyte vacuolation
Hemosiderin deposits
Hepatocyte degeneration
Control
n=15
0
1
3
5
0
0
n=15
0
5
0
5
0
0
40 |j,g/kg-day
n=15
1
2
0
5
0
0
n=15
0
8
0
5
0
0
200 ng/kg-day
n=15
0
4
0
6
0
1
n=15
0
8
0
11
1
1
1000 ng/kg-day
n=15
0
15
1
O
15
14
n=15
1
14
1
8
14
9
33
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.2.1.3.2. Cyanobacterial Extracts
2
3 Falconer et al. (1994) administered dried bloom materials in the drinking water of pigs
4 for 44 days. Plasma samples collected over 56 days showed dose- and time-dependent increases
5 in GOT, ALP and total bilirubin, as well as a decrease in plasma albumin. Dose-related changes
6 in the incidence and severity of histopathological changes of the liver were also observed,
7 including cytoplasmic degeneration, hepatic cord disruption, single cell necrosis, periacinar
8 degeneration, congestion and Kupffer cell proliferation.
9
10 4.2.1.4. Chronic Studies
11
12 4.2.1.4.1. Purified Microcystins
13
14 Ueno et al. (1999) evaluated the toxicity of MCLR in mice chronically exposed via
15 drinking water. Two hundred 6-week-old female BALB/c mice were randomly assigned to
16 receive either no treatment or drinking water (ad libitum) containing 20 [ig/L MCLR for 7
17 days/week. The MCLR had been isolated from lyophilized algal bloom materials from Lake
18 Suwa in Nagano, Japan and had been characterized as 95% pure by HPLC. Twenty animals
19 from each group were sacrificed at 3, 6 and 12 months, while the remaining 40 animals were
20 retained for chronic toxicity evaluation and sacrificed at 18 months.
21
22 Weekly estimates of food and water consumption and daily observations for clinical
23 signs of toxicity, morbidity and mortality were recorded. Body weights were recorded weekly
24 for the first 2 months, biweekly up until the first year and monthly until sacrifice. At 3, 6, 12 and
25 18 months, blood was obtained from 20 animals from each group. Samples from 10 animals per
26 group were used for hematological evaluation, and samples from 10 additional animals were
27 used for serum biochemistry evaluation. At each scheduled sacrifice time, complete necropsy of
28 10 animals per group was conducted. Animals from the chronic toxicity group were necropsied
29 when moribund or dead (prior to scheduled sacrifice) or upon sacrifice at 18 months. Relative
30 and absolute organ weights (liver, kidneys, spleen, thymus, adrenal, ovaries, brain, heart and
31 uterus) were recorded for 9-10 animals per group at each scheduled sacrifice, and
32 histopathological evaluation of these and numerous other organs was conducted. Finally, three
33 to five animals per group were subjected to immunohistochemistry of the liver upon sacrifice to
34 determine the distribution of MCLR in the liver.
35
36 Based on weekly estimates of water consumption, the authors calculated the average total
37 intake of MCLR over 18 months to be 35.5 |j,g/mouse. No clinical signs of toxicity were
38 observed in either of the groups, and survival in the control and chronic treatment groups was
39 similar. No statistically significant differences in body weight, food consumption, water
40 consumption or hematology were observed; however, hematology data from the 3-month
41 sacrifice were lost due to sampling errors. Treated mice were reported to have a statistically
42 significant decrease in ALP at month 12 (13%) and a significant increase in cholesterol at month
43 18 (22%). Neither effect was considered by the authors to be lexicologically significant in the
44 absence of other treatment-related effects; however, the increase in cholesterol could be related
45 to the interaction of MCLR with bile acid transport in the liver.
34 DRAFT - DO NOT CITE OR QUOTE
-------
1 A decrease in heart weight among treated mice sacrificed at month 12 was not considered
2 treatment-related in the absence of histopathological changes. Sporadic changes in absolute and
3 relative thymus weight in treated mice were observed, but histological and morphometric
4 evaluation of the thymus revealed no abnormalities attributable to exposure. In contrast to other
5 studies, the authors observed no difference in the incidence of liver histopathology between
6 treated and control mice. Immunohistochemistry of the liver revealed no accumulation of
7 MCLR. This study identifies a free-standing NOAEL of 2.7 |j,g/kg-day in female mice
8 (calculated assuming a 24.5 g body weight and an exposure duration of 548 days).
9
10 Ito et al. (1997b) evaluated the carcinogenicity and liver toxicity of chronic gavage doses
11 of MCLR. A water bloom from Lake Suwa, Japan served as the source of the MCLR, which
12 was isolated and dissolved in ethanol and saline for dosing. The purity of the isolated MCLR
13 was not specified. Twenty-two ICR mice (13 weeks old) were given either 80 or 100 gavage
14 doses of 80 ng/kg MCLR over the course of 28 weeks. Ten mice were sacrificed after 80
15 treatments, five were sacrificed after 100 treatments and seven were withdrawn from treatment
16 after 100 doses and sacrificed 2 months later. Three mice served as untreated control. Although
17 the authors did not specify the nature of the postmortem examinations, it appears that the liver
18 was the only organ examined. No change in mean liver weight was observed in the MCLR-
19 treated animals compared with controls. The authors reported "light" injuries to hepatocytes in
20 the vicinity of the central vein in 8 of 15 mice sacrificed immediately after treatment, and in 5 of
21 7 mice that were withdrawn from treatment for 2 months after exposure. No fibrous changes or
22 neoplastic nodules were observed. Analysis for MCLR and its metabolites by
23 immunohistochemistry failed to detect either the parent compound or any metabolites in the
24 livers of mice sacrificed immediately after treatment.
25
26 Thiel (1994) briefly reported the results of a chronic toxicity study of MCLA in vervet
27 monkeys. The report is a brief summary published in the proceedings of an international
28 workshop; a published version of this study was not located. According to the summary, three
29 monkeys were given increasing intragastric doses of MCLA for 47 weeks, while three other
30 monkeys served as controls. Doses increased from 20 ng/kg at the commencement of the study
31 to 80 ng/kg at study termination. The rate of dosage increase was not reported. Monthly
32 measures of body weight and clinical signs (respiration, pulse, temperature) showed no effect of
33 treatment. Blood was withdrawn monthly; hematological parameters examined were hematocrit,
34 bilirubin, hemoglobin, erythrocyte and leukocyte count and platelet count. No statistically
35 significant changes in hematological parameters were observed. No changes were observed in
36 serum biochemistry analyses (albumin, globulins and electrolytes, as well as AST, LDH, ALP,
37 ALT and GGT). Histopathological examination of the liver and other organs (not specified) did
38 not show any differences in treated monkeys when compared with controls.
39
40 4.2.1.4.2. Cyanobacterial Extracts
41
42 Falconer et al. (1988) conducted a chronic exposure experiment using an extract of aM
43 aeruginosa water bloom in Swiss Albino mice. A concentration-dependent increase in mortality,
44 reduced body weight and a concentration-dependent increase in ALT levels were observed
45 among groups of mice receiving serial dilutions of the extract as their drinking water for a year.
46 There was some evidence that bronchopneumonia incidence was related to concentration of
3 5 DRAFT - DO NOT CITE OR QUOTE
-------
1 extract. No significant differences in liver histopathology were observed, although the observed
2 liver changes were slightly more prevalent in treated animals. The data showed some indication
3 of sex differences in susceptibility; male mice showed effects (including mortality and enzyme
4 level increases) at lower concentrations than females.
5
6 4.2.1.5. Initiation/Promotion Studies - Cyanobacterial Extracts
7
8 Falconer (1991) and Falconer and Buckley (1989) reported evidence of skin tumor
9 promotion by extracts of Microcystis. Microcystis extract was administered via drinking water
10 to mice pretreated topically with an initiating dose of dimethylbenzanthracene (DMB A). The
11 total skin tumor weight in mice drinking Microcystis extract was significantly higher than that of
12 mice receiving only water after initiation. The number of tumors per mouse was only slightly
13 increased in mice receiving extract; the weight difference was largely due to the weight of
14 individual tumors (Falconer and Buckley, 1989). The total weight of tumors in this group also
15 exceeded that of mice pretreated with DMBA and subsequently treated with topical croton oil,
16 with or without concurrent consumption of Microcystis extract. Details of the tumor incidence
17 in the mice were not provided by the authors. When Microcystis extract was provided in the
18 drinking water of mice pretreated with two oral doses of N-methyl-N-nitroso-urea, no evidence
19 of promotion of lymphoid or duodenal adenomas and adenocarcinomas was observed. No
20 primary liver tumors were observed (Falconer and Humpage, 1996).
21
22 Humpage et al. (2000) administered M. aeruginosa extract in drinking water to mice
23 pretreated with azoxymethane. Mice were sacrificed at intervals up to 31 weeks after
24 commencement of extract exposure. Enzyme analysis showed a concentration-dependent
25 increase in ALP and decrease in albumin in mice treated with extract. The authors observed a
26 concentration-dependent increase in the mean area of aberrant crypt foci of the colon, although
27 the number of foci per colon and the number of crypts per focus were not different among the
28 groups. The authors proposed that increased cell proliferation caused the increase in size of foci.
29 Histological examination of the livers of mice treated with extract showed more leukocyte
30 infiltration in animals treated with the highest concentration of extract compared to those
31 receiving a low concentration.
32
33 4.2.2. Inhalation Exposure
34
35 All available studies of inhalation exposure used MCLR, and there were no inhalation
36 studies using cyanobacterial extracts.
37
38 4.2.2.1. Acute Studies
39
40 Fitzgeorge et al. (1994) conducted experiments in CBA/BALBc mice with MCLR
41 administered via intranasal instillation and inhalation. This study is poorly described, giving few
42 details of study design and findings. A single experiment with mice (number unspecified)
43 inhaling a fine aerosol (particle size 3-5 |j,m) with 50 |j,g/L MCLR for an unspecified duration of
44 time did not result in any deaths, clinical signs of toxicity or histopathological changes. The
45 nature of the examinations was not reported. The authors estimated the delivered dose of MCLR
46 to be very small (about 0.0005 ng/kg).
36 DRAFT - DO NOT CITE OR QUOTE
-------
1
2 A brief abstract describes a study of acute MCLR exposure via inhalation (Creasia,
3 1990). Details of study design and results were not reported. The LCso for mice exposed to a
4 MCLR aerosol (nose only) for 10 minutes was reported to be 18 |J,g/L (mg/m3) air with a 95%
5 confidence interval of 15.0-22.0 |j,g/L (mg/m3). Based on studies of lung deposition after
6 exposure of mice to the LCso concentration, an LD50 of 43 ng/kg body weight was estimated.
7 The authors reported that histological lesions in mice killed by aerosol exposure were similar to
8 those in mice dosed intravenously with MCLR.
9
10 4.2.2.2. Short-Term Studies
11
12 Benson et al. (2005) exposed groups of six male BALB/c mice to monodisperse
13 submicron aerosols of MCLR via nose-only inhalation for 30, 60 or 120 minutes each day for 7
14 consecutive days. The concentration of MCLR was 260-265 ng/m3 and doses deposited in the
15 respiratory tract were estimated to be 3, 6 and 12.5 |J,g/kg body weight. Control mice were
16 exposed to the aerosolized vehicle (20% ethanol in water). Clinical signs were recorded daily.
17 Sacrifice by injection of Euthasol occurred the day after the last exposure. Blood was collected
18 by cardiac puncture and serum was subjected to clinical chemistry analysis (blood urea nitrogen
19 [BUN], creatinine, total bilirubin, ALP, AST, ALT, total protein, albumin and globulin). Organ
20 weights (adrenals, lung, liver, kidney, spleen and thymus) were recorded and histopathological
21 examinations of the liver, respiratory tract tissues, adrenals, kidney, spleen, thymus,
22 gastrointestinal tract and testes were conducted.
23
24 No clinical signs or effects on body weight or organ weights were observed following
25 exposure to MCLR aerosol. Histopathological examination revealed treatment-related lesions in
26 the nasal cavity only. Lesions were not observed in the liver, other organs or in other parts of the
27 respiratory tract. As indicated in Table 4-7, the incidence and severity of nasal lesions increased
28 with length of the daily exposure period. The lesions consisted primarily of necrosis or
29 inflammation of respiratory epithelial cells and degeneration, necrosis and atrophy of olfactory
30 epithelial cells. Necrotic lesions of olfactory epithelial cells were generally larger patches, while
31 few cells were involved in respiratory epithelial cell necrosis.
32
33 4.2.2.3. Subchronic and Chronic Studies
34
35 No subchronic or chronic animal studies evaluating the inhalation route of exposure were
36 identified in the materials reviewed for this document.
37 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-7. Incidence and Severity of Nasal Cavity Lesions in Mice Inhaling Microcystin
Aerosol for 7 Days (Benson et al., 2005)
Lesion
Severity
Control
Daily Exposure Period (minutes)
30
60
120
Respiratory Epithelial Necrosis
Turbinate 1
Turbinate 2
Minimal
Mild
Moderate
Mild
Moderate
0/6
0/6
0/6
0/6
0/6
1/6
0/6
0/6
0/6
0/6
0/6
6/6
0/6
6/6
0/6
0/6
0/6
2/6
3/6
3/6
Respiratory Epithelial Inflammation
Turbinate 1
Turbinate 2
Mild
Mild
0/6
0/6
1/6
1/6
0/6
0/6
1/6
0/6
Olfactory Epithelial Degeneration, Necrosis and Atrophy
Turbinate 1
Turbinate 2
Turbinate 3
Mild
Moderate
Mild
Moderate
Mild
Moderate
Marked
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
0/6
6/6
0/6
6/6
0/6
0/6
4/6
1/6
0/6
6/6
0/6
4/6
2/6
38
DRAFT - DO NOT CITE OR QUOTE
-------
1 4.3. REPRODUCTIVE/DEVELOPMENTAL STUDIES - ORAL AND INHALATION
2
3 4.3.1. Oral Exposure
4
5 4.3.1.1. Purified Microcystins
6
7 Fawell et al. (1999) conducted acute, subchronic and developmental toxicity studies of
8 MCLR given via gavage to Crl :CD-1(ICR)BR (VAF plus) mice (age not specified). MCLR (0,
9 200, 600 or 2000 ng/kg) was administered to groups of 26 mice on days 6-15 of pregnancy. The
10 mice were sacrificed on day 18 and necropsied. Weight and sex of the fetuses were recorded,
11 and external, visceral and skeletal examinations performed. Seven of 26 dams receiving 2000
12 Hg/kg died and 2 others were sacrificed prematurely due to morbidity. Altered liver appearance
13 was noted during gross examination of these animals. Surviving dams in this group did not
14 display any clinical signs of toxicity or differences in body weight or food consumption. The
15 authors reported that fetal body weight was significantly lower than controls and there was
16 delayed skeletal ossification at the highest dose; however, the data were not presented in the
17 publication. These effects may be associated with maternal toxicity. According to the authors,
18 no effects on resorption or litter size were observed, nor were there increases in external, visceral
19 or skeletal abnormalities in fetuses of any treatment group. Data on reproductive and
20 developmental parameters were not provided. Based on the authors' description of the findings,
21 it seems evident that the 600 |j,g/kg-day dose represents a NOAEL for both maternal and
22 developmental toxicity in mice. Further, the deaths of seven dams receiving 2000 |j,g/kg-day
23 clearly identify this dose as a maternal PEL. However, in the absence of the data showing
24 incidences of reproductive and developmental parameters, it is difficult to determine
25 conclusively whether the high dose also represents a LOAEL for developmental toxicity.
26
27 4.3.1.2. Cyanobacterial Extracts
28
29 Falconer et al. (1988) conducted a limited study of reproductive effects using an extract
30 from anM aeruginosa bloom sample. Eight female mice that had been given l/4th dilution of
31 the extract as drinking water (estimated to contain 14 |j,g/mL of unspecified microcystin toxin)
32 since weaning were mated with similarly treated males. No difference in number of litters, pups
33 per litter, sex ratio or litter weight were observed. Reduced brain size was reported to occur in 7
34 of 73 pups from treated parents and in none of 67 pups from controls. The litter distribution of
35 the affected pups was not reported by the authors. One of the small brains was examined
36 histologically, revealing extensive damage to the hippocampus.
37
38 4.3.2. Inhalation Exposure
39
40 No reports of developmental or reproductive toxicity by the inhalation route of exposure
41 were identified in the materials reviewed for this document.
42
39 DRAFT - DO NOT CITE OR QUOTE
-------
1 4.4. OTHER STUDIES
2
3 4.4.1. Neurological Effects
4
5 Neurological effects have been reported to occur with acute lethal doses of MCLR.
6 Clinical signs such as hypoactivity and piloerection have been observed in mice and rats exposed
7 to lethal doses of MCLR by oral or i.p. administration (Hooser et al., 1989a; Fawell et al., 1999).
8 These signs were observed in animals that subsequently succumbed to the lethal exposures. No
9 reports of neurotoxicity by the inhalation route of exposure were identified in the materials
10 reviewed for this document.
11
12 Maidana et al. (2006) reported that long-term memory retrieval (as assessed by step-
13 down inhibitory avoidance task) was impaired in rats receiving intrahippocampal injection of
14 0.01 or 20 |J,g/L of a microcystin extract from Microcystis strain RST 9501. Exposure at 0.01
15 |j,g/L also impaired spatial learning (as assessed by performance on the radial arm maze), but
16 exposure at the higher concentration did not. The authors indicated that the primary microcystin
17 produced by this strain is [D-Leu1] MCLR, a variant of MCLR. Oxidative damage, as measured
18 by lipid peroxides and DNA damage, was increased in tissue homogenates of the hippocampus
19 from treated animals.
20
21 Foxall and Sasner (1981) conducted limited in vitro studies on the neurological effects of
22 a crude extract from a bloom of M. aeruginosa. Little detail on experiment design was reported;
23 frog and mouse heart, frog sartorius muscle, frog sciatic nerve and mouse ileum were used in the
24 experiments. The authors reported that the extract had no effect on electrical or mechanical
25 events.
26
27 4.4.2. Immunological Effects
28
29 Evaluation of the immunotoxicity of microcystins in vivo was reported in only two
30 studies, both of which used a cyanobacterial cell extract rather than purified microcystins. Shirai
31 et al. (1986) reported that mice, immunized i.p. with either live or sonicated cells from a
32 Microcystis water bloom, developed delayed-type hypersensitivity when challenged 2 weeks
33 later with a subcutaneous injection of sonicated cells. Delayed hypersensitivity was assessed by
34 footpad swelling, which was increased approximately 2-fold over controls at the highest doses of
35 cells. It is not clear whether an endotoxin in the bloom sample was responsible for the
36 development of hypersensitivity, or whether the antigenic epitope existed on other components
37 of the sample.
38
39 Shen et al. (2003) assessed the effect of cyanobacterial cell extract on immune function.
40 Mice received 14 daily i.p. injections containing a cell-free extract from a water bloom
41 dominated by M. aeruginosa. Doses were reported as 16, 32 and 64 mg lyophilized cells/kg
42 body weight or as 4.97, 9.94 and 19.88 ng/kg microcystin equivalents. HPLC analysis indicated
43 that the microcystin content of the extract was 79.53%, although specific congeners in the extract
44 were not reported. The following immunotoxicity endpoints were examined: phagocytosis,
45 lymphocyte proliferation and antibody production in response to sheep red blood cells.
46 Phagocytic capacity was reduced at the two highest doses, but percentage phagocytosis was not
40 DRAFT - DO NOT CITE OR QUOTE
-------
1 affected. B-lymphocyte proliferation was significantly reduced (33% compared to controls at 32
2 mg/kg), while changes in T-lymphocyte proliferation were mild, and deemed biologically
3 insignificant. Finally, humoral immune response, as measured by antibody-forming plaques,
4 was reduced in a dose-dependent manner in treated mice.
5
6 Chen et al. (2004, 2005) evaluated the role of nitric oxide generation and macrophage
7 related cytokines on the reduced phagocytic capacity induced by pure MCLR. A dose-dependent
8 inhibition of nitric oxide production was observed in activated macrophages, and a repressive
9 effect was seen in cytokine formation at the mRNA level (e.g., IL-lp, TNF-a, GM-CSF, IFN-y)
10 after either 24 hour (Chen et al., 2004) or 6 hour treatment (Chen et al., 2005). Hernandez et al.
11 (2000) indicated that MCLR enhanced the early spontaneous polymorphonuclear leukocyte
12 (PMN) adherence (not late or PMN stimulated early or late) at low concentrations, suggesting
13 that microcystins may affect the immune system.
14
15 4.4.3. Hematological Effects
16
17 Several studies have noted thrombocytopenia in laboratory animals treated with
18 microcystins or bloom extracts purportedly containing microcystins (Slatkin et al., 1983; Adams
19 et al., 1985, 1988; Takahashi et al., 1995). Early investigations explored whether microcystins
20 had a direct effect on platelets, and whether platelets might be responsible for pulmonary
21 thrombi (Slatkin et al., 1983; Jones, 1984). However, in vitro studies have shown that MCLR
22 neither induces nor impedes the aggregation of platelets (Adams et al., 1985). Pulmonary
23 thrombi apparently consist of necrotic hepatocytes circulating in the blood (see Section
24 4.4.5.2.1). More recent information supports the hypothesis that hematological effects observed
25 in animals acutely exposed to microcystins are secondary effects of liver hemorrhage (Takahashi
26 etal., 1995).
27
28 Takahashi et al. (1995) reported dose-dependent reductions in erythrocyte count,
29 leukocyte count, hemoglobin concentration, hematocrit and coagulation parameters 1 hour after
30 rats were exposed to MCLR (100 and 200 ng/kg i.p). None of these parameters changed until
31 after massive liver hemorrhage commenced. Further, hematological changes such as increased
32 prothrombin time and fibrin deposition in the renal glomeruli were not observed. The authors
33 concluded that the depletion of blood components occurred as a result of liver hemorrhage.
34
35 Interestingly, mild thrombocytopenia was reported in 1-week-old mice treated with a
36 large i.p. dose of MCLR, even though none of these mice died (Adams et al., 1985).
37
38 4.4.4. Effects by Dermal Exposure
39
40 No animal studies evaluating the effects in animals of dermal exposure to purified
41 microcystins were identified in the materials reviewed for this document; only one study using a
42 cyanobacterial extract was located. Davidson (1959) applied a crude extract from a bloom of TV.
43 rivulare to the shaved backs of three mice every 2 hours for a total of 12 hours. Heavy scales
44 were observed on the treated areas; the scales were gone within 4 days and hair regrowth
45 occurred in the following weeks. Dermal application of an aqueous extract or aqueous filtrate
46 did not result in any effects.
41 DRAFT - DO NOT CITE OR QUOTE
-------
1
2 4.4.5. Effects by Parenteral Exposure
3
4 4.4.5.1. Effects in Humans after Parenteral Exposure
5
6 In February of 1996, unfinished water from a reservoir with a cyanob acted al bloom was
7 used at a hemodialysis center in Caruaru Brazil, leading to numerous deaths among the patients
8 treated with the water (Jochimsen et al., 1998; Pouria et al., 1998; Carmichael et al., 2001;
9 Azevedo et al., 2002). The first report of this incident was published by Jochimsen et al. (1998),
10 and Pouria et al. (1998) published follow-up information on the status of the patients. Azevedo
11 et al. (2002) provided the most up-to-date information on patient status after the incident. Using
12 water samples from the reservoir, serum and tissue samples from patients, and a variety of
13 methods, each publication identified microcystins as the primary causative factor in the deaths
14 and disease among patients. Carmichael et al. (2001) used analytical results from tissue samples
15 with dialysate volumes to estimate the concentration of microcystins in the water to which
16 patients were exposed, as a direct measure of exposure concentration was not available.
17
18 Of the 131 patients undergoing dialysis during the 4 days when unfinished water was
19 supplied to the center, 116 experienced symptoms, including visual disturbances, nausea,
20 vomiting and muscle weakness; 100 developed acute liver failure; and 52 had died as of
21 December, 1996 (Azevedo et al., 2002). The acute presentation of the disease included malaise,
22 weakness, dizziness, vertigo, visual disturbances and blindness, nausea, vomiting and abdominal
23 pain. Clinical signs included hepatomegaly and jaundice. Biochemistry showed high
24 concentrations of bilirubin and ALP, moderate increases in AST and ALT, hypoglycemia,
25 hypoalbuminemia and severe hypertrigliceridemia. Major hematology findings were slightly
26 low platelet count (within normal range) and reduced platelet aggregation, as well as red blood
27 cell abnormalities (anisocytosis, acanthocytosis and schistocytosis) in some patients. Liver
28 biopsy and autopsy samples showed severe, diffuse individual hepatocyte necrosis throughout
29 the liver lobule, with cell-plate disruption and apoptosis; however, no intrahepatic hemorrhage
30 was observed. Leukocyte infiltration and canalicular cholestasis were also observed.
31
32 Microcystins were implicated as the major contributing factor to patient death and
33 morbidity (Jochimsen et al., 1998; Pouria et al., 1998; Carmichael et al., 2001; Azevedo et al.,
34 2002). Evidence for the role of microcystins was derived from a variety of sources discussed by
35 Azevedo et al. (2002) and Carmichael et al. (2001). Quantitative analysis of the phytoplankton
36 in the reservoir from which the water was supplied to the dialysis center showed that
37 cyanobacteria represented about 99% of the phytoplankton in the reservoir, although the species
38 present during the outbreak were not identified. Analyses of filter systems in the dialysis center
39 showed microcystins; the carbon filter also had cylindrospermopsin. Analyses of patient sera
40 and liver samples provided additional evidence. Patient sera were analyzed for other potential
41 toxins (chlorines, chloramines, trace elements, heavy metals, agricultural compounds and
42 pesticides), but none were found (Pouria et al., 1998). Both serum and liver analyses for
43 microcystins revealed MCYR, MCLR and MCAR. None of the biological samples contained
44 cylindrospermopsin. Finally, physiological effects observed in the patients closely mirrored
45 effects observed in laboratory animals exposed to microcystins. Specifically, the liver damage
46 observed in patients was similar to that observed in mice (Jochimsen et al., 1998). Using
42 DRAFT - DO NOT CITE OR QUOTE
-------
1 microcystin concentrations measured in patients' livers and typical dialysis volume, Carmichael
2 et al. (2001) estimated the average concentration of microcystins in the dialysate to be 19.5 |j,g
3 microcystins/L.
4
5 Scares et al. (2006) reported another incident involving human exposure to microcystins
6 via dialysate. In November, 2001, 44 patients of a renal dialysis center in Rio de Janeiro, Brazil,
7 were assumed to be exposed to microcystins after a bloom of Microcystis andAnabaena
8 occurred in the reservoir supplying water to the center. The concentration of microcystins in the
9 drinking water was 0.4 [ig/L by ELISA. In the water used at the center (after treatment by
10 activated carbon column), the concentration was 0.32 [ig/L. Of the 44 patients exposed, 90%
11 had serum microcystin concentrations above the limit of detection (0.16 ng/mL). Twelve of the
12 patients were selected for 2-month follow-up monitoring of serum levels. Over the follow-up
13 period, serum concentrations ranged from <0.16 to 0.96 ng/mL. The highest serum
14 concentrations occurred 1 month after initial exposure. The authors did not provide any
15 information on health effects from the exposure.
16
17 Pilotto et al. (2004) reported that about 20% of 114 volunteers subjected to skin-patch
18 testing using cell suspensions and extracts of various cyanobacterial cultures (including two
19 strains of M aeruginosd) experienced a mild dermal skin reaction (erythema). The reaction did
20 not vary with cyanobacterial species.
21
22 4.4.5.2. Effects in Animals after Parenteral Exposure
23
24 4.4.5.2.1. Acute Studies with Parenteral Exposure
25
26 The acute toxicity of microcystins administered i.p. has been extensively studied. The
27 primary target organ for acute microcystin toxicity is the liver; effects have also been observed
28 in the kidney, lungs and gastrointestinal tract. Some of the effects observed in organs other than
29 the liver are believed to be secondary to liver effects.
30
31 A number of references report LDso estimates for injected MCLR (Slatkin et al., 1983;
32 LeClaire et al., 1988; Lovell et al., 1989a; Hermansky et al., 1990c; Miura et al., 1991; Stotts et
33 al., 1993; Gupta et al., 2003); two report LD50 values for MCRR and MCYR (Stotts et al., 1993;
34 Gupta et al., 2003). Table 4-8 summarizes the available estimates of microcystin LD50 values
35 after injection exposure to purified microcystins.
36
37 As the table shows, the LDso for MCLR in mice ranges between 30 and 60 ng/kg. The
38 acute lethality of MCYR is slightly lower than MCLR; LD50 estimates for MCYR were 111 and
39 171 ng/kg (Gupta et al., 2003 and Stotts et al., 1993, respectively). The LD50 for MCRR is
40 higher still, with LDso values estimated as 235 ng/kg (Gupta et al., 2003) and 650 |J,g/kg (Stotts
41 et al., 1993). In rats, the LDso for MCLR was similar to that in mice. There is some evidence
42 that
43 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-8. LD50 Values of Purified Microcystin Congeners by Intraperitoneal Administration
Sex/Strain
Purity
Vehicle
LD50 (95% CI)
Duration
Comments
Reference
MCLR-Mice
Male/Balb/C
Female/Swiss
albino Hale-
Stoner
Female/NIH
non- Swiss
Male/Swiss
Webster
Not specified/
CBA/Balbc
Female/Swiss
albino
Male/Swiss
albino
75%
NS
>95%
NA
NS
NS
NS
NS*
NS
Distilled
water
0.09% saline
NS
NS
Methanol and
PBS
32.6(±1.2)ng/kg
60 ng/kg
61 ng/kg
<100 ng/kg
250 ng/kg
43. 0|ig/kg (37.5-49.4)
43 ng/kg (37.5-49.4)
24 hours
NS
NS
NS
NS
24 hours
24 hours
Impurities tested for
toxicity at high doses
with negative results
LD50 by up and down
method
LD50 by up and down
method
LDso by up and down
method
Lovell et al.,
1989a
Slatkin et al.,
1983
Hermansky et
al., 1990c
Stotts et al.,
1993
Fitzgeorge et
al., 1994
Gupta et al.,
2003
Rao et al.,
2005
44
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-8 cont.
Sex/Strain
Purity
Vehicle
LD50 (95% CI)
Duration
Comments
Reference
MCLR-Rats
Male/Fischer
Male/Fischer
344
NS
>95%
NS
saline
50 ng/kg (36-68)
Fed rats: 122 ng/kg (106-141)
Fasted rats: 72 ng/kg (60-83)
72 hours
25 hours
Intraarterial injection.
Abstract only
LeClaire et al.,
1988
Miura et al.,
1991
MCRR-Mice
Male/Swiss
Webster
Female/Swiss
albino
NS
NS
0.09% saline
NS
-650 ng/kg
235.40 ng/kg (202.3-272.8)
NS
24 hours
LD50 by up and down
method
LDso by up and down
method
Stotts et al.,
1993
Gupta et al.,
2003
MCYR-Mice
Male/Swiss
Webster
Female/Swiss
albino
NS
NS
0.09% saline
NS
-171 ng/kg
1 10.6 ng/kg (81.7-149.6)
NS
24 hours
LDso by up and down
method
LD50 by up and down
method
Stotts et al.,
1993
Gupta et al.,
2003
2 * Not specified.
45
DRAFT - DO NOT CITE OR QUOTE
-------
1 the LD50 for MCLR was higher in fed rats (122 ng/kg) than in fasted rats (72 ng/kg) (Miura et
2 al., 1991).
3
4 In general, death occurs quickly in animals receiving a lethal injected dose of MCLR.
5 Mice typically die within 1-2 hours of a lethal i.p. dose of MCLR (Adams et al., 1988; Gupta et
6 al., 2003). Mean time to death for mice exposed to a lethal dose of MCRR or MCYR is also
7 within 2 hours (Gupta et al., 2003). In mice, an i.p. dose of 100 ng/kg MCLR typically results in
8 100% mortality (Adams et al., 1988; Hooser et al., 1989a; Hermansky et al., 1990c). Hooser et
9 al. (1989a) compared the effects of MCLR in mice and rats and observed significant differences
10 in survival time; both male and female rats given less than 240 ng/kg survived between 20 and
11 32 hours; rats receiving higher doses died within 8 hours. In contrast, female mice receiving 100
12 Hg/kg died within 1.5 hours (Hooser et al., 1989a). Miura et al. (1991) administered MCLR i.p.
13 to fed and fasted rats and reported a protective effect of feeding. Median time to death for rats
14 given 100 ng/kg MCLR was 32 hours in fed rats and less than 2 hours in fasted rats.
15
16 The sequence of events leading to death in laboratory rodents has been extensively
17 studied (Slatkin et al., 1983; Adams et al., 1988; Hooser et al., 1989a,b, 1990; Takahashi et al.,
18 1995). In general, similar effects have been reported in both rats and mice, but effects occur
19 later in rats than in mice (Hooser et al., 1989a). Gross and microscopic changes in swine treated
20 intravenously are similar to those observed in rodents (Lovell et al., 1989b).
21
22 Ten minutes after mice received a lethal i.p. dose, clinical signs, enzyme changes and
23 liver weight changes were generally absent (Slatkin et al., 1983; Adams et al., 1988). Beginning
24 approximately 20 minutes after dosing in mice, liver weights increased as the livers became
25 suffused with blood (Slatkin et al., 1983; Adams et al., 1988; Hermansky et al., 1990c). Adams
26 et al. (1988) estimated that as much as 44% of the total blood volume was located in the liver 30
27 minutes after a lethal dose of MCLR in mice. Similarly, in swine treated intravenously, blood
28 volume lost to the liver was estimated to be about 40% (Beasley et al., 2000). At 30 minutes,
29 there were isolated areas of hepatic necrosis, and at 45 minutes, there was marked liver
30 congestion and widespread hepatic necrosis (Hermansky etal., 1990c). Pulmonary thrombi
31 observed at the time of death were generally believed to result from necrotic hepatocytes (Adams
32 et al., 1988). Other pulmonary effects observed at 30-60 minutes after exposure included
33 congestion, bronchial epithelial hyperplasia and necrosis, edema and hemorrhage (Gupta et al.,
34 2003). In general, hepatic enzyme levels show little or no change until 20-30 minutes after
35 dosing, when hepatic hemorrhage is beginning (Adams et al., 1988; Hooser et al., 1989a;
36 Hermansky et al., 1990c; Takahashi et al., 1995).
37
38 At 60 minutes, there was severe disassociation of hepatocytes, hepatocyte loss and
39 hemorrhage, with disintegration of liver architecture (Hooser et al., 1991b; Guzman et al., 2003).
40 Thrombocytopenia coincided with hepatic hemorrhage as blood accumulated in the liver (Slatkin
41 et al., 1983; Adams et al., 1988; Takahashi et al., 1995).
42
43 The available studies demonstrate a very steep dose-response curve for MCLR acute
44 toxicity. In female NIH non-Swiss outbred mice, the only change observed after i.p.
45 administration of 50 |j,g MCLR/kg was Kupffer-cell hyperplasia, while all mice receiving 100
46 DRAFT - DO NOT CITE OR QUOTE
-------
1 Hg/kg died (Hermansky et al., 1990c). Hooser et al. (1989a) reported that male rats dosed i.p.
2 with 20, 40 or 80 ng/kg and females dosed with 40 |J,g/kg MCLR showed no clinical signs of
3 toxicity nor gross or microscopic lesions in the liver or other organs examined, while 120 ng/kg
4 was lethal in some rats and 160 |J,g/kg was consistently lethal. Lovell et al. (1989a) administered
5 a sublethal i.p. dose (about 25 ng/kg) of MCLR to male mice and reported a significant increase
6 in liver weight (8.7%), but no clinical signs or hepatic lesions.
7
8 Induction of apoptosis is believed to be one mechanism for liver damage in acutely
9 exposed animals (Hooser, 2000). Yoshida et al. (1998) treated mice with single sublethal doses
10 of MCLR (20, 30 and 45 ng/kg i.p.) and observed them for 7 days. Multiple apoptotic cells were
11 noted in the centrilobular regions of the livers of these mice. Hooser (2000) reported widespread
12 apoptosis in most hepatocytes after rats were treated with a single i.p. dose of 500 ng/kg MCLR.
13 Yoshida et al. (1998) reported the occurrence of two types of apoptotic hepatocytes in mice
14 given sublethal doses of MCLR; one that revealed MCLR by immunohistochemistry and one
15 that did not. The authors suggested that the latter type may contain MCLR that had lost the
16 antigenic epitope, or may have become apoptotic via other means, including ischemia or
17 hypoxia. Apoptosis induction is not restricted to the liver; Botha et al. (2004) reported
18 significantly increased apoptotic indices in the gastrointestinal tract of BALB/c mice as early as
19 8 hours after a single 75% LD50 dose (specific dose not reported) of MCLR i.p. The authors
20 observed immunohistochemical evidence of MCLR in the lamina propria and postulated that
21 MCLR was involved in the induction of apoptosis. The role of apoptosis in microcystin-induced
22 liver toxicity is further discussed in Section 4.4.7.5 (Mechanistic Studies).
23
24 Effects outside the liver have been reported after acute injection exposure to MCLR. It
25 remains unclear whether such effects may be indirectly related to hepatotoxicity. Some studies
26 have shown increases in kidney weight (Hooser et al., 1989a; Lovell et al., 1989a) or other signs
27 of kidney damage (LeClaire et al., 1988; Zhang et al., 2002) in rats and mice after injection of
28 sublethal doses of MCLR. Lovell et al. (1989a) observed dilation of cortical tubules and
29 eosinophilic granular or fibrillar material in the cortical tubules after MCLR administration. In
30 addition to reports of kidney effects, there are scattered reports of cardiac effects, including
31 degeneration and necrosis of myocardial cells after i.p. or i.v. exposure to MCLR (LeClaire et
32 al., 1988; Zhang et al., 2002). LeClaire et al. (1995) reported that mechanisms such as reflex
33 tachycardia and increased cardiac output, which typically would allow the heart to compensate
34 for the acute hypotension caused by blood pooling in the liver, were impaired in rats given a
35 lethal dose of MCLR. The authors suggested a cardiogenic component to the toxicity of MCLR.
36 Oishi and Watanabe (1986) observed tachycardia in mice 20 minutes after i.p. injection of
37 lyophilized cells fromM aeruginosa. Finally, a few studies suggest that acute i.p. exposure to
38 MCLR can affect lipid peroxidation levels in both the intestinal mucosa (Moreno et al., 2003)
39 and liver (Towner et al., 2002) of rats.
40
41 Two injection studies support the finding by Ito et al. (1997a) by oral exposure that
42 young animals are not as susceptible to the acute toxic effects of microcystins as older animals.
43 Adams et al. (1985) administered lethal doses of MCLR via i.p. injection to 1-, 2- and 3-week-
44 old mice. None of the 1- or 2-week-old mice died, whereas 23 of 31 mice aged 3 weeks died
45 within 2 hours. The 3-week-old mice that survived were rechallenged with MCLR a week later,
47 DRAFT - DO NOT CITE OR QUOTE
-------
1 and all died. Rao et al. (2005) reported that time to death decreased with age in mice treated i.p.
2 with MCLR.
3
4 Guzman and Solter (2002) evaluated the acute effects of repeated injection of MCLR.
5 Male BALB/c mice were injected with 45 ng/kg MCLR daily for 2, 4 or 7 days. Livers of mice
6 receiving four or seven daily doses were pale and moderately enlarged, with an accentuated
7 reticular pattern; absolute and relative liver weights were statistically increased over controls.
8 Histopathology revealed apoptotic hepatocytes in the centrilobular region of mice receiving two
9 doses, and marked hepatocytomegaly, disorganized hepatocytes, multinucleated hepatocytes and
10 cytoplasmic vacuolation in mice receiving four or seven doses. Guzman et al. (2003) reported
11 immunostaining of some centrilobular hepatocytes following two doses of 45 ng/kg; however,
12 protein phosphatase activity was not affected.
13
14 Acute toxicity of bloom extracts is highly variable, likely reflecting the variable toxin
15 content of algal blooms. LDso values estimated for various bloom extracts range from 14 mg dry
16 weight cells/kg body weight to 1924 mg dry weight cells/kg (see Table 4-9). These studies used
17 a variety of test materials, including lyophilized cyanobacterial cells, cell-free lysates, etc.
18
19 Early studies of purified but unidentified toxins firomM aeruginosa show essentially the
20 same pattern of acute hepatotoxicity and time to death after i.p. injection of lethal doses in rats
21 and mice that is seen with purified MCLR (Elleman et al., 1978; Foxall and Sasner, 1981;
22 Falconer et al., 1981; Theiss et al., 1984, 1985, 1988; Jones and Carmichael, 1984; Siegelman et
23 al., 1984; Dabholkar and Carmichael, 1987). In addition, many studies of cyanobacterial bloom
24 extracts (primarily M. aeruginosa) administered via i.p. injection to laboratory rodents show
25 similar effects (Ashworth and Mason, 1945; Ohtake et al., 1989; Rao et al., 1994; Porfmo et al.,
26 1999; Sabour et al., 2002).
27
28 Jackson et al. (1984) also observed liver pathology in sheep exposed intraruminally to a
29 bloom sample identified asM aeruginosa. Time to death ranged from 18 to 48 hours post
30 exposure. In animals that succumbed, the livers were hemorrhagic and necrotic.
31
32 Effects on organs other than the liver have been reported in some studies of bloom
33 extracts. Bhattacharya et al. (1997) observed changes suggesting distal tubular dysfunction
34 (proteinurea, as well as decreases in kidney LDH and AST levels) in rats injected with LD50
35 doses of M aeruginosa extract, but no histopathological changes. Dose-dependent increases in
36 urea and creatinine and decreases in total protein and albumin were also observed. Picanco et al.
37 (2004) reported that i.p. injection of an extract from a culture of M aeruginosa (strain NPJB-1)
38 into either young or mature mice resulted in increased alveolar collapse and increased number of
39 polymorphonuclear and mononuclear cell infiltrations when compared with saline-treated
40 controls. There was a very low concentration of contaminating bacteria in the culture (i.e., the
41 culture was not axenic) used in this study, and the authors acknowledged that materials in the
42 extract other than microcystins may have contributed to the pulmonary effects.
43
48 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-9. Intraperitoneal LD50 Values for Bloom Extracts
Microcystin Source
Solution of lyophilized
cells of M. aemginosa
Solution of lyophilized
cells of M aemginosa
Solution of lyophilized
cells of M ichthyoblabe
Cell-free lysate of M.
aemginosa
Solution of lyophilized
bloom sample
Solution of lyophilized
bloom sample
Cell-free lysate of M.
aemginosa
Purified toxin fromM.
aemginosa
purified toxin of M.
aemginosa
Species
Rats
Mice
Mice
Mice
Mice
Mice
Mice
Mice
Mice
Sex/Strain
Male/Jcl:
Wistar
Male/Jcl:ICR
Male/Swiss
Male/Swiss
Not specified/
Balb/C
Male/Swiss
albino
Male/Swiss
albino
Not specified/
white, strain
unspecified
Male/white,
strain
unspecified
Vehicle
Saline
Saline
0.9% saline
0.9% saline
Saline
0.9% saline
Not
specified
Ethanol and
water
Not
specified
LD50 (95% CI)
67.4 mg dry weight
cells/kg, 2 hours
14.4 mg dry weight
cells/kg, 1 hour
502-1 924 mg cell dry
weight/ kg body weight
43 1 mg/kg dry weight, 24
hours
25-250 mg dry weight
phytoplankton sample/kg
154.28 mg algae/kg, 48
hours
3.5 g extract/kg, 24 hours
466+13 |ag/kg
56 |ag/kg (43-60)
Duration
2 hours
1 hour
Not
specified
24 hours
Not
specified
48 hours
24 hours
Not
specified
Not
specified
Comments
None of mice surviving
past 1 hour died within 1
week
Microcystin content
ranged from 0.73-0.78
Hg/g
LD50 by up and down
method
Bloom dominated by M.
aemginosa. MCLR
content of samples ranged
from 53-952 |ag/gDW
biomass
M. aemginosa was 95%
of bloom biomass
MCLR dose
approximately 9.625
mg/kg
Congener not
identified/specified
Congener not
identified/specified
Reference
Oishi and
Watanabe, 1986
Oishi and
Watanabe, 1986
Sabouretal.,
2002
Raoetal., 1994
Tarczynska et
al., 2000
Porfino et al.,
1999
Raoetal., 2005
Bishop etal.,
1959
Ellemanetal.,
1978
49
DRAFT - DO NOT CITE OR QUOTE
-------
1 In further support for the findings of age-dependent liver effects from purified
2 microcystins, Foxall and Sasner (1981) reported that neonatal and juvenile (age not specified)
3 mice had no clinical symptoms or liver pathology after i.p. injections of a crude lysate from a
4 bloom of M. aeruginosa. Details of the experiment and results were not provided. The authors
5 reported that the mice did not die until they were 20 days of age, but it is not clear from the
6 publication whether the mice treated as neonates died later or only mice that were at least 20
7 days old when treated died. Mature mice were also treated for comparison, but the results were
8 not reported. The authors concluded that "young animals were not sensitive to the toxin but
9 developed sensitivity as they matured".
10
11 4.4.5.2.2. Short-Term Studies with Parenteral Exposure
12
13 Guzman and Solter (1999) and Solter et al. (1998, 2000) evaluated the effects on rats of
14 short-term administration of MCLR via continuous i.p. infusion. After 28 days of exposure at
15 16, 32 or 48 |j,g/kg-day, there were dose-dependent increases in serum levels of sorbitol
16 dehydrogenase (SDH), AST, GOT, ALP and bile acids, while a dose-dependent decrease in
17 serum albumin and a decrease in ALT synthesis were also observed (Solter et al., 1998, 2000).
18
19 Immunohistochemistry on the liver showed evidence of bioaccumulation of MCLR in
20 liver cytosol, with measured liver concentrations increasing at a greater rate than the
21 administered dose (Solter et al., 1998). Apoptotic cells and cytoplasmic vacuolation were
22 observed in the livers of rats receiving 32 and 48 |j,g/kg-day. A later study exposing rats at the
23 same doses showed evidence for oxidative damage in the liver, as measured by dose-dependent
24 increases in malondialdehyde, a lipid peroxidation byproduct (Guzman and Solter, 1999). This
25 observation is consistent with evidence for oxidative stress after short-term exposure to MCLR.
26 4.4.5.2.3. Subchronic Studies with Parenteral Exposure
27
28 Shi et al. (2002; Chinese publication, only abstract reviewed) reported oxidative stress in
29 rats injected with i.p. doses of 4, 8 or 12 |j,g/kg-day MCLR for 35 days. Serum GGT and whole
30 blood glutathione were decreased, while LDH and AST increased after exposure, with no change
31 in ALT levels. Hepatocyte proliferation and apoptosis were also observed. Oxidative stress and
32 apoptosis are discussed further in Section 4.4.7 (Mechanistic Studies).
33
34 Elleman et al. (1978) administered daily i.p. injections of a purified toxin from a bloom
35 of M aeruginosa to white male rats (strain unspecified) for 6 weeks. Doses were reported as
36 fractions of the LDioo (0.75, 0.5, 0.25 and 0). These doses correspond to 52.5, 35 and 17.5 ng/kg
37 based on the reported LDioo (70 ng/kg). Two mice from each group were sacrificed weekly for
38 necropsy and histopathological examination of the liver, kidney, heart, lung, spleen and brain.
39 Eleven of 16 high-dose and 5 of 14 mid-dose mice died prior to scheduled sacrifice; none of the
40 mice receiving the low dose died early. Of the 11 high dose animals that died prematurely, nine
41 died during the first week with symptoms of acute toxicity and liver hemorrhage. In the
42 remaining groups, progressive liver changes were seen with each week, and dose-dependent
43 pathology was observed. The authors noted numerous mitotic figures in hepatocytes of the low-
44 dose mice early on. Other histopathological findings in the liver were hepatocyte degeneration,
50 DRAFT - DO NOT CITE OR QUOTE
-------
1 scattered necrosis, fibrosis and mononuclear cellular infiltration; details of the severity of these
2 findings in each group were not reported.
o
4 4.4.5.2.4. Chronic Studies with Parenteral Exposure
5
6 Milutinovic et al. (2002, 2003) evaluated the kidney effects of chronic i.p. administration
7 of MCLR and MCYR in rats. Doses of 10 ng/kg were administered to groups of five male
8 Wistar rats every other day for 8 months. After sacrifice under CO2 anesthesia, the kidneys were
9 removed, fixed, sectioned and stained. During exposure, the treated rats exhibited clinical signs
10 of toxicity and reduced body weight. Microscopic examination of the kidneys of treated animals
11 showed collapsed glomeruli and dilated tubules with eosinophilic casts and some cytoplasmic
12 vacuolation. The interstitial space was infiltrated with lymphocytes. More renal corpuscles
13 were significantly damaged in the MCLR-treated group than the MCYR-treated group.
14 Cytoskeletal abnormalities and DNA damage typical of apoptosis or necrosis were also observed
15 in tubular epithelial cells (Milutinovic et al., 2002, 2003). Although they did not report details of
16 the liver pathology in this study, the authors noted that the kidneys were more damaged than the
17 livers, suggesting that adaptation to exposure may have occurred in the livers.
18
19 4.4.5.2.5. Initiation/Promotion Studies with Parenteral Exposure
20
21 Nishiwaki-Matsushima et al. (1992) demonstrated that i.p administration of MCLR could
22 enhance the number and area of glutathione S-transferase (placental form; GST-P) positive foci
23 in a medium-term rat liver bioassay. In male F344 rats pretreated with 200 mg/kg
24 diethylnitrosamine (DEN) and partially hepatectomized, the number of GST-P positive foci was
25 significantly increased when the rats were subsequently treated with 10 ng/kg MCLR i.p. twice a
26 week. In a follow-up experiment, rats were pretreated with DEN and then given twice weekly
27 i.p. doses of 10, 25 or 50 ng/kg MCLR. A dose-dependent increase in the number and area of
28 GST-P positive foci was observed in the animals treated with MCLR (Nishiwaki-Matsushima et
29 al., 1992).
30
31 Ohta et al. (1994) also used the two-stage rat liver bioassay model to evaluate the
32 promotion capability of MCLR in rats pretreated with DEN, but not subjected to partial
33 hepatectomy. After treatment with 200 mg/kg DEN, rats given twice weekly i.p. injections of 25
34 Hg/kg MCLR had significantly more GST-P positive foci and significant increases in the area of
35 such foci when compared with DEN pretreatment alone. MCLR alone had negligible initiating
36 capability.
37
38 Hu et al. (2002) reported significant enhancement of gamma-GT foci in a two-stage
39 medium-term rat bioassay. Microcystin treatment (congener not specified) in DEN-pretreated
40 rats resulted in 100% incidence of gamma-GT foci, while DEN treatment alone resulted in foci
41 in only 22% of rats. Immunohistochemistry showed that microcystin exposure reduced
42 expression of the bax gene and increased expression of the bcl-2 gene. The Bax protein induces
43 apoptosis, while the Bcl-2 protein inhibits apoptosis (Klassen, 2001). This finding suggests that
44 apoptosis is inhibited by low doses of microcystin; in contrast, higher doses appear to induce
45 apoptosis (see Section 4.4.7.5 below).
51 DRAFT - DO NOT CITE OR QUOTE
-------
1 Sekijima et al. (1999) conducted a similar experiment using either DEN or aflatoxin Bl
2 (AFB1) as an initiator (with partial hepatectomy) and MCLR or a combination of AFB1 and
3 MCLR for promotion. In rats pretreated with 200 mg/kg DEN and subsequently given 10 ng/kg
4 MCLR i.p., there were increases in both the number and area of GST-P positive foci, but this
5 difference did not reach statistical significance. In rats pretreated with DEN and subsequently
6 given either AFB1 alone or a combination of MCLR and AFB1, a statistically significant
7 increase in number and area of foci was observed. The effect on GST-P positive foci of
8 combined treatment with MCLR and AFB1 was not synergistic, however; the number and area
9 of foci in animals treated with both was not larger than the sum of the foci induced by each
10 compound individually. When altered hepatic foci of all types (including basophilic and
11 eosinophilic/clear, rather than only GST-P positive foci) were analyzed, the number of foci was
12 significantly greater in rats treated with both MCLR and AFB1 than in those treated with either
13 toxin alone. To assess the effect of MCLR on initiation by AFB1, Sekijima et al. (1999)
14 pretreated rats with AFB1 followed by twice weekly i.p. injections of 1 or 10 |J,g/kg MCLR for 6
15 weeks. The number and area of GST-P positive foci were significantly increased in animals
16 given MCLR compared with controls. There was no difference in number or area of foci
17 between the two doses of MCLR. The authors suggested that the higher dose may have had
18 cytotoxic effects on hepatocytes.
19
20 Ito et al. (1997b) treated 13 ICR mice with 100 i.p. injections (5 times/week) of 20 |ig/kg
21 MCLR over 28 weeks. Five mice were sacrificed immediately after the last injection, while
22 eight mice were withdrawn from treatment and sacrificed 2 months later. Three untreated mice
23 served as controls. Relative liver weights were 4.7% of body weight in the control mice, 9% in
24 mice sacrificed immediately after the last treatment, and 6.8% in those withdrawn from treatment
25 and sacrificed later. Statistical comparisons among the relative liver weights were not provided,
26 nor were the data with which to perform these comparisons. Neoplastic nodules were observed
27 in the livers of all mice of both treatment groups. The nodules ranged in size up to 5 mm in
28 diameter. The mean numbers of nodules in the treated animals (7.7 and 9.9 nodules per cm2 area
29 in the groups sacrificed immediately and 2 months later, respectively) were not significantly
30 different from the controls (control data not provided). The incidence of nodule development in
31 the few control mice was not reported. The types of nodules were characterized as A) weakly
32 staining with hematoxylin and eosin and with small nuclei; B) intensely staining with eosin but
33 not with PAS or C) mainly occupied with fat droplets. The small number of animals in the
34 treatment and control groups limits the usefulness of these data.
35
36 4.4.5.2.6. Developmental/Reproductive Studies with Parenteral Exposure
37
38 Chernoff et al. (2002) investigated the developmental toxicity of MCLR in CD-I mice.
39 Pregnant mice were treated with i.p. or subcutaneous doses of MCLR (95% pure) on gestation
40 days (GDs) 7-8, 9-10 or 11-12. Doses of 0, 32, 64 and 128 ng/kg were administered i.p., while
41 only the 128 |J,g/kg dose level was administered subcutaneously. Mice were sacrificed by CO2
42 inhalation on GD 17. After litter and gravid uterus weights were recorded, fetuses were
43 examined for gross malformations and preserved for skeletal examination. Livers of the dams
44 were examined grossly and subjected to histopathology. No effects on maternal weight gain,
45 litter size, average fetal weight or incidence of gross or skeletal abnormalities were observed.
46 Histological examination of the maternal livers showed no effects of treatment.
52 DRAFT - DO NOT CITE OR QUOTE
-------
1
2 In a separate experiment by the same researchers (Chernoff et al., 2002), pregnant mice
3 treated with 32-128 ng/kg MCLR (via i.p. injection on GD 7-8, 9-10 or 11-12) were allowed to
4 give birth, and the growth and viability of the offspring were followed for 5 days. A different,
5 apparently more potent, lot of MCLR was used in this experiment. Mortality exceeded 50% (19
6 of 35 dams) in the 64 |J,g/kg dose group and only 1 of 34 animals in each of the 96 and 128 |J,g/kg
7 dose groups survived treatment. Among the surviving animals, there were no effects on
8 viability, birth weight or growth of litters during the brief follow-up period.
9
10 Experiments with rabbit whole embryo cultures in vitro suggest that low concentrations
11 of MCLR (10-20 |jM) can alter the organization of actin filaments and microtubules, although
12 cell morphology is not significantly affected (Frangez et al., 2003; Zuzek et al., 2003). At high
13 concentrations (100 |jM), MCLR causes cell rounding and loss of adhesion properties, with
14 consequent cell detachment and dispersion. Frangez et al. (2003) showed that the zona pellucida
15 (a glycoprotein envelope surrounding the ovum) forms an effective barrier against the effects of
16 MCLR, as rabbit whole embryos embedded in zona pellucida were not affected even at high
17 concentrations of MCLR.
18
19 Development of mouse embryos in culture was inhibited by a purified toxin from a
20 bloom dominated by Microcystis (Sepulveda et al., 1992). At a concentration of 120 |j,g/mL of
21 toxin in the culture medium, development of two-cell embryos was halted and cytolysis occurred
22 in some embryos. Disruption of the actin cortex was also observed in these embryos. At 60
23 Hg/mL, two-thirds of the embryos divided once more, and the remainder did not develop further.
24 In eight-cell embryos treated with 120 |j,g/mL toxin, compaction was prevented or reversed.
25 Embryos at this stage treated with 240 |j,g/mL did not develop further; cells were rounded and
26 lysed. The specific toxin used in this experiment was not identified.
27
28 Conflicting results have been observed in teratogenicity testing of purified microcystins
29 in the Frog Embryo Teratogenicity Assay-Xenopus (FETAX) assay with X. laevis embryos. In
30 one study, MCLR at concentrations of 25-250 |J,g/L was shown to induce both skeletal and soft
31 tissue malformations mX. laevis embryos (Dvorakova et al., 2002). In contrast, Fischer and
32 Dietrich (2000) reported no effects on mortality, malformation or growth in these embryos after
33 exposure to either MCLR or MCRR at concentrations up to 2000 |J,g/L. Dvorakova et al. (2002)
34 attributed the differing results to interlaboratory variability or variability in sensitivity of the
35 embryos. O'Brien et al. (2003) reported no effects of MCYR on X. laevis embryo mortality,
36 malformation or growth.
37
38 Interestingly, Dvorakova et al. (2002) demonstrated that biomass from & Microcystis
39 species (wesenbergii) that does not produce microcystins could induce malformations in
40 Xenopus embryos. O'Brien et al. (2003) also tested two extracts each from Plantothrix
41 rubescens andM aeriigmosa, reporting that all four extracts resulted in facial narrowing and
42 growth retardation in 96-hour Xenopus embryos, while purified MCYR had no effects.
43
53 DRAFT - DO NOT CITE OR QUOTE
-------
1 4.4.6. Effects by Intratracheal or Intranasal Instillation
2
3 Ito et al. (2001) evaluated the distribution of MCLR after intratracheal instillation of
4 lethal doses in male ICR mice and included a limited description of toxic effects. MCLR in
5 saline solution was instilled at various doses (50, 75, 100, 150 and 200 ng/kg) into 34 mice; 3
6 mice were sham-exposed as controls. Mortality was 100% in 12 mice receiving doses of 100
7 Hg/kg and greater. At 75 |J,g/kg, two of four mice died, while no deaths occurred in 18 mice
8 given 50 ng/kg intratracheally. The time course of hepatotoxicity was further evaluated in eight
9 mice given intratracheal doses of 100 ng/kg. One mouse was sacrificed at each of 5, 10, 20, 30,
10 45, 60, 90 and 120 minutes. Immunostaining for MCLR showed the toxin in the lungs within 5
11 minutes and in the liver after 60 minutes. Hemorrhage in the liver was observed after 90 minutes
12 and became severe by 120 minutes.
13
14 Fitzgeorge et al. (1994) conducted experiments in CBA/BALBc mice with MCLR
15 administered via intranasal instillation and inhalation. This study is poorly described, giving few
16 details of study design and findings. The LD50 for intranasal instillation of MCLR was equal to
17 the i.p. LD50 (250 ng/kg). Liver and kidney weights were increased in the animals receiving
18 MCLR intranasally (41.6 and 7.5% respectively). The authors further evaluated the relationship
19 between dose and liver weight increase after intranasal instillation of MCLR. At single
20 intranasal doses of 31.3, 62.5, 125, 250 and 500 ng/kg, liver weight increased proportionally (0,
21 1.5, 24.4, 37.4 and 87%). Seven daily intranasal doses of 31.3 ng/kg, a dose that had produced
22 no liver weight change after a single dose, resulted in a liver weight increase of 75%. Fitzgeorge
23 et al. (1994) reported histopathological findings, but failed to specify which findings resulted
24 from single doses and which resulted from the multiple-dose experiment reported in the same
25 publication. Findings included necrosis of respiratory and olfactory epithelium in the nasal
26 mucosa and centrilobular necrosis with hemorrhage in the liver. Early changes in the liver
27 included vacuolar degeneration and necrosis of hepatocytes near the central vein. The adrenal
28 glands showed effects as well, with vacuolation and necrosis of the inner cortex, as well as
29 congestion of medullary blood vessels. No histopathological changes were observed in the
30 trachea, lungs, esophagus, pancreas, spleen, lymph nodes, kidneys or brain.
31
32 4.4.7. Mechanistic Studies
33
34 Many mechanistic studies have been conducted to characterize the toxicology of
35 microcystins. These studies include in vivo investigations in laboratory animals, in situ studies
36 in isolated perfused organ systems and in vitro assays in isolated cell preparations. Mechanistic
37 studies have evaluated many aspects of microcystin toxicity, including: 1) the reason for target
38 organ and cell type specificity of microcystins, 2) description of the subcellular effects that occur
39 in susceptible cells, 3) interaction with serine and threonine protein phosphatases (i.e., PP1 and
40 PP2A) as the molecular target for microcystins, 4) the role of cytoskeletal effects, 5) the
41 importance of oxidative stress and apoptosis as a mode of toxic action and 6) the use of
42 chemoprotectants to reduce toxicity. Each of these topics is discussed in further detail below.
43 Mechanistic data related to the genotoxicity of microcystins is presented below in Section 4.4.8
44 (Genotoxicity and Cell Proliferation).
45
54 DRAFT - DO NOT CITE OR QUOTE
-------
1 4.4.7.1. Target Organ/Cell Type Specificity
2
3 Oral and injection studies in laboratory animals have demonstrated that the liver is the
4 primary target organ for microcystin toxicity (see Section 4.2). Mechanistic studies suggest that
5 the target organ specificity is directly related to the limited ability of microcystins to cross cell
6 membranes in the absence of an active transport system, such as the bile acid transporter in
7 hepatocytes. Evidence of the importance of the bile acid transporter to liver toxicity is provided
8 by studies that used bile acids and bile acid transport inhibitors (Runnegar et al., 1981, 1993,
9 1995b; Runnegar and Falconer, 1982; Eriksson et al., 1990a; Hermansky et al., 1990a,b;
10 Hermansky et al., 1991). These studies demonstrated that the liver toxicity produced by in vitro
11 or in vivo exposures to microcystins was reduced or eliminated by inhibition of hepatocellular
12 uptake using bile acid transport inhibitors (e.g., antamanide, sulfobromophthalein and
13 rifampicin) and bile salts (i.e., cholate and taurocholate). Additional discussion of the cellular
14 uptake of microcystins is provided in Section 3.2 (Distribution).
15
16 Runnegar et al. (1993) demonstrated that i.p. injection of mice with MCYM or MCLR
17 caused inhibition of liver protein phosphatase activity followed by evidence of liver toxicity (i.e.,
18 increased liver weight). Kidney protein phosphatase activity was unchanged following the in
19 vivo exposure even at lethal doses. In vitro exposure of kidney extracts to microcystins did
20 result in a decrease in kidney phosphatase activity, and no difference in sensitivity was observed
21 between liver and kidney phosphatase inhibition. This result suggests that target organ
22 specificity is most likely due to slower intercellular uptake of microcystins in the kidney.
23
24 The cell type specificity of microcystins was investigated using isolated rat hepatocytes,
25 rat renal epithelial cells (ATCC 1571) and rat skin fibroblasts (ATCC 1213) (Khan et al., 1995;
26 Wickstrom et al., 1995). The time course of light microscopic and ultrastructural effects was
27 examined following in vitro exposure to MCLR (Khan et al., 1995). Effects were noted after 4
28 minutes in hepatocytes, 1 hour in renal cells and 8 hours in fibroblasts. Similar lesions observed
29 in all cell types included blebbing, loss of cell-cell contact, clumping and rounding, cytoplasmic
30 vacuolization and redistribution of cellular organelles. Effects that were seen only in
31 hepatocytes include loss of microvilli, whirling of rough ER, dense staining and dilated cristae of
32 mitochondria and pinching off of membrane blebs. The nuclear changes typical of apoptosis
33 were seen in renal cells and fibroblasts. Cell type differences may be related to the specific
34 proteins that were overphosphorylated within each cell type. The authors postulated that the lack
35 of apoptotic changes in hepatocytes in this study might be related to the short exposure duration
36 or the failure of their transmission electron microscopic method to examine severely damaged
37 cells that had detached from the coverslips. Wickstrom et al. (1995) evaluated the changes in
38 cytoskeletal morphology after MCLR exposure in these cell types. High concentrations and long
39 incubation times were required for cytoskeletal changes in kidney and skin cells; however, the
40 nature of the changes was similar in all cell types (e.g., actin aggregation). Wickstrom et al.
41 (1995) suggested that microcystins may enter kidney cells and fibroblasts via pinocytosis.
42
43 McDermott et al. (1998) treated several cell types with MCLR (primary rat hepatocytes,
44 human fibroblasts, human endothelial cells, human epithelial cells and rat promyelocytes).
45 Hepatocytes underwent membrane blebbing, cell shrinkage, organelle redistribution, chromatin
55 DRAFT - DO NOT CITE OR QUOTE
-------
1 condensation and, in some cells, DNA fragmentation. Similar changes were observed in the
2 other cell types, but a longer duration of exposure was required.
3
4 Matsushima et al. (1990) demonstrated that injection of MCYR into mouse skin
5 epithelial cells and human fibroblasts resulted in morphological changes in cell shape (i.e.,
6 spindle shape to round forms). These cells were thought to be resistant to microcystins;
7 however, microcystins YR, LR and RR produced a dose-dependent inhibition of protein
8 phosphatase activity using the partially purified enzyme derived from mouse skin cell cytosol.
9 The authors suggest that absence of a direct effect in these cells is due to lack of penetration
10 through the cell membrane.
11
12 Many cell types and established cell lines have been evaluated for potential susceptibility
13 to microcystin uptake and toxicity. Primary isolated hepatocytes have been shown to be the
14 most sensitive to cytoxicity, due to the presence of the organic ion/bile acid transport system
15 (Eriksson et al., 1987, 1990b). Uptake was negligible in human hepatocarcinoma cells (Hep
16 G2), mouse fibroblasts (NIH-3T3), erythrocytes and human neuroblastoma cells (SH-SY5Y).
17 Hepatic endothelial cells have also been shown to be resistant to microcystin toxicity (Solow et
18 al., 1989; Runnegar et al., 1994). Primary cultures of liver cells cease to express these bile acid
19 transport proteins after 2-3 days of being maintained in culture. Therefore, established liver cell
20 lines are not generally useful for evaluating microcystin toxicity (Eriksson et al., 1994; Battle et
21 al., 1997; Heinze et al., 2001).
22
23 4.4.7.2. Characterization of Subcellular Effects in the Liver
24
25 The liver effects that occur following in vivo exposures to microcystins are generally
26 discussed in Section 4.2 (Animal Studies). Many additional mechanistic studies describe liver
27 histopathology, ultrastructural changes and biomarkers of cytotoxicity in either isolated perfused
28 rat liver (Berg et al., 1988; Pace et al., 1991; Runnegar et al., 1995b) or primary isolated
29 hepatocytes (Runnegar et al., 1981; Runnegar and Falconer, 1982; Aune and Berg, 1986;
30 Runnegar and Falconer, 1986; Berg and Aune, 1987; Runnegar et al., 1987; Falconer and
31 Runnegar, 1987a,b; Thompson et al., 1988; Solow et al., 1989; Mereish et al., 1989; Mereish and
32 Solow, 1990; Eriksson et al., 1990a; Boe et al., 1991; Khan et al., 1995; Runnegar et al., 1995b;
33 Yea et al., 2001; Batista et al., 2003).
34
35 Toxicological effects of microcystins in the isolated perfused rat liver were similar to
36 those demonstrated following in vivo exposure (Pace et al., 1991). During a 60-minute exposure,
37 MCLR caused liver engorgement and cessation of bile flow. Electron microscopy revealed loss
38 of sinusoidal architecture, dilation of bile canaliculi and the space of Disse and decreased
39 intracellular contact. Mitochondrial swelling, disruption of endoplasmic reticulum and
40 formation of whorls and loss of desmosomal intermediate filaments were also observed.
41 Mitochondrial function was impaired, with inhibition of state 3 respiration and a decrease in the
42 respiratory control index.
43
44 Runnegar et al. (1995b) demonstrated a decrease in protein phosphatase activity in
45 perfused rat liver exposed to MCYM. Cessation of bile flow, increased perfusion pressure,
46 decreased protein secretion and decreased glucose secretion were also observed. Histological
56 DRAFT - DO NOT CITE OR QUOTE
-------
1 changes included hepatocytes swelling, loss of sinusoidal architecture, pyknotic nuclei and
2 extensive necrosis. Exposure to high concentrations of toxin extracts in the isolated perfused
3 liver produced loss of cord architecture due to hepatocyte disassociation, membrane damage and
4 cytolysis and nuclear effects (pyknosis, karyokinesis, karyolysis) (Berg et al., 1988).
5 Ultrastructural effects included swollen mitochondria, vacuoles, necrosis, abnormal nuclei, bile
6 canaliculi lacking microvilli and whorls of rough endoplasmic reticulum.
7
8 Studies in primary isolated hepatocytes have described the morphological and
9 histopathological changes induced by microcystins that relate to loss of sinusoidal architecture
10 and cytotoxicity (Runnegar et al., 1981; Runnegar and Falconer, 1982; Aune and Berg, 1986;
11 Runnegar and Falconer, 1986; Berg and Aune, 1987; Runnegar et al., 1987; Falconer and
12 Runnegar, 1987a,b; Thompson et al., 1988; Solow et al., 1989; Mereish et al., 1989; Mereish and
13 Solow, 1990; Eriksson et al., 1990a; Boe et al., 1991; Runnegar et al., 1995b; Khan et al., 1995;
14 Yea et al., 2001; Batista et al., 2003). Microcystin exposure to hepatocytes in suspension or
15 cultured in a monolayer results in membrane blebbing that becomes more pronounced and
16 localized in one region of the cell surface. Cells are observed to be rounded in appearance and
17 become dissociated from one another. Microfilaments are reorganized as a compact spherical
18 body in the vicinity of the blebbing, while the rest of cell is depleted of filamentous actin.
19 MCLR disrupts hepatocellular morphology within minutes, leading to loss of sinusoidal
20 architecture and hemorrhage. Morphological changes in hepatocytes (i.e., blebbing, rounding)
21 have been shown to occur prior to any effect on cell membrane integrity (measured as LDH
22 leakage or release of radiolabeled adenine nucleotides) or cell viability (generally measured as
23 decreased trypan blue exclusion) (Runnegar et al., 1981; Runnegar and Falconer, 1982; Aune
24 and Berg, 1986; Ding et al., 2000a).
25
26 Thompson et al. (1988) described the time course of cellular effects of microcystins (type
27 not specified) on cultured rat hepatocytes. Disintegration of attachment matrix occurred by 15
28 minutes, followed by cells clustered in groups with no extracellular material at 1 hour, and
29 release of cells from plates between 2 and 4 hours. LDH release did not occur until after these
30 visual effects, but was dose-related.
31
32 Similar toxicological effects were observed in isolated human hepatocytes (Yea et al.,
33 2001; Batista et al., 2003). MCLR produced blebbing, fragmentation and hepatocyte
34 disassociation. Cytotoxicity, as measured by LDH leakage, occurred after morphological
35 changes were evident. Yea et al. (2001) indicated that cytotoxicity in human hepatocytes was
36 observed at a concentration (1 |jM) that did not affect rat hepatocytes. Batista et al. (2003) also
37 reported a slightly higher susceptibility to microcystin-induced morphological change in human
38 hepatocytes as compared to rat hepatocytes.
39
40 The ultrastructural effects observed following microcystin exposure in isolated rat
41 hepatocytes (i.e., condensation of chromatin, segregation of organelles, separated by apoptotic
42 microbodies, decrease in cell volume and increase in cell density) suggest that hepatocyte cell
43 death is apoptosis and not necrosis (Boe et al., 1991). Microcystin exposure in hepatocytes
44 causes cell shrinkage, externalization of membrane phosphatidylserine, DNA fragmentation and
45 chromatin condensation, indicating a rapid apoptosis (Ding et al., 2000b). Apoptosis is
46 discussed in further detail below as a possible mode of action for microcystin liver toxicity.
57 DRAFT - DO NOT CITE OR QUOTE
-------
1 Several studies have suggested that microcystins may increase the release and decrease the
2 reincorporation of arachadonic acid into cellular membranes (Adams et al., 1985; Naseem et al.,
3 1990, 1991; Nobre et al., 2001).
4
5 4.4.7.3. Molecular Target: Inhibition of Type 1 and 2A Protein Phosphatases
6
7 The primary molecular target of microcystins has been identified as serine and threonine
8 protein phosphatases PP1/PP2A. Protein phosphatases dephosphorylate proteins while protein
9 kinases phosphorylate them. Together, protein kinases and phosphatases maintain the balance of
10 phosphorylation and dephosphorylation of key proteins involved in cell cycle regulation.
11 Because more than 97% of protein phosphates occur at serine and threonine residues (Gehringer,
12 2004), the PP1 and PP2A protein phosphatases are particularly important. Inhibition of these
13 enzymes results in the increased phosphorylation of a number of regulatory proteins.
14 Importantly, PP1 is believed to be the major phosphorylase a phosphatase in the liver (Runnegar
15 et al., 1993). PP2A, the major soluble serine/threonine phosphatase, regulates several mitogen-
16 activated protein kinases (Gehringer, 2004).
17
18 Microcystins bind to these enzymes under both in vivo and in vitro study conditions,
19 resulting in an inhibition of enzyme activity leading to an increase in protein phosphorylation.
20 Microcystins have been shown to directly inhibit the activity of PP1 and PP2A derived from
21 several different species (i.e., fish, mammals, plants) and cell types (Honkanen et al., 1990;
22 MacKintosh et al., 1990; Matshushima et al., 1990; Yoshizawa et al., 1990; Sim and Mudge,
23 1993; Xu et al., 2000; Leiers et al., 2000; Becchetti et al., 2002). Microcystins do not alter
24 protein kinase activity, suggesting the balance of phosphorylation and dephosphorylation is
25 related to protein phosphatase inhibition alone. Microcystins have been used as a tool to
26 investigate the importance of serine and threonine phosphorylation to specific cellular functions.
27 The regulatory effects of phosphorylation on sodium channel opening in renal cells (Becchetti et
28 al., 2002), smooth and skeletal muscle contraction (Hayakawa and Kohama, 1995; Knapp et al.,
29 2002) and insulin secretion (Leiers et al., 2000) have been studied.
30
31 Runnegar et al. (1993) demonstrated the inhibition of PP1 and PP2A activity in the liver
32 following i.p. injection of MCYM and MCLR. Increased protein phosphorylation preceded the
33 observed increase in liver weight and was correlated with hepatotoxicity. Decreased
34 phosphatase activity was also demonstrated in the isolated perfused liver (Runnegar et al.,
35 1995b).
36 The relationship between phosphatase inhibition by microcystins and changes in
37 cytoskeletal structure and cell morphology has been reviewed (Eriksson and Golman, 1993).
38 Inhibition of protein phosphatase activity by MCLH and 7-dmMCRR was associated with an
39 increased phosphorylation of cytoskeletal and cytosolic proteins (Eriksson et al., 1990b).
40 Concentrations that produce a marked increase in protein phosphorylation were accompanied by
41 a complete reorganization of microfilament network. The cytoskeletal effects of microcystins
42 are discussed in further detail below. Microcystin LH and 7-dmMCRR were equipotent
43 inhibitors of purified PP1 and PP2A; however, higher concentrations of 7-dmMCRR were
44 required to increase protein phosphorylation. Table 4-10 shows studies with comparative data
45 on inhibition of protein phosphatases (IC50s) by MCLR, MCYR, MCRR and MCLA.
58 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-10. Studies Comparing Protein Phosphatase Inhibition Activity of Microcystin
Congeners
Reference
IC50 (nM)
MCLR
MCLA
MCYR
MCRR
PP2A Inhibition
Craig etal., 1996
Nishiwaki-Matsushima et al., 1991
Matsushima et al., 1990
0.15
0.28
7.6
0.16
4.5
0.78
5.8
PP1 Inhibition
MacKintosh et al., 1995
0.2
0.2
Mixture of PPs
Yoshizawa et al., 1990
1.6
1.4
3.4
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
The molecular interaction between microcystins and protein phosphatases has been
evaluated using immunoprecipitation, autoradiography, reverse phase liquid chromatography,
X-ray crystallography, nuclear magnetic resonance (NMR) solution structures, and molecular
dynamics simulation (Runnegar et al., 1995b; MacKintosh et al., 1995; Goldberg et al., 1995;
Craig et al., 1996; Bagu et al., 1997; Mattila et al., 2000; Mikhailov et al., 2003; Maynes et al.,
2004, 2006). Molecular modeling and molecular dynamics simulations have indicated that
microcystins bind in a Y-shaped groove containing the catalytic site on the surface of PP1
(Mattila et al., 2000). Studies with PP1 suggest that the C-terminal P12-P13 loop of PP1
(containing residues 268-281) is important for microcystin-protein phosphatase interactions as
well as for substrate recognition (Maynes et al., 2004, 2006). Information available to date
indicates that the binding process primarily involves the amino acids Glu, Adda, Leu and MDha
of microcystins. Figure 4-2 shows a schematic representation of the interactions between
microcystin-LR and protein phosphatase 1; these interactions are discussed further below.
Microcystins LR, LA and LL interact with the catalytic subunits of PP1 and PP2A in two
phases. The first phase occurs within minutes and consists of rapid inactivation of the
phosphatase. The second, slower phase of interaction represents a covalent interaction that takes
place within several hours (Craig et al., 1996). The initial binding and inactivation of protein
phosphatases appears to result from several non-covalent interactions that are still being
elucidated. Mattila et al. (2000) demonstrated an interaction of the Glu amino acid (reported as
59
DRAFT - DO NOT CITE OR QUOTE
-------
1 IGlu in the publication) carboxyl group of MCLR with a metal ion (Fe, Mn) in the PP1 catalytic
2 site. Glu appears to be an important component because esterification eliminates toxicity
3 (Namikoshi et al., 1993; Rinehart et al., 1994). Herfmdal and Selheim (2006), in a review of the
4 mechanisms of microcystin toxicity, indicated that the Adda side chain is involved in a
5 hydrophobic interaction between the Trp 206 and IlelSO residues in the hydrophobic groove of
6 PPL Mattila et al. (2000) suggested that the long side chain of the Adda residue may contribute
7 to orienting the toxin into the catalytic site. The Adda amino acid residue of microcystins plays
8 an important role in the inhibition of protein phosphatase activity (Nishiwaki-Matsushima et al.,
9 1991; Gulledge et al., 2002, 2003a,b). Isomerization of the diene from 4E,6E to 4E,6Z on the
10 Adda chain (see Figure 2-1) eliminates the toxic activity of microcystins (Harada et al., 1990;
11 Nishiwaki-Matsushima et al., 1991; Stotts et al., 1993). Microcystin analogues containing only
12 Adda and one additional amino acid are capable of substantial inhibition of PP1 and PP2A, while
13 modifications to the Adda structure abolished the inhibition (Gulledge et al., 2003b). Finally,
14 Herfmdal and Selheim (2006) indicated that the L-Leu of MCLR participates in a hydrophobic
15 interaction with Tyr 272 of PP1 (on the 012-013 loop).
16
17 The second phase of interaction between microcystins and protein phosphatase consists
18 of covalent bonding (Craig et al., 1996). Immunoprecipitation and autoradiography methods
19 indicate that covalent bonds result from the interaction between the methylene of the MDha
20 residue of microcystins and the thiol of Cys273 located at the C-terminal of PPL NMR solution
21 structures and X-ray crystallography data on the MCLR/PP1 complex illustrate the covalent
22 linkage at Cys-273 (Goldberg et al., 1995; Bagu et al., 1997). Site-directed mutagenesis
23 replacing Cys273 in PP1 results in a loss of microcystin binding (MacKintosh et al., 1995;
24 Maynes et al., 2004). Based on sequence similarity between PP1 and PP2A, it has been
25 suggested that Cys-266 is the site of equivalent covalent linkage between PP2A and microcystins
26 (Craig etal., 1996).
27
28 Microcystin analogues containing a reduced MDha residue are not capable of covalent
29 binding to protein phosphatases. MacKintosh et al. (1995) reported that a reduction of the MDha
30 residue of MCYR by ethanethiol abolished covalent binding to PPL Likewise, Craig et al.
31 (1996) showed that reduction of the MDha residue of MCLA abolished the covalent binding
32 phase with PP2A. Maynes et al. (2006) confirmed the lack of covalent interaction by
33 determining the crystal structure of dihydroMCLA bound to PP1. Their work showed that the
34 012-013 loop of PP1 takes on a different conformation when the covalent bond is absent, and
35 that other interactions (including hydrogen bonding) are responsible for the bond between
36 dihydroMCLA and PP1.
37
38 The importance of covalent bonding between microcystins and protein phosphatases to
39 toxicity resulting from the enzyme inhibition is uncertain, as other interactions are apparently
40 responsible for the rapid inactivation of the enzymes (Herfmdal and Selheim, 2006).
41 Modifications to either molecule (microcystin or protein phosphatase) to prevent covalent
42 bonding generally decrease, but do not eliminate, the toxic action (Meriluoto et al., 1990;
43 MacKintosh et al., 1995; Hastie et al., 2005).
60 DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
P12-P13loop
I
Figure 4-2. Schematic Representation of Interactions between Microcystin-LR and the Catalytic Site of Protein Phosphatase 1
61
DRAFT - DO NOT CITE OR QUOTE
-------
1 Microcystins may also bind to other molecular targets in addition to protein
2 phosphatases. Chen et al. (2006) used bioinformatic approaches to identify human liver aldehyde
3 dehydrogenase 2 (ALDH2) as a potential molecular target of MCLR. After screening a phage
4 display library to identify potential ligands specific for MCLR, Chen et al. (2006) used
5 molecular docking studies to show that MCLR could bind to ALDH2. The authors postulated
6 that this interaction could lead to aldehyde-induced reactive oxygen species (ROS) and
7 apoptosis.
8
9 4.4.7.4. Cytoskeletal Effects
10
11 The cytoskeletal effects of microcystins in the liver have been visually demonstrated in
12 several studies using light, electron and fluorescent microscopy (Runnegar and Falconer, 1986;
13 Eriksson et al., 1989; Hooser et al., 1989a,b, 1991b; Falconer and Yeung, 1992). Ultrastructural
14 changes in rats given a lethal dose of microcystin A1 include the following: a widening of
15 intracellular spaces; progressive disassociation followed by rounding, blebbing and invagination
16 of hepatocytes; loss of microvilli in the space of Disse; breakdown of the endothelium;
17 hemorrhage; and loss of lobular architecture (Hooser et al., 1989b). No effects were noted in
18 endothelial cells or Kupffer cells. In isolated hepatocytes, actin aggregates were seen at the base
19 of the membrane blebs. As membrane blebs grew larger and were drawn toward one pole of the
20 cell, the microfilaments were organized toward the same pole, resulting in rosette formation with
21 a condensed band of microfilaments at the center (Runnegar and Falconer, 1986; Eriksson et al.,
22 1989; Hooser et al.,1991b; Falconer and Yeung, 1992; Wickstrom et al., 1995; Ding et al.,
23 2000a). Frangez et al. (2003) also demonstrated cytoskeletal changes in rabbit primary whole
24 embryo cultured cells. Actin and microtubule disorganization was demonstrated to lead to
25 detachment and cellular toxicity.
26
27 The observed reorganization of microfilaments that leads to alteration of hepatocyte
28 morphology was not shown to be due to effects on actin polymerization (Runnegar and Falconer,
29 1986; Eriksson et al., 1989; Falconer and Yeung, 1992). Instead, microcystins cause an increase
30 in the phosphorylation of cytokeratin intermediate filament proteins (Falconer and Yeung, 1992;
31 Ohta et al., 1992; Wickstrom et al., 1995; Blankson et al., 2000). Toivola et al. (1997) evaluated
32 the effects of MCLR on hepatic keratin intermediate filaments in primary hepatocytes cultures.
33 A disruption of the desmoplakin2 organization at the cell surface (disorganization of
34 desmosomes) is followed by a dramatic reorganization of the intermediate filament and
35 microfilament networks, resulting in intermediate filaments being organized around a condensed
36 actin core. The major target proteins for microcystin-induced hyperphosphorylation include
37 keratins 8 and 18 and desmoplakin (DP) I/II. Keratins 8 and 18 are the major proteins of
38 intermediate filaments in hepatocytes; DP I and II attach keratin filaments in epithelial cells to
39 desmosomes. Hyperphosphorylation of DPI/II leads to loosening of cell junction and loss of
40 interactions with cytoplasmic intermediate filaments. The hyperphosphorylation of keratin
41 proteins leads to increased solubility (caused by disassembly or prevention of subunit
42 polymerization), leading to the observed morphological effects. Phosphopeptide mapping shows
1 The authors refer to the test compound as microcystin-A, which may reflect an old nomenclature no longer in use.
Available information is insufficient to identify the congener with current nomenclature.
2 Desmoplakin is the principal plaque protein in a desmosome, which is a localized thickening of membrane that
serves as an adhesion junction connecting contiguous cells.
62 DRAFT - DO NOT CITE OR QUOTE
-------
1 four specific tryptic peptides in soluble keratin 18 that are highly phosphorylated; however, no
2 specific phosphorylation sites have been identified for keratin 8. A Ca2+/calmodulin-dependent
3 kinase may be involved in regulating the serine-specific phosphorylation of keratin proteins 8
4 and 18.
5
6 Some investigators have suggested that generation of reactive oxygen species may play a
7 role in the cytoskeletal changes induced by microcystins. Ding et al. (2001) illustrated
8 generation of superoxide and hydrogen peroxide radicals preceding microfilament
9 disorganization and cytotoxicity. Hepatocellular glutathione levels were affected by
10 microcystins, and administration of N-acetylcysteine was shown to protect against cytoskeletal
11 alterations (Ding et al., 2000a).
12
13 4.4.7.5. Apoptosis
14
15 The ultrastructural changes observed in hepatocytes after microcystin exposure suggest
16 that cell death is related to apoptosis and not necrosis. These changes include cell shrinkage
17 (decreased volume and increased density), condensation of chromatin and segregation of
18 organelles separated by apoptotic microbodies (Boe et al., 1991; Fladmark et al., 1998;
19 McDermott et al., 1998; Ding et al., 2000b; Mankiewicz et al., 2001). The effects of
20 microcystins on the signaling pathways involved in rapid apoptosis have been investigated in
21 several studies (Ding et al., 1998a,b, 2000b, 2001, 2002; Ding and Ong, 2003). Mitochondrial
22 permeability transition (MPT) is considered to be a critical rate-limiting event in apoptosis.
23 Oxidative stress may play a role in the induction of MPT and the onset of apoptosis. In cultured
24 hepatocytes exposed to microcystins, an increase in the generation of ROS preceded the onset of
25 MPT, mitochondrial depolarization and apoptosis. A dose- and time-dependent increase in ROS
26 and lipid peroxidation, measured as malondialdehyde formation, was shown to precede
27 morphological changes in hepatocytes and release of LDH. The addition of deferoxamine or
28 cyclosporine A inhibited the formation of ROS and delayed the onset of MPT and cell death.
29 The addition of superoxide dismutase prevented collapse of cytoskeleton and release of LDH
30 from isolated hepatocytes. An early surge of mitochondrial Ca2+ was shown to occur prior to
31 MPT and cell death. Prevention of this Ca2+ surge by one of several methods (i.e., chelation of
32 intracellular Ca2+, blockage of the mitochondrial Ca2+ uniporter or use of mitochondrial
33 uncoupler) prevented MPT and cell death. Electron transport chain inhibitors (e.g., rotenone,
34 actinomycin A, oligomycin or carbonyl cyanide m-chlorophenylhydrazone) also inhibited the
35 onset of MPT. MCLR caused the release of cytochrome c through MPT, which is considered
36 universal in mitochondrial apoptosis; however, caspases -9 and -3 were not activated. The
37 increase in intracellular Ca2+ may instead facilitate the activation of calpain, which occurred
38 following exposure to microcystins (Ding and Ong, 2003). Botha et al. (2004) demonstrated that
39 apoptosis and oxidative stress can be induced in nonhepatic cells by microcystins. LDH leakage
40 and increased apoptotic indices were observed in the human colon carcinoma cell line (CaCo2)
41 and MCF-7 cells (deficient in pro-caspase-3). These changes were accompanied by increased
42 H2O2 formation and increased calpain activity.
43
44 Western blot analysis has been used to show that MCLR increases the expression of p53
45 and the pro-apoptotic Bax protein in both cultured rat hepatocytes treated with MCLR in vitro
46 and rat liver after in vivo exposure (Fu et al., 2005). Expression of the anti-apoptotic protein
63 DRAFT - DO NOT CITE OR QUOTE
-------
1 Bcl-2 was decreased in vitro, but in vivo MCLR treatment did not lead to a difference in the
2 expression of this protein. This finding suggests that MCLR may induce apoptosis through other
3 mechanisms in addition to the inhibition of protein phosphatases.
4
5 4.4.7.6. Lipid Peroxidation
6
7 Several studies have investigated the role of glutathione homeostasis and lipid
8 peroxidation in microcystin-induced liver toxicity (Runnegar et al., 1987; Eriksson et al., 1989;
9 Bhattacharya et al., 1996; Ding et al., 2000a; Towner et al., 2002; Gehringer et al., 2003a,b,
10 2004; Boua'icha and Maatouk, 2004). Ding et al. (2000a) indicated that microcystin exposure in
11 isolated hepatocytes resulted in an initial increase in glutathione synthesis followed by a later
12 depletion of glutathione. MCLR was shown to induce the de novo synthesis of glutathione in
13 mice exposed to a toxic sublethal dose (75% of the LD50) (Gehringer et al., 2004). Increased
14 transcription of glutathione-S-transferase was also demonstrated in his study. Gehringer et al.
15 (2004) suggest that increased lipid peroxidation induced by microcystins is accompanied by an
16 increase in glutathione peroxidase, transcriptional regulation of glutathione-S-transferase and
17 glutathione peroxidase and de novo synthesis of glutathione. Boua'icha and Maatouk (2004) also
18 reported that a low noncytotoxic concentration (2 ng/mL) of MCLR in primary rat hepatocytes
19 caused an initial increase in ROS formation and an increase in glutathione; however, a decrease
20 in lipid peroxidation was observed in this study. Electron spin resonance (ESR) spin trapping
21 techniques have demonstrated the formation of two possible lipid-derived free radical
22 metabolites in rat liver following in vivo exposure to MCLR (Towner et al., 2002). Vitamin E
23 and selenium supplementation in mice provided some protection against liver toxicity and
24 lethality by MCLR (Gehringer et al., 2003a,b). Measures of liver toxicity included serum
25 enzyme determination, lipid peroxidation, glutathione levels and histopathology. Hermansky et
26 al. (1991) reported that membrane active antioxidants (i.e., vitamin E, silymarin and glutathione)
27 provided some protection from microcystin toxicity (i.e., LDH leakage) and lethality; however,
28 free radical scavengers and water soluble antioxidants were ineffective (see below).
29
30 Several studies have reported MCLR-induced increases in lipid peroxidation as well as
31 decreases in antioxidant enzymes (Moreno et al., 2005; Jayaraj et al., 2006). Jayaraj et al. (2006)
32 measured oxidative stress in mice treated i.p. with an LDso dose of MCLR. Significant increases
33 in heat shock protein-70 and hepatic lipid peroxidation were observed. Further, GSH was
34 depleted, and there were decreases in the activity of glutathione peroxidase, superoxide
35 dismutase, catalase, glutathione reductase and glutathione-S-transferase in the animals treated at
36 the LD50. Similarly, Moreno et al. (2005) reported significant reductions in glutathione
37 peroxidase, glutathione reductase, superoxide dismutase and catalase, along with increases in
38 lipid peroxidation, in both the liver and kidney of rats treated intraperitoneally with single doses
39 of MCLR.
40
41 Some studies report the absence of lipid peroxidation during microcystin-induced
42 hepatotoxicity. A time-dependent leakage of LDH, ALT and AST was observed in liver slices
43 with no change observed for glutathione content or lipid peroxidation (Bhattacharya et al., 1996).
44 In addition, Runnegar et al. (1987) indicated that glutathione depletion did not occur until after
45 morphological changes (i.e., blebbing) were observed. Eriksson et al. (1989) indicated that rapid
64 DRAFT - DO NOT CITE OR QUOTE
-------
1 deformation of isolated rat hepatocytes by MCLR was not associated with alterations in
2 glutathione homeostasis.
o
4 4.4.7.7. Prevention of Liver Toxicity and Lethality
5
6 Several types of agents have been evaluated as potential chemoprotectants against
7 microcystin toxicity, including inhibitors of bile acid transport, microsomal enzyme inducers,
8 calcium channel blockers, free radical scavengers, water-soluble antioxidants and membrane
9 active antioxidants. It was initially reported that preincubation of hepatocytes with chemicals
10 that interfere with uptake of bile acids (sulfobromophthalein, rifampicin, sodium cholate and
11 sodium deoxycholate) also prevents hepatocyte deformation (Runnegar et al., 1981; Runnegar
12 and Falconer, 1982).
13
14 Hermansky et al. (1991) evaluated several possible chemoprotectants by measuring LDH
15 leakage and lethality following i.p. injection of MCLR. No protective effect was observed using
16 calcium channel blockers, free-radical scavengers or water-soluble antioxidants administered
17 prior to MCLR. Membrane active antioxidants, such as vitamin E, silymarin and glutathione,
18 provided some protection from microcystin toxicity and lethality. Phenobarbital provided partial
19 protection; however, tetrachlorodibenzo-^-dioxin did not afford protection, suggesting that the
20 protective effect may not be related to microsomal enzyme induction. Rifampicin and
21 cyclosporine A provided complete protection by blocking uptake of microcystins at the bile acid
22 transporter. Hermansky et al. (1990a,b) reported that rifampicin can be given 15 minutes after
23 MCLR injection and still prevent lethality, while cyclosporine A prevents lethality only if given
24 0.5-3 hours before MCLR injection.
25
26 Silymarin and dithioerythritol, both antioxidants, were shown to reduce MCLR toxicity,
27 as measured by LDH and adenine nucleotide release and light microscopy in primary cultures of
28 adult rat hepatocytes (Mereish and Solow, 1990). Dithioerythritol and silymarin have both been
29 shown to increase the content of reduced thiols (i.e., glutathione). Silymarin has additionally
30 been shown to stabilize membranes, inhibit lipoxygenase, reduce leukotrienes, scavenge free
31 radicals and increase protein synthesis.
32
33 Mereish et al. (1991) indicated that silymarin pretreatment inhibited microcystin liver
34 toxicity in mice (evidenced by histopathology and serum enzyme levels) following i.p. injection,
35 but not oral administration. As discussed above, vitamin E and selenium supplementation in
36 mice provided some protection against liver toxicity and lethality by MCLR (Gehringer et al.,
37 2003a,b).
38
39 Thompson and Pace (1992) evaluated several types of agents for their ability to protect
40 against MCLR toxicity in cultured hepatocytes. Toxicity was measured as morphology under
41 light microscopy, LDH release and protein synthesis inhibition. The uptake of MCLR into
42 hepatocytes was also measured. Cytochalasins D&E, fungal metabolites that interfere with actin
43 polymerization into microfilaments, were shown to protect against LDH release and provided
44 moderate protection from rounding and clustering of cells; however, these compounds produced
45 cytotoxicity themselves at concentrations that were required for protection against microcystin
46 toxicity. Cholate and deoxycholate are competitive inhibitors for the bile acid transporter.
65 DRAFT - DO NOT CITE OR QUOTE
-------
1 These compounds provided some protection against LDH leakage, but were also cytotoxic.
2 Trypan blue and trypan red also provided some protection related to blocking microcystin uptake
3 into hepatocytes. The antibiotic rifampicin was shown to prevent microcystin uptake and
4 toxicity at low non-cytotoxic concentrations, suggesting a possible therapeutic use in
5 microcystin poisoning.
6
7 Rao et al. (2004) confirmed that pretreatment with cyclosporine-A, rifampin and
8 silymarin each provided 100% protection against a lethal dose of MCLR (although the route was
9 not specified, it is assumed to be i.p., based on the lethal dose). Protected animals had
10 significantly reduced glutathione and increased hepatic lipid peroxidation up to 7 days after
11 treatment, but levels were returned to normal by 14 days.
12
13 Adams et al. (1985) demonstrated that MCLR lethality following i.p. injection in mice
14 was reduced by pretreatment with a single subcutaneous injection of carbon tetrachloride.
15 Lethality was also reduced in young mice in this study (no deaths at 1 and 2 weeks of age, 23/31
16 deaths at 3 weeks of age), suggesting that normal hepatic function is necessary for the uptake
17 and hepatotoxicity of MCLR. The microsomal enzyme inhibitors SKF525A and cobalt chloride
18 produced no effect on MCLR lethality, indicating that microsomal metabolism is not critical for
19 MCLR toxicity. The administration of hydrocortisone was also shown to protect against MCLR
20 lethality in mice, possibly due to a decrease in the release of arachadonic acid from membrane
21 phospholipids. Naseem et al. (1990) demonstrated that pretreatment of cultured rat hepatocytes
22 with glucocorticoids (flucinolone, dexamethasone and hydrocortisone) reduced the release of
23 arachadonic acid and metabolites caused by MCLR.
24
25 p-Carotene and lutein inhibited the effect of MCLR on hepatocyte morphology in mouse
26 primary hepatocyte cultures (Matsushima-Nishiwaki et al., 1995). P-Carotene protected the
27 cytokeratin network from disassembly and suppressed the hyperphosphorylation of cytokeratins
28 8 and 18. Several carotenoid analogs were evaluated, and the protective effect appeared to be
29 related to the number of trans configured double-bonds in the carotenoid.
30
31 4.4.7.8. Extra-Hepatic Effects of Microcystins
32
33 An isolated perfused kidney model was used to evaluate the kidney toxicity of MCLR
34 (Nobre et al., 1999, 2001). MCLR produced vascular, glomerular and tubular effects in the
35 exposed kidney. An increase in perfusion pressure was followed by an increase in the
36 glomerular filtration rate (GFR), increased urinary flow rate and a reduction in tubular transport
37 at the proximal tubules. Histopathological changes included protein in the urinary spaces, but
38 were not further described. Dexamethazone and indomethacin were shown to antagonize the
39 effects of MCLR on perfusion pressure, renal vascular resistance (RVR), GFR and urinary flow.
40 These results suggest a role for phospholipase A2 and cyclooxygenase in the kidney toxicity of
41 microcystins. Nobre et al. (2003) utilized rat peritoneal macrophages exposed to MCLR to
42 further investigate the role of inflammatory mediators in the isolated perfused kidney model.
43 Macrophage supernatants from exposed rats caused an increase in RVR, GFR and urinary flow
44 and reduced Na+ transport. These effects were reduced by cyclohexamide, dexamethasone and
45 quinacrine, further suggesting the involvement of PLA2 and other inflammatory mediators in
46 microcystin-induced kidney toxicity.
66 DRAFT - DO NOT CITE OR QUOTE
-------
1
2 Moreno et al. (2003) investigated the effects of MCLR on intestinal physiology following
3 an i.p. injection of 100 ng/kg in rats. Lipid peroxidation was increased in both the serum and the
4 intestinal mucosa of treated rats. With the exception of sucrase, intestinal brush border enzymes
5 were unaffected by MCLR exposure. An increase in the specific activity of acid phosphatase
6 and succinate dehydrogenase in intestinal homogenates suggests an effect of MCLR on
7 lysosomal and mitochondrial membranes, respectively. Nobre et al. (2004) used perfused rat
8 ileal segments and ligated intestinal loops to evaluate the effect of MCLR on electrolyte and
9 water secretion. MCLR caused significant secretion of water, sodium, potassium and chloride.
10 Aziz (1974) observed that a dialyzable component of whole cell lysate fromM aeruginosa
11 caused fluid accumulation in the ligated small intestine of guinea pigs.
12
13 Sicihska et al. (2006) evaluated the effects of MCLR on human erythrocytes in vitro.
14 MCLR exposure resulted in the formation of echinocytes, hemolysis, conversion of
15 oxyhemoglobin to methemoglobin, and a decrease in membrane fluidity. In addition, measures
16 of oxidative stress were affected in treated erythrocytes; glutathione reductase and superoxide
17 dismutase activity were decreased, while ROS and lipid peroxidation were increased.
18
19 Several studies have evaluated the effects of MCLR on immune system components in
20 vitro (Lankoff et al., 2004; Teneva et al., 2005; Chen et al., 2005; Kujbida et al., 2006). Lankoff
21 et al. (2004) reported that MCLR inhibited B-cell proliferation in human and chicken peripheral
22 blood lymphocytes at all concentrations tested, and decreased T-cell proliferation only at the
23 highest concentration. Apoptosis was enhanced in both human and chicken lymphocytes
24 (Lankoff et al., 2004). Similarly, MCLR was cytotoxic to mouse splenocytes, and caused
25 apoptosis in B-cells but not in T-cells (Teneva et al., 2005).
26
27 Kujbida et al. (2006) assessed the effects of MCLR and [Asp3]-MCLR on human
28 polymorphonuclear lymphocytes (PMNs) in vitro. Both compounds caused migration of
29 neutrophils in a chemotaxis chamber, suggesting that PMNs may migrate from the blood stream
30 to organs that concentrate microcystins, such as the liver. In addition, both caused a dose-related
31 increase in ROS production as measured by chemiluminescence of PMN degranulation products
32 that accompany ROS production. The phagocytosis of Candida albicans by PMNs was
33 increased after exposure to either compound, but only MCLR increased the intracellular killing
34 of C. albicans. These findings suggest the possibility that PMNs may mediate some of the toxic
35 effects of microcystins.
36
37 4.4.8. Genotoxicity and Cell Proliferation
38
39 Available data give conflicting results when purified MCLR has been tested for
40 mutagenicity. Pure MCLR did not induce mutations in the Ames assay either with or without
41 metabolic activation, although microcystin-containing extracts did induce mutations (Ding et al.,
42 1999). A crude toxin extracted from M. aeruginosa did not induce mutations in the Ames assay
43 (Grabow et al., 1982). In contrast, Suzuki et al. (1998) observed increased ouabain resistance
44 mutation frequency in human embryo fibroblast cells treated with MCLR (purity not specified).
45 Similarly, Zhan et al. (2004) reported a 5-fold increase over control in the frequency of
46 thymidine kinase mutations when human lymphoblastoid TK6 cells were treated with
67 DRAFT - DO NOT CITE OR QUOTE
-------
1 commercially-obtained MCLR. More slow-growing mutants were observed than fast-growing
2 mutants, suggesting that the mutation damage was larger than the TK locus, and that MCLR
3 induced large deletions, recombinations or rearrangements. Repavich et al. (1990) reported that
4 Ames assays (using strains TA98, TA100 and TA102) of a purified hepatotoxin (supplied by
5 Wright State University and presumed to be microcystin) were negative, as were Bacillus subtilis
6 multigene sporulation assays.
7
8 The conflicting information on mutagenicity may be related to differences in the cell
9 uptake of MCLR. For example, the failure of MCLR to induce mutations in bacterial cells may
10 be related to poor uptake. Zhan et al. (2004) reported that MCLR is not taken up by many cell
11 types, including bacteria; however, the authors did not provide references to support this
12 assertion. While hepatocytes take up MCLR at a significant rate, other cell types show limited
13 or no uptake unless measures are taken to enhance the penetration of the cells by MCLR. The
14 cellular uptake of microcystins is discussed in detail in Sections 3.2 (Distribution) and 4.4.7
15 (Mechanistic Studies).
16
17 A number of studies have reported DNA damage after MCLR treatment in vivo (Rao and
18 Bhattacharya, 1996), and in primary rat hepatocytes (Ding et al., 1999) and human hepatoma
19 cells (Zegura et al., 2003, 2004). Recent studies suggest that apoptosis may be intimately linked
20 to observations of DNA damage in cells treated with MCLR. Lankoff et al. (2004) showed a
21 strong correlation between DNA damage, as measured by the comet assay, and the induction of
22 apoptosis, as measured by the terminal deoxynucleotidyl transferase-mediated dUTP-biotin nick
23 end-labeling (TUNEL) assay, in human lymphocytes. Other evidence has suggested that the
24 comet assay can give a false positive measure of DNA damage when apoptosis is induced, as
25 DNA fragmentation is one consequence of apoptosis induction (Lankoff et al., 2004). The
26 authors postulated that earlier reports of DNA damage measured by the comet assay may have
27 been related to early stages of apoptosis due to cytotoxicity rather than a direct effect on DNA.
28 The induction of apoptosis appears to be dose-related. Humpage and Falconer (1999) showed
29 that low (picomolar) concentrations of commercially-obtained MCLR induced cytokinesis and
30 inhibited apoptosis in primary mouse hepatocytes, while higher (nanomolar) concentrations
31 resulted in the inverse effects.
32
33 Boua'icha et al. (2005) reported that noncytotoxic concentrations of MCLR did not cause
34 the formation of hydrophobic DNA adducts in primary cultured rat hepatocytes treated in vitro,
35 but did decrease the amount of endogenous hydrophobic adducts. MCLR was also shown to
36 cause a dose- and time-dependent increase in the formation of 8-oxo-7,8-dihydro-2'-
37 deoxyguanosine (a measure of oxidative DNA damage) both in cultured hepatocytes and in rat
38 liver cells after in vivo treatment via i.p. injection (Maatouk et al., 2004; Boua'icha et al., 2005).
39
40 Conflicting results have been reported in studies of MCLR-induced clastogenicity.
41 MCLR (commercially-obtained) has induced micronuclei in human lymphoblastoid cells and
42 mouse bone marrow erythrocytes (Ding et al., 1999; Zhan et al., 2004). Lankoff et al. (2004)
43 observed no effect of MCLR on the incidence of chromosomal aberrations in human peripheral
44 blood lymphocytes. Observations of polyploidy in MCLR-treated cells (Humpage and Falconer,
45 1999; Lankoff et al., 2003) may be related to its effects on cytokinesis. Lankoff et al. (2003)
46 showed that MCLR, through its effect on microtubules, damages the mitotic spindle, leading to
68 DRAFT - DO NOT CITE OR QUOTE
-------
1 the formation of polyploid cells. Repavich et al. (1990) reported a dose-related increase in
2 chromosome breakage in human lymphocytes exposed to a purified hepatotoxin (presumed to be
3 a microcystin).
4
5 Mechanistic evidence provides support for the hypothesis that MCLR can act as a
6 promoter at low doses. Zhu et al. (2005) reported that MCLR can transform immortalized
7 colorectal crypt cells, resulting in anchorage-independent growth and enhanced proliferation.
8 MCLR has been shown to increase the expression of the bcl-2 protein (that inhibits apoptosis)
9 and decrease the expression of the bax protein (that induces apoptosis) (Hu et al., 2002).
10 Further, MCLR upregulates the transcription factors c-fos and c-jun, leading to abnormal
11 proliferation (Zhao and Zhu, 2003). Gehringer (2004) reviewed the molecular mechanisms
12 leading to promotion by MCLR and the related tumor promoter, okadaic acid. Gehringer (2004)
13 reported that MCLR inhibits protein phosphatase PP2A, which regulates several mitogen-
14 activated protein kinases (MAPK). The MAPK cascade regulates transcription of genes required
15 for cell proliferation, including c-jun and c-fos. In addition, activation of the MAPK cascade has
16 been postulated to inhibit apoptosis and thus increase cell proliferation. Finally, Gehringer
17 (2004) noted that MCLR has been reported to increase phosphorylation of p53, which is
18 involved in the regulation of the cell cycle and apoptosis as well as the carcinogenic process.
19
20 4.4.9. Structure-Activity Relationships
21
22 With a few exceptions, microcystin congeners exhibit i.p. LDso values between 50 and
23 300 ng/kg in mice (Rinehart et al., 1994; WHO, 1999). MCLR is one of the most potent
24 congeners (i.p. LD50 approximately 50 ng/kg). Limited comparative testing of in vitro protein
25 phosphatase inhibition (IC5o) of MCLR, -RR and -YR resulted in IC5o values of 1.6, 3.4 and 1.4
26 nM, respectively (Yoshizawa et al., 1990), indicating that microcystin congeners may be
27 relatively similar in protein inhibition potency. Pharmacokinetic differences among the various
28 microcystin congeners may be at least partially responsible for observed variations in lethal
29 potency (Ito et al., 2002). Microcystin congeners of varying hydrophobicity were shown to
30 interact differently with lipid monolayers (Vesterkvist and Meriluoto, 2003). Effects on
31 membrane fluidity could alter the cellular uptake of these toxins.
32
33 Wolf and Frank (2002) proposed toxicity equivalency factors (TEFs) for the four major
34 microcystin congeners based on LD50 values obtained after i.p. administration. The proposed
35 TEFs, using MCLR as the index compound (TEF=1.0) were 1.0 for MCLA and MCYR and 0.1
36 for MCRR. The application of TEFs based on i.p. LDso values to assessment of risk from oral or
37 dermal exposure is questionable given that differences in liphophilicity and polarity of the
38 congeners may lead to variable absorption by non-injection routes of exposure.
39
40 The molecular interaction between microcystins and the catalytic subunits of protein
41 phosphatases has been extensively studied (see Section 4.4.7.3 for more detail). The interaction
42 was shown to occur in two phases. The first phase occurs within minutes and results in rapid
43 inactivation of the phosphatase (Craig et al., 1996). The amino acids Glu and Adda appear to be
44 important for the rapid inactivation of the protein phosphatases and for subsequent toxicity
45 (Harada et al., 1990; Nishiwaki-Matsushima et al., 1991; Namikoshi et al., 1993; Rinehart et al.,
46 1994; Gulledge et al., 2002, 2003a,b). The carboxyl group of the Glu residue in MCLR
69 DRAFT - DO NOT CITE OR QUOTE
-------
1 apparently interacts with a metal (Fe, Mn) ion in the PP1 catalytic site (Mattila et al., 2000). The
2 Adda side chain is involved in a hydrophobic interaction between the Trp 206 and IlelSO
3 residues in the hydrophobic groove of PP1 (Herfindal and Selheim, 2006). The few apparently
4 non-toxic (i.p. LDso values >1000 ng/kg) microcystin congeners exhibit structural alterations in
5 the Adda or Glu regions (Harada et al., 1990; Stotts et al., 1993; Rinehart et al., 1994).
6
7 The second, slower phase of interaction represents a covalent interaction between the
8 methylene of the MDha residue of microcystins and the thiol of Cys273 located at the C-terminal
9 of PP1 that takes place over several hours (Craig et al., 1996). Microcystin analogues containing
10 a reduced MDha residue are not capable of covalent binding to protein phosphatases
11 (MacKintosh et al., 1995; Craig et al., 1996; Maynes et al., 2006). The importance of covalent
12 bonding to the toxic effect of microcystins is uncertain, as other interactions are apparently
13 responsible for the rapid inactivation of the enzymes (Herfindal and Selheim, 2006).
14 Modifications to either molecule (microcystin or protein phosphatase) to prevent covalent
15 bonding generally decrease, but do not eliminate the toxic action (Meriluoto et al., 1990;
16 MacKintosh et al., 1995; Hastie et al., 2005; Herfindal and Selheim, 2006).
17
18 4.5. MODE OF ACTION - NONCANCER AND CANCER
19
20 Microcystins appear to result in different cellular effects depending on dose. In a review
21 of the mechanistic data on microcystins, Gehringer (2004) postulated a dualistic response for
22 microcystins and okadaic acid (another potent inhibitor of PP1 and PP2A). Gehringer (2004)
23 outlined evidence suggesting that at high doses, microcystins cause alterations in cellular
24 structure and function that may lead to cell death via apoptosis or necrosis, while at low doses,
25 microcystins inhibit apoptosis and cause cell proliferation. The high-dose effects are likely to be
26 responsible for the acute toxicity and lethality of microcystins and are discussed below.
27 4.5.1. Target Organ Specificity
28
29 The liver is the primary site of toxicological action for microcystins after oral, i.v., i.p.
30 and intranasal instillation exposure. Acute and short-term exposures to microcystins have
31 resulted in intrahepatic hemorrhage in both rats and mice (Ito et al., 1997a; Fawell et al., 1999;
32 Heinze, 1999). One subchronic study in mice also showed liver effects (Fawell et al., 1999),
33 including hepatocyte degeneration, chronic inflammation and hemosiderin deposits, but no
34 hemorrhage. Liver hemorrhage, resulting from apoptosis and necrosis of hepatocytes leading to
35 disintegration of hepatic architecture, appears to be the most prominent effect observed in
36 available toxicological studies,. The main reason for this target organ specificity is the greater
37 cellular uptake of microcystins by hepatocytes compared with other cells. Microcystins are
38 actively transported into hepatocytes by the bile acid transporter system, while uptake by other
39 cell types is limited by the lack of an active transport system. In vitro studies demonstrate that
40 preincubation of hepatocytes with compounds that block the uptake of bile acids prevent damage
41 to the hepatocyte from microcystin exposure (Runnegar et al., 1981, 1993, 1995a; Runnegar and
42 Falconer, 1982; Eriksson et al., 1990a; Hermansky et al., 1990a,b, 1991). Treatment of various
43 cell types with microcystins results in rapid damage to hepatocytes, while changes in other cell
44 types occur after much longer exposure (Khan et al., 1995; Wickstrom et al., 1995; McDermott
45 etal., 1998).
70 DRAFT - DO NOT CITE OR QUOTE
-------
1
2 4.5.2. Key Events in the Mode of Action for Liver Toxicity and Hemorrhage
3
4 Available mechanistic data indicate that the hepatotoxic effects of microcystins begin
5 with one or two molecular effects: inhibition of protein phosphatases and induction of oxidative
6 stress. At present, it is not clear whether the two effects are linked, whether they lead
7 independently to similar cellular effects (e.g., cytoskeletal damage and apoptosis), or whether
8 one effect is the predominant cause of hepatocellular damage. In both cases, the initial cellular
9 effects are associated with cytoskeletal changes and the induction of apoptosis. These alterations
10 in the structural integrity and function of hepatocytes lead to profound hepatotoxicity and
11 hemorrhage. In summary, the key events in the hepatotoxicity of microcystins appear to be:
12
13 1. Molecular events (inhibition of protein phosphatase and/or induction of oxidative stress);
14 2. Cellular effects (cytoskeletal damage and/or apoptosis); and
15 3. Tissue damage (altered liver structure and function, and intrahepatic hemorrhage).
16
17 The molecular events and cellular effects leading to tissue damage are discussed further
18 below.
19
20 4.5.2.1. Molecular Events
21
22 Protein Phosphatase Inhibition. As discussed above in Section 4.4.7.3, microcystins are
23 potent inhibitors of serine and threonine PP1 and PP2A. Inhibition of these protein phosphatases
24 results in aberrant phosphorylation of a number of cellular proteins, with the potential for
25 multiple effects on the cell. Current data suggest that the inhibition of PP1/PP2A by
26 microcystins can trigger cytoskeletal damage and apoptosis.
27 One outcome of microcystin-induced inhibition of PP1 and PP2A is the
28 hyperphosphorylation of cytokeratin intermediate filament proteins (Falconer and Yeung, 1992;
29 Ohta et al., 1992; Wickstrom et al., 1995; Blankson et al., 2000). Specifically, microcystin
30 exposure results in hyperphosphorylation of keratins 8 and 18 and desmoplakin I/II (Toivola et
31 al., 1997). It has been suggested that protein kinases PKC, PKA or the calcium/calmodulin-
32 dependent kinase may play a role in the hyperphosphorylation of these proteins (Gehringer,
33 2004). The hyperphosphorylation of desmoplakin I/II results in the loosening of cell junctions
34 and loss of interactions with cytoplasmic intermediate filaments, while the hyperphosphorylation
35 of keratins 8 and 18 leads to increased solubility. Some of the morphological changes observed
36 in hepatocytes (e.g., blebbing, rounding) may result from the hyperphosphorylation of
37 cytokeratin intermediate filament proteins.
38
39 Guzman et al. (2003) reported an increase in the phosphorylation of p53 in rat livers after
40 i.p. exposure to MCLR. PP1 and PP2A help to regulate the activity of p53 through
41 dephosphorylation. Thus, inhibition of these enzymes can result in hyperphosphorylation of
42 p53. Increases in the phosphorylation of p53 can cause an increase in the transcription of
43 p21WAF1, which in turn inhibits cyclin D-, E- and A-dependent kinases. The result of the latter
44 inhibition is to stall the cell cycle in Gl phase (Gehringer, 2004). This disruption of cell cycling
45 can allow for repair of DNA damage or for apoptosis to occur. For MCLR, there is evidence of
71 DRAFT - DO NOT CITE OR QUOTE
-------
1 hyperphosphorylation of p53, but other steps in this cascade of events have not yet been
2 investigated.
3
4 Oxidative Stress. A number of studies indicate that oxidative stress may play a role in
5 microcystin-induced hepatotoxicity. As noted above in Section 4.4.7.5, studies of
6 chemoprotectants show that several antioxidants can provide protection against the toxicity of
7 microcystins. Among the antioxidants shown to protect against the effects of microcystin, either
8 in vitro or in vivo, are vitamin E, silymarin, dithioerythritol, desferoxamine, N-acetylcysteine,
9 superoxide dismutase and glutathione. Further, dose and time-dependent increases in reactive
10 oxygen species have been shown to precede morphological changes in hepatocytes, and the
11 addition of superoxide dismutase prevents the cytoskeletal collapse caused by microcystins.
12 Ding and Ong (2003) have proposed two primary pathways by which microcystins increase
13 oxidative stress leading to cell death. First, microcystins may deplete glutathione, leading to
14 oxidative damage and cell death. Second, microcystins may increase the production of ROS by
15 disrupting the mitochondrial electron transport chain, leading to mitochondrial permeability
16 transition and apoptosis.
17
18 Microcystins may enhance oxidative stress by altering glutathione homeostasis; however,
19 the importance of glutathione homeostasis in MCLR-induced hepatotoxicity is not clear.
20 Glutathione serves as an intracellular antioxidant, by scavenging free radicals, by serving as a
21 substrate for the reduction of hydrogen peroxide by glutathione peroxidase and by detoxifying
22 xenobiotics. In addition, depletion of glutathione can disrupt microfilament structures in some
23 cell types (Ding and Ong, 2003). MCLR lethality has been prevented in mice by pretreatment
24 with glutathione (Hermansky et al., 1991), and hepatocytes pretreated with a glutathione
25 precursor were likewise protected from MCLR toxicity. Some studies have shown glutathione
26 depletion after microcystin exposure (Runnegar et al., 1987); however, depletion did not occur
27 until after membrane blebbing had been observed. Other studies have reported an increase in
28 glutathione after MCLR exposure (Ding et al., 2000a; Bou'iacha and Maatouk, 2004). Finally,
29 Eriksson et al. (1989) indicated that the rapid deformation of rat hepatocytes after MCLR
30 exposure was not associated with changes in glutathione levels. Thus, the role of glutathione
31 homeostasis in MCLR-induced hepatotoxicity has not yet been determined.
32
33 A variety of studies have demonstrated the importance of mitochondrial permeability
34 transition in the apoptotic cascade induced by MCLR (see Ding and Ong, 2003; Gehringer,
35 2004). In particular, studies have shown that pretreating hepatocytes with cyclosporin A, a
36 specific inhibitor of MPT, prevented cell death from microcystin exposure (Ding and Ong, 2003;
37 Gehringer, 2004). Ding and Ong (2003) outlined the following pathways for MCLR-induced
38 apoptosis via MPT. First, microcystin disrupts the mitochondrial electron transport chain,
39 leading to the release of reactive oxygen species from mitochondria and mitochondrial
40 permeability transition. MPT triggers a release of cytochrome c and mitochondrial calcium.
41 Cytochrome c may activate one or more caspases that trigger apoptosis, although neither
42 caspase-9 nor caspase-3 appear to be involved based on current information (Ding and Ong,
43 2003; Gehringer, 2004). The release of mitochondrial calcium activates calpain and
44 calcium/calmodulin-dependent protein kinase II, both of which also lead to apoptosis.
45
72 DRAFT - DO NOT CITE OR QUOTE
-------
1 4.5.2.2. Cellular Effects
2
3 Cytoskeletal Changes. Morphological changes observed in hepatocytes treated in vitro
4 with MCLR include membrane blebbing, cell rounding and dissociation. Membrane blebs
5 become localized in one region of the cell, and microfilaments are reorganized as a compact
6 spherical body near the blebbing (Runnegar and Falconer, 1986; Eriksson et al., 1989; Hooser et
7 al., 1991b; Falconer and Yeung, 1992; Wickstrom et al., 1995; Ding et al., 2000a). These
8 morphological changes occur before cell viability or cell membrane integrity is affected.
9 Electron microscopy of isolated perfused rat liver showed that these cellular effects led to loss of
10 sinusoidal architecture, dilation of bile canaliculi and the space of Disse and decreased
11 intercellular contact (Pace et al., 1991). Intrahepatic hemorrhage results from the breakdown of
12 liver structure, and the liver is rapidly engorged with blood.
13
14 Apoptosis. Membrane blebbing is also a characteristic of the apoptotic process
15 (Gehringer, 2004). A growing body of evidence indicates that microcystin exposure can trigger
16 apoptosis. Hooser (2000) used several visualization (light and electron microscopy) and
17 analytical techniques (TUNEL and electrophoresis to evaluate DNA laddering) to demonstrate
18 widespread apoptosis in the livers of rats 3 hours after an i.p. dose of MCLR. Characteristic
19 apoptotic changes including cell rounding, shrinkage, disassociation, loss of microvilli and
20 chromatin margination and condensation were observed in a majority of hepatocytes. The author
21 postulated that the rapidity with which the apoptotic process occurred overwhelmed the
22 phagocytic capacity of the liver, such that apoptotic hepatocytes depleted their energy stores and
23 later underwent necrosis. In mice, apoptotic hepatocytes have also been observed, but not to the
24 degree reported in rats. Hooser (2000) postulated that intrahepatic hemorrhage and death
25 occurred so quickly in mice that cellular changes characteristic of apoptosis did not have time to
26 develop.
27 4.5.3. Conclusion
28
29 The mechanisms by which microcystins induce hepatic damage have not been fully
30 elucidated. Available evidence suggests roles for both protein phosphatase inhibition and
31 oxidative stress as important molecular events, since chemoprotectant studies show that
32 pretreatment with compounds that inhibit either of these effects can protect against
33 hepatotoxicity in MCLR-treated animals. It is possible, even likely, that microcystin exposure
34 triggers a series of independent or linked events that cause cytoskeletal damage and/or apoptosis,
35 given the numerous cellular functions controlled by PP1 and PP2A, as well as the number of
36 effects triggered by oxidative stress. These cytoskeletal and apoptotic changes apparently lead
37 to the altered hepatic structure/function and intrahepatic hemorrhage observed in animal studies.
73 DRAFT - DO NOT CITE OR QUOTE
-------
1 4.6. SYNTHESIS AND EVALUATION OF MAJOR NONCANCER EFFECTS
2
3 The preponderance of toxicological data on the effects of microcystins is restricted to the
4 MCLR congener. A single, poorly-described study, reported only in a secondary source is
5 available for the LA congener. Data on the YR and RR congeners are limited to i.p. LD50 values
6 and measures of relative inhibition of protein phosphatases. As a result, this section largely
7 describes the available information on the toxic effects of MCLR, with limited reference to other
8 congeners.
9
10 Anecdotal reports indicate that, in humans, exposure to cyanobacterial blooms (including
11 microcystin-producing genera) can result in neurological, gastrointestinal and dermatological
12 symptoms, such as headache; muscle weakness; eye, ear and throat irritation; nausea; stomach
13 pain; diarrhea; blistering around the mouth; and hay-fever like symptoms (Dillenberg and
14 Dehnel, 1960; Billings, 1981; Turner et al., 1990; Teixeira et al., 1993; el Saadi and Cameron,
15 1993). Effects were reported in persons exposed via recreational contact (swimming, boating)
16 and drinking water. Turner et al. (1990) also reported pneumonia in army recruits exposed to a
17 cyanobacterial bloom. Symptoms occurring after exposure to cyanobacteria cannot be directly
18 attributed to microcystin toxins (or other endotoxins); some effects may result from exposure to
19 the cyanobacterial cells themselves, or from exposure to multiple toxins in the bloom.
20
21 The primary noncancer health effect of exposure to MCLR is liver damage. The liver is
22 targeted largely because hepatocytes are among only a few cell types that actively take up
23 microcystins, which do not readily cross the cell membrane. Severe liver damage (diffuse
24 individual hepatocyte necrosis, cell-plate disruption and apoptosis) occurred in dialysis patients
25 exposed to microcystins3 in dialysate (Jochimsen et al., 1998; Pouria et al., 1998; Carmichael et
26 al., 2001; Azevedo et al., 2002). At high acute doses in laboratory animals, MCLR caused
27 potentially fatal hemorrhaging within the liver. While the liver is the usual target of microcystin
28 toxicity, there have been some reports of effects in other systems, including hematological,
29 kidney, cardiac, neurological and gastrointestinal effects. It has been suggested that some effects
30 in other organs observed after high doses of MCLR may result from ischemia or hypoxia caused
31 by hepatic hemorrhage. However, some effects outside the liver have been observed in the
32 absence of hemorrhage.
33
34 Much of the toxicological data on microcystins are limited to reports of liver effects after
35 single lethal or sublethal doses administered via i.p. injection. These studies indicate that
36 injected doses of 50-200 |J,g/kg MCLR or MCYR are usually lethal in mice and rats within a few
37 hours (Adams et al., 1988; Hooser et al., 1989a; Hermansky et al., 1990c; Stotts et al., 1993;
38 Gupta et al., 2003; Rao et al., 2005). Miura et al. (1991) showed that the median time to death is
39 greatly increased in fed rats (32 hours) when compared with fasted rats (less than 2 hours). The
40 authors suggest that fasting may increase the sensitivity of animals to the mitochondrial toxicity
41 of microcystins, although this could not be conclusively demonstrated. In the liver, MCLR
42 destroys the cytoskeleton of hepatocytes, leading to hepatocyte disassociation, degeneration,
3 Exposure was to untreated water containing cyanobacteria. The presence of microcystins was confirmed in patient
biopsy samples; however, it is possible that the patients may also have been exposed to other microbial or chemical
contaminants.
74 DRAFT - DO NOT CITE OR QUOTE
-------
1 apoptosis and necrosis (Hermansky et al., 1990c; Hooser et al., 1991b). Hepatic hemorrhage and
2 disintegration of the liver architecture follow quickly (Hooser et al., 1991b). Effects reported to
3 occur outside the liver include pulmonary thrombi derived from necrotic hepatocytes, kidney
4 effects such as dilation of cortical tubules and eosinophilic material in the cortical tubules, and
5 degeneration and necrosis of myocardial cells (Adams et al., 1988; LeClaire et al., 1988; Zhang
6 et al., 2002). As previously stated, some of these effects may occur secondary to hepatic
7 hemorrhage.
8
9 Injection studies suggest a very steep dose-response curve for acute liver effects from
10 microcystin exposure. In several studies, mice and rats receiving single i.p. doses of 20-40
11 Hg/kg MCLR showed no clinical toxicity and few or no gross or microscopic effects in the liver
12 or other organs (Hooser et al., 1989a; Lovell et al., 1989a; Hermansky et al., 1990c), while i.p.
13 doses of 50-200 ng/kg are usually lethal within a few hours (Hooser et al., 1989a; Hermansky et
14 al., 1990c; Stotts et al., 1993; Gupta et al., 2003).
15
16 4.6.1. Oral
17
18 Table 4-11 provides a summary of the noncancer effects from repeated-dose oral studies
19 of MCLR toxicity in laboratory animals. The table includes all of the studies that used purified
20 microcystins as the test substance. As the table indicates, the toxicological database for effects
21 of microcystins after oral exposure is limited.
22
23 Liver Effects. One study of human exposure to drinking water before, during and after a
24 bloom of M. aeruginosa reported a significant increase in GGT levels during the bloom
25 compared with levels before the bloom (Falconer et al., 1983). The study population consisted
26 of all persons subjected to liver function tests in the area served by the affected drinking water
27 supply; as such, it is not representative of the general population. The liver is the primary target
28 organ when laboratory animals are exposed to high doses of MCLR. Oral exposure to single 500
29 Hg/kg doses of MCLR caused diffuse hemorrhage in the liver of mice and rats; more pronounced
30 liver damage occurred at higher doses (Ito et al., 1997a; Fawell et al., 1999). Young mice (5
31 weeks old) did not develop signs of hepatotoxicity at 500 ng/kg MCLR, while aged mice (32
32 weeks old) developed clear signs (Ito et al., 1997a). This difference may result in part from
33 differences in gastrointestinal absorption of microcystins, but cannot be entirely explained by
34 absorption differences, since similar age-dependent effects were reported after i.p. exposure
35 (Adams et al., 1985; Rao et al., 2005).
36
37 A single 28-day study of oral exposure to 50 or 150 ng/kg MCLR in drinking water
38 showed increased liver weight, slight to moderate liver lesions with hemorrhages and increased
39 ALP and LDH in rats exposed at 50 |j,g/kg-day (Heinze, 1999). A subchronic study in mice
40 using a similar dose range identified a LOAEL of 200 ng/kg (Fawell et al., 1999). At this dose,
41 mild liver lesions including chronic inflammation, hemosiderin deposits and single hepatocyte
42 degeneration were observed, as well as increased ALT and AST in male animals. The 40 ng/kg
43 dose was identified as a NOAEL. Mild hepatocyte injury was reported in mice given 80 or 100
44 gavage doses of 80 ng/kg each over 28 weeks, corresponding to time-weighted average doses of
45 33-41 |j,g/kg-day (Ito et al., 1997b). Based on the report, it appears that a limited postmortem
75 DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-11. Summary Noncancer Results in All Animal Studies of Oral Exposure to Purified Microcystin-LR
Species
Sex
Average
Daily Dose
(|ag/kg-day)
Exposure
NOAEL
Og/kg-
day)
LOAEL
(Hg/kg-
day)
Responses
Comments
Reference
Acute Exposure
Rat
Mouse
Mouse
Mouse
M/F
M/F
F
M
500, 1580,
5000
500, 1580,
5000
8000,
10000,
12500
0,500
Single
gavage
Single
gavage
Single
gavage
Single
gavage
ND
ND
ND
ND
5000*
1580*
12500*
500 (aged
mice
only)
Mortality; diffuse hepatic
hemorrhage at lower doses
Mortality; diffuse hepatic
hemorrhage at lower dose
Mortality (2/2); hypertrophic
hepatocytes, fibrosis in
centrilobular and midzonal
regions at lower doses
Centrilobular hepatic
hemorrhage and necrosis;
necrosis of intestinal mucosa
and duodenal damage
No untreated controls. Dose-
dependent increase in
hepatotoxicity
No untreated controls. Dose-
dependent increase in
hepatotoxicity
No untreated controls. 1-2
animals/dose group.
Effects observed in aged (32
week-old) mice; no effects on
liver or gastrointestinal tract in
young (5 week-old) mice
Fawell et al.,
1999
Fawell et al.,
1999
Yoshidaetal.,
1997
Ito et al.,
1997a
Short-Term Exposure
Rat
M
0, 50, 150
Drinking
water, 28
day
ND
50
Slight to moderate
degenerative and necrotic
hepatocytes with
hemorrhages; increased serum
enzymes (ALP and LDH)
Heinze, 1999
76
DRAFT - DO NOT CITE OR QUOTE
-------
Table 4-11. cont.
Species
Sex
Average
Daily Dose
(|ag/kg-day)
Exposure
NOAEL
(|4,g/kg-
day)
LOAEL
(Hg/kg-
day)
Responses
Comments
Reference
Subchronic Exposure
Mouse
M/F
0, 40, 200,
1000
Daily
gavage,
13 weeks
40
200
Minimal/slight chronic
inflammation with
hemosiderin deposits and
single hepatocyte
degeneration; increased serum
enzymes (ALT and AST)
Fawell el al.,
1999
Chronic Exposure
Mouse
Mouse
F
Not
given
0,3
Not
available
Drinking
water, 18
months
Gavage,
80 (o,g/kg,
80-100
times over
28 weeks
3
ND
ND
ND
No effects on survival, body
weight, hematology, serum
biochemistry, organs or
histopathology
Light injuries to hepatocytes
in the vicinity of the central
vein
Minor changes in ALP and
cholesterol not considered
lexicologically significant by
researchers
Only liver examined; only Ihree
conlrol animals; dosing
frequency unclear
Ueno el al.,
1999
Ilo el al.,
1997b
Developmental Toxicity
Mouse
F
0, 200, 600,
2000
Gavage,
GD 6-15
600
2000*
Maternal mortality (7/26) and
morbidity (2/26 humanely
sacrificed); reduced fetal body
weight, delayed skeletal
ossification
Aulhors defined 600 |ag/kg-day
as NOAEL bul did nol presenl
dala on reproductive or
developmental parameters lo
support identification of LOAEL
Fawell el al.,
1999
: Frank Effect Level (PEL)
77
DRAFT - DO NOT CITE OR QUOTE
-------
1 examination was conducted in this study, which was primarily aimed at evaluating
2 carcinogenicity. No liver or other toxicity was reported in female mice given approximately 3
3 |j,g/kg-day MCLR in drinking water for 18 months (Ueno et al., 1999).
4
5 Neurological Effects. The database contains scattered reports of neurological symptoms
6 after exposure to high doses of MCLR. Dialysis patients exposed to microcystins in dialysate
7 reported symptoms such as visual disturbance, blindness, vertigo, headache and muscle
8 weakness (Jochimsen et al., 1998). Clinical signs in mice and rats orally exposed to lethal doses
9 (about 5000 |J,g/kg) include hypoactivity and piloerection (Fawell et al., 1999).
10
11 Other Organs. Gastrointestinal effects (necrosis, duodenal damage) were observed in
12 aged mice exposed orally to single 500 ng/kg doses of MCLR (Ito et al., 1997a). Kidney effects
13 including eosinophilic materials in the Bowman's spaces were observed in two mice exposed to
14 a lethal dose of 12.5 mg/kg (Yoshida et al., 1997). Female mice exposed subchronically to 1000
15 Hg/kg had slight increases in hemoglobin concentration, erythrocyte count and packed cell
16 volume (Fawell et al., 1999). Milutinovic et al. (2002, 2003) briefly reported that kidney effects
17 are more pronounced than liver effects in rats chronically exposed to i.p. doses of MCLR and
18 MCYR (time weighted average dose, 5 ng/kg for 8 months). Details of the liver examinations
19 were not reported in this study, limiting the usefulness of these data.
20
21 Developmental Effects. A single oral study of developmental toxicity in mice reported
22 maternal toxicity, liver effects and deaths in some dams treated at the highest dose of MCLR
23 (2000 |J,g/kg during GD 6-15), along with reduced fetal body weight and delayed skeletal
24 ossification. No effects on reproductive or developmental parameters were observed in other
25 treatment groups, and 600 ng/kg was identified as a NOAEL for developmental toxicity (Fawell
26 et al., 1999). One study of developmental toxicity after i.p. injection of 32-238 ng/kg MCLR in
27 mice confirmed the lack of developmental or reproductive effects in the absence of maternal
28 toxicity (Chernoff et al., 2002). A study in which an extract ofM aeruginosa (estimated to
29 contain about 14 |J,g/L unspecified toxin) was administered in the drinking water to mice before
30 and during pregnancy revealed small brains in 7 of 73 pups from treated parents and none in
31 untreated controls (Falconer et al., 1988). The litter distribution of the affected pups was not
32 reported by the authors. It is not possible to attribute this effect to microcystin exposure, as the
33 extract may have contained other compounds.
34 In vitro studies suggest that MCLR can disrupt the cytoskeleton of embryonic cells,
35 causing cell detachment, retarding division or causing cytolysis (Sepulveda et al., 1992; Frangez
36 et al., 2003; Zuzek et al., 2003). MCLR effects on these and other cell types may be limited by
37 the degree of uptake. Frangez et al. (2003) showed that an intact zona pellucida prevented
38 effects in rabbit whole embryo cultures.
39
40 4.6.2. Inhalation
41
42 Very limited information is available on the toxicity of MCLR via inhalation exposure.
43 The available data indicate that short-term inhalation of a low concentration of MCLR can cause
44 local damage to the epithelial cells of the nasal cavity. A single study of inhalation exposure in
78 DRAFT - DO NOT CITE OR QUOTE
-------
1 mice revealed dose-dependent damage to the respiratory and olfactory epithelial cells of the
2 nasal cavity (Benson et al., 2005). Exposure occurred over 7 days at 260 ng/m3 for 30, 60 and
3 120 minutes per day (authors estimated the deposited doses as 3, 6 and 12 ng/kg). No effects on
4 the liver or other organs were observed.
5
6 Several limited lines of evidence suggest that high doses of MCLR via respiratory
7 exposure routes can lead to systemic uptake with subsequent liver effects. Systemic uptake of
8 MCLR by respiratory routes of exposure has been demonstrated in studies of acute, high-dose
9 exposure (Creasia, 1990; Fitzgeorge et al., 1994; Ito et al., 2001). Importantly, the LD50 for
10 MCLR given via either intranasal or intratracheal instillation is similar to that of MCLR given
11 via i.p. injection (Fitzgeorge et al., 1994; Ito et al., 2001). As with i.p. and oral exposure, liver
12 hemorrhage is the proximate cause of death in animals lethally dosed via intranasal or
13 intratracheal instillation (Fitzgeorge et al., 1994; Ito et al., 2001). Further evidence of systemic
14 effects comes from a brief abstract describing lethality in mice exposed via inhalation (nose
15 only) to MCLR aerosols. Creasia (1990) reported an LC50 for MCLR of 18 mg/m3 air for 10
16 minutes (authors estimated the deposited dose as 45 ng/kg), and indicated that histopathological
17 findings in deceased mice were similar to those reported after i.v. dosing. Ito et al. (2001)
18 suggested that MCLR could enter the bloodstream either via local damage to the nasal mucosa
19 leading to exposure of the nasal blood vessels, or through transport to the lung and absorption
20 into alveolar capillaries.
21
22 4.7. WEIGHT-OF-EVIDENCE EVALUATION AND CANCER
23 CHARACTERIZATION
24
25 4.7.1. Summary of Overall Weight of Evidence
26
27 Applying the Guidelines for Carcinogen Risk Assessment (U.S. EPA, 2005a), there is
28 inadequate information to assess carcinogenic potential of microcystins by the oral, dermal or
29 inhalation routes of exposure. One poorly-described long-term carcinogenicity bioassay found
30 no increase in neoplastic liver nodules after gavage administration in mice (Ito et al., 1997b).
31 The few available epidemiological studies that suggest a positive association between liver or
32 colorectal cancers and microcystins are limited by ecological study design, poor measures of
33 exposure, potential coexposure to other microbial or chemical contaminants and, in some cases,
34 failure to control for known liver and colorectal risk factors. MCLR has been shown to have a
35 promotional effect in two-stage rat liver bioassays using i.p. administration; however, the
36 relevance of this effect to environmental exposures is uncertain. Mechanistic information
37 provides some support for a possible promotional effect of MCLR.
38
39 4.7.2. Synthesis of Human, Animal and Other Supporting Evidence
40
41 Several human epidemiological studies have reported an association between
42 consumption of drinking water containing cyanobacteria and microcystins and liver or colon
43 cancer in certain areas of China (Yu et al., 1989 and Yu, 1989 as cited in Ueno et al., 1996; Zhou
44 et al., 2002). In all of these studies, the use of a surface drinking water supply was used as a
45 surrogate for exposure to microcystins. Individual exposure to microcystins was not estimated.
79 DRAFT - DO NOT CITE OR QUOTE
-------
1 Further, it is not clear whether these studies adequately controlled for confounding factors, such
2 as hepatitis infection or aflatoxin exposure.
3
4 Ito et al. (1997b) conducted the only study of oral carcinogenicity of a purified
5 microcystin. In this study, chronic gavage doses of MCLR over 28 weeks failed to induce
6 neoplastic nodules of the liver in mice. Limited information from two-stage, medium-term rat
7 liver bioassays where MCLR was administered i.p. suggest that MCLR can act as a promoter,
8 increasing the number and/or size of GST (placental form) positive foci in livers of rats
9 pretreated with an initiating agent (Nishiwaki-Matsushima et al., 1992; Ohta et al., 1994;
10 Sekijima et al., 1999; Hu et al., 2002). In one such study, MCLR alone showed no initiating
11 activity (Ohta et al., 1994). Ito et al. (1997b) observed an increase in the size of neoplastic liver
12 nodules in mice given 100 i.p. injections of MCLR without an initiating agent; however, the
13 numbers of treated and control animals were small.
14
15 Studies of cyanobacterial extract also suggest a possible promotional effect. In mice
16 given an extract of M. aeruginosa in drinking water, the mean area of aberrant crypt foci of the
17 colon was significantly increased, although the number of foci was not affected (Humpage et al.,
18 2000). Similarly, the total weight of skin tumors was increased in mice given an extract of
19 Microcystis in drinking water after topical DMBA pretreatment (Falconer and Buckley, 1989;
20 Falconer, 1991). It is not possible to determine whether the observed effects resulted from
21 exposure to microcystins or to other contaminants in the extracts.
22
23 Mechanistic data indicate that at low doses, MCLR may increase cell proliferation.
24 MCLR has been shown to increase the expression of the bcl-2 protein (that inhibits apoptosis)
25 and decrease the expression of the bax protein (that induces apoptosis) (Hu et al., 2002).
26 Further, MCLR upregulates the transcription factors c-fos and c-jun, leading to abnormal
27 proliferation (Zhao and Zhu, 2003). Gehringer (2004) reviewed the molecular mechanisms
28 leading to promotion by MCLR and the related tumor promoter, okadaic acid. Gehringer (2004)
29 reported that MCLR inhibits protein phosphatase PP2A, which regulates several MAPKs. The
30 MAPK cascade regulates transcription of genes required for cell proliferation, including c-jun
31 and c-fos. In addition, activation of the MAPK cascade has been postulated to inhibit apoptosis
32 and thus increase cell proliferation. Finally, Gehringer (2004) noted that MCLR has been
33 reported to increase phosphorylation of p53, which is involved in the regulation of the cell cycle
34 and apoptosis.
35 Genotoxicity studies of MCLR have given conflicting results, with negative findings in
36 Ames assays (Grabow et al., 1982; Ding et al., 1999) while positive results were observed with
37 human cell lines (Suzuki et al., 1998; Zhan et al., 2004). Evidence for MCLR-induced DNA
38 damage as measured by the comet assay has been called into question by the finding that
39 apoptosis can lead to false positive findings in this assay (Lankoff et al., 2004). There is some
40 evidence for a clastogenic effect of MCLR (Ding et al., 1999; Zhan et al., 2004).
41
80 DRAFT - DO NOT CITE OR QUOTE
-------
1 4.8. SUSCEPTIBLE POPULATIONS AND LIFE STAGES
2
3 Little information is available on potentially susceptible populations. Studies in
4 laboratory rodents suggest that the acute effects of MCLR may be more pronounced in adult or
5 aged animals than in juvenile animals (Adams et al., 1985; Ito et al., 1997a; Rao et al., 2005). In
6 these studies, young animals showed little or no effect at MCLR doses lethal to adult animals.
7 Age-dependent differences in toxicity were observed after both oral and i.p. exposure,
8 suggesting that differences in gastrointestinal uptake were not entirely responsible for the effect
9 of age. The relevance of these age-related differences to acute toxicity in humans is unclear.
10
11 Available information does not suggest any pronounced gender differences in response to
12 microcystins. Studies with algal extracts suggest the possibility that male mice may be more
13 sensitive than female mice to oral exposure to algal extracts (Falconer et al., 1988). However,
14 the relevance of this finding to human microcystin exposure is uncertain given the potential for
15 coexposure to other contaminants in algal extracts.
16
17 Because microcystins inhibit the action of protein phosphatases (PP1 and PP2A),
18 coexposure to other compounds that inhibit these enzymes (for example, okadaic acid) may
19 enhance the toxicological effects of microcystins.
81 DRAFT - DO NOT CITE OR QUOTE
-------
1 5. DOSE RESPONSE ASSESSMENTS
2
3
4 5.1. NARRATIVE DESCRIPTION OF THE EXTENT OF THE DATABASE
5
6 The available information on the toxicokinetic behavior of microcystins in humans or
7 animals after oral or inhalation exposure is limited to a single study of the organ distribution of
8 dihydro-MCLR. No other data are available on the absorption, distribution, metabolism or
9 elimination of microcystins via environmentally relevant exposure routes. Acute lethality data
10 show a significant difference in lethal doses via injected and oral routes of exposure, suggesting
11 that the toxicokinetic behavior of microcystins is an important determinant of health effects after
12 oral exposure; thus, the deficiencies in this category of data are significant.
13
14 The mode by which microcystins affects its primary target organ, the liver, is remarkably
15 well-studied. There are abundant mechanistic data ranging from target organ specificity down to
16 molecular targets. In vitro studies using human hepatocytes show effects similar to those in
17 animal hepatocytes, indicating that the mode of toxicological action is similar. Human
18 hepatocytes appear to be more susceptible to the action of MCLR than rat hepatocytes (Yea et
19 al., 2001; Batista et al., 2003). Further evidence for the relevance of this mode of toxicological
20 action to humans comes from reports of human exposure. Liver histopathology on humans
21 exposed to MCLR via dialysate showed effects similar to those seen in animals, although
22 intrahepatic hemorrhage was not observed (Azevedo et al., 2002).
23
24 The toxicological database for microcystins is almost exclusively limited to data on a
25 single congener, MCLR. Data on the other congeners is restricted to in vitro studies of protein
26 phosphatase inhibition, i.p. LD50 measures and a single, poorly-described toxicological
27 evaluation of MCLA in primates. The database on the oral toxicity of MCLR is adequate to
28 support the derivation of RfD values. Human data on the oral toxicity of MCLR are limited by
29 potential co-exposure to other cyanobacterial toxins and microorganisms. There are three
30 studies of acute oral exposure to MCLR in two laboratory animal species; however, none of
31 these identified a NOAEL, and the minimum dose tested was the same for all three. Further,
32 neither of the two experiments published in Fawell et al. (1999) nor the experiment by Yoshida
33 et al. (1997) used an untreated control group. One animal study evaluated the oral toxicity of
34 MCLR after short-term (<30 days) exposure and and one after subchronic (30-90 days)
35 exposure. The short-term study (Heinze, 1999) used a small number of animals (10/dose) and
36 did not identify a NOAEL, but was otherwise of good study quality. The subchronic study
37 (Fawell et al., 1999) used an adequate number of animals (30/dose) and identified both a
38 NOAEL and LOAEL. Two chronic exposure studies are available, but one of these (Ito et al.,
39 1997b) apparently conducted only a limited examination of the liver for toxicity, and the other
40 (Ueno et al., 1999) used a single dose and did not identify a LOAEL. A single, well-conducted
41 developmental toxicity study in the mouse is available; however, the results are presented only
42 briefly and without any supporting data in the publication (Fawell et al., 1999). The noncancer
43 database is missing a chronic toxicity study in a second species, as well as a multigeneration
44 reproductive toxicity study and neurotoxicity study.
45
82 DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
The database on the inhalation toxicity of MCLR is inadequate for the derivation of any
RfC. There are no human data on the inhalation of MCLR. There is a single well-reported
animal study addressing inhalation exposure to MCLR for 7 days (Benson et al., 2005). This
study used only one exposure concentration with daily exposure for 30, 60 or 120 minutes and,
as such, is not adequate for short-term RfC derivation.
The available data on carcinogenicity are inadequate for carcinogenicity assessment.
There is no well-conducted long-term carcinogenicity bioassay for microcystin. Several studies
using an initiation-promotion protocol are available, as are limited mechanistic data suggesting a
potential promoting effect of microcystins.
5.2. ORAL REFERENCE DOSE
Data considered in deriving oral reference dose for each exposure duration are
summarized in the following exposure-response array (Figure 5-1) as well as in Table 5-1 below.
Due to the limited toxicological database for microcystins, both the table and the figure include
all studies in the published literature that examined the oral toxicity of purified MCLR in
laboratory animals, with one exception. The publication by Ito et al. (1997b) did not provide the
dosing frequency or information to estimate an average daily dose; thus, this study is not
included in the table or figure.
100000
10000
100
10
1
Ra
1
1
ts
Mi
1
ce
• FEL
• LOAEL
ONOAEL
Fawell etal.
(1999)
Fawell etal.
(1999)
Ac
The vertical lines
represent the range
^ of dos^s t^st^d in a
given study.
Mice
* °
Aged ^
i'licc Mice
Rats Mice
o
Mice
Yoshidaet Ito etal. Heinze Fawell etal. Ueno et al. Fawell etal.
al. (1997) (1997a) (1999) (1999) (1999) (1999)
Lite Short-term Longer-term Chronic Develop-
mental
22 Figure 5-1. Exposure Response Array for Oral Exposure: All Studies of Purified Microcystin-
23 LR
83
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-1. Available Dose-Response Information for Oral Exposure to Purified MCLR
Species
Sex
Average Daily
Dose
Og/kg-day)
Exposure
NOAEL
(Hg/kg-
day)
LOAELorFEL
Og/kg-day)
Reference
Acute Exposure
Rat
Mouse
Mouse
Mouse
M/F
M/F
F
M
500, 1580, 5000
500, 1580, 5000
8000, 10000,
12500
0,500
Single gavage
Single gavage
Single gavage
Single gavage
ND
ND
ND
ND
5000*
1580*
12500*
500 (aged mice
only)
Fawelletal., 1999
Fawelletal., 1999
Yoshidaetal.,
1997
Itoetal., 1997a
Short-Term Exposure
Rat
M
0, 50, 150
Drinking
water, 28 d
ND
50
Heinze, 1999
Subchronic Exposure
Mouse
M/F
0, 40, 200, 1000
Daily gavage,
13 weeks
40
200
Fawelletal., 1999
Chronic Exposure
Mouse
F
0,3
Drinking
water, 18
months
o
J
ND
Uenoetal., 1999
Developmental Toxicity
Mouse
F
0, 200, 600, 2000
Gavage,
GD 6-15
600
2000*
Fawell et al., 1999
2
3
4
5
6
7
8
9
10
11
12
13
: Frank Effect Level (PEL)
5.2.1. Acute Oral RfD
The acute oral data for MCLR are inadequate for the derivation of an acute RfD. There
are four studies of acute exposure to MCLR (Yoshida et al., 1997; Ito et al., 1997a; Fawell et al.,
1999); however, none of the available studies identified a NOAEL. Yoshida et al. (1997)
conducted a LDso determination using a small number of mice (5) treated with a single gavage
dose, and no untreated controls. Histopathology of surviving mice (one at 8 mg/kg and two at 10
mg/kg) showed liver lesions (hypertrophic hepatocytes with centrilobular and midzonal fibrosis).
The high dose in this study was an PEL based on deaths of the 2 treated animals. Fawell et al.
84
DRAFT - DO NOT CITE OR QUOTE
-------
1 (1999) administered single gavage doses of 500, 1580 or 5000 ng/kg MCLR to groups of rats
2 and mice (10/dose/species). There were no untreated control groups in this study. In both
3 species, diffuse hepatic hemorrhage was observed at the low dose (500 ng/kg); however, in the
4 absence of untreated controls for comparison, this dose cannot be identified as a LOAEL. Ito et
5 al. (1997a) administered single gavage doses of 500 ng/kg to young (5 weeks old) and aged (32
6 weeks old) mice. Centrilobular hemorrhage and hepatocyte necrosis, as well as gastrointestinal
7 lesions, were observed in the aged mice, while no effect was observed in the young mice. This
8 study identifies a freestanding LOAEL. The absence of an acute study of adequate quality
9 precludes derivation of an acute oral RfD.
10
11 5.2.2. Short-Term Oral RfD
12
13 5.2.2.1. Choice of Principal Study and Critical Effect
14
15 A single short-term study and a single developmental toxicity study of orally-
16 administered MCLR are available. Heinze (1999) evaluated the effects of MCLR in drinking
17 water in 11-week-old male hybrid rats. Groups of 10 rats (5 of each sex) were given
18 approximate doses of 0, 50 or 150 ng/kg body weight for 28 days. Serum biochemistry showed
19 significantly increased mean levels of ALP and LDH in both treatment groups (84 and 100%
20 increase in LDH, 34 and 33% increase in ALP in low and high doses, respectively). A dose-
21 dependent increase in relative liver weights was observed at both dose levels (17 and 26% at the
22 low and high doses, respectively). Liver lesions were observed in both treatment groups, but the
23 severity of the damage was increased in the 150 ng/kg dose group. Moderate to severe
24 degenerative and necrotic hepatocytes with hemorrhage was observed in 0 of 10 controls, 6 of 10
25 low-dose and 9 of 10 high-dose rats.
26
27 Fawell et al. (1999) evaluated the developmental toxicity of MCLR administered via
28 gavage to mice at doses of 0, 200, 600 and 2000 ng/kg on GDs 6-15. Seven of 26 dams
29 receiving 2000 ng/kg died and two others were sacrificed prematurely due to morbidity. At this
30 dose, fetal body weight was significantly lower than controls and there was delayed skeletal
31 ossification; these effects may have been associated with maternal toxicity. Data on
32 reproductive and developmental parameters were not provided in the reference; thus, a LOAEL
33 for developmental toxicity could not be determined. This study identified a NOAEL of 600
34 |j,g/kg-day for both developmental and maternal effects. The high dose of 2000 ng/kg was an
35 FEL based on maternal mortality. The study by Heinze (1999) identified a lower LOAEL and
36 more sensitive effect (hepatotoxicity) than the developmental toxicity study did; thus, this study
37 was chosen as the basis for the short-term RfD.
38
39 5.2.2.2. Methods of Analysis
40
41 Liver toxicity observed by Heinze (1999) included liver lesions, serum enzyme changes,
42 and changes in relative liver weight. All three of these endpoints were considered for
43 determining the point of departure for RfD derivation.
44
85 DRAFT - DO NOT CITE OR QUOTE
-------
1 All quantal models in U.S. EPA's Benchmark Dose (BMD) software (version 1.3.2) were
2 fit to the incidence data for liver lesions in rats (males and females combined) exposed to MCLR
3 in the drinking water for 28 days (Heinze, 1999). The incidence data for liver lesions are
4 reported in Table 4-4 (Section 4.2.1.2.1). Of the liver lesions observed, the category of
5 degenerative and necrotic hepatocytes with hemorrhage showed a strong dose-related trend with
6 greater incidence and severity with higher dose, and no control animals were affected. As the
7 table shows, at 50 ng/kg, 4/10 rats had slight lesions and 6/10 had lesions of moderate severity.
8 At 150 ng/kg, 6/10 rats had moderate lesions and 3/10 had intensive damage. For BMD
9 modeling, the moderate and severe lesion categories were collapsed into one. The data modeled
10 are shown in Table 5-2.
11
Table 5-2. Incidence of Liver Lesions Used for BMD Modeling (Heinze, 1999)
Lesion incidence
0 ng/kg-day
0/10
50 |j,g/kg-day
6/10
150 |j,g/kg-day
9/10
12
13
14 In accordance with the U.S. EPA (2000c) BMD methodology, the default benchmark
15 response (BMR) of 10% increase in extra risk was used. The high response rate (60%) at the
16 lowest dose with a positive response precludes the use of a lower BMR for this analysis. Models
17 were run using the default restrictions on parameters built into the BMDS. The modeling results
18 are shown in Table 5-3. Adequate fits were achieved with all models, except the quantal
19 quadratic. While the gamma, multistage, quantal linear and Weibull all converged on the same
20 model, the log probit model gave the best fit, as assessed by AIC. Figure 5-2 shows the fit of the
21 log probit model to the data. Appendix A contains the full model outputs. The BMD and
22 BMDL estimated by the log probit model for the liver lesion data are 11.0 and 6.4 |j,g/kg-day,
23 respectively.
24
25 The linear model for continuous data was fit to the increased LDH and ALP levels
26 reported in Heinze (1999). These data are shown in Table 4-3 (Section 4.2.1.2.1). With only
27 three observations in each of these datasets, there were not enough data points to use the
28 remaining models (polynomial, power, or Hill), which each have more than three parameters.4
29
30 The linear model did not provide adequate fit to either dataset as measured by goodness-
31 of-fit criteria (see Appendix A for model outputs). The linear model was also fit to the relative
32 liver weight changes (also three observations) reported in Heinze (1999). These data are shown
33 in Table 4-3 (Section 4.2.1.2.1). In accordance with the U.S. EPA methodology, the default
34 BMR of one standard deviation change from the control mean was used, and the polynomial
35 coefficients were restricted to be positive. The linear model provided an adequate fit to the data
36 (see Appendix A for model output). The BMD and BMDL estimated by the linear model for the
37 relative liver weight data are 85 and 58 |j,g/kg-day, respectively.
4 The number of parameters describing the shape of the dose-response curve cannot exceed the number of dose
groups (U.S. EPA, 2000c).
86 DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-3. BMD Modeling Results for Heinze (1999) Liver Lesion Data
Model
Log-probit (slope >1)
Gamma (power >1)
Multistage (degree=l)*
Quantal Linear
Weibull (power >1)
Log-logistic (slope >1)
Quantal Quadratic
Logistic
Probit
Degrees
of
Freedom
2
2
2
2
2
1
2
1
1
x2
0.01
0.09
0.09
0.09
0.09
0.00
5.77
3.43
3.50
% Goodness
of Fit
p-Va\ue
0.99
0.96
0.96
0.96
0.96
1
0.06
0.06
0.06
AIC
21.97
22.05
22.05
22.05
22.05
23.96
26.17
28.31
28.43
BMD
Og/kg-
day)
11.04
6.31
6.31
6.31
6.31
10.14
24.81
19.43
19.69
BMDL
(Hg/kg-day)
6.38
3.92
3.92
3.92
3.92
1.24
19.01
11.40
12.31
* Degree of polynomial initially set to (n-1) where n= number of dose groups including control;
model selected is lowest degree model providing adequate fit. Betas restricted to >0.
87
DRAFT - DO NOT CITE OR QUOTE
-------
1
3
5
7
9
11
13
15
17
19
21
23
25
27
29
31
33
35
37
39
41
43
44
45
46
47
48
49
50
0.
Probit Model with 0.95 Confidence Level
Probit
BMD
BM
4 6 8 10 12 14 16
dose
15:0607/142005
Figure 5-2. Probit Model Fit to Liver Lesion Incidence Data from Heinze (1999)
88
DRAFT - DO NOT CITE OR QUOTE
-------
1 5.2.2.3. RfD Derivation
2
3 The BMDL of 6 |ig/kg-day from the Heinze (1999) data is the lowest BMDL among the
4 modeled datasets and was used as the point of departure (POD) for the short-term RfD. Dividing
5 the BMDL of 6 |j,g/kg-day by a composite uncertainty factor (UF) of 1000 results in a short-term
6 RfD for MCLR of 6xlO'6 mg/kg-day.
7
8 Short-term RfD = BMDL-UF
9 =6 ng/kg-day - 1000
10 = 0.000006 mg/kg-day or 6x106 mg/kg-day
11
12 The composite UF of 1000 includes a factor of 10 for interspecies extrapolation, a factor
13 of 10 to account for interindividual variability in the human population and a factor of 10 for
14 database limitations, as follows.
15
16 • A default 10-fold UF for intraspecies differences was used to account for potentially
17 susceptible individuals in the human population. There is insufficient information on the
18 toxicity of microcystins in exposed humans. Cases of human poisoning have been
19 attributed to ingestion of water containing microcystin-producing cyanobacteria, but no
20 dose-response information is available. There is no information on the degree to which
21 humans of varying gender, age, health status or genetic makeup might vary in the
22 disposition of, or response to, ingested microcystins. There are some data to suggest that
23 adult or aged rodents may be more susceptible than young rodents to the acute toxicity of
24 MCLR (Ito et al., 1997a). Further, studies with algal extracts suggest the possibility that
25 male mice may be more sensitive than female mice to oral exposure to algal extracts
26 (Falconer et al., 1988); however, the relevance of this finding to human microcystin
27 exposure is unclear.
28 • An interspecies UF of 10 was used to account for differences in response between
29 laboratory rodents and humans. No information is available on the toxicity of purified
30 microcystins in humans, and data on toxicokinetic differences between animals and
31 humans in the disposition of ingested microcystins are not available. Limited data from
32 in vitro studies suggests that human hepatocytes may be more susceptible to the effects of
33 MCLR than rat hepatocytes (Yea et al., 2001; Batista et al., 2003), supporting the use of a
34 full 10-fold UF.
35 • A 10-fold UF is used to account for deficiencies in the database. Database deficiencies
36 include the lack of a detailed developmental toxicity study, a neurotoxicity study, a
37 multi-generation reproductive toxicity study and supporting information on systemic
38 toxicity in a second species.
89 DRAFT - DO NOT CITE OR QUOTE
-------
1 5.2.3. Subchronic Oral RfD
2
3 5.2.3.1. Choice of Principal Study and Critical Effect
4
5 A single subchronic oral toxicity study is available for MCLR. Fawell et al. (1999)
6 identified both a NOAEL (40 |j,g/kg-day) and a LOAEL (200 |j,g/kg-day for slight liver injury)
7 after subchronic exposure of male and female mice. Fawell et al. (1999) administered daily oral
8 gavage doses of 0, 40, 200 or 1000 |j,g MCLR per kg body weight to groups of 15 male and 15
9 female mice for 13 weeks. Histopathological changes in the liver and serum enzyme changes
10 were reported in the mid- and high-dose groups. Both the histopathology and the serum
11 enzymes showed dose-dependent changes. The authors considered the liver changes in the 200
12 |j,g/kg-day dose group to represent a minimal effect.
13
14 The NOAEL from this 90-day study (40 |j,g/kg-day) is only slightly lower than the
15 LOAEL identified in the 28-day study above (50 |j,g/kg-day). Further, Heinze (1999) observed
16 more severe effects in rats exposed to 50 |j,g/kg-day for 28 days in drinking water than Fawell et
17 al. (1999) observed in mice exposed via gavage doses of 200 |j,g/kg-day for 90 days. The reason
18 for this difference in response is not clear. Both studies used a commercially-produced test
19 material from the same manufacturer. In the drinking water study, MCLR was dissolved in
20 ethanol and diluted to a stock solution that was subsequently used to prepare drinking water. It
21 is possible that the small intake of ethanol may have potentiated the hepatic effects of MCLR;
22 however, there are no data to determine whether this is likely or not. In the gavage study, test
23 solutions were prepared with distilled water and the concentration was confirmed by HPLC with
24 UV detection (Fawell et al., 1999). The accuracy of dosing in the gavage study was likely to be
25 greater than in the drinking water study. The authors of the drinking water study indicated that
26 the MCLR solution was prepared daily, and water consumption was measured daily. Between 3
27 and 7% of the water solution administered over the 28 days was not consumed, and the dose
28 estimates were not corrected for this loss (Heinze, 1999). This loss of administered dose would
29 lead to a small overestimate of the LOAEL in the drinking water study, leading to a further
30 discrepancy in the results of the two studies.
31
32 The drinking water study used smaller group sizes (10 males/dose) than the gavage study
33 (15/sex/dose or 30/dose). However, the incidence of liver lesions (with necrosis and
34 hemorrhage) in the drinking water study increased from 0% to 100% (including slight, moderate,
35 and intensive lesions) between the control and low-dose group, and there was a dose-related
36 change in the severity of the lesions, leaving little question that the effect was treatment-related.
37
38 These studies appear to contradict evidence from acute parenteral studies indicating that
39 mice are more sensitive to the acute effects of MCLR. Typically, mice die within a few hours of
40 a lethal injected dose, while rats may survive 24-48 hours. Species-specific differences in oral
41 absorption of MCLR do not appear to account for the discrepancy between these studies; in an
42 acute study of orally-administered MCLR using both mice and rats, mortality occurred at a lower
43 dose in mice (1580 ng/kg) than in rats (5000 ng/kg; Fawell et al., 1999). It is possible that the
44 more mild effects in the mice in the subchronic study resulted from an adaptive response to
45 MCLR exposure. The longer exposure duration may have allowed for liver regeneration and
90 DRAFT - DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
repair that was not possible in the shorter-term study; however, there is no information to support
this hypothesis. In fact, Ito et al. (1997) reported that light injuries to hepatocytes were still
evident in five of seven mice 2 months after treatment with MCLR had ceased.
5.2.3.2 Methods of Analysis
The data from both Heinze (1999) and Fawell et al. (1999) were considered for
identifying the point of departure for the subchronic RfD derivation. Results of the modeling for
Heinze (1999) are reported in Section 5.2.2.2. Among the liver lesions reported by Fawell et al.
(1999), only chronic inflammation showed evidence of a dose-response relationship.
Consequently, all quantal models in U.S. EPA's Benchmark Dose Software (BMDS) were fit to
the incidence data for chronic liver inflammation in male and female mice reported by Fawell et
al. (1999). The incidence data are reported in Table 4-6. In accordance with the U.S. EPA
methodology, the default BMR of 10% increase in extra risk was used. Models were run using
the default restrictions on parameters built into the BMDS. Adequate fits were achieved with all
models. For both male and female mice, the probit model provided the best fit, as assessed by
AIC. Table 5-4 gives the results for the best fit models; Appendix A contains the full outputs for
all models.
Table 5-4. BMD Modeling Results for Fawell et al. (1999) Chronic Liver Inflammation Data
Male:
Probit Model
Female:
Probit Model
Degrees
of
Freedom
2
2
X2
0.21
0.94
Goodne
ss of Fit
p-Va\ue
0.90
0.63
AIC
40.75
72.84
BMD
Og/kg-
day)
107.59
86.34
BMDL
Og/kg-
day)
66.45
56.92
20
21
22
23
24
25
26
27
28
29
30
Fawell et al. (1999) also reported significant increases in ALP, AST, and ALT in high-
dose animals. These data are shown in Table 4-5 (Section 4.2.1.3.1). Of these, only the increase
in ALT in male mice showed a dose-response trend amenable to modeling. All continuous
models but the Hill model5 were fit to the ALT data for male mice reported in Fawell et al.
(1999). In accordance with the U.S. EPA methodology, the default BMR of one standard
deviation change from the control mean was used, and the polynomial coefficients were
restricted to be positive. Only the linear model with a nonhomogenous variance provided an
adequate fit to the data. Table 5-5 gives the results from the linear model (see Appendix A for
model output). The BMD and BMDL estimated by the linear model (nonhomogenous variance)
for the ALT increases in male mice are 82 and 58 |j,g/kg-day, respectively.
5 There were too few dose groups to apply the Hill model.
91
DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-5. BMD Modeling Results for Fawell et al. (1999) ALT Data in Male Mice
Linear
model, non-
homogenous
variance
Degrees
of
Freedom
2
Goodness
of Fit
/>-Value
0.10
AIC
447.68
BMD
Og/kg-
day)
81.84
BMDL
(Hg/kg-day)
58.37
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
The BMDL from the 28-day drinking water study (6 |j,g/kg-day) is approximately an
order of magnitude lower than any of the BMDL values from the 90-day gavage study (57-66
Hg/kg-day). Details of the BMD modeling and analysis of the data from Heinze (1999) are
provided above in Section 5.2.2.2, Method of Analysis, under Short-Term Oral RfD.
5.2.3.3. RfD Derivation
The BMDL of 6 ng/kg-day from the Heinze (1999) data is used as the POD for the
subchronic RfD. A composite UF of 1000 is used to derive the subchronic RfD, including a
factor of 10 for interspecies extrapolation, a factor of 10 for interindividual variability, and a
factor of 10 for database limitations (see Section 5.2.2.3 above for details). Although the BMDL
comes from a 28-day study, a UF for exposure duration is not proposed, based on the lower
toxicity observed in the 90-day gavage study conducted by Fawell et al. (1999). The subchronic
RfD is, therefore, set equal to the short-term RfD of 0.006 |j,g/kg-day or 6xlO"6 mg/kg-day.
Subchronic RfD
5.2.4. Chronic Oral RfD
BMDL -H UF
6 ng/kg-day - 1000
0.006 ng/kg-day or 6x10 ° mg/kg-day
-6
5.2.4.1. Choice of Principal Study and Critical Effect
Two chronic studies of the oral toxicity of MCLR were identified. Ito et al. (1997b)
conducted a chronic gavage study in mice with 80 to 200 doses (80 |j,g/kg-day) given over 28
weeks; however, the control group was very small (3 animals) and the postmortem examination
was apparently limited to the liver. This study was not given further consideration for RfD
development given these study quality concerns.
Ueno et al. (1999) evaluated the toxicity of MCLR in female mice chronically exposed
via drinking water 7 days/week. The authors conducted a comprehensive postmortem
examination. No treatment-related effects were identified, and the authors observed no
92
DRAFT - DO NOT CITE OR QUOTE
-------
1 difference in the incidence of liver histopathology between treated and control mice. It is
2 important to note that immunohistochemistry of the liver revealed no accumulation of MCLR.
3 This study identified a free-standing NOAEL of approximately 3 |j,g/kg-day in female mice.
4
5 Although Ueno et al. (1999) used only a single dose level and identified a freestanding
6 NOAEL, it was chosen for RfD derivation because it was a well-conducted chronic study using a
7 relevant exposure route (drinking water). The BMDL of 6 |j,g/kg-day from modeling (see
8 Section 5.2.2.2) of Heinze (1999) compares favorably with the free-standing NOAEL of 3
9 |j,g/kg-day reported by Ueno et al. (1996), providing support for the use of the NOAEL from
10 Ueno etal. (1999).
11
12 5.2.4.2. RfD Derivation
13
14 As noted above, the NOAEL of 3 |j,g/kg-day from the study by Ueno et al. (1999) is used
15 as the POD for the chronic RfD. Because this study used a single dose, it was not possible to use
16 BMD modeling to identify the POD. A composite UF of 1000 is used to derive the chronic RfD,
17 including a factor of 10 for interspecies extrapolation, a factor of 10 for interindividual
18 variability and a factor of 10 for database limitations (see Section 5.2.2.3 above for details).
19 Dividing the NOAEL of 3 |j,g/kg-day by a composite UF of 1000 results in a chronic RfD for
20 MCLR of 3xlO"6 mg/kg-day.
21
22 Chronic RfD = NOAEL-UF
23 =3 ng/kg-day - 1000
24 = 0.003 ng/kg-day or 3x106 mg/kg-day
25
26 In 1999, the WHO published a provisional Tolerable Daily Intake (TDI) for MCLR based
27 on the subchronic gavage study later published by Fawell et al. (1999). The WHO used the
28 NOAEL of 40 ng/kg-day with a composite UF of 1000 to derive a TDI of 0.04 ng/kg-day or
29 4xlO"5 mg/kg-day. The composite uncertainty factor included UFs of 10-fold each for
30 interindividual variability, interspecies extrapolation, and database deficiencies (WHO
31 specifically cited the lack of chronic toxicity and carcinogenicity studies). WHO (1999) did not
32 evaluate either Ueno et al. (1999) or Heinze (1999), which may not have been published at the
33 time.
34
35 Table 5-6 provides a summary of the RfD values derived for MCLR in this report.
36
37 5.3. INHALATION REFERENCE CONCENTRATION
38
39 The available data do not provide adequate information for the derivation of inhalation
40 RfCs for MCLR. Two acute inhalation studies were identified in the literature. In a poorly
41 described study, Fitzgeorge et al. (1994) conducted a single experiment with mice (number
42 unspecified) inhaling a fine aerosol (particle size 3-5 |jm) with 50 |jg/L MCLR for an unspecified
43 duration of time. There were apparently no deaths, clinical signs of toxicity or histopathological
44 changes; however, the authors gave few details of study design and findings. A brief abstract
45 describes a study of acute microcystin exposure via inhalation (Creasia, 1990). The LCso for
93 DRAFT - DO NOT CITE OR QUOTE
-------
Table 5-6. Summary of Reference Dose Values
Acute
Short-term
Sub chronic
Chronic
RfD (mg/kg-day)
NA
6xlO'6
6xlO'6
3 x ID'6
Critical Effect
Hepatotoxicity
Hepatotoxicity
No effects observed
Principal Study
Heinze, 1999
Heinze, 1999
Uenoetal., 1999
94
DRAFT - DO NOT CITE OR QUOTE
-------
1 mice exposed to a MCLR aerosol (nose only) for 10 minutes was reported to be 18 (ig/L (mg/m3)
2 air with a 95% confidence interval of 15.0-22.0 |j,g/L (mg/m3). The authors reported that
3 histological lesions in mice killed by aerosol exposure were similar to those in mice dosed
4 intravenously with MCLR. Neither of these studies provides adequate basis for an acute RfC.
5
6 Only one well-conducted study of inhalation exposure to MCLR was identified. Benson
7 et al. (2005) exposed groups of six male BALB/c mice to monodisperse submicron aerosols of
8 MCLR via nose-only inhalation for 30, 60 or 120 minutes each day for 7 consecutive days. The
9 concentration of MCLR was 260-265 |J,g/m3. Histopathological examination revealed treatment-
10 related lesions in the nasal cavity only. The incidence and severity of nasal lesions increased
11 with daily exposure duration. This study used only one exposure concentration, and as such, the
12 data are of limited utility for RfC derivation. Further, extrapolation of the effects from this study
13 for the purpose of deriving a short-term RfC would be associated with substantial uncertainty
14 given the brief exposure time (30-120 minutes/day) and duration (7 days). There are no
15 subchronic or chronic animal studies evaluating the inhalation route of exposure.
16
17 Route-to-route extrapolation is not considered appropriate for microcystins based on
18 current data. Limited available information indicates that inhalation exposure to microcystins
19 may cause point-of-entry effects (Benson et al., 2005), while oral exposure leads to
20 hepatotoxicity. Data from intratracheal and intranasal instillation studies show hepatic effects
21 after exposure via these routes; however, the relevance of this information to inhalation
22 exposures is uncertain.
23
24 5.4. CANCER ASSESSMENT
25
26 No dose-response or other information is available regarding the carcinogenicity of pure
27 microcystins.
95 DRAFT - DO NOT CITE OR QUOTE
-------
1 6. MAJOR CONCLUSIONS IN THE CHARACTERIZATIONS OF HAZARD
2 AND DOSE-RESPONSE
3
4
5 6.1. HUMAN HAZARD POTENTIAL
6
7 Microcystins are a group of naturally occurring hepatotoxins produced by freshwater
8 cyanobacteria. No studies of the absorption, distribution, metabolism or elimination of
9 microcystins LR, YR, RR or LA in vivo have been conducted. Acute lethality data suggest a
10 significant difference in the lethal dose after oral exposure to MCLR when compared with
11 injection routes of exposure, suggesting low absorption of orally-administered MCLR. The
12 cellular uptake and distribution of MCLR has been extensively studied, showing preferential
13 uptake of MCLR by hepatocytes due to the presence of a bile acid transporter.
14
15 Data on human exposures to microcystin-producing cyanobacteria have shown
16 gastrointestinal and dermal effects; however, it is not clear whether these are effects of
17 microcystins, other endotoxins, or the microorganisms themselves. The preponderance of the
18 available toxicological studies in animals employed the MCLR congener. Both oral and
19 parenteral exposure studies in laboratory animals point to the liver as the primary target organ of
20 MCLR, and parenteral exposure studies suggest a steep dose-response curve for the hepatotoxic
21 effects. The toxicological database is limited to only a few studies using oral exposure to
22 purified microcystins; all of these used the MCLR congener, and most were conducted in mice.
23 The database includes four studies of acute exposure, one 28-day and one 90-day study, and two
24 chronic studies. In all of the studies where a toxicological effect was observed, the primary
25 target organ was the liver (one study also identified gastrointestinal lesions). The liver effects
26 included hepatocyte degeneration and necrosis, inflammation, fibrosis, hypertrophy and
27 hemorrhage. Humans exposed to microcystins via dialysate suffered acute liver failure and, in
28 many cases, death. Liver biopsies conducted on the decedents showed hepatocyte necrosis and
29 apoptosis, but no intrahepatic hemorrhage.
30
31 A single well-described study of short-term (7 days) inhalation exposure identified the
32 upper respiratory tract as a target organ. Damage to the respiratory and olfactory epithelial cells
33 of the nasal cavity was observed in mice in this study. Studies using intratracheal or intranasal
34 instillation have shown systemic effects, including liver toxicity, after these exposures.
35
36 A single oral developmental toxicity study with inadequate data reporting indicated that
37 developmental effects in the absence of maternal toxicity are not likely, and that developmental
38 effects, if any, would occur at much higher doses than liver effects.
39
40 Mechanistic studies indicate that the inhibition of protein phosphatases 1 and 2A (a well-
41 established molecular effect of microcystin exposure) and/or oxidative stress play a role in the
42 hepatotoxicity of MCLR. Cellular effects in hepatocytes include membrane blebbing, cell
43 rounding and dissociation, and apoptosis. These cellular effects lead to alterations in the liver
44 structure and function as well as intrahepatic hemorrhage.
45
96 DRAFT - DO NOT CITE OR QUOTE
-------
1 6.2. DOSE RESPONSE
2
3 The available oral data were sufficient for derivation of short-term, subchronic and
4 chronic oral RfDs for MCLR. Based on a BMDL of 6 |j,g/kg-day for hepatotoxicity in rats
5 exposed to MCLR in drinking water for 28 days (Heinze, 1999), an RfD of 0.006 |j,g/kg-day
6 (6xlO"6 mg/kg-day) was derived for short-term and subchronic exposure durations. A UF of
7 1000 was used to derive the RfD. The UF comprises component factors of 10 for interspecies
8 extrapolation, 10 for interindividual variability and 10 for database deficiencies. The subchronic
9 RfD did not include an additional UF for extrapolating from a 28-day study because a subchronic
10 (90-day) gavage study identified a higher NOAEL. A chronic RfD of 0.003 |ig/kg-day (3xlO"6
11 mg/kg-day) was derived from a free-standing NOAEL of 3 |j,g/kg-day in female mice chronically
12 exposed via drinking water. A composite UF of 1000 was used, with factors of 10 each for
13 interindividual variability, interspecies extrapolation and database deficiencies. Inhalation RfCs
14 were not derived since there were no studies of adequate quality for this purpose. There is
15 inadequate evidence to evaluate the carcinogenicity of microcystins LR, RR, YR and LA.
97 DRAFT - DO NOT CITE OR QUOTE
-------
1 7. REFERENCES
2
3
4 Adams, W.H., R.D. Stoner, D.G. Adams et al. 1985. Pathophysiologic effects of a toxic peptide
5 fromMicrocystis aeruginosa. Toxicon. 23(3):441-447.
6 Adams, W.H., J.P. Stone, B. Sylvester et al. 1988. Pathophysiology of cyanoginosin-LR, in
7 vivo and in vitro studies. Toxicol. Appl. Pharmacol. 96(2):248-257.
8 Ashworth, C.T. and M.F. Mason. 1945. Observations on the pathological changes produced by
9 a toxic substance present in blue-green algae (Microcystis aeruginosa). Am. J. Pathol.
10 22:369-383.
11 Aune, T. and K. Berg. 1986. Use of freshly prepared rat hepatocytes to study toxicity of blooms
12 of the blue-green algae Microcystis aeruginosa and cyanotoxin Oscillatoria agardhii. J.
13 Toxicol. Environ. Health. 19(3):325-336.
14 Azevedo, S.M.F.O., W.W. Carmichael, E.M. Jochimsen et al. 2002. Human intoxication by
15 microcystins during renal dialysis treatment in Caruaru, Brazil. Toxicology. 181-182:441-446.
16 Aziz, K.M.S. 1974. Diarrhea toxin obtained from a waterbloom-producing species, Microcystis
17 aeruginosaKutzing. Science. 183:1206-1207.
18 Bagu, J.R., B.D. Sykes, M.M. Craig and C.F.B. Holmes. 1997. A molecular basis for different
19 interactions of marine toxins with protein phosphatase-1. Molecular models for bound
20 motuporin, microcystins, okadaic acid, and calyculin A. J. Biol. Chem. 272(8):5087-5097.
21 Batista, T., G. de Sousa, J.S. Suput et al. 2003. Microcystin-LR causes the collapse of actin
22 filaments in primary human hepatocytes. Aquat. Toxicol. 65(1):85-91.
23 Battle, T., C. Touchard, HJ. Moulsdale et al. 1997. New cell substrates for in vitro evaluation
24 of microcystin hepatocytoxicity. Toxicol. In Vitro. ll(5):557-567.
25 Beasley, V.R., R.A. Lovell, K.R. Holmes et al. 2000. Microcystin-LR decreases hepatic and
26 renal perfusion and causes circulatory shock, severe hyperglycemia, and terminal hyperkalemia
27 in intravascularly dosed swine. J. Toxicol. Environ. Health-Part A. 61(4):281-303.
28 Becchetti, A., B. Malik, G. Yue et al. 2002. Phosphatase inhibitors increase the open
29 probability of ENaC in A6 cells. Am. J. Physiol. Renal Physiol. 283(5):F1030-F1045.
30 Benson, J.M., J. A. Hutt, K. Rein et al. 2005. The toxicity of microcystin LR in mice following
31 7 days of inhalation exposure. Toxicon. 45(6):691-698.
32 Berg, K. and T. Aune. 1987. Freshly prepared rat hepatocytes used in screening the toxicity of
33 blue-green algal blooms. J. Toxicol. Environ. Health. 20(1/2): 187-197.
98 DRAFT - DO NOT CITE OR QUOTE
-------
1 Berg, K., J. Wyman, W.W. Carmichael and A.S. Dabholkar. 1988. Isolated rat liver perfusion
2 studies with cyclic heptapeptide toxins of Microcystis and Oscillatoria (freshwater
3 cyanobacteria). Toxicon. 26(9):827-837.
4 Bhattacharya, R., P.V.L. Rao, A.S.B. Bhaskar et al. 1996. Liver slice culture for assessing
5 hepatoxicity of freshwater cyanobacteria. Human Exp. Toxicol. 15(2): 105-110.
6 Bhattacharya, R., K. Sugendran, R.S. Dangi and P.V.L. Rao. 1997. Toxicity evaluation of
7 freshwater cyanobactenum Microcystis aeruginosaPCC-7806. 2. Nephrotoxicity in rats.
8 Biomed. Environ. Sci. 10(1):93-101.
9 Billings, W.H. 1981. Water-associated human illness in northeast Pennsylvania and its
10 suspected association with blue-green algae blooms. In: The Water Environment: Algal Toxins
11 and Health, W.W. Carmichael, Ed. Plenum Press, New York, NY. p. 243-255.
12 Bishop, C.T., E.F.L.J. Anet and P.R. Gorham. 1959. Isolation and identification of the fast-
13 death factor in Microcystis aeruginosa NRC-l. Can. J. Biochem. Physiol. 37(3):453-471.
14 Blankson, H., E.M. Grotterod and P.O. Seglen. 2000. Prevention of toxin-induced cytoskeletal
15 disruption and apoptotic liver cell death by the grapefruit flavonoid, narigin. Cell Death Diff.
16 7(8):739-746.
17 Boe, R., B.T. Gjersten, O.K. Vintermyr et al. 1991. The protein phosphatase inhibitor okadaic
18 acid induces morphological changes typical of apoptosis in mammalian cells. Exp. Cell Res.
19 195(l):237-246.
20 Botha, N, M. van de Venter, T.G. Downing et al. 2004. The effect of intraperitoneally
21 administered microcystin-LR on the gastrointestinal tract of Balb/c mice. Toxicon.
22 43(3):251-254.
23 Boua'icha, N. and I. Maatouk. 2004. Microcystin-LR and nodularin induce intracellular
24 glutathione alteration, reactive oxygen species production and lipid peroxidation in primary
25 cultured rat hepatocytes. Toxicol. Lett. 148(l-2):53-63.
26 Boua'icha, N., I. Maatouk, MJ. Plessis and F. Perin. 2005. Genotoxic potential of microcystin-
27 LR and nodularin in vitro in primary cultured rat hepatocytes and in vivo in rat liver. Environ.
28 Toxicol. 20(3):341-347.
29 Brooks, W.P. and G.A. Codd. 1987. Immunological and toxicological studies on Microcystis
30 aeruginosa peptide toxin. Br. Phycol. J. 22(3):301.
31 Carmichael, W.W., S.M.F.O. Azevedo, IS. An et al. 2001. Human fatalities from
32 cyanobacteria: Chemical and biological evidence for cyanotoxins. Environ. Health Perspect.
33 109(7):663-668.
34 Chen, J., T.C. Campbell, J.Li and R.Peto. 1991. Diet, Life-style and Mortality in China. A
35 Study of the Characteristics of 65 Chinese Counties, Cornell University Press, Ithaca, NY. (As
36 cited in Health Canada, 2002).
99 DRAFT - DO NOT CITE OR QUOTE
-------
1 Chen, K., Y.Z. Shen and G.F. Shen. 1994. Study on incidence rate of some cancer in areas with
2 difference in drinking water sources. Chin. J. Public Health. 12(3): 146-148 (As cited in Zhou et
3 al., 2002). (Chinese)
4 Chen, T., X. Zhao, Y Liu et al. 2004. Analysis of immunomodulating nitric oxide, iNOS and
5 cytokines MRNA in mouse macrophages induced by microcystin-LR. Toxicology.
6 197(l):67-77.
7 Chen, T., P. Shen, J. Zhang et al. 2005. Effects of microcystin-LR on patterns of iNOS and
8 cytokine mRNA expression in macrophages in vitro. Environ. Toxicol. 20(1):85-91.
9 Chen, T., J. Cui, Y. Liang et al. 2006. Identification of human liver mitochondrial aldehyde
10 dehydrogenase as a potential target for microcystin-LR. Toxicology. 220(1):71-80.
11 Chernoff, N, E.S. Hunter III, L.L. Hall et al. 2002. Lack of teratogenicity of microcystin-LR in
12 the mouse and toad. J. Appl. Toxicol. 22(1):13-17.
13 Chong, M.W.K., K.D. Gu, P.K.S. Lam et al. 2000. Study of the cytotoxicity of microcystin-LR
14 on cultured cells. Chemosphere. 41(1-2):143-147.
15 Cote, L-M., R. A. Lovell, E.H. Jeffrey et al. 1986. Failure of blue-green algae (Microcystis
16 aeruginosd) hepatotoxin to alter in vitro mouse liver enzymatic activity. J. Toxicol.- Toxin Rev.
17 52(2):256.
18 Craig, M., H.A. Luu, T.L. McCready et al. 1996. Molecular mechanisms underlying the
19 interaction of motuporin and microcystins with type-1 and type-2A protein phosphatases.
20 Biochem. Cell Biol. 74(4):569-578.
21 Creasia, D.A. 1990. Acute inhalation toxicity of microcystin-LR with mice. Toxicon.
22 28(6):605.
23 Dabholkar, A.S. and W.W. Carmichael. 1987. Ultrastructural changes in the mouse liver
24 induced by hepatoxin from the freshwater cyanobacterium Microcystis aeruginosa strain 7820.
25 Toxicon. 25(3):285-292.
26 Dahlem, A.M., A.S. Hassan, S.P. Swanson et al. 1989. A model system for studying the
27 bioavailability of intestinally administered microcystin-LR, a hepatotoxic peptide from the
28 cyanobacterium Microcystis aeruginosa. Pharmacol. Toxicol. 64(2): 177-181.
29 Davidson, F.F. 1959. Poisoning of wild and domestic animals by a toxic waterbloom of Nostoc
30 rivulare Kuetz. J. Am. Water Works Assoc. 51:1277-1287.
31 Dillenberg, H.O. and M.K. Dehnel. 1960. Toxic water bloom in Saskatchewan, 1959. Can.
32 Med. Assoc. J. 83:1151-1154.
33 Ding, W.X. and C.N. Ong. 2003. Role of oxidative stress and mitochondrial changes in
34 cyanobacteria-induced apoptosis and hepatotoxicity. FEMS Microbiol. Lett. 220(1): 1-7.
100 DRAFT - DO NOT CITE OR QUOTE
-------
1 Ding, W.X., H.M. Shen, Y. Shen et al. 1998a. Microcystic cyanobacteria causes mitochondrial-
2 membrane potential alteration and reactive oxygen species formation in primary cultured rat
3 hepatocytes. Environ. Health Perspect. 106(7):409-413.
4 Ding, W.X., H.M. Shen, H.G. Zhu and C.N. Ong. 1998b. Studies on oxidative damage induced
5 by cyanobacteria extract in primary cultured rat hepatocytes. Environ. Res. 78(1): 12-18.
6 Ding, W.X., H.M. Shen, H.G. Zhu, B.L. Lee and C.N. Ong. 1999. Genotoxicity of microcystic
7 cyanobacterial extract of a water source in China. Mutat Res. 442(2):69-77.
8 Ding, W.X., H.M. Shen and C.N. Ong. 2000a. Microcystic cyanobacteria extract induces
9 cytoskeletal disruption and intracellular glutathione alteration in hepatocytes. Environ. Health
10 Perspect. 108(7):605-609.
11 Ding, W.X., H.M. Shen and C.N. Ong. 2000b. Critical role of reactive oxygen species and
12 mitochondrial permeability transition in microcystin-induced rapid apoptosis in rat hepatocytes.
13 Hepatology. 32(3):547-555.
14 Ding, W.X., H.M. Shen and C.N. Ong. 2001. Critical role of reactive oxygen species formation
15 in microcystin-induced cytoskeleton disruption in primary cultured hepatocytes. J. Toxicol.
16 Environ. Health Part A. 64(6):507-519.
17 Ding, W.X., H.M. Shen and C.N. Ong. 2002. Calpain activation after mitochondrial
18 permeability transition in microcystin-induced cell death in rat hepatocytes. Biochem. Biophys.
19 Res. Commun. 291(2):321-331.
20 Duy, T.N., P.K.S. Lam, G.R. Shaw et al. 2000. Toxicology and risk assessment of freshwater
21 cyanobacterial (blue-green algal) toxins in water. Rev. Environ. Contam. Toxicol. 163:113-185.
22 Dvorakova, D., K. Dvorakova, L. Blaha et al. 2002. Effects of cyanobacterial biomass and
23 purified microcystins on malformations in Xenopus laevis: Teratogenesis assay (FETAX).
24 Environ. Toxicol. 17(6):547-555.
25 Elleman, T.C., I.R. Falconer, J.R. Jackson andM.T.C. Runnegar. 1978. Isolation,
26 characterization and pathology of the toxin from aMicrocystis aeruginosa (=Anacystis cyanea)
27 bloom. Aust. J. Biol. Sci. 31(3):209-218.
28 el Saadi, O. and A.S. Cameron. 1993. Illness associated with blue-green algae. Med. J. Aust.
29 158(11):792-793.
30 el Saadi, O., AJ. Esterman, S. Cameron and D.M. Roder. 1995. Murray River water raised
31 cyanobacterial cell counts, and gastrointestinal and dermatological symptoms. Med. J. Aust.
32 162(3): 122-125.
33 Eriksson, I.E. and R.D. Golman. 1993. Protein phosphatase inhibitors alter cytoskeletal
34 structure and cellular morphology. Adv. Prot. Phosphatases. 7:335-357.
101 DRAFT - DO NOT CITE OR QUOTE
-------
1 Eriksson, I.E., J.A.O. Meriluoto and T. Lindholm. 1987. Cyanobacterial toxins - physiological
2 and ecological effects. In: Rapport till Finlans Akademi. p. 1-12.
3 Eriksson, I.E., G.I.L. Paatero, J.A.O. Meriluoto et al. 1989. Rapid microfilament reorganization
4 induced in isolated rat hepatocytes by microcystin-LR, a cyclic peptide toxin. Exp. Cell Res.
5 185(1):86-100.
6 Eriksson, I.E., L. Gronberg, S. Nygard et al. 1990a. Hepatocellular uptake of
7 3H-dihydromicrocystin-LR, a cyclic peptide toxin. Biochim. Biophys. Acta. 1025(l):60-66.
8 Eriksson, I.E., D. Toivola, J.A.O. Meriluoto et al. 1990b. Hepatocyte deformation induced by
9 cyanobacterial toxins reflects inhibition of protein phosphatases. Biochem. Biophys. Res.
10 Commun. 173(3): 1347-1353.
11 Eriksson, I.E., D.M. Toivola, M. Reinikainen et al. 1994. Testing of toxicity in cyanobacteria
12 by cellular assays. In: Detection Methods for Cyanobacterial Toxins, G.A. Codd, T.M. Jefferies,
13 C.W. Keevil and E. Potter, Eds. Royal Society of Chemistry, Cambridge, p. 75-84.
14 Falconer, I.R. 1991. Tumor promotion and liver injury caused by oral consumption of
15 cyanobacteria. Environ. Toxicol. Water Qual. 6(2):177-184.
16 Falconer, I.R. and T.H. Buckley. 1989. Tumour promotion by Microcystis sp., a blue-green alga
17 occurring in water supplies. Med. J. Aust. 150(6):351.
18 Falconer, I.R. and A.R. Humpage. 1996. Tumour promotion by cyanobacterial toxins.
19 Phycologia. 35:74-79.
20 Falconer, L.R. and M.T.C. Runnegar. 1987a. Effects of the peptide toxin from Microcystis
21 aeruginosa on intracellular calcium, pH and membrane integrity in mammalian cells. Chem-
22 Biol. Interact. 63(3):215-225.
23 Falconer, L.R. and M.T.C. Runnegar. 1987b. Toxic peptide from the blue-green alga
24 Microcystis aeruginosa- effects on hepatocytes and thymocytes. Biochem. Soc. Trans.
25 15(3):468-469.
26 Falconer, I.R. and S.K. Yeung. 1992. Cytoskeletal changes in hepatocytes induced by
27 Microcystis toxins and their relation to hyperphosphorylation of cell proteins. Chem-Biol.
28 Interact. 81(1-2): 181-196.
29 Falconer, I.R., A.R.B. Jackson, J. Langley and M.T.C. Runnegar. 1981. Liver pathology in
30 mice in poisoning by the blue-green alga Microcystis aeruginosa. Aust. J. Biol. Sci. 34(2): 174-
31 187.
32 Falconer, I.R., A.M. Beresford and M.T.C. Runnegar. 1983. Evidence of liver damage by toxin
33 from a bloom of the blue-green alga, Microcystis aeruginosa. Med. J. Aust. 1(11):511-514.
102 DRAFT - DO NOT CITE OR QUOTE
-------
1 Falconer, I.R., T. Buckley and M.T.C. Runnegar. 1986. Biological half-life organ distribution
• 125
2 and excretion of I-labeled toxic peptide from the blue-green alga Microcystis aeruginosa.
3 Aust.J. Biol. Sci. 39(1): 17-21.
4 Falconer, L.R., J.V. Smith, A.R.B. Jackson et al. 1988. Oral toxicity of a bloom of the
5 cyanobacterium Microcystis aeruginosa administered to mice over periods of up to one year. J.
6 Toxicol. Environ. Health. 24(3):291-305.
7 Falconer, I.R., M.D. Burch, D.A. Steffensen, M. Choice and O.R. Coverdale. 1994. Toxicity of
8 the blue-green alga (Cyanobacterium) Microcystis aeruginosa in drinking water to growing pigs,
9 as an animal model for human injury and risk assessment. Environ. Toxicol. Water Qual.
10 9:131-139.
11 Fawell, J.K., R.E. Mitchell, D.J. Everett and R.E. Hill. 1999. The toxicity of cyanobacterial
12 toxins in the mouse. 1. Microcystin-LR. Human Exp. Toxicol. 18(3):162-167.
13 Feitz, A.J., T. Lukondeh, M.C. Moffitt et al. 2002. Absence of detectable levels of the
14 cyanobacterial toxin (microcystin-LR) carry-over into milk. Toxicon. 40(8): 1173-1180.
15 Fischer, WJ. and D.R. Dietrich. 2000. Toxicity of the cyanobacterial cyclic heptapeptide toxins
16 microcystin-LR and -RR in early lifestages of the African clawed frog. Aquat. Toxicol.
17 49(3):189-198.
18 Fitzgeorge, N.L.M., S. A. Clark and C.W. Kelvin. 1994. Routes of intoxication. In: Detection
19 Methods for Cyanobacterial (Blue-Green Algal) Toxins. G.A. Codd, T.M. Jeffreies, C.W.
20 Kelvin and E. Potter, Eds. Royal Society of Chemistry, Cambridge, U.K. p. 69-74.
21 Fladmark, K.E., M.H. Serres, N.L. Larsen et al. 1998. Sensitive detection of apoptogenic toxins
22 in suspension cultures of rat and salmon hepatocytes. Toxicon. 36(8):1101-1114.
23 Fleming, L.E., C. Rivero, J. Burns et al. 2002. Blue green algal (cyanobacterial) toxins, surface
24 drinking water and liver cancer in Florida. Harmful Algae. 1(2): 157-168.
25 Fleming, L.E., C. Rivero, J. Burns, C. Williams, J.A. Bean and W.B. Stephan. 2004.
26 Cyanobacteria exposure, drinking water and colorectal cancer. In: Harmful Algae 2002.
27 Proceedings of the Xth International Conference on Harmful Algae. K.A. Steidinger, J.H.
28 Landsberg, C.R. Tomas and G.A. Vargo, Eds. Florida Fish and Wildlife Conservation
29 Commission and Intergovernmental Oceanographic Commission of UNESCO, Tallahassee, FL.
30 p. 470-472.
31 Foxall, T.L. and JJ. Sasner, Jr. 1981. Effects of a hepatic toxin from the cyanophyte
32 Microcystis aeruginosa. In: The Water Environment: Algal Toxins and Health, W.W.
33 Carmichael, Ed. Plenum Press, New York, NY. p. 365-387.
34 Frangez, R., M.C. Zuzek, J. Mrkun et al. 2003. Microcystin-LR affects cytoskeleton and
35 morphology of rabbit primary whole embryo cultured cells in vitro. Toxicon. 41(8):999-1005.
103 DRAFT - DO NOT CITE OR QUOTE
-------
1 Fu, W.Y., J.P. Chen, X.M. Wang and L.H. Xu. 2005. Altered expression of p53, Bcl-2 and Bax
2 induced by microcystin-LR in vivo and in vitro. Toxicon. 46(2): 171-177.
3 Gehringer, M.M. 2004. Microcystin-LR and okadaic acid-induced cellular effects: A dualistic
4 response. FEES Lett. 557(l-3):l-8.
5 Gehringer, M.M., S. Govender, M. Shaw and T.G. Downing. 2003 a. An investigation of the
6 role of vitamin E in the protection of mice against microcystin toxicity. Environ. Toxicol.
7 18(2): 142-148.
8 Gehringer, M.M., K.S. Downs, T.G. Downing et al. 2003b. An investigation into the effects of
9 selenium supplementation on microcystin hepatotoxicity. Toxicon. 41(4):451-458.
10 Gehringer, M.M., E.G. Shephard, T.G. Downing et al. 2004. An investigation into the
11 detoxification of microcystin-LR by the glutathione pathway in Balb/c mice. Int. J. Biochem.
12 CellBiol. 36(5):931-941.
13 Goldberg, J., H.B. Huang, Y.G. Kwon, et al. 1995. Three-dimensional structure of the catalytic
14 subunit of protein serine/threonine phosphatase-1. Nature 376(6543)745-753.
15 Grabow, W.O.K., W.C. Du Randt, O.W. Prozesky and W.E. Scott. 1982. Microcystis
16 aeruginosa toxin: Cell culture toxicity, hemolysis, and mutagenicity assays. Appl. Environ.
17 Microbiol. 43(6): 1425-1433.
18 Gulledge, B.M., J.B. Aggen, H.B. Huang et al. 2002. The microcystins and nodularins: cyclic
19 polypeptide inhibitors of PP1 and PP2A. Curr. Med. Chem. 9(22): 1991-2003.
20 Gulledge, B.M., J.B. Aggen and A.R. Chamberlin. 2003a. Linearized and truncated microcystin
21 analogues as inhibitors of protein phosphatases 1 and 2A. Bioorg. Medicinal Chem.. Lett.
22 13(17):2903-2906.
23 Gulledge, B.M., J.B. Aggen, H. Engetal. 2003b. Microcystin analogues comprised only of
24 Adda and a single additional amino acid retain moderate activity as PP1/PP2A inhibitors.
25 Bioorg. Medicinal Chem. Lett. 13(17):2907-2911.
26 Gupta, N., S.C. Pant, R. Vijayaraghavan and P.V. Rao. 2003. Comparative toxicity evaluation
27 of cyanobacterial cyclic peptide toxin microcystin variants (LR, RR, YR) in mice. Toxicology.
28 188(2-3):285-296.
29 Guzman, R.E. and P.F. Solter. 1999. Hepatic oxidative stress following prolonged sublethal
30 microcystin-LR exposure. Toxicologic Pathol. 27(5):582-588.
31 Guzman, R.E. and P.F. Solter. 2002. Characterization of sublethal microcystin-LR exposure in
32 mice. Vet. Pathol. 39(l):17-26.
33 Guzman, R.E., P.F. Solter and M.T. Runnegar. 2003. Inhibition of nuclear protein phosphatase
34 activity in mouse hepatocytes by the cyanobacterial toxin microcystin-LR. Toxicon.
35 41(7):773-781.
104 DRAFT - DO NOT CITE OR QUOTE
-------
1 Harada, K., K. Ogawa, K. Matsuura et al. 1990. Structural determination of geometrical
2 isomers of microcystins LR and RR from cyanobacteria by two-dimensional NMR spectroscopic
3 techniques. Chem. Res. Toxicol. 3(5):473-481.
4 Hastie, C.J., E.B. Borthwick, L.F. Morrison et al. 2005. Inhibition of several protein
5 phosphatases by a non-covalently interacting microcystin and a novel cyanobacterial peptide,
6 nostocyclin. Biochim. Biophys. Acta. 1726:187-193.
7 Hayakawa, K. and K. Kohama. 1995. Reversible effects of okadaic acid and microcystin-LR on
8 the ATP-dependent interaction between actin and myosin. J. Biochem. 117(3):509-514.
9 Health Canada. 2002. Guidelines for Canadian Drinking Water Quality: Supporting
10 Documentation — Cyanobacterial Toxins-Microcystin-LR. Water Quality and Health Bureau,
11 Healthy Environments and Consumer Safety Branch, Health Canada, Ottawa, Ontario.
12 Available at http://www.hc-sc.gc.ca/ewh-semt/pubs/water-eau/doc_sup-appui/index_e.html.
13 Heinze, R. 1999. Toxicity of the cyanobacterial toxin microcystin-LR to rats after 28 days
14 intake with the drinking water. Environ. Toxicol. 14(1):57-60.
15 Heinze, R., J. Fastner, U. Neumann and I. Chorus. 2001. Testing cyanobacterial toxicity with
16 primary rat hepatocyte and cell-line assays. In: Cyanotoxins: Occurrence, Causes,
17 Consequences, I. Chorus, Ed. Springer-Verlag, New York, NY. p. 317-324.
18 Herfindal, L. and F. Selheim. 2006. Microcystin produces disparate effects on liver cells in a
19 dose dependent manner. Mini. Rev. Med. Chem. 6(3):279-285.
20 Hermansky, S.J., PJ. Casey and SJ. Stohs. 1990a. Cyclosporin A—a chemoprotectant against
21 microcystin-LR toxicity. Toxicol. Lett. 54(2-3):279-285.
22 Hermansky S.J., S.N. Wolff and SJ. Stohs. 1990b. Use of rifampin as an effective
23 chemoprotectant and antidote against microcystin-LR toxicity. Pharmacology. 41(4):231-236.
24 Hermansky, S.J., SJ. Stohs, R.S. Markin and WJ. Murray. 1990c. Hepatic lipid peroxidation,
25 sulfhydryl status, and toxicity of the blue-green algal toxin microcystin-LR in mice. J. Toxicol.
26 Environ. Health. 31(1):71-91.
27 Hermansky, S J., SJ. Stohs, Z.M. Eldeen et al. 1991. Evaluation of potential chemoprotectants
28 against microcystin-LR hepatotoxicity in mice. J. Appl. Toxicol. ll(l):65-73.
29 Hernandez, M., M. Macia, C. Padilla and F.F. Del Campo. 2000. Modulation of human
30 polymorphonuclear leukocyte adherence by cyanopeptide toxins. Environ. Res. 84(l):64-68.
31 Hilborn, E.D., W.W. Carmichael, M. Yuan et al. 2005. Serologic evaluation of human
32 microcystin exposure. From: Interagency International Symposium on Cyanobacterial Harmful
33 Algal Blooms, Research Triangle Park, NC.
105 DRAFT - DO NOT CITE OR QUOTE
-------
1 Honkanen, R.E., J. Zwiller, R.E. Moore et al. 1990. Characterization of microcystin-LR, a
2 potent inhibitor of type 1 and type 2A protein phosphatases. J. Biol. Chem.
3 265(32): 19401-19404.
4 Hooser, S.B. 2000. Fulminant hepatocyte apoptosis in vivo following microcystin-LR
5 administration to rats. Toxicol. Pathol. 28(5):726-733.
6 Hooser, S.B., V.R. Beasley, R.A. Lovell et al. 1989a. Toxicity of microcystin LR, a cyclic
7 heptapeptide hepatoxin from Microcystis aeruginosa to rats and mice. Vet. Pathol.
8 26(3):246-252.
9 Hooser, S.B., L.L. Waite, V.R. Beasley et al. 1989b. Microcystin-A induces morphologic and
10 cytoskeletal hepatocyte changes in vitro. Toxicon 27(1):50-51.
11 Hooser, S.B., V.R. Beasley, EJ. Basgall, W.W. Carmichael, and W.M. Haschek. 1990.
12 Microcystin-LR-induced ultrastructural changes in rats. Vet. Pathol. 27(1):9-15.
13 Hooser, S.B., M.S. Kuhlenschmidt, A.M. Dahlem et al. 1991a. Uptake and subcellular
14 localization of tritiated dihydro-microcystin-LR in rat liver. Toxicon. 29(6):589-601.
15 Hooser, S.B., V.R. Beasley, L.L. Waite et al. 1991b. Actin filament alterations in rat
16 hepatocytes induced in vivo and in vitro by microcystin-LR, a hepatoxin from the blue-green
17 alga, Microcystis aeruginosa. Vet. Pathol. 28(4):259-266.
18 Hu, Z., H. Chen, Y. Li et al. 2002. [The expression of bcl-2 and bax genes during microcystin
19 induced liver tumorigenesis.] Zhonghua Yu Fang Yi Xue ZaZhi. 36(4):239-242. (Chinese)
20 Humpage, A.R. and LR. Falconer. 1999. Microcystin-LR and liver tumor promotion: Effects on
21 cytokinesis, ploidy, and apoptosis in cultured hepatocytes. Environ. Toxicol. 14(l):61-75.
22 Humpage, A.R., SJ. Hardy, EJ. Moore et al. 2000. Microcystins (cyanobacterial toxins) in
23 drinking water enhance the growth of aberrant crypt foci in the mouse colon. J. Toxicol.
24 Environ. Health Part A. 61(3):155-165.
25 Ito, E., F. Kondo andK.-I. Harada. 1997a. Hepatic necrosis in aged mice by oral administration
26 of microcystin-LR. Toxicon. 35(2):231-239.
27 Ito, E., F. Kondo, K. Terao and K.-I. Harada. 1997b. Neoplastic nodular formation in mouse
28 liver induced by repeated intraperitoneal injections of microcystin-LR. Toxicon.
29 35(9):1453-1457.
30 Ito, E., F. Kondo and K. Harada. 2001. Intratracheal administration of microcystin-LR, and its
31 distribution. Toxicon. 39(2-3):265-271.
32 Ito, E., A. Takai, F. Kondo et al. 2002. Comparison of protein phosphatase inhibitory activity
33 and apparent toxicity of microcystins and related compounds. Toxicon. 40(7): 1017-1025.
106 DRAFT - DO NOT CITE OR QUOTE
-------
1 Jackson, A.R.B., A. Mclnnes, I.R. Falconer and M.T.C. Runnegar. 1984. Clinical and
2 pathological changes in sheep experimentally poisoned by the blue-green alga Microcysstis
3 aeruginosa. Vet. Pathol. 21(1): 102-113.
4 Jayaraj, R., T. AnandandP.V. Rao. 2006. Activity and gene expression profile of certain
5 antioxidant enzymes to microcystin-LR induced oxidative stress in mice. Toxicology.
6 220(2-3): 136-146.
7 Jiao, D.A., G.F. Shen, Y.Z. Shen and G.M. Zheng. 1985. The case-control study of colorectal
8 cancer. Chin. J. Epidemio. 6:285-288 (As cited in Zhou etal., 2002). (Chinese)
9 Jochimsen, E.M., W.W. Carmichael, J.S. An et al. 1998. Liver failure and death after exposure
10 to microcystins at a hemodialysis center in Brazil. New Engl. J. Med. 338(13):873-878.
11 Jones, C.L.A. 1984. Biochemical, hematological, and hepatotoxicological studies of
12 hepatotoxins from Microcystis aeruginosa strain 7820, andAnabaenaflos-aquae strain S-23-G.
13 M.S. Thesis, Wright State Univ., Dayton, OH.
14 Jones, C.L.A. and W.W. Carmichael. 1984. Comparison of hepatotoxins from the cyanobacteria
15 Anabaenaflos-aquae and Microcystis aeruginosa. Fed. Proc. 43(3):A1716.
16 Kalbe, L. 1984. Animal experiments on the oral toxicity of blue-green algae waterblooms.
17 Limnol. (Berlin) 15(2):559-562.
18 Khan, S.A., S. Ghosh, M.L. Wickstrom et al. 1995. Comparative pathology of microcystin-LR
19 in cultured hepatocytes, fibroblasts and renal epithelial cells. Natural Toxins. 3(3): 119-128.
20 Klassen, C.D., Ed. 2001. Casarett and Doull's Toxicology: The Basic Science of Poisons, 6th
21 ed. McGraw-Hill, New York, NY.
22 Knapp J., S. Aleth, F. Balzer et al. 2002. Calcium-independent activation of the contractile
23 apparatus in smooth muscle of mouse aorta by protein phosphatase inhibition. Naunyn
24 Schmiedebergs Arch. Pharmacol. 366(6):562-569.
25 Kondo, F., Y. Ikai, H. Oka et al. 1992. Formation, characterization, and toxicity of the
26 glutathione and cysteine conjugates of toxic heptapeptide microcystins. Chem. Res. Toxicol.
27 5(5):591-596.
28 Kondo, F., H. Matsumoto, S. Yamada et al. 1996. Detection and identification of metabolites of
29 microcystins formed in vivo in mouse and rat livers. Chem. Res. Toxicol. 9(8): 1355-1359.
30 Kujbida, P., E. Hatanaka, A. Campa, P. Colepicolo and E. Pinto. 2006. Effects of microcystins
31 on human polymorphonuclear leukocytes. Biochem. Biophys. Res. Commun. 341(l):273-277.
32 Lankoff, A., A. Banasik, G. Obe et al. 2003. Effect of microcystin-LR and cyanobacterial
33 extract from Polish reservoir drinking water on cell cycle progression, mitotic spindle, and
34 apoptosis in CHO-K1 cells. Toxicol. Appl. Pharmacol. 189(3):204-213.
107 DRAFT - DO NOT CITE OR QUOTE
-------
1 Lankoff, A., L. Krzowski, J. Glab et al. 2004. DNA damage and repair in human peripheral
2 blood lymphocytes following treatment with microcystin-LR. Mutat. Res. 559(1-2): 131-142.
3 LeClaire, R.D., W.B. Lawrence, K. A. Bostian and K. A. Mereish. 1988. Acute toxicity of
4 microcystin-LR in the rat: A comparative dose-response study using serum chemistries and
5 mortality as indices. Toxicologist. 8(1):221.
6 LeClaire, R.D., G.W. Parker and D.R. Franz. 1995. Hemodynamic and calorimetric changes
7 induced by microcystin-LR in the rat. J. Appl. Toxicol. 15(4):303-311.
8 Leiers, T., A. Bihlmayer, H.P.T. Ammon and M.A. Wahl. 2000. [Ca2+](i)- and insulin-
9 stimulating effect of the non-membranepermeable phosphatase-inhibitor microcystin-LR in
10 intact insulin-secreting cells (RINmSF). Br. J. Pharmacol. 130(6):1406-1410.
11 Lin, J.R. and F.S. Chu. 1994. Kinetics of distribution of microcystin-LR in serum and liver
12 cytosol of mice: an immunochemical analysis. J. Agric. Food Chem. 42(4): 1035-1040.
13 Lovell, R.A., DJ. Schaeffer, S.B. Hooser et al. 1989a. Toxicity of intraperitoneal doses of
14 microcystin-LR in two strains of male mice. J. Environ. Pathol. Toxicol. Oncol. 9(3):221-237.
15 Lovell, R.A., K.R. Holmes, W.W. Carmichael and V.R. Beasley. 1989b. Hemodynamic and
16 pathologic effects of microcystin-LR administered intravenously to swine. Toxicon. 27(1):60.
17 (Abstract)
18 Maatouk, L, N. Bouaicha, MJ. Plessis and F. Perin. 2004. Detection by 32P-postlabelling of
19 8-oxo-7,8-dihydro-2'-deoxyguanosine in DNA as biomarker of microcystin-LR- and nodularin-
20 induced DNA damage in vitro in primary cultured rat hepatocytes and in vivo in rat liver. Mutat.
21 Res. 564(1):9-20.
22 MacKintosh, C., K.A. Beattie, S. Klumpp et al. 1990. Cyanobacterial microcystin-LR is a
23 potent and specific inhibitor of protein phosphatases 1 and 2A from both mammals and higher
24 plants. FEES Lett. 264(2): 187-192.
25 MacKintosh, R.W., K.N. Dalby, D.G. Campbell, P.T. Cohen, P. Cohen and C. MacKintosh.
26 1995. The Cyanobacterial toxin microcystin binds covalently to cysteine-273 on protein
27 phosphatase 1. FEES Lett. 371(3):236-240.
28 Maidana, M., V. Carlis, F.G. Galhardi et al. 2006. Effects of microcystins over short- and long-
29 term memory and oxidative stress generation in hippocampus of rats. Chem. Biol. Interact.
30 159(3):223-234.
31 Mankiewicz, J., M. Tarczynska, K.E. Fladmark et al. 2001. Apoptotic effect of Cyanobacterial
32 extract on rat hepatocytes and human lymphocytes. Environ. Toxicol. 16(3):225-233.
33 Matsushima, R., S. Yoshizawa, M.F. Watanabe et al. 1990. In vitro and in vivo effects of
34 protein phosphatase inhibitors, microcystins and nodularin on mouse skin and fibroblasts.
35 Biochem. Biophys. Res. Commun. 171(2):867-874.
108 DRAFT - DO NOT CITE OR QUOTE
-------
1 Matsushima-Nishiwaki, R., Y. Shidoji, S. Nishiwaki et al. 1995. Suppression by carotenoids of
2 microcystin-induced morphological changes in mouse hepatocytes. Lipids. 30(11): 1029-1034.
3 MattilaK., A. Annila and T.T. Rantala. 2000. Metal ions mediate the binding of cyanobacterial
4 toxins to human protein phosphatase I: A computational study. Oulu University Library, Oulun
5 Yliopisto, Oulu.
6 Maynes JT, K.R. Perreault, M.M. Cherney, H.A. Luu, M.N. James and C.F. Holmes. 2004.
7 Crystal structure and mutagenesis of a protein phosphatase-1 :calcineurin hybrid elucidate the
8 role of the P12-P13 loop in inhibitor binding. J. Biol. Chem. 279(41):43198-43206.
9 Maynes, J.T., H.A. Luu, M.M. Cherney et al. 2006. Crystal structures of protein phosphatase-1
10 bound to motuporin and dihydromicrocystin-LA: Elucidation of the mechanism of enzyme
11 inhibition by cyanobacterial toxins. J. Mol. Biol. 356(1):111-120.
12 McDermott, C.M., C.W. Nho, W. Howard and B. Holton. 1998. The cyanobacterial toxin,
13 microcystin-LR can induce apoptosis in a variety of cell types. Toxicon. 36(12):1981-1996.
14 Mereish, K. A. and R. Solow. 1990. Effect of antihepatotoxic agents against microcystin-LR
15 toxicity in cultured rat hepatocytes. Pharmacol. Res. 7(3):256-259.
16 Mereish, K.A., R. Solow, Y. Singh et al. 1989. Comparative toxicity of cyclic polypeptides and
17 depsipeptides on cultured rat hepatocytes. Toxicologist. 9(1):68.
18 Mereish, K.A., D.L. Bunner, D.R. Ragland and D.A. Creasia. 1991. Protection against
19 microcystin-LR induced hepatoxicity by Silymarin: Biochemistry, histopathology and lethality.
20 Pharm. Res. 8(2):273-277.
21 Meriluoto, J.A., S.E. Nygard, A.M. Dahlem and I.E. Eriksson. 1990. Synthesis, organotropism
22 and hepatocellular uptake of two tritium-labeled epimers of dihydromicrocystin-LR, a
23 cyanobacterial peptide toxin analog. Toxicon. 28(12): 1439-1446.
24 Metcalf J.S., K.A. Beattie, S. Pflugmacher and G.A. Codd. 2000. Immuno-crossreactivity and
25 toxicity assessment of conjugation products of the cyanobacterial toxin, microcystin-LR. FEMS
26 Microbiol. Lett. 189(2):155-158.
27 Mikhailov, A., A.S. Harmala-Brasken, J. Hellman et al. 2003. Identification of ATP-synthase as
28 a novel intracellular target for microcystin-LR. Chem-Biol. Inter. 142(3):223-237.
29 Milutinovic, A., B. Sedmak, I. Horvat-Znidarsic and D. Suput. 2002. Renal injuries induced by
30 chronic intoxication with microcystins. Cell Mol. Biol. Lett. 7(1):139-141.
31 Milutinovic A, M. Zivin, R. Zore-Pleskovic, B. Sedmak and D. Suput. 2003. Nephrotoxic
32 effects of chronic administration of microcystins-LR and-YR. Toxicon. 42(3):281-288.
33 Miura, G.A., N.A. Robinson, W.B. Lawrence and J.G. Page. 1991. Hepatoxicity of
34 microcystin-LR in fed and fasted rats. Toxicon. 29(3):337-346.
109 DRAFT - DO NOT CITE OR QUOTE
-------
1 Moreno, I.M., A. Mate, G. Repetto et al. 2003. Influence of microcystin-LR on the activity of
2 membrane enzymes in rat intestinal mucosa. J. Physiol. Biochem. 59(4):293-299.
3 Moreno, I, S. Pichardo, A. Jos et al. 2005. Antioxidant enzyme activity and lipid peroxidation
4 in liver and kidney of rats exposed to microcystin-LR administered intraperitoneally. Toxicon.
5 45(4):395-402.
6 Namikoshi M., BW Choi, F Sun, KL Rinehart, WR Evans and W.W. Carmichael. 1993.
7 Chemical characterization and toxicity of dihydro derivatives of nodularin and microcystin-LR,
8 potent cyanobacterial cyclic peptide hepatotoxins. Chem. Res. Toxicol. 6(2):151-158.
9 Naseem, S.M., Hines, H.B. andD.A. Creasia. 1990. Inhibition of microcystin-induced release
10 of cyclooxygenase products from rat hepatocytes by anti-inflammatory steroids. Proc. Soc. Exp.
11 Biol. Med. 195(3):346-349.
12 Naseem, S.M., K.A. Mereish, R. Solow and H.B. Hines. 1991. Microcystin-induced activation
13 of prostaglandin synthesis and phospholipids metabolism in rat hepatocytes. Toxicol. In Vitro.
14 5(4):341-345.
15 National Research Council. 1983. Risk Assessment in the Federal Government: Managing the
16 Process. National Academy Press, Washington, DC.
17 Nishiwaki R., T. Ohta, E. Sueoka et al. 1994. Two significant aspects of microcystin-LR:
18 specific binding and liver specificity. Cancer Lett. 83(l-2):283-289.
19 Nishiwaki-Matsushima, R., S. Nishiwaki, T. Ohta et al. 1991. Structure-function relationships
20 of microcystins, liver tumor promoters, in interaction with protein phosphatase. Jpn. J. Cancer
21 Res. 82(9):993-996.
22 Nishiwaki-Matsushima, R., T. Ohta, S. Nishiwaki et al. 1992. Liver tumor promotion by the
23 cyanobacterial cyclic peptide toxin microcystin-LR. J. Cancer Res. Clin. Oncol.
24 118(6):420-424.
25 Nobre, A.C.L., M.C.M. Jorge, D.B. Menezes et al. 1999. Effects of microcystin-LR in isolated
26 perfused rat kidney. Brazilian J. Med. Biol. Res. 32(8):985-988.
27 Nobre, A.C.L., G.R. Coelho, M.C.M. Coutinho et al. 2001. The role of phospholipase A(2) and
28 cyclooxygenase in renal toxicity induced by microcystin-LR. Toxicon. 39(5):721-724.
29 Nobre, A.C.L., A.M.C. Martins, A. Havt et al. 2003. Renal effects of supernatant from rat
30 peritoneal macrophages activated by microcystin-LR: Role protein mediators. Toxicon.
31 41(3):377-381.
32 Nobre, A.C.L., S.M. Nunes-Monteiro, M.C.S.A. Monteiro et al. 2004. Microcystin-LR promote
33 intestinal secretion of water and electrolytes in rats. Toxicon. 44:555-559.
34 O'Brien, E., S. Altheimer, N. Kreke and D.R. Dietrich. 2003. Teratogenicity of cyanobacterial
35 extracts to FETAX embryos. Toxicol. Sci. 72(Suppl. 1):246.
110 DRAFT - DO NOT CITE OR QUOTE
-------
1 Ohta, T., R. Nishiwaki, J. Yatsunami et al. 1992. Hypersphosphorylation of cytokeratins 8 and
2 18 by microcystin-LR, a new liver tumor promoter, in primary cultured rat hepatocytes.
3 Carcinogenesis. 13(12): 2443-2447.
4 Ohta, T., E. Sueoka, N. lida et al. 1994. Nodularin, a potent inhibitor of protein phosphatases 1
5 and 2A, is a new environmental carcinogen in male F344 rat liver. Cancer Res.
6 54(24):6402-6406.
7 Ohtake, A., M. Shirai, T. Aida et al. 1989. Toxicity of Microcystis species isolated from natural
8 blooms and purification of the toxin. Appl. Environ. Microbiol. 55(12):3202-3207.
9 Oishi, S. andM.F. Watanabe. 1986. Acute toxicity of Microcystis aeruginosa and its
10 cardiovascular effects. Environ. Res. 40(2):518-524.
11 Orr, P.T., GJ. Jones, R.A. Hunter et al. 2001. Ingestion of toxic Microcystis aeumginosa by
12 dairy cattle and implications for microcystin contamination of milk. Toxicon.
13 39(12):1847-1854.
14 Orr, P.T., GJ. Jones, R.A. Hunter and K. Berger. 2003. Exposure of beef cattle to sub-clinical
15 doses of Microcystis aeruginosa: Toxin bioaccumulation, physiological effects and human
16 health risk assessment. Toxicon. 41(5):613-620.
17 Pace, J.G., N.A. Robinson, G.A. Miura et al. 1991. Toxicity and kinetics of [3H]microcystin-LR
18 in isolated perfused rat livers. Toxicol. Appl. Pharmacol. 107(3):391-401.
19 Pican90 M.R., R.M. Scares, V.R. Cagido, S.M.F.O. Azevedo, P.R.M. Rocco and W.A. Zin.
20 2004. Toxicity of cyanobacterial extract containing microcystins to mouse lungs. Braz. J. Med.
21 Biol. Res. 37(8): 1225-1229.
22 Pilotto, L.S., R.M. Douglas, M.D. Burch et al. 1997. Health effects of exposure to
23 cyanobacteria (blue-green algae) during recreational water-related activities. Aust. NZ J. Pub.
24 Health. 21(6):562-566.
25 Pilotto, L.S.,E.V. Klewer, R.D. Davies, M.D. Burch and R.G. Attewell. 1999. Cyanobacterial
26 (blue-green algae) contamination in drinking water and perinatal outcomes. Aust. NZ J. Pub.
27 Health. 23(2): 154-158.
28 Pilotto L., P. Hobson, M.D. Burch, G. Ranmuthugala, R. Attewell and W. Weightman. 2004.
29 Acute skin irritant effects of cyanobacteria (blue-green algae) in healthy volunteers. Aust. NZ J.
30 Pub. Health. 28(3):220-224.
31 Porfmo, Z., M.P. Ribeiro, C.S. Esstevam et al. 1999. Hepatospenomegaly caused by an extract
32 of cyanobacterium Microcystis aeurignosa bloom collected in the Manguaba lagoon, Alagoas-
33 Brazil. Rev. Mi crobiol. 30(3):278-285.
34 Pouria, S, A. de Andrade, J. Barbosa et al. 1998. Fatal microcystin intoxication in
35 haemodialysis unit in Caruaro, Brazil. Lancet. 352:21-26.
111 DRAFT - DO NOT CITE OR QUOTE
-------
1 Rao, P.V.L. and R. Bhattacharya. 1996. The cyanobacterial toxin microcystin-LR induced
2 DNA damage in mouse liver in vivo. Toxicology. 114(l):29-36.
3 Rao, P.V.L., R. Bhattacharya and S.D. Gupta. 1994. Isolation, culture, and toxicity of the
4 cyanobacterium (blue-green alga) Microcystis aeruginosa from a freshwater source in India.
5 Bull. Environ. Contam. Toxicol. 52(6):878-885.
6 Rao, P.V., R. Jayaraj and A.S. Bhaskar. 2004. Protective efficacy and the recovery profile of
7 certain chemoprotectants against lethal poisoning by microcystin-LR in mice. Toxicon.
8 44(7):723-730.
9 Rao P.V.L., N. Gupta, R. Jayaraj et al. 2005. Age-dependent effects on biochemical variables
10 and toxicity induced by cyclic peptide toxin microcystin-LR in mice. Comp. Biochem. Physiol.
11 Part C: Toxicol. Pharmacol. 140(1):11-19.
12 Repavich, W.M., W.C. Sonzogni, J.H. Standridge, R.E. Wedepohl and L.F. Meisner. 1990.
13 Cyanobacteria (blue-green algae) in Wisconsin waters: Acute and chronic toxicity. Water Res.
14 24(2):225-231.
15 Rinehart K.L., M. Namikoshi and B.W. Choi. 1994. Structure and biosynthesis of toxins from
16 blue-green algae (cyanobacteria). J. Appl. Phycol. 6:159-176.
17 Robinson, N.A., G. A. Miura, C.F. Matson et al. 1989. Characterization of chemically tritiated
18 microcystin-LR and its distribution in mice. Toxicon. 27(9): 1035-1042.
19 Robinson, N.A., J.G. Pace, C.F. Matson et al. 1991a. Tissue distribution, excretion and hepatic
20 biotransformation of microcystin-LR in mice. J. Pharmacol. Exp. Therapeut. 256(1): 176-182.
21 Robinson, N.A., C.F. Matson and J.G. Pace. 1991b. Association of microcystin-LR and its
22 biotransformation product with a hepatic-cystolic protein. J. Biochem. Toxicol. 6(3):171-180.
23 Rogers, E.H., E.S. Hunter, V.C. Moseretal. 2005. Potential developmental toxicity of
24 anatoxin-a, a cyanobacterial toxin. J. Appl. Toxicol. 25(6):527-534.
25 Runnegar, M.T.C. and I.R. Falconer. 1982. The in vivo and in vitro biological effects of the
26 peptide hepatotoxin from the blue-green alga Microcystis aeruginosa. S. Afr. J. Sci.
27 78:363-366.
28 Runnegar, M.T.C. and I.R. Falconer. 1986. Effect of toxin from the cyanobacterium
29 Microcystis aeruginosa on ultrastructural morphology and actin polymerization in isolated
30 hepatocytes. Toxicon. 24(2): 109-115.
31 Runnegar, M.T.C., I.R. Falconer and J. Silver. 1981. Deformastion of isolated rat hepatocytes
32 by a peptide hepatoxin from the blue-green alga Microcystis aeruginosa. Nayn-Schmied Arch.
33 Pharmacol. 317(3):268-272.
112 DRAFT - DO NOT CITE OR QUOTE
-------
1 Runnegar, M.T.C., I.R. Falconer, T. Buckley and A.R.B. Jackson. 1986. Lethal potency and
2 tissue distribution of 125I-labelled toxic peptides from the blue-green alga Microcystis
3 aeruginosa. Toxicon. 24(5):506-509.
4 Runnegar, M.T.C., J. Andrews, R.G. Gerdes and I.R. Falconer. 1987. Injury to hepatocytes
5 induced by a peptide toxin from the cyanobacterium Microcystis aeruginosa. Toxicon.
6 25(11):1235-1239.
7 Runnegar, M.T.C., N. Berndt and N. Kaplowitz. 1991. Identification of hepatic protein
8 phosphatases as novel critical targets for the hepatotoxicity of microcystin in vivo. Hepatology.
9 14(4):A159.
10 Runnegar, M.T.C., S. Kong and N. Berndt. 1993. Protein phosphatase inhibition and in vivo
11 hepatotoxicity of microcystins. Am. J. Physiol. 265(2):G224-G230.
12 Runnegar, M.T.C., L.D. Deleve and N. Berndt. 1994. The effects of the protein phosphatase
13 inhibitors and microcystin and calyculin A differ in hepatocytes and hepatic endothelial cells.
14 FASEBJ. 8(7): 1231.
15 Runnegar, M., N. Berndt, S.M. Kong, E.Y. Lee and L. Zhang. 1995a. In vivo and in vitro
16 binding of microcystin to protein phosphatases 1 and 2A. Biochem. Biophys. Res. Commun.
17 216(1):162-169.
18 Runnegar, M.T.C., N. Berndt and N. Kaplowitz. 1995b. Microcystin uptake and inhibition of
19 protein phosphatases: effects of chemoprotectants and self-inhibition in relation to known
20 hepatic transporters. Toxicol. Appl. Pharmacol. 134(2):264-272.
21 Sabour, B., M. Loudiki, B. Oudra et al. 2002. Toxicology of & Microcystis ichthyoblabe
22 waterbloom from Lake Oued Mellah (Morocco). Environ. Toxicol. 17(1):24-31.
23 Schaeffer, D.J., P.B. Malpas and L.L. Barton. 1999. Risk assessment of microcystin in dietary
24 Aphanizomenon flos-aquae. Ecotoxicol. Environ. Safety. 44(1):73-80.
25 Sekijima, M., T. Tsutsumi, T. Yoshida et al. 1999. Enhancement of glutathione S-transferase
26 placental-form positive liver cell foci development by microcystin-LR in aflatoxin Bi-initiated
27 rats. Carcinogenesis. 20(1): 161-165.
28 Sepulveda, M.S., M. Rojas and F. Zambrano. 1992. Inhibitory effect of ^Microcystis sp.
29 (cyanobacteria) toxin on development of preimplantation mouse embryos. Comp. Biochem.
30 Physiol.-C Pharmacol. Toxicol. Endocrinol. 102(3):549-553.
31 Shen, X., G.R. Shaw, G.A. Codd et al. 2003. DNA microarray analysis of gene expression in
32 mice treated with the cyanobacterial toxin, cylindrospermopsin. Can. Tech. Rep. Fish Aquat.
33 Sci. 2498:49-51.
34 Shi, W., H. Zhu, X. Yan and Z. Zhou. 2002. [Sub-acute hepatotoxicity of low doses of
35 microcystins.] Huan Jing Ke Xue. 23(5):47-51. (Chinese)
113 DRAFT - DO NOT CITE OR QUOTE
-------
1 Shirai, M., Y. Takamura, H. Sakuma et al. 1986. Toxicity and delayed type hypersensitivity
2 caused by Microcystis blooms from Lake Kasumigaura. Microbiol. Immunol. 30(7):731 -73 5.
3 Sicihska, P., B. Bukowska, J. Michalowicz and W. Duda. 2006. Damage of cell membrane and
4 antioxidative system in human erythrocytes incubated with microcystin-LR in vitro. Toxicon.
5 47(4):387-397.
6 Siegelman, H.W., W.H. Adams, R.D. Stoner and D.N. Slatkin. 1984. Toxins of Microcystis
7 aeruginosa and their hematological and histopathological effects. In: Seafood Toxins, E.P.
8 Ragelis, Ed. American Chemical Society, Washington, DC. p. 407-413.
9 Sim, A.T.R. and L.M. Mudge. 1993. Protein phosphatase activity in cyanobacteria:
10 consequences for microcystin toxicity analysis. Toxicon. 31(9): 1179-1186.
11 Slatkin, D.N., R.D. Stoner, W.H. Adams et al. 1983. Atypical pulmonary thrombosis caused by
12 a toxic cyanobacterial peptide. Science. 220:1383-1385.
13 Scares, R.M., M. Yuan, J.C. Servaites et al. 2006. Sublethal exposure from microcystins to
14 renal insufficiency patients in Rio de Janeiro, Brazil. Environ. Toxicol. 21(2):95-103.
15 Solow, R., K. Mereish, G.W. Anderson, Jr. et al. 1989. Effect of microcystin-LR on cultured rat
16 endothelial cells. Toxicologist. 9(1): 160.
17 Solter, P.P., G.K. Wollenberg, X. Huang et al. 1998. Prolonged sublethal exposure to the
18 protein phosphatase inhibitor microcystin-LR results in multiple dose-dependent hepatotoxic
19 effects. Toxicol. Sci. 44(l):87-96.
20 Solter, P., Z.L. Lui and R. Guzman. 2000. Decreased hepatic ALT synthesis is an outcome of
21 subchronic microcystin-LR toxicity. Toxicol. Appl. Pharmacol. 164(2):216-220.
22 Stotts, R.R., M. Namikoshi, W.M. Haschek et al. 1993. Structural modifications imparting
23 reduced toxicity in microcystins from Microcystis spp. Toxicon. 31(6):783-789.
24 Stotts, R.R., A.R. Twardock, W.M. Haschek et al. 1997a. Distribution of tritiated
25 dihydromicrocystin in swine. Toxicon. 35(6):937-953.
26 Stotts, R.R., A.R. Twardock, G.D. Koritz et al. 1997b. Toxicokinetics of tritiated
27 dihydromicrocystin-LR in swine. Toxicon. 35(3):455-465.
28 Suzuki, H., M.F. Watanabe, Y.P. Yu et al. 1998. Mutagenicity of microcystin-LR in human
29 RSA cells. Int. J. Mol. Med. 2(1): 109-112.
30 Takahashi, O., S. Oishi and M.F. Watanabe. 1995. Defective blood coagulation is not causative
31 of hepatic haemorrhage induced by microcystin-LR. Pharmacol. Toxicol. 76(4):250-254.
32 Takenaka, S. 2001. Covalent glutathione conjugation to cyanobacterial hepatotoxin
33 microcystin-LR by F344 rat cytosolic and microsomal glutathione S-transferases. Environ.
34 Toxicol. Pharmacol. 9(4):135-139.
114 DRAFT - DO NOT CITE OR QUOTE
-------
1 Tarczynska M., G. Nalecz-Jawecki, M. Brzychcy et al. 2000. The toxicity of cyanobacterial
2 blooms as determined by microbiotests and mouse assays. In: New Microbiotests for Routine
3 Toxicity Screening and Biomonitoring, Persoone et al., Eds. Kluwer Academic/Plenum
4 Publishers, New York, NY. pp. 527-532.
5 Teixeira, M.G.L.C., M.C.N. Costa, V.L.P. Carvalho et al. 1993. Gastroenteritis epidemic in the
6 area of the Itaparica Dam, Bahia, Brazil. Bulletin of PAHO. 27(3):244-253.
7 Teneva, I, R. Mladenov, N. Popov and B. Dzhambazov. 2005. Cytotoxicity and apoptotic
8 effects of microcystin-LR and anatoxin-a in mouse lymphocytes. Folia. Biol. (Praha).
9 51(3):62-67.
10 Theiss, W.C., W.W. Carmichael, M.M. Mullins and R.H. Bruner. 1984. Hepatotoxicity of the
11 blue-green algae (cyanobacteria)Microcystis aeruginosa. Toxicologist. 4(1): 12.
12 Theiss, W.C., W.W. Carmichael, W.W. Mullins et al. 1985. Direct rapid hepatotoxicity of the
13 cyanobacteria Microcystis aeruginosa. Toxicon. 23(1):40-41.
14 Theiss, W.C., W.W. Carmichael, J. Wyman and R. Bruner. 1988. Blood pressure and
15 hepatocellular effects of the cyclic heptapeptide toxin produced by Microsystis aeruginosa strain
16 PCC-7820. Toxicon. 26(7):603-613.
17 Thiel, P. 1994. The South African contribution to studies on the toxic cyanobacteria and their
18 toxins. In: Toxic Cyanobacteria: Current Status of Research and Management. Proceedings of
19 an International Workshop. Adelaide, Australia, March 22-26. D.A. Steffensen and B.C.
20 Nicholson, Ed. Australian Centre for Water Quality Research, Salisbury, Australia, pp. 23-27.
21 Thomspon, W.L. and J.G. Pace. 1992. Substances that protect cultured hepatocytes from the
22 toxic effects of microcystin-LR. Toxicol. In Vitro. 6(6):579-587.
23 Thompson, W.L., M.B. Allen and K.A. Bostian. 1988. The effects of microcystin on
24 monolayers of primary rat hepatocytes. Toxicon. 26(1):44.
25 Toivola, D., I.E. Eriksson and D.L. Brautigan. 1994. Identification of protein phosphatase 2A
26 as the primary target for microcystin-LR in rat liver homogenates. FEES Lett. 344(2-3): 175-
27 180.
28 Toivola, D.M., R.D. Goldman, D.R. Garrod and I.E. Eriksson. 1997. Protein phosphatases
29 maintain the organization and structural interactions of hepatic keratin intermediate filaments. J.
30 CellSci. 110(Pt. l):23-33.
31 Towner, R. A., S. A. Sturgeon and K.E. Hore. 2002. Assessment of in vivo oxidative lipid
32 metabolism following acute microcystin-LR-induced hepatotoxicity in rats. Free Radical Res.
33 36(1):63-71.
34 Turner, P.C., AJ. Gammie, K. Hollinrake and G.A. Codd. 1990. Pneumonia associated with
35 contact with cyanobacteria. Br. Med. J. 300(6737):1440-1441.
115 DRAFT - DO NOT CITE OR QUOTE
-------
1 Ueno, Y., S. Nagata, T. Tsutsumi et al. 1996. Detection of microcystins, a blue-green algal
2 hepatotoxin, in drinking water sampled in Haimen and Fusui, endemic areas of primary liver
3 cancer in China, by highly sensitive immunoassay. Carcinogenesis. 17(6):1317-1321.
4 Ueno, Y., Y. Makita, S. Nagata et al. 1999. No chronic oral toxicity of a low-dose of
5 microcystin-LR, a cyanobacterial hepatoxin, in female Balb/C mice. Environ. Toxicol.
6 14(l):45-55.
7 U.S. EPA. 1986a. Guidelines for the Health Risk Assessment of Chemical Mixtures. Fed. Reg.
8 51(185):34014-34025.
9 U.S. EPA. 1986b. Guidelines for Mutagenicity Risk Assessment. Fed. Reg.
10 51(185):34006-34012.
11 U.S. EPA. 1988. Recommendations for and Documentation of Biological Values for Use in
12 Risk Assessment. U.S. Environmental Protection Agency, Office of Research and Development,
13 Washington, DC. EPA 600/6-87/008. NTIS PB88-179874/AS.
14 U.S. EPA. 1991. Guidelines for Developmental Toxicity RiskAssessment. Fed. Reg.
15 56(234):63798-63826.
16 U.S. EPA. 1994a. Interim Policy for Particle Size and Limit Concentration Issues in Inhalation
17 Toxicity Studies. Fed. Reg. 59(206):53799.
18 U.S. EPA. 1994b. Methods for Derivation of Inhalation Reference Concentrations and
19 Application of Inhalation Dosimetry. U.S. Environmental Protection Agency, Office of
20 Research and Development, Washington, DC. EPA/600/8-90/066F. Available at
21 http://www.epa.gov/iris/backgr-d.htm.
22 U.S. EPA. 1995. Use of the Benchmark Dose Approach in Health Risk Assessment. U.S.
23 Environmental Protection Agency. U.S. Environmental Protection Agency, Office of Research
24 and Development, Washington, DC. EPA/630/R-94/007. Available at
25 http://www.epa.gov/iris/backgr-d.htm.
26 U.S. EPA. 1996. Guidelines for Reproductive Toxicity Risk Assessment. Fed. Reg.
27 61(212):56274-56322.
28 U.S. EPA. 1998a. Guidelines for Neurotoxicity Risk Assessment. Fed. Reg.
29 63(93):26926-26954.
30 U.S. EPA. 1998b. Science Policy Council Handbook: Peer Review. Prepared by the Office of
31 Science Policy, Office of Research and Development, Washington, DC. EPA/100/B-98/001.
32 Available at http://www.epa.gov/iris/backgr-d.htm.
33 U.S. EPA. 2000a. Science Policy Council Handbook: Peer Review, 2nd ed. U.S. Environmental
34 Protection Agency, Office of Research and Development, Office of Science Policy, Washington,
35 DC. EPA/100/B-00/001. Available at http://www.epa.gov/iris/backgr-d.htm.
116 DRAFT - DO NOT CITE OR QUOTE
-------
1 U.S. EPA. 2000b. Science Policy Council Handbook: Risk Characterization. U.S.
2 Environmental Protection Agency, Office of Research and Development, Office of Science
3 Policy, Washington, DC. EPA/100/B-00/002. Available at
4 http://www.epa.gov/iris/backgr-d.htm.
5 U.S. EPA. 2000c. Benchmark Dose Technical Guidance Document: External Review Draft.
6 U.S. Environmental Protection Agency, Office of Research and Development, Office of Science
7 Policy, Washington, DC. EPA/630/R-00/001. Available at
8 http://www.epa.gov/iris/backgr-d.htm.
9 U.S. EPA. 2000d. Supplemental Guidance for Conducting for Health Risk Assessment of
10 Chemical Mixtures. U.S. Environmental Protection Agency, Office of Research and
11 Development, Office of Science Policy, Washington, DC. EPA/630/R-00/002. Available at
12 http://www.epa.gov/iris/backgr-d.htm.
13 U.S. EPA. 2002. A Review of the Reference Dose and Reference Concentration Processes. U.S.
14 Environmental Protection Agency, Risk Assessment Forum, Washington, DC.
15 EPA/630/P-02/0002F. Available at http://www.epa.gov/iris/backgr-d.htm.
16 U.S. EPA. 2005a. Guidelines for Carcinogen Risk Assessment. U.S. Environmental Protection
17 Agency, Risk Assessment Forum, Washington, DC. EPA/630/P-03/001B. Available at
18 http://www.epa.gov/iris/backgr-d.htm.
19 U.S. EPA. 2005b. Supplemental Guidance for Assessing Susceptibility from Early-Life
20 Exposure to Carcinogens. U.S. Environmental Protection Agency, Risk Assessment Forum,
21 Washington, DC. EPA/630/R-03/003F. Available at http://www.epa.gov/iris/backgr-d.htm.
22 U.S. EPA. 2005c. Peer Review Handbook. 3rd edition. Review draft. Science Policy Council,
23 Washington, DC. Available at: http://intranet.epa.gov/ospintra/scipol/prhndbk05.doc.
24 Vesterkvist, P.S. and J.A. Meriluoto. 2003. Interaction between microcystins of different
25 hydrophobicities and lipid monolayers. Toxicon. 41(3): 349-355.
26 WHO (World Health Organization). 1999. Toxic Cyanobacteria in Water: A Guide to their
27 Public Health Consequences, Monitoring, and Management, I. Chorus and J. Bartram, Eds.
28 E&FN Spon, London, UK
29 Wickstrom, M.L., S.A. Khan, W.M. Haschek et al. 1995. Alterations in microtubules,
30 intermediate filaments and microfilaments induced by microcystin-LR in cultured cells.
31 Toxicologic. Pathol. 23(3):326-337.
32 Wolf, H.-U. and C. Frank. 2002. Toxicity assessment of cyanobacerial toxin mixtures.
33 Environ. Toxicol. 17(4): 395-399.
34 Xu, L., P.K.S. Lam, J. Chen et al. 2000. Comparative study on in vitro inhibition of grass carp
35 (Ctenopharyngodon idellus) and mouse protein phosphatases by microcystins. Environ. Toxicol.
36 15(2):71-75.
117 DRAFT - DO NOT CITE OR QUOTE
-------
1 Yea, S.S., Y.I. Yang, W.H. Jang and K.H. Paik. 2001. Microcystin-induced proinflammatory
2 cytokines expression and cell death in human hepatocytes. Hepatology. 34(4 Pt. 2 Suppl.):516A
3 Yoshida, T., Y. Makita, S. Nagata et al. 1997. Acute oral toxicity of microcystin-LR, a
4 cyanobacterial hepatotoxin in mice. Nat. Toxins. 5:91-95.
5 Yoshida, T., Y. Makita, T. Tsutsumi et al. 1998. Immunohistochemical localization of
6 microcystin-LR in the liver of mice: A study on the pathogenesis of microcystin-LR-induced
7 hepatotoxicity. Toxicologic. Pathol. 26(3):411-418.
8 Yoshizawa, S., R. Matsushima, M.F. Watanabe et al. 1990. Inhibition of protein phosphatases
9 by Microcystis and nodularin associated with hepatotoxicity. J. Cancer Res. Clin. Oncol.
10 116(6):609-614.
11 Yu, S.-Z. 1989. Drinking water and primary liver cancer. In: Primary Liver Cancer, Z.Y. Tang,
12 M.C. Wu and S.S. Xia, Ed. China Academic Publishers, New York, NY. p. 30-37 (as cited in
13 Ueno et al., 1996 and Health Canada, 2002).
14 Yu, S.-Z., Z.-Q. Chen, Y.-K. Liu, Z.-Y. Huang and Y.-F. Zhao. 1989. The aflatoxins and
15 contaminated water in the etiological study of primary liver cancer. In: Mycotoxins and
16 Phycotoxins'88, S. Natori, K. Hashimoto and Y. Ueno, Eds. Elsevier, Amsterdam, p. 37-44 (as
17 cited in Ueno et al., 1996 and Health Canada, 2002).
18 Yu, S.Z., X.E. Huang, T. Koide et al. 2002. Hepatitis B and C viruses infection, lifestyle and
19 genetic polymorphisms as risk factors for hepatocellular carcinoma in Haimen, China. Jpn. J.
20 Cancer Res. 93(12): 1287-1292.
21 Zegura, B., B. Sedmak and M. Filipic. 2003. Microcystin-LR induces oxidative DNA damage
22 in human hepatoma cell line HepG2. Toxicon. 41(l):41-48.
23 Zegura, B., T.T. Lah and M. Filipic. 2004. The role of reactive oxygen species in microcystin-
24 LR-induced DNA damage. Toxicology. 200(l):59-68.
25 Zhan, L., M. Sakamoto, M. Sakuraba et al. 2004. Genotoxicity of microcystin-LR in human
26 lymphoblastoid TK6 cells. Mutat. Res. 557(l):l-6.
27 Zhang, Z., S. Kang, C. Chen et al. 2002. [The acute toxic effects of microcystin-LR in SD rats.]
28 Zhonghua Yu Fang YiXueZaZhi. 36(5):295-297. (Chinese)
29 Zhao, J.M. and H.G. Zhu. 2003. [Effects of microcystins on cell cycle and expressions of c-fos
30 andc-jun.] Zhonghua Yu Fang Yi Xue Za Zhi. 37(l):23-25. (Chinese)
31 Zhou, L., H. Yu and K. Chen. 2002. Relationship between microcystin in drinking water and
32 colorectal cancer. Biomed. Environ. Sci. 15(2):166-171.
33 Zhu, Y., X. Zhong, S. Zheng, Z. Ge, Q. Du and S. Zhang. 2005. Transformation of
34 immortalized colorectal crypt cells by microcystin involving constitutive activation of Akt and
35 MAPK cascade. Carcinogenesis. 26(7): 1207-1214.
118 DRAFT - DO NOT CITE OR QUOTE
-------
1 Zuzek, M.C., M. Kosec, J. Mrkum et al. 2003. Microcystin-LR causes reorganization of actin
2 filaments and microtubules in rabbit whole embryo cultures. Toxicol. Lett. 144(Suppl 1):568.
119 DRAFT - DO NOT CITE OR QUOTE
-------
1 APPENDIX A
2
3
4 Contents: Benchmark Dose Modeling Output Files for:
5
6 1) Heinze, 1999 Liver Lesions
7 2) Fawell et al., 1999 Male Chronic Liver Inflammation
8 3) Fawell et al., 1999 Female Chronic Liver Inflammation
9 4) Fawell et al., 1999 Male and Female (combined) Chronic Liver Inflammation
10 5) Heinze, 1999 Relative Liver Weight Changes
11 6) Heinze, 1999 Lactate Dehydrogenase Changes
12 7) Heinze, 1999 Alkaline Phosphatase Changes
13 8) Fawell et al., 1999 Male Alanine Aminotransferase Changes
A-l DRAFT: DO NOT CITE OR QUOTE
-------
1 Heinze,1999. Liver Lesions
2 ====================================================================
3 $Revision: 2.2 $ $Date: 2001/03/14 01:17:00 $
4 Input Data File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.(d)
5 Gnuplot Plotting File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.plt
6 Thu Jul 14 15:03:08 2005
7 ====================================================================
8
9 BMDS MODEL RUN
11
12 The form of the probability function is:
13
14 P[response]= backgrounds- (1-background)*CumGamma[slope*dose,power] ,
15 where CumGamma(.) is the cummulative Gamma distribution function
16
17
18 Dependent variable = COLUMN2
19 Independent variable = COLUMN1
20 Power parameter is restricted as power >=1
21
22 Total number of observations = 3
23 Total number of records with missing values = 0
24 Maximum number of iterations = 250
25 Relative Function Convergence has been set to: le-008
26 Parameter Convergence has been set to: le-008
27
28
29
30 Default Initial (and Specified) Parameter Values
31 Background = 0.0454545
32 Slope = 0.0153804
33 Power = 1.02976
34
35
36 Asymptotic Correlation Matrix of Parameter Estimates
37
38 ( *** The model parameter(s) -Background -Power
39 have been estimated at a boundary point, or have been
40 specified by the user,
41 and do not appear in the correlation matrix )
42
43 Slope
44
45 Slope 1
46
47
48
49 Parameter Estimates
50
51 Variable Estimate Std. Err.
52 Background 0 NA
53 Slope 0.0166997 0.00500499
54 Power 1 NA
55
56 NA - Indicates that this parameter has hit a bound
57 implied by some inequality constraint and thus
58 has no standard error.
59
60
61
62 Analysis of Deviance Table
63
64 Model Log(likelihood) Deviance Test DF P-value
A-2 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Full model -9.98095
Fitted model -10.0255 0.089063 2
Reduced model -20.7944 21.6269 2
AIC: 22.051
Goodness of Fit
0.9564
<.0001
Scaled
Dose Est._Prob. Expected Observed Size Residual
0.0000 0.0000 0.000 0
50.0000 0.5661 5.661 6
150.0000 0.9183 9.183 9
10 0
10 0.2162
10 -0.2115
Chi-square = 0.09 DF = 2 P-value = 0.9553
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 6.30914
BMDL = 3.92229
Gamma Multi-Hit Model with 0.95 Confidence Level
1
0.8
1 0.6
£=
g
| 0.4
0.2
0
Gamma Multi-Hit
^--^^^
<,,//
; ./
/
~ /
/
:
:
:
~
:
BMDL BMD :
0 20 40 60 80 100 120 140
dose
160
31
15:0307/142005
A-3
DRAFT: DO NOT CITE OR QUOTE
-------
1
2 ====================================================================
3 Logistic Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:20 $
4 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY DOCUMENTS\_CYANO
5 TOX REV\HEINZE_LIVER_LESIONS.(d)
6 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
7 DOCUMENTS\_CYANO TOX REV\HEINZE_LIVER_LESIONS.pit
8 Thu Mar 09 11:41:20 2006
9 ====================================================================
10
11 BMDS MODEL RUN
12
13
14 The form of the probability function is:
15
16 P[response] = I/[1+EXP(-intercept-slope*dose)]
17
18
19 Dependent variable = COLUMN2
20 Independent variable = COLUMN1
21 Slope parameter is not restricted
22
23 Total number of observations = 3
24 Total number of records with missing values = 0
25 Maximum number of iterations = 250
26 Relative Function Convergence has been set to: le-008
27 Parameter Convergence has been set to: le-008
28
29
30
31 Default Initial Parameter Values
32 background = 0 Specified
33 intercept = -2.28075
34 slope = 0.0300564
35
36
37 Asymptotic Correlation Matrix of Parameter Estimates
38
39 ( *** The model parameter(s) -background
40 have been estimated at a boundary point, or have been
41 specified by the user,
42 and do not appear in the correlation matrix )
43
44 intercept slope
45
46 intercept 1 -0.75
47
48 slope -0.75 1
49
50
51
52 Parameter Estimates
53
54 Variable Estimate Std. Err.
55 intercept -2.02314 0.772069
56 slope 0.0344016 0.0123743
57
58
59
60 Analysis of Deviance Table
61
62 Model Log(likelihood) Deviance Test DF P-value
63 Full model -9.98095
64 Fitted model -12.1529 4.34392 1 0.03714
A-4 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
_L\7
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Reduced model
AIC:
Dose Est
0.0000 0
50.0000 0
150.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
-20.7944 21.6269 2 <.
28.3058
Goodness of Fit
._Prob. Expected Observed Size
.1168 1.168 0 10
.4248 4.248 6 10
.9584 9.584 9 10
3.43 DF = 1 P-value = 0.0639
Computation
0.1
Extra risk
0.95
19.4327
11.4032
.0001
Scaled
Residual
-1.15
1.121
-0.9248
30
31
32
Logistic Model with 0.95 Confidence Level
0.6
0.4
Logistic
BMDL
BMD
40
60
80
dose
100
120
140
160
11:41 03/092006
A-5
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Logistic Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:20 $
3 Input Data File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.(d)
4 Gnuplot Plotting File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.plt
5 Thu Jul 14 15:04:46 2005
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
14
15
16 Dependent variable = COLUMN2
17 Independent variable = COLUMN1
18 Slope parameter is restricted as slope >= 1
19
20 Total number of observations = 3
21 Total number of records with missing values = 0
22 Maximum number of iterations = 250
23 Relative Function Convergence has been set to: le-008
24 Parameter Convergence has been set to: le-008
25
26
27
28 User has chosen the log transformed model
29
30
31 Default Initial Parameter Values
32 background = 0
33 intercept = -5.97477
34 slope = 1.63093
35
36
37 Asymptotic Correlation Matrix of Parameter Estimates
38
39 ( *** The model parameter(s) -background
40 have been estimated at a boundary point, or have been
41 specified by the user,
42 and do not appear in the correlation matrix )
43
44 intercept slope
45
46 intercept 1 -0.99
47
48 slope -0.99 1
49
50
51
52 Parameter Estimates
53
54 Variable Estimate Std. Err.
55 background 0 NA
56 intercept -5.97477 4.77026
57 slope 1.63093 1.12507
58
59 NA - Indicates that this parameter has hit a bound
60 implied by some inequality constraint and thus
61 has no standard error.
62
63 Warning: Likelihood for the fitted model larger than the Likelihood for the
64 full model.
A-6 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
_L /
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
0.0000 0
50.0000 0
150.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Analysis of Deviance Table
Log (likelihood) Deviance Test DF P-value
-9.98095
-9.98095 -3.55271e-015 1 -1
-20.7944 21.6269 2 <.0001
23.9619
Goodness of Fit
Scaled
._Prob. Expected Observed Size Residual
.0000 0.000 0 10 0
.6000 6.000 6 10 -3.44e-015
.9000 9.000 9 10 -1.123e-014
0.00 DF = 1 P-value = 1.0000
Computation
0.1
Extra risk
0.95
10.137
1.23783
A-7
DRAFT: DO NOT CITE OR QUOTE
-------
3 0.6
0.4
Log-Logistic Model with 0.95 Confidence Level
Log-Logistic
BMDL
BMD
20 40
60
80
dose
100 120 140 160
15:0407/142005
A-8 DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Multistage Model. $Revision: 2.1 $ $Date: 2000/08/21 03:38:21 $
3 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY DOCUMENTS\_CYANO
4 TOX REV\HEINZE_LIVER_LESIONS.(d)
5 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
6 DOCUMENTS\_CYANO TOX REV\HEINZE_LIVER_LESIONS.pit
7 Thu Mar 09 11:37:17 2006
8 ====================================================================
9
10 BMDS MODEL RUN
12
13 The form of the probability function is:
14
15 P[response] = background + (1-background)*[1-EXP(
16 -betal*dose^l-beta2*dose^2) ]
17
18 The parameter betas are restricted to be positive
19
20
21 Dependent variable = COLUMN2
22 Independent variable = COLUMN1
23
24 Total number of observations = 3
25 Total number of records with missing values = 0
26 Total number of parameters in model = 3
27 Total number of specified parameters = 0
28 Degree of polynomial = 2
29
30
31 Maximum number of iterations = 250
32 Relative Function Convergence has been set to: le-008
33 Parameter Convergence has been set to: le-008
34
35
36
37 Default Initial Parameter Values
38 Background = 0.0617654
39 Beta(l) = 0.015138
40 Beta (2) = 0
41
42
43 Asymptotic Correlation Matrix of Parameter Estimates
44
45 ( *** The model parameter(s) -Background -Beta(2)
46 have been estimated at a boundary point, or have been specified by the
47 user, and do not appear in the correlation matrix )
48
49 Beta(l)
50
51 Beta(l) 1
52
53
54
55 Parameter Estimates
56
57 Variable Estimate Std. Err.
58 Background 0 NA
59 Beta(l) 0.0166997 0.00582148
60 Beta(2) 0 NA
61
62 NA - Indicates that this parameter has hit a bound
63 implied by some inequality constraint and thus
64 has no standard error.
A-9 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
_L /
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
i: 1
0.0000 0.
i: 2
50.0000 0.
i: 3
150.0000 0.
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Analysis of Deviance Table
Log (likelihood) Deviance Test DF
-9.98095
-10.0255 0.089063 2
-20.7944 21.6269 2
22.051
Goodness of Fit
._Prob. Expected Observed
0000 0.000 0
5661 5.661 6
9183 9.183 9
0.09 DF = 1 P-value =
Computation
0.1
Extra risk
0.95
6.30914
3.92229
P-value
0.9564
<.0001
Size Chi^2 Res.
10 -1.000
10 0.138
10 -0.244
0.7623
A-10
DRAFT: DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
0.6
o 0.4
0.2
Multistage
BiylDL BMD
0
20 40 60 80 100 120 140
dose
160
11:37 03/092006
A-l 1 DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Multistage Model. $Revision: 2.1 $ $Date: 2000/08/21 03:38:21 $
3 Input Data File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.(d)
4 Gnuplot Plotting File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.plt
5 Thu Jul 14 15:05:31 2005
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = background + (1-background)*[1-EXP(
14 -betal*dose^l) ]
15
16 The parameter betas are restricted to be positive
17
18
19 Dependent variable = COLUMN2
20 Independent variable = COLUMN1
21
22 Total number of observations = 3
23 Total number of records with missing values = 0
24 Total number of parameters in model = 2
25 Total number of specified parameters = 0
26 Degree of polynomial = 1
27
28
29 Maximum number of iterations = 250
30 Relative Function Convergence has been set to: le-008
31 Parameter Convergence has been set to: le-008
32
33
34
35 Default Initial Parameter Values
36 Background = 0.0617654
37 Beta(l) = 0.015138
38
39
40 Asymptotic Correlation Matrix of Parameter Estimates
41
42 ( *** The model parameter(s) -Background
43 have been estimated at a boundary point, or have been
44 specified by the user,
45 and do not appear in the correlation matrix )
46
47 Beta(l)
48
49 Beta(l) 1
50
51
52
53 Parameter Estimates
54
55 Variable Estimate Std. Err.
56 Background 0 NA
57 Beta(l) 0.0166997 0.00582148
58
59 NA - Indicates that this parameter has hit a bound
60 implied by some inequality constraint and thus
61 has no standard error.
62
63
64
A-12 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
IT-
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
i : 1
0.0000 0.
i: 2
50.0000 0.
i: 3
150.0000 0.
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Analysis of Deviance Table
Log (likelihood) Deviance Test DF
-9.98095
-10.0255 0.089063 2
-20.7944 21.6269 2
22.051
Goodness of Fit
. Prob . Expected Observed
0000 0.000 0
5661 5.661 6
9183 9.183 9
0.09 DF = 2 P-value =
Computation
0.1
Extra risk
0.95
6.30914
3.92229
P-value
0.9564
<.0001
Size Chi^2 Res.
10 0.000
10 0.138
10 -0.244
0.9553
Multistage Model with 0.95 Confidence Level
0.8
Multistage
40
60
80
dose
100
120
140
160
37
15:0507/142005
A-13
DRAFT: DO NOT CITE OR QUOTE
-------
A-14 DRAFT: DO NOT CITE OR QUOTE
-------
1
2 ====================================================================
3 Probit Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:53 $
4 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY DOCUMENTS\_CYANO
5 TOX REV\HEINZE_LIVER_LESIONS.(d)
6 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
7 DOCUMENTS\_CYANO TOX REV\HEINZE_LIVER_LESIONS.pit
8 Thu Mar 09 11:49:47 2006
9 ====================================================================
10
11 BMDS MODEL RUN
12
13
14 The form of the probability function is:
15
16 P[response] = CumNorm(Intercept+Slope*Dose) ,
17
18 where CumNorm(.) is the cumulative normal distribution function
19
20
21 Dependent variable = COLUMN2
22 Independent variable = COLUMN1
23 Slope parameter is not restricted
24
25 Total number of observations = 3
26 Total number of records with missing values = 0
27 Maximum number of iterations = 250
28 Relative Function Convergence has been set to: le-008
29 Parameter Convergence has been set to: le-008
30
31
32
33 Default Initial (and Specified) Parameter Values
34 background = 0 Specified
35 intercept = -1.57069
36 slope = 0.0201403
37
38
39 Asymptotic Correlation Matrix of Parameter Estimates
40
41 ( *** The model parameter(s) -background
42 have been estimated at a boundary point, or have been
43 specified by the user,
44 and do not appear in the correlation matrix )
45
46 intercept slope
47
48 intercept 1 -0.75
49
50 slope -0.75 1
51
52
53
54 Parameter Estimates
55
56 Variable Estimate Std. Err.
57 intercept -1.209 0.433188
58 slope 0.0190194 0.00553748
59
60
61
62 Analysis of Deviance Table
63
64 Model Log(likelihood) Deviance Test DF P-value
A-15 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
Full model
Fitted model
Reduced model
AIC:
-9.98095
-12.2154 4.46887
-20.7944 21.6269
28.4308
Goodness of Fit
0.03452
<.0001
Dose
0.0000
50.0000
150.0000
Est ._Prob.
0.1133
0.3982
0.9499
Expected
1.133
3.982
9.499
Observed
0
6
9
Size
10
10
10
Scaled
Residual
-1.131
1.304
-0.7234
Chi-square =
3.50
DF = 1
P-value = 0.0613
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 19.6901
BMDL = 12.3138
Probit Model with 0.95 Confidence Level
S 0.6
0.4
Probit
BMDL
BMD
40 60 80 100 120 140
dose
160
33
11:4903/092006
A-16
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Probit Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:53 $
3 Input Data File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.(d)
4 Gnuplot Plotting File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.plt
5 Thu Jul 14 15:06:15 2005
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = Background
14 + (1-Background) * CumNorm(Intercept+Slope*Log(Dose)),
15
16 where CumNorm(.) is the cumulative normal distribution function
17
18
19 Dependent variable = COLUMN2
20 Independent variable = COLUMN1
21 Slope parameter is restricted as slope >= 1
22
23 Total number of observations = 3
24 Total number of records with missing values = 0
25 Maximum number of iterations = 250
26 Relative Function Convergence has been set to: le-008
27 Parameter Convergence has been set to: le-008
28
29
30
31 User has chosen the log transformed model
32
33
34 Default Initial (and Specified) Parameter Values
35 background = 0
36 intercept = -3.68466
37 slope = 1
38
39
40 Asymptotic Correlation Matrix of Parameter Estimates
41
42 ( *** The model parameter(s) -background -slope
43 have been estimated at a boundary point, or have been
44 specified by the user,
45 and do not appear in the correlation matrix )
46
47 intercept
48
49 intercept 1
50
51
52
53 Parameter Estimates
54
55 Variable Estimate Std. Err.
56 background 0 NA
57 intercept -3.68338 0.323658
58 slope 1 NA
59
60 NA - Indicates that this parameter has hit a bound
61 implied by some inequality constraint and thus has no standard error.
62
63
64
A-17 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
0.0000 0
50.0000 0
150.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Analysis of Deviance Table
Log (likelihood) Deviance Test DF P-value
-9.98095
-9.98638 0.0108688 2 0.9946
-20.7944 21.6269 2 <.0001
21.9728
Goodness of Fit
Scaled
. Prob . Expected Observed Size Residual
.0000 0.000 0 10 0
.5904 5.904 6 10 0.06155
.9078 9.078 9 10 -0.08513
0.01 DF = 2 P-value = 0.9945
Computation
0.1
Extra risk
0.95
11.0433
6.37572
Probit Model with 0.95 Confidence Level
0.6
0.4
0.2
Probit
BMDL BMD
0 20 40 60 80 100 120 140 160
dose
36
37
15:0607/142005
A-18
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Quantal Linear Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
Input Data File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.(d)
Gnuplot Plotting File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.plt
Thu Jul 14 15:07:01 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-slope*dose)]
Dependent variable = COLUMN2
Independent variable = COLUMN1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.0454545
Slope = 0.0129727
Power = 1 Specified
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Background -Power
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
Slope
Slope
1
Variable
Background
Slope
Parameter Estimates
Estimate
0
0.0166997
Std. Err.
NA
0.00500498
NA -
Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Deviance Test DF
Log(likelihood)
-9.98095
-10.0255
-20.7944
22.051
P-value
0.089063
21.6269
0.9564
<.0001
A-19
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
/
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Dose Est
0.0000 0
50.0000 0
150.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
. Prob . Expected Observed Size
.0000 0.000 0 10
.5661 5.661 6 10
.9183 9.183 9 10
0.09 DF = 2 P-value = 0.9553
Computation
0.1
Extra risk
0.95
6.30914
3.92229
Scaled
Residual
0
0.2162
-0.2115
26
27
Quantal Linear Model with 0.95 Confidence Level
0.6
0.4
0.2
Quantal Linear
40
60
80
dose
100
120
140
160
15:07 07/142005
A-20
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Quantal Quadratic Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
Input Data File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.(d)
Gnuplot Plotting File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.plt
Thu Jul 14 15:07:40 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-slope*dose^2)]
Dependent variable = COLUMN2
Independent variable = COLUMN1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.0454545
Slope = 8.64849e-005
Power = 2 Specified
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Background -Power
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
Slope
Slope
1
Variable
Background
Slope
Parameter Estimates
Estimate
0
0.000171114
Std. Err.
NA
5.74454e-005
NA -
Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Deviance Test DF
Log(likelihood)
-9.98095
-12.0873
-20.7944
26 . 1745
P-value
4.21262
21.6269
0.1217
<.0001
A-21
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
/
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Dose Est
0.0000 0
50.0000 0
150.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
. Prob . Expected Observed Size
.0000 0.000 0 10
.3480 3.480 6 10
.9787 9.787 9 10
5.77 DF = 2 P-value = 0.0558
Computation
0.1
Extra risk
0.95
24.814
19.01
Scaled
Residual
0
1.673
-1 .725
26
27
Quantal Quadratic Model with 0.95 Confidence Level
0.8
1 0.6
•S
c
.2
1 0.4
L_
0.2
Quantal Quadratic
BMDL
BMD
20
40
60
80
dose
100
120
140
160
15:07 07/142005
A-22
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Weibull Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
Input Data File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.(d)
Gnuplot Plotting File: C:\BMDS\HEINZE_MOD_AND_INT_WITH_HEMORRHAGE.plt
Thu Jul 14 15:08:20 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-slope*dose^power)]
Dependent variable = COLUMN2
Independent variable = COLUMN1
Power parameter is restricted as power >=1
Total number of observations = 3
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.0454545
Slope = 0.0129727
Power = 1
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Background -Power
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
Slope
Slope
1
Variable
Background
Slope
Power
Parameter Estimates
Estimate
0
0.0166997
1
Std. Err.
NA
0.00500498
NA
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Model
Full model
Fitted model
Analysis of Deviance Table
Log(likelihood) Deviance Test DF
-9.98095
-10.0255 0.089063 2
P-value
0.9564
A-23
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
_L\7
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Reduced model
AIC:
Dose Est
0.0000 0
50.0000 0
150.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
-20.7944 21.6269 2 <.
22.051
Goodness of Fit
._Prob. Expected Observed Size
.0000 0.000 0 10
.5661 5.661 6 10
.9183 9.183 9 10
0.09 DF = 2 P-value = 0.9553
Computation
0.1
Extra risk
0.95
6.30914
3.92229
.0001
Scaled
Residual
0
0.2162
-0.2115
31
32
Weibull Model with 0.95 Confidence Level
3 0.6
0.4
Weibull
40 60 80 100 120 140
dose
160
15:0807/142005
A-24
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Fawell et al.,1999 Male Chronic Inflammation
$Revision: 2.2 $ $Date: 2001/03/14 01:17:00 $
Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:45:24 2005
BMDS MODEL RUN
The form of the probability function is:
P[response]= background+(1-background)*CumGamma[slope*dose,power],
where CumGamma(.) is the cummulative Gamma distribution function
Dependent variable = COLUMN2
Independent variable = COLUMN1
Power parameter is restricted as power >=1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.09375
Slope = 0.00193589
Power = 1.3
Asymptotic Correlation Matrix of Parameter Estimates
Background Slope Power
Background 1 0.037 0.041
Slope 0.037 1 1
Power 0.041 1 1
Variable
Background
Slope
Power
Model
Full model
Fitted model
Reduced model
Parameter Estimates
Estimate Std. Err.
0.0999826 0.054818
0.0276638 0.488219
8.09035 116.456
Analysis of Deviance Table
Log(likelihood)
-18.2628
-18.4512
-39.4295
Deviance Test DF
P-value
0.376765
42.3333
0.5393
<.0001
A-25
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
«
o
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
AIC:
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 1
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
42.9024
Goodness of Fit
. Prob . Expected Observed Size
.1000 1.500 1 15
.1000 1.500 2 15
.2667 4.001 4 15
.0000 15.000 15 15
0.37 DF = 1 P-value = 0.5429
Computation
0.1
Extra risk
0.95
170.825
53.1183
Scaled
Residual
-0.4301
0.4304
-0.0002949
0.00692
30
31
Gamma Multi-Hit Model with 0.95 Confidence Level
£ 0.6
0.4
0.2
Gamma Multi-Hit
BMDL
BMD
200
400 600
dose
800
1000
15:45 12/282005
A-26
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Logistic Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:20 $
3 Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
5 Thu Mar 09 11:56:58 2006
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = I/[1+EXP(-intercept-slope*dose)]
14
15
16 Dependent variable = COLUMN2
17 Independent variable = COLUMN1
18 Slope parameter is not restricted
19
20 Total number of observations = 4
21 Total number of records with missing values = 0
22 Maximum number of iterations = 250
23 Relative Function Convergence has been set to: le-008
24 Parameter Convergence has been set to: le-008
25
26
27
28 Default Initial Parameter Values
29 background = 0 Specified
30 intercept = -2.07771
31 slope = 0.00552538
32
33
34 Asymptotic Correlation Matrix of Parameter Estimates
35
36 ( *** The model parameter(s) -background
37 have been estimated at a boundary point, or have been
38 specified by the user,
39 and do not appear in the correlation matrix )
40
41 intercept slope
42
43 intercept 1 -0.72
44
45 slope -0.72 1
46
47
48
49 Parameter Estimates
50
51 Variable Estimate Std. Err.
52 intercept -2.49527 0.612166
53 slope 0.00805129 0.00337645
54
55
56
57 Analysis of Deviance Table
58
59 Model Log(likelihood) Deviance Test DF P-value
60 Full model -18.2628
61 Fitted model -18.4278 0.329996 2 0.8479
62 Reduced model -39.4295 42.3333 3 <.0001
63
64 AIC: 40.8556
A-27 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
/
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
Goodness of Fit
c
Dose Est . Prob . Expected Observed Size I
0.0000 0.0762 1.143 1 15
40.0000 0.1022 1.533 2 15
200.0000 0.2921 4.382 4 15
1000.0000 0.9962 14.942 15 15
Chi-square = 0.28 DF = 2 P-value = 0.8680
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 111.719
BMDL = 70.686
Scaled
Residual
-0.139
0.3983
-0.217
0.2408
28
29
Logistic Model with 0.95 Confidence Level
0.6
0.4
Logistic
BMDL
BMD
200
400 600
dose
800
1000
11:5603/092006
A-28
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Logistic Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:20 $
Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:47:11 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
Dependent variable = COLUMN2
Independent variable = COLUMN1
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.0666667
intercept = -10.0178
slope = 1.86367
Asymptotic Correlation Matrix of Parameter Estimates
background intercept slope
background 1 0.0022 -0.0023
intercept 0.0022 1 -1
slope -0.0023 -1 1
Variable
background
intercept
slope
Model
Full model
Fitted model
Reduced model
AIC:
Parameter Estimates
Estimate
0.0999997
-65.9875
12.1748
Std. Err.
0.0547725
5466.64
1031.77
Analysis of Deviance Table
Deviance Test DF
Log(likelihood)
-18.2628
-18.4512
-39.4295
42.9024
P-value
0.376844
42.3333
0.5393
<.0001
A-29
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
/
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 1
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
. Prob . Expected Observed Size
.1000 1.500 1 15
.1000 1.500 2 15
.2667 4.000 4 15
.0000 15.000 15 15
0.37 DF = 1 P-value = 0.5428
Computation
0.1
Extra risk
0.95
188.582
94.487
Scaled
Residual
-0.4303
0.4303
-4.206e-005
0.0004285
Log-Logistic Model with 0.95 Confidence Level
0.6
0.4
Log-Logistic
BMDL
BMD
200
400 600
dose
800
1000
27 15:4712/282005
28
A-30
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Multistage Model. $Revision: 2.1 $ $Date: 2000/08/21 03:38:21 $
Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
Thu Mar 09 11:59:44 2006
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l-beta2*dose^2-beta3*dose^3)]
The parameter betas are restricted to be positive
Dependent variable = COLUMN2
Independent variable = COLUMN1
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 0
Beta(2) = 0
Beta(3) = 1.00264e+011
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2)
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
Background
Background 1
Beta(l) -0.65
Beta(3) 0.14
Variable
Background
Beta(l)
Beta(2)
Beta(3)
Beta(l)
-0.65
1
-0.42
Parameter
Estimate
0.0782274
0.000933123
0
7.10432e-009
Beta(3)
0.14
-0.42
1
Estimates
Std. Err
0.200883
0.00219636
NA
1.57018e-008
A-31
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
i: 1
0.0000 0.
i: 2
40.0000 0.
i: 3
200.0000 0.
i: 4
1000.0000 0.
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Analysis of Deviance Table
Log (likelihood) Deviance Test DF
-18.2628
-18.3175 0.10941 1
-39.4295 42.3333 3
42.635
Goodness of Fit
._Prob. Expected Observed
0782 1.173 1
1124 1.686 2
2774 4.161 4
9997 14.996 15
0.11 DF = 1 P-value =
Computation
0.1
Extra risk
0.95
104.279
37.4815
P-value
0.7408
<.0001
Size Chi^2 Res.
15 -0.160
15 0.210
15 -0.054
15 1.000
0.7438
A-32
DRAFT: DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
o 0.6
0.4
Multistage
BMDL 3MD
200
400 600
dose
800
1000
11:5903/092006
A-33 DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Multistage Model. $Revision: 2.1 $ $Date: 2000/08/21 03:38:21 $
3 Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
5 Thu Mar 09 12:00:48 2006
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = background + (1-background)*[1-EXP(
14 -betal*dose^l-beta2*dose^2) ]
15
16 The parameter betas are restricted to be positive
17
18
19 Dependent variable = COLUMN2
20 Independent variable = COLUMN1
21
22 Total number of observations = 4
23 Total number of records with missing values = 0
24 Total number of parameters in model = 3
25 Total number of specified parameters = 0
26 Degree of polynomial = 2
27
28
29 Maximum number of iterations = 250
30 Relative Function Convergence has been set to: le-008
31 Parameter Convergence has been set to: le-008
32
33
34
35 Default Initial Parameter Values
36 Background = 0
37 Beta(l) = 0
38 Beta(2) = 1.01264e+014
39
40
41 Asymptotic Correlation Matrix of Parameter Estimates
42
43 Background Beta(l) Beta (2)
44
45 Background 1 -0.65 0.33
46
47 Beta(l) -0.65 1 -0.75
48
49 Beta(2) 0.33 -0.75 1
50
51
52
53 Parameter Estimates
54
55 Variable Estimate Std. Err.
56 Background 0.0889744 0.205309
57 Beta(l) 8.80277e-005 0.00292558
58 Beta(2) 5.76916e-006 6.80122e-006
59
60
61
62 Analysis of Deviance Table
63
64 Model Log(likelihood) Deviance Test DF P-value
A-34 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
1 1
_L J.
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Full model
Fitted model
Reduced model
AIC:
Dose Est
i: 1
0.0000 0.
i: 2
40.0000 0.
i: 3
200.0000 0.
i: 4
1000.0000 0.
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
-18.2628
-18.4529 0.380206 1
-39.4295 42.3333 3
42.9058
Goodness of Fit
._Prob. Expected Observed Size
0890 1.335 1 15
1005 1.508 2 15
2893 4.340 4 15
9974 14.961 15 15
0.35 DF = 1 P-value = 0.5556
Computation
0.1
Extra risk
0.95
127.726
39.8667
0.5375
<.0001
Chi^2 Res.
-0.275
0.363
-0.110
1.003
Multistage Model with 0.95 Confidence Level
o 0.6
0.4
Multistage
BMDL
BMD
200
400
600
800
1000
36
dose
12:0003/092006
A-35
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Multistage Model. $Revision: 2.1 $ $Date: 2000/08/21 03:38:21 $
Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:47:43 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l) ]
The parameter betas are restricted to be positive
Dependent variable = COLUMN2
Independent variable = COLUMN1
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 2
Total number of specified parameters = 0
Degree of polynomial = 1
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0
Beta(l) = 1.04991e+017
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l)
Background 1 -0.37
Beta(l) -0.37 1
Parameter Estimates
Variable
Background
Beta(l)
Model
Full model
Fitted model
Reduced model
AIC:
Estimate
0.048937
0.0025684
Std. Err.
0.16314
0.000904753
Analysis of Deviance Table
Deviance Test DF
Log(likelihood)
-18.2628
-20.3189
-39.4295
44.6378
P-value
4.11219
42.3333
0.128
<.0001
A-36
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
Dose Est
i: 1
0.0000 0.
i: 2
40.0000 0.
i: 3
200.0000 0.
i: 4
1000.0000 0.
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
._Prob. Expected
0489 0.734
1418 2.127
4310 6.465
9271 13.906
2 . 94 DF = 2
Computation
0.1
Extra risk
0.95
41.0219
25.943
Observed Size Chi^2 Res.
1 15 0.381
2 15 -0.070
4 15 -0.670
15 15 1.079
P-value = 0.2298
Multistage Model with 0.95 Confidence Level
o 0.6
0.4
Multistage
200
400
600
800
1000
30
31
dose
15:47 12/282005
A-37
DRAFT: DO NOT CITE OR QUOTE
-------
1 ===================================================================
2 Probit Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:53 $
3 Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
5 Thu Mar 09 12:02:26 2006
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P [response] = CumNorm(Intercept+Slope*Dose) ,
14
15 where CumNorm(.) is the cumulative normal distribution function
16
17
18 Dependent variable = COLUMN2
19 Independent variable = COLUMN1
20 Slope parameter is not restricted
21
22 Total number of observations = 4
23 Total number of records with missing values = 0
24 Maximum number of iterations = 250
25 Relative Function Convergence has been set to: le-008
26 Parameter Convergence has been set to: le-008
27
28
29
30 Default Initial (and Specified) Parameter Values
31 background = 0 Specified
32 intercept = -1.32259
33 slope = 0.00345221
34
35
36 Asymptotic Correlation Matrix of Parameter Estimates
37
38 ( *** The model parameter(s) -background
39 have been estimated at a boundary point, or have been
40 specified by the user,
41 and do not appear in the correlation matrix )
42
43 intercept slope
44
45 intercept 1 -0.65
46
47 slope -0.65 1
48
49
50
51 Parameter Estimates
52
53 Variable Estimate Std. Err.
54 intercept -1.42508 0.306506
55 slope 0.00435347 0.00171942
56
57
58
59 Analysis of Deviance Table
60
61 Model Log(likelihood) Deviance Test DF P-value
62 Full model -18.2628
63 Fitted model -18.3773 0.229031 2 0.8918
64 Reduced model -39.4295 42.3333 3 <.0001
A-38 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
AIC:
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
40.7546
Goodness of Fit
. Prob . Expected Observed Size
.0771 1.156 1 15
.1055 1.582 2 15
.2897 4.345 4 15
.9983 14.974 15 15
0.21 DF = 2 P-value = 0.9002
Computation
0.1
Extra risk
0.95
107.589
66.4468
Scaled
Residual
-0.151
0.3512
-0.1963
0.16
Probit Model with 0.95 Confidence Level
£ 0.6
0.4
0.2
Probit
BMDL BMD
200
400
600
800
1000
31
32
dose
12:02 03/092006
A-39
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Probit Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:53 $
3 Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
5 Wed Dec 28 15:48:13 2005
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = Background
14 + (1-Background) * CumNorm(Intercept+Slope*Log(Dose)),
15
16 where CumNorm(.) is the cumulative normal distribution function
17
18
19 Dependent variable = COLUMN2
20 Independent variable = COLUMN1
21 Slope parameter is restricted as slope >= 1
22
23 Total number of observations = 4
24 Total number of records with missing values = 0
25 Maximum number of iterations = 250
26 Relative Function Convergence has been set to: le-008
27 Parameter Convergence has been set to: le-008
28
29
30
31 User has chosen the log transformed model
32
33
34 Default Initial (and Specified) Parameter Values
35 background = 0.0666667
36 intercept = -5.60926
37 slope = 1.03389
38
39
40 Asymptotic Correlation Matrix of Parameter Estimates
41
42 background intercept slope
43
44 background 1 0.0018 -0.002
45
46 intercept 0.0018 1 -1
47
48 slope -0.002 -1 1
49
50
51
52 Parameter Estimates
53
54 Variable Estimate Std. Err.
55 background 0.1 0.0547723
56 intercept -22.1534 1087.37
57 slope 4.01215 205.229
58
59
60
61 Analysis of Deviance Table
62
63 Model Log(likelihood) Deviance Test DF P-value
64 Full model -18.2628
A-40 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
9
10
1 1
_L J.
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
Fitted model
Reduced model
AIC:
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 1
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
-18.4512 0.376844 1
-39.4295 42.3333 3
42.9024
Goodness of Fit
._Prob. Expected Observed
.1000 1.500 1
.1000 1.500 2
.2667 4.000 4
.0000 15.000 15
0.37 DF = 1 P-value =
Computation
0.1
Extra risk
0.95
181.665
95.2612
0.5393
<.0001
Scaled
Size Residual
15 -0.4303
15 0.4303
15 -1.174e-005
15 0.0004249
0.5428
Probit Model with 0.95 Confidence Level
31
32
0.8
B 0.6
I
c
.2
S 0.4
0.2
Probit
BMDL
15:48 12/282005
BMD
200
400 600
dose
800
1000
A-41
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Quantal Linear Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
3 Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
5 Wed Dec 28 15:48:41 2005
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = background + (1-background)*[1-EXP(-slope*dose)]
14
15
16 Dependent variable = COLUMN2
17 Independent variable = COLUMN1
18
19 Total number of observations = 4
20 Total number of records with missing values = 0
21 Maximum number of iterations = 250
22 Relative Function Convergence has been set to: le-008
23 Parameter Convergence has been set to: le-008
24
25
26
27 Default Initial (and Specified) Parameter Values
28 Background = 0.09375
29 Slope = 0.0033673
30 Power = 1 Specified
31
32
33 Asymptotic Correlation Matrix of Parameter Estimates
34
35 ( *** The model parameter(s) -Power
36 have been estimated at a boundary point, or have been
37 specified by the user,
38 and do not appear in the correlation matrix )
39
40 Background Slope
41
42 Background 1 -0.23
43
44 Slope -0.23 1
45
46
47
48 Parameter Estimates
49
50 Variable Estimate Std. Err.
51 Background 0.048937 0.0439067
52 Slope 0.0025684 0.000746108
53
54
55
56 Analysis of Deviance Table
57
58 Model Log(likelihood) Deviance Test DF P-value
59 Full model -18.2628
60 Fitted model -20.3189 4.11219 2 0.128
61 Reduced model -39.4295 42.3333 3 <.0001
62
63 AIC: 44.6378
64
A-42 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
._Prob. Expected
.0489 0.734
.1418 2.127
.4310 6.465
.9271 13.906
2.94 DF = 2
Computation
0.1
Extra risk
0.95
41.0219
25.943
Scaled
Observed Size Residual
1 15 0.3183
2 15 -0.09393
4 15 -1.285
15 15 1.086
P-value = 0.2298
Quantal Linear Model with 0.95 Confidence Level
0.6
0.4
Quantal Linear
BMDL BMD
200
400
600
800
1000
27
28
dose
15:48 12/282005
A-43
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Quantal Quadratic Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
3 Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
5 Wed Dec 28 15:49:14 2005
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = background + (1-background)*[1-EXP(-slope*dose^2)]
14
15
16 Dependent variable = COLUMN2
17 Independent variable = COLUMN1
18
19 Total number of observations = 4
20 Total number of records with missing values = 0
21 Maximum number of iterations = 250
22 Relative Function Convergence has been set to: le-008
23 Parameter Convergence has been set to: le-008
24
25
26
27 Default Initial (and Specified) Parameter Values
28 Background = 0.09375
29 Slope = 3.3673e-006
30 Power = 2 Specified
31
32
33 Asymptotic Correlation Matrix of Parameter Estimates
34
35 ( *** The model parameter(s) -Power
36 have been estimated at a boundary point, or have been
37 specified by the user,
38 and do not appear in the correlation matrix )
39
40 Background Slope
41
42 Background 1 -0.32
43
44 Slope -0.32 1
45
46
47
48 Parameter Estimates
49
50 Variable Estimate Std. Err.
51 Background 0.091177 0.0532315
52 Slope 6.03482e-006 3.53595e-006
53
54
55
56 Analysis of Deviance Table
57
58 Model Log(likelihood) Deviance Test DF P-value
59 Full model -18.2628
60 Fitted model -18.4542 0.382862 2 0.8258
61 Reduced model -39.4295 42.3333 3 <.0001
62
63 AIC: 40.9085
64
A-44 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Goodness of Fit
Dose Est._Prob. Expected
0.0000 0.0912 1.368
40.0000 0.0999 1.499
200.0000 0.2861 4.291
1000.0000 0.9978 14.967
Chi-square = 0.36 DF = 2
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 132.132
BMDL = 85.4708
Scaled
Observed Size Residual
1 15 -0.3298
2 15 0.4317
4 15 -0.1665
15 15 0.1808
P-value = 0.8372
Quantal Quadratic Model with 0.95 Confidence Level
0.6
0.4
Quantal Quadratic
BMDL BMD
200
400
600
800
1000
27
28
dose
15:49 12/282005
A-45
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Weibull Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
Input Data File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_MALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:49:46 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-slope*dose^power)]
Dependent variable = COLUMN2
Independent variable = COLUMN1
Power parameter is restricted as power >=1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.09375
Slope = 4.78181e-007
Power = 2.28256
Asymptotic Correlation Matrix of Parameter Estimates
Background Slope Power
Background 1 -0.33 0.31
Slope -0.33 1 -1
Power 0.31 -1 1
Parameter Estimates
Variable
Background
Slope
Power
Estimate
0.0959027
.27341e-006
2.27222
Std. Err.
0.0577365
1.17833e-005
1.67024
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Deviance Test DF
Log(likelihood)
-18.2628
-18.4284
-39.4295
42.8569
P-value
0.331283
42.3333
0.5649
<.0001
A-46
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
,J
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
._Prob. Expected
.0959 1.439
.1009 1.514
.2712 4.067
.9998 14.997
0.33 DF = 1
Computation
0.1
Extra risk
0.95
145.973
48.796
Scaled
Observed Size Residual
1 15 -0.3845
2 15 0.4168
4 15 -0.03917
15 15 0.05665
P-value = 0.5678
Weibull Model with 0.95 Confidence Level
£ 0.6
0.4
0.2
Weibull
BMDL
BMD
200
400
600
800
1000
26
27
dose
15:49 12/282005
A-47
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Fawell et al., 1999 Female Chronic Inflammation
$Revision: 2.2 $ $Date: 2001/03/14 01:17:00 $
Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:50:58 2005
BMDS MODEL RUN
The form of the probability function is:
P[response]= background+(1-background)*CumGamma[slope*dose,power],
where CumGamma(.) is the cummulative Gamma distribution function
Dependent variable = COLUMN2
Independent variable = COLUMN1
Power parameter is restricted as power >=1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.34375
Slope = 0.00408479
Power = 2.14573
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Power
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
Background
Background 1
Slope -0.42
Slope
-0.42
1
Parameter Estimates
Variable
Background
Slope
Power
Estimate
0.385764
0.0020737
1
Std. Err.
0.0913888
0.000852726
NA
NA -
Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
A-48
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
12
13
14
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Analysis of Deviance Table
Log (likelihood) Deviance Test DF
-33.9494
-34.4584 1.01801 2
-40.7516 13.6045 3
72.9167
Goodness of Fit
. Prob . Expected Observed
.3858 5.786 5
.4347 6.520 8
.5943 8.914 8
.9228 13.842 14
1.02 DF = 2 P-value =
Computation
0.1
Extra risk
0.95
50.8081
27.7368
P-value
0.6011
0.003496
Scaled
Size Residual
15 -0.4172
15 0.7709
15 -0.4808
15 0.1531
0.5996
Gamma Multi-Hit Model with 0.95 Confidence Level
0.6
0.4
0.2
Gamma Multi-Hit
200
400
600
800
1000
36
dose
15:50 12/282005
A-49
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Logistic Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:20 $
Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
Thu Mar 09 12:16:37 2006
BMDS MODEL RUN
The form of the probability function is:
P[response] = I/[1+EXP(-intercept-slope*dose)]
Dependent variable = COLUMN2
Independent variable = COLUMN1
Slope parameter is not restricted
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
background = 0 Specified
intercept = -0.342416
slope = 0.00261455
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -background
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
intercept
intercept 1
slope -0.51
Variable
intercept
slope
slope
-0.51
1
Parameter Estimates
Estimate
-0.374017
0.00300128
Std. Err.
0.335179
0.00110665
Model
Full model
Fitted model
Reduced model
AIC:
Analysis of Deviance Table
Deviance Test DF
Log(likelihood)
-33.9494
-34.4216
-40.7516
72.8432
P-value
0.944503
13.6045
0.6236
0.003496
A-50
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
/
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
Goodness of Fit
Dose Est . Prob . Expected
0.0000 0.4076 6.114
40.0000 0.4368 6.553
200.0000 0.5563 8.345
1000.0000 0.9326 13.989
Chi-square = 0.94 DF = 2
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 80.3245
BMDL = 48.7137
Scaled
Observed Size Residual
5 15 -0.5851
8 15 0.7534
8 15 -0.1792
14 15 0.0114
P-value = 0.6243
Logistic Model with 0.95 Confidence Level
29
30
31
0.6
0.4
Logistic
BMDL BMD
200
400
600
800
1000
dose
12:1603/092006
A-51
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Logistic Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:20 $
Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:51:45 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background+(1-background)/[1+EXP(-intercept-slope*Log(dose))]
Dependent variable = COLUMN2
Independent variable = COLUMN1
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial Parameter Values
background = 0.333333
intercept = -5.18041
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
background intercept slope
background 1 -0.51 0.47
intercept -0.51 1 -0.99
slope 0.47 -0.99 1
Variable
background
intercept
slope
Model
Full model
Fitted model
Reduced model
Parameter Estimates
Estimate
0.426119
-12.3243
2.07233
Std. Err.
0.0985863
8.90473
1.36766
Analysis of Deviance Table
Log(likelihood)
-33.9494
-34.5387
-40.7516
Deviance Test DF
P-value
1.17872
13.6045
0.2776
0.003496
A-52
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
AIC:
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
75.0774
Goodness of Fit
. Prob . Expected Observed Size
.4261 6.392 5 15
.4314 6.471 8 15
.5448 8.172 8 15
.9310 13.966 14 15
1.17 DF = 1 P-value = 0.2789
Computation
0.1
Extra risk
0.95
132.532
11.3311
Scaled
Residual
-0.7267
0 .7971
-0.08904
0.03513
Log-Logistic Model with 0.95 Confidence Level
0.6
0.4
Log-Logistic
BMDL
BMD
200
400
600
800
1000
28
29
15:51 12/282005
dose
A-53
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Multistage Model. $Revision: 2.1 $ $Date: 2000/08/21 03:38:21 $
Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
Thu Mar 09 12:17:59 2006
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l-beta2*dose^2-beta3*dose^3)]
The parameter betas are restricted to be positive
Dependent variable = COLUMN2
Independent variable = COLUMN1
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 4
Total number of specified parameters = 0
Degree of polynomial = 3
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.417955
Beta(l) = 0.00124572
Beta(2) = 0
Beta(3) = 9.2085e-010
Asymptotic Correlation Matrix of Parameter Estimates
( *** The model parameter(s) -Beta(2)
have been estimated at a boundary point, or have been
specified by the user,
and do not appear in the correlation matrix )
Background
Background 1
Beta(l) -0.66
Beta(3) 0.55
Variable
Background
Beta(l)
Beta(2)
Beta(3)
Beta(l)
-0.66
1
-0.93
Parameter
Estimate
0.403311
0.00148059
0
7 . 06214e-010
Beta(3)
0.55
-0.93
1
Estimates
Std. Err
0.162784
0.00267062
NA
2.70652e-009
A-54
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Model
Full model
Fitted model
Reduced model
AIC:
Dose Est
i: 1
0.0000 0.
i: 2
40.0000 0.
i: 3
200.0000 0.
i: 4
1000.0000 0.
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Analysis of Deviance Table
Log (likelihood) Deviance Test DF
-33.9494
-34.4013 0.903921 1
-40.7516 13.6045 3
74.8026
Goodness of Fit
. Prob . Expected Observed
4033 6.050 5
4376 6.565 8
5587 8.381 8
9330 13.995 14
0.90 DF = 1 P-value =
Computation
0.1
Extra risk
0.95
70.9904
28.0638
P-value
0.3417
0.003496
Size Chi^2 Res.
15 -0.291
15 0.389
15 -0.103
15 0.005
0.3421
A-55
DRAFT: DO NOT CITE OR QUOTE
-------
Multistage Model with 0.95 Confidence Level
1
0.9
0.8
T3 0.7
"o
< 0.6
£=
g
"o
<5 0.5
0.4
0.3
0.2
MultistE
Q6 —
^-^^^ :
,^~" \
BMDL
/^
^^ J
/^" \
^ :
j
.
H
BMD :
0 200 400 600 800 1000
dose
12:17 03/092006
A-56
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Multistage Model. $Revision: 2.1 $ $Date: 2000/08/21 03:38:21 $
Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:52:16 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(
-betal*dose^l-beta2*dose^2) ]
The parameter betas are restricted to be positive
Dependent variable = COLUMN2
Independent variable = COLUMN1
Total number of observations = 4
Total number of records with missing values = 0
Total number of parameters in model = 3
Total number of specified parameters = 0
Degree of polynomial = 2
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
Background = 0.417495
Beta(l) = 0.00110779
Beta(2) = 1.05844e-006
Asymptotic Correlation Matrix of Parameter Estimates
Background Beta(l) Beta(2)
Background 1 -0.67 0.58
Beta(l) -0.67 1 -0.95
Beta(2) 0.58 -0.95 1
Variable
Background
Beta(l)
Beta(2)
Parameter Estimates
Estimate
0.402286
0.00142259
Std. Err.
0.165802
0.00323206
7.46169e-007
3.28803e-006
Model
Analysis of Deviance Table
Log(likelihood) Deviance Test DF
P-value
A-57
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
1 1
_L J.
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
Full model
Fitted model
Reduced model
AIC:
Dose Est
i: 1
0.0000 0.
i: 2
40.0000 0.
i: 3
200.0000 0.
i: 4
1000.0000 0.
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
-33.9494
-34.4149 0.93098
-40.7516 13.6045
74 .8297
Goodness of Fit
._Prob. Expected Observed
4023 6.034 5
4360 6.540 8
5635 8.453 8
9317 13.975 14
0.93 DF = 1 P-value
Computation
0.1
Extra risk
0.95
71.3892
27.9852
1 0.3346
3 0.003496
Size Chi^2 Res.
15 -0.287
15 0.396
15 -0.123
15 0.026
= 0.3347
Multistage Model with 0.95 Confidence Level
1
0.9
0.8
T3 0.7
S
0
< 0.6
£=
g
~o
<5 0.5
0.4
0.3
0.2
Multista
9e
H
^— ""^
^^--"~"~^ -
(
/
BMDL
/
^^-^ :
//^ :
,/ :
// J
-
H
H
H
BMD :
0 200 400 600 800 1000
dose
36
37
15:52 12/282005
A-58
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Multistage Model. $Revision: 2.1 $ $Date: 2000/08/21 03:38:21 $
3 Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
5 Thu Mar 09 12:20:10 2006
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = background + (1-background)*[1-EXP(
14 -betal*dose^l) ]
15
16 The parameter betas are restricted to be positive
17
18
19 Dependent variable = COLUMN2
20 Independent variable = COLUMN1
21
22 Total number of observations = 4
23 Total number of records with missing values = 0
24 Total number of parameters in model = 2
25 Total number of specified parameters = 0
26 Degree of polynomial = 1
27
28
29 Maximum number of iterations = 250
30 Relative Function Convergence has been set to: le-008
31 Parameter Convergence has been set to: le-008
32
33
34
35 Default Initial Parameter Values
36 Background = 0.377465
37 Beta(l) = 0.00221127
38
39
40 Asymptotic Correlation Matrix of Parameter Estimates
41
42 Background Beta(l)
43
44 Background 1 -0.45
45
46 Beta(l) -0.45 1
47
48
49
50 Parameter Estimates
51
52 Variable Estimate Std. Err.
53 Background 0.385756 0.1375
54 Beta(l) 0.00207378 0.000944904
55
56
57
58 Analysis of Deviance Table
59
60 Model Log(likelihood) Deviance Test DF P-value
61 Full model -33.9494
62 Fitted model -34.4584 1.01801 2 0.6011
63 Reduced model -40.7516 13.6045 3 0.003496
64
A-59 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
/
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
AIC:
Dose Est
i: 1
0.0000 0.
i: 2
40.0000 0.
i: 3
200.0000 0.
i: 4
1000.0000 0.
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
72.9167
Goodness of Fit
. Prob . Expected Observed Size
3858 5.786 5 15
4347 6.520 8 15
5943 8.914 8 15
9228 13.842 14 15
1.02 DF = 2 P-value = 0.5996
Computation
0.1
Extra risk
0.95
50.8061
27.7368
Chi^2 Res.
-0.221
0.402
-0.253
0.148
Multistage Model with 0.95 Confidence Level
-a
~o
£
<
£=
g
"o
TO
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
Mult
SI8Q6 —
^ — — j
^^~ \
/^ -
/'
BMDL
0
/
x-/ ;
j
^
'-_
BMD :
200 400 600 800 1000
dose
32
12:2003/092006
A-60
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Probit Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:53 $
3 Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
5 Thu Mar 09 12:21:25 2006
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P [response] = CumNorm(Intercept+Slope*Dose) ,
14
15 where CumNorm(.) is the cumulative normal distribution function
16
17
18 Dependent variable = COLUMN2
19 Independent variable = COLUMN1
20 Slope parameter is not restricted
21
22 Total number of observations = 4
23 Total number of records with missing values = 0
24 Maximum number of iterations = 250
25 Relative Function Convergence has been set to: le-008
26 Parameter Convergence has been set to: le-008
27
28
29
30 Default Initial (and Specified) Parameter Values
31 background = 0 Specified
32 intercept = -0.216871
33 slope = 0.00162326
34
35
36 Asymptotic Correlation Matrix of Parameter Estimates
37
38 ( *** The model parameter(s) -background
39 have been estimated at a boundary point, or have been
40 specified by the user,
41 and do not appear in the correlation matrix )
42
43 intercept slope
44
45 intercept 1 -0.53
46
47 slope -0.53 1
48
49
50
51 Parameter Estimates
52
53 Variable Estimate Std. Err.
54 intercept -0.223386 0.20693
55 slope 0.00172875 0.000560405
56
57
58
59 Analysis of Deviance Table
60
61 Model Log(likelihood) Deviance Test DF P-value
62 Full model -33.9494
63 Fitted model -34.421 0.943277 2 0.624
64 Reduced model -40.7516 13.6045 3 0.003496
A-61 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
AIC: 72.842
Goodness of Fit
Dose
0.0000
40.0000
200.0000
1000.0000
Est. Prob.
0.4116
0.4387
0.5487
0.9339
Expected
6.174
6.581
8.230
14.008
Observed
5
8
8
14
Size
15
15
15
15
Scaled
Residual
-0.6161
0.7385
-0.1196
-0.00859
Chi-square =
0.94
DF = 2
P-value = 0.6252
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 86.341
BMDL = 56.9234
A-62
DRAFT: DO NOT CITE OR QUOTE
-------
e 0.6
0.4
0.2
Probit
BMDL 3MD
Probit Model with 0.95 Confidence Level
200
400
600
800
1000
dose
12:21 03/092006
A-63 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Probit Model $Revision: 2.1 $ $Date: 2000/02/26 03:38:53 $
Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:52:51 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = Background
+ (1-Background) * CumNorm(Intercept+Slope*Log(Dose)
where CumNorm(.) is the cumulative normal distribution function
Dependent variable = COLUMN2
Independent variable = COLUMN1
Slope parameter is restricted as slope >= 1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
User has chosen the log transformed model
Default Initial (and Specified) Parameter Values
background = 0.333333
intercept = -5.62294
slope = 1
Asymptotic Correlation Matrix of Parameter Estimates
background intercept slope
background 1 -0.46 0.43
intercept -0.46 1 -0.99
slope 0.43 -0.99 1
Variable
background
intercept
slope
Parameter Estimates
Estimate
0.431473
-7.732
1.2906
Std. Err.
0.0937984
4.97551
0.75516
Model
Full model
Analysis of Deviance Table
Log(likelihood) Deviance Test DF
-33.9494
P-value
A-64
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
Fitted model
Reduced model
AIC:
-34.5602 1.22166
-40.7516 13.6045
75.1204
Goodness of Fit
0.269
0.003496
Dose
0.0000
40.0000
200.0000
1000.0000
Est ._Prob.
0.4315
0.4323
0.5370
0.9327
Expected 01
6.472
6.485
8.055
13.991
^served £
5
8
8
14
3ize
15
15
15
15
Scaled
Residual
-0.7674
0.7897
-0.02871
0.009746
Chi-square =
1.21
DF = 1
P-value = 0.2706
Benchmark Dose Computation
Specified effect = 0.1
Risk Type = Extra risk
Confidence level = 0.95
BMD = 148.118
BMDL = 48.5214
Probit Model with 0.95 Confidence Level
t 0.6
0.4
Probit
BMDL
BMD
200
400 600
dose
800
1000
15:52 12/282005
A-65
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Quantal Linear Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
3 Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
5 Wed Dec 28 15:54:53 2005
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = background + (1-background)*[1-EXP(-slope*dose)]
14
15
16 Dependent variable = COLUMN2
17 Independent variable = COLUMN1
18
19 Total number of observations = 4
20 Total number of records with missing values = 0
21 Maximum number of iterations = 250
22 Relative Function Convergence has been set to: le-008
23 Parameter Convergence has been set to: le-008
24
25
26
27 Default Initial (and Specified) Parameter Values
28 Background = 0.34375
29 Slope = 0.00194591
30 Power = 1 Specified
31
32
33 Asymptotic Correlation Matrix of Parameter Estimates
34
35 ( *** The model parameter(s) -Power
36 have been estimated at a boundary point, or have been
37 specified by the user,
38 and do not appear in the correlation matrix )
39
40 Background Slope
41
42 Background 1 -0.42
43
44 Slope -0.42 1
45
46
47
48 Parameter Estimates
49
50 Variable Estimate Std. Err.
51 Background 0.385756 0.0913888
52 Slope 0.00207382 0.000852767
53
54
55
56 Analysis of Deviance Table
57
58 Model Log(likelihood) Deviance Test DF P-value
59 Full model -33.9494
60 Fitted model -34.4584 1.01801 2 0.6011
61 Reduced model -40.7516 13.6045 3 0.003496
62
63 AIC: 72.9167
64
A-66 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
._Prob. Expected
.3858 5.786
.4347 6.520
.5943 8.914
.9228 13.842
1.02 DF = 2
Computation
0.1
Extra risk
0.95
50.8051
27.7368
Scaled
Observed Size Residual
5 15 -0.4171
8 15 0.771
8 15 -0.4808
14 15 0.153
P-value = 0.5996
Quantal Linear Model with 0.95 Confidence Level
0.6
0.4
0.2
Quantal Linear
200
400
600
800
1000
26
27
dose
15:54 12/282005
A-67
DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Quantal Quadratic Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
3 Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
4 Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
5 Wed Dec 28 15:55:22 2005
6 ====================================================================
7
8 BMDS MODEL RUN
10
11 The form of the probability function is:
12
13 P[response] = background + (1-background)*[1-EXP(-slope*dose^2)]
14
15
16 Dependent variable = COLUMN2
17 Independent variable = COLUMN1
18
19 Total number of observations = 4
20 Total number of records with missing values = 0
21 Maximum number of iterations = 250
22 Relative Function Convergence has been set to: le-008
23 Parameter Convergence has been set to: le-008
24
25
26
27 Default Initial (and Specified) Parameter Values
28 Background = 0.34375
29 Slope = 1.94591e-006
30 Power = 2 Specified
31
32
33 Asymptotic Correlation Matrix of Parameter Estimates
34
35 ( *** The model parameter(s) -Power
36 have been estimated at a boundary point, or have been
37 specified by the user,
38 and do not appear in the correlation matrix )
39
40 Background Slope
41
42 Background 1 -0.23
43
44 Slope -0.23 1
45
46
47
48 Parameter Estimates
49
50 Variable Estimate Std. Err.
51 Background 0.449003 0.0779228
52 Slope 2.16129e-006 1.00801e-006
53
54
55
56 Analysis of Deviance Table
57
58 Model Log(likelihood) Deviance Test DF P-value
59 Full model -33.9494
60 Fitted model -34.6163 1.33385 2 0.5133
61 Reduced model -40.7516 13.6045 3 0.003496
62
63 AIC: 73.2326
64
A-68 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
._Prob. Expected
.4490 6.735
.4509 6.764
.4946 7.420
.9365 14.048
1.32 DF = 2
Computation
0.1
Extra risk
0.95
220.792
153.94
Scaled
Observed Size Residual
5 15 -0.9007
8 15 0.6416
8 15 0.2998
14 15 -0.05091
P-value = 0.5181
Quantal Quadratic Model with 0.95 Confidence Level
0.6
0.4
0.2
Quantal Quadratic
BMDL
BMD
200
400
600
800
1000
26
dose
15:55 12/282005
A-69
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Weibull Model $Revision: 2.2 $ $Date: 2000/03/17 22:27:16 $
Input Data File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.(d)
Gnuplot Plotting File: C:\BMDS\FAWELL_FEMALE_CHRONIC_INFLAMMATION.plt
Wed Dec 28 15:55:56 2005
BMDS MODEL RUN
The form of the probability function is:
P[response] = background + (1-background)*[1-EXP(-slope*dose^power)]
Dependent variable = COLUMN2
Independent variable = COLUMN1
Power parameter is restricted as power >=1
Total number of observations = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial (and Specified) Parameter Values
Background = 0.34375
Slope = 0.000217656
Power = 1.31712
Asymptotic Correlation Matrix of Parameter Estimates
Background Slope Power
Background 1 -0.77 0.76
Slope -0.77 1 -1
Power 0.76 -1 1
Variable
Background
Slope
Power
Parameter Estimates
Estimate
0.396812
0.00106435
1.09925
Std. Err.
0.136488
0.00719029
0.999439
Analysis of Deviance Table
Model
Full model
Fitted model
Reduced model
AIC:
Log(likelihood)
-33.9494
-34.4529
-40.7516
74.9057
Deviance Test DF
P-value
1.00699
13.6045
0.3156
0.003496
A-70
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
Dose Est
0.0000 0
40.0000 0
200.0000 0
1000.0000 0
Chi -square =
Benchmark Dose
Specified effect
Risk Type
Confidence level
BMD
BMDL
Goodness of Fit
._Prob. Expected
.3968 5.952
.4327 6.491
.5792 8.689
.9271 13.906
1.01 DF = 1
Computation
0.1
Extra risk
0.95
65.3741
27.7679
Scaled
Observed Size Residual
5 15 -0.5025
8 15 0.7864
8 15 -0.3601
14 15 0.09327
P-value = 0.3151
Weibull Model with 0.95 Confidence Level
27
28
29
30
0.6
0.4
0.2
Weibull
BMDL BMD
200
400
600
800
1000
dose
15:55 12/282005
A-71
DRAFT: DO NOT CITE OR QUOTE
-------
1 Heinze, 1999 Relative Liver Weight Changes
2
4 Polynomial Model. Revision: 2.2 Date: 9/12/2002
5 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY DOCUMENTS \_CYANO
6 TOX REV\MODELING\HEINZE_ENZYMES_AND_LIVER_WT. (d)
7 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
8 DOCUMENTS \_CYANO TOX REV\MODELING\HEINZE_ENZYMES_AND_LIVER_WT .pit
9 Thu May 18 09:37:38 2006
10 ====================================================================
11
12 BMDS MODEL RUN
13 _____________________________________________________________________
14
15 The form of the response function is:
16
17 Y[dose] = beta_0 + beta_l*dose + beta_2*dose^2 + ...
18
19
20 Dependent variable = MEAN
21 Independent variable = COLUMN1
22 rho is set to 0
23 The polynomial coefficients are restricted to be positive
24 A constant variance model is fit
25
26 Total number of dose groups = 3
27 Total number of records with missing values = 0
28 Maximum number of iterations = 250
29 Relative Function Convergence has been set to: le-008
30 Parameter Convergence has been set to: le-008
31
32
33
34 Default Initial Parameter Values
35 alpha = 0.1466
36 rho = 0 Specified
37 beta_0 = 2.84857
38 beta_l = 0.00447143
39
40
41
42 Parameter Estimates
43
44 95.0% Wald Confidence
45 Interval
46 Variable Estimate Std. Err. Lower Conf. Limit
47 Upper Conf. Limit
48 alpha 0.143276 0.0369936 0.0707695
49 0.215782
50 beta_0 2.84857 0.101163 2.6503
51 3.04685
52 beta_l 0.00447143 0.00110819 0.00229942
53 0.00664343
54
55
56 Asymptotic Correlation Matrix of Parameter Estimates
57
58 alpha beta_0 beta_l
59 alpha 1 5.6e-009 1.9e-009
60 beta_0 5.6e-009 1 -0.73
61 beta_l 1.9e-009 -0.73 1
62
63
64 Table of Data and Estimated Values of Interest
A-72 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Dose
Res.
0
50
150
Obs Mean
Obs Std Dev Est Mean
Est Std Dev
10
10
10
2.75
3.22
3.47
0.29
0.34
0.49
2.85
3.07
3.52
0.379
0.379
0.379
-0.824
1.24
-0.412
Model Descriptions for likelihoods calculated
Model Al: Yij
Var{e(ij) }
Model A2: Yij
Var{e(ij)
= Mu (i) + e(ij)
Mu(i) + e(ij )
Sigma(i)^2
Model R: Yi = Mu + e(i)
Var{e(i) } =
Likelihoods of Interest
Model Log(likelihood) DF
Al 15.381120 4
A2 16.880747 6
fitted 14.144766 2
R 7.133405 2
AIC
-22.762240
-21.761494
-24.289533
-10.266809
Test 1:
levels
Test 2:
Test 3 :
Does response and/or variances differ among dose
(A2 vs. R)
Are Variances Homogeneous (Al vs A2)
Does the Model for the Mean Fit (Al vs. fitted)
Test
Test 1
Test 2
Test 3
Tests of Interest
-2*log(Likelihood Ratio) Test df
19.4947
2.99925
2.47271
4
2
1
p-value
<.0001
0.2232
0.1158
The p-value for Test 1 is less than .05. There appears
to be a
difference between response and/or variances among the
dose levels.
It seems appropriate to model the data
The p-value for Test 2 is greater than .05. A
homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is greater than .05.
chosen appears
to adequately describe the data
Benchmark Dose Computation
The model
A-73
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Specified effect =
Risk Type
Confidence level =
BMD =
BMDL =
3.8
3.6
3.4
c
o
Q_
0 3.2
ce
c
ro
0
2.8
2.6
Linear
09:3705/182006
Estimated standard deviations from the control mean
0.95
84.6525
57.9321
Linear Model with 0.95 Confidence Level
BMD
BMD
20
40
60
80
dose
100
120
140
160
A-74
DRAFT: DO NOT CITE OR QUOTE
-------
1 Heinze, 1999 Lactate Dehydrogenase Changes
2
4 Polynomial Model. Revision: 2.2 Date: 9/12/2002
5 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY DOCUMENTS \_CYANO
6 TOX REV\MODELING\HEINZE_ENZYMES_AND_LIVER_WT. (d)
7 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
8 DOCUMENTS \_CYANO TOX REV\MODELING\HEINZE_ENZYMES_AND_LIVER_WT .pit
9 Thu May 18 09:40:36 2006
10 ====================================================================
11
12 BMDS MODEL RUN
13 _____________________________________________________________________
14
15 The form of the response function is:
16
17 Y[dose] = beta_0 + beta_l*dose + beta_2*dose^2 + ...
18
19
20 Dependent variable = MEAN
21 Independent variable = COLUMN1
22 rho is set to 0
23 The polynomial coefficients are restricted to be positive
24 A constant variance model is fit
25
26 Total number of dose groups = 3
27 Total number of records with missing values = 0
28 Maximum number of iterations = 250
29 Relative Function Convergence has been set to: le-008
30 Parameter Convergence has been set to: le-008
31
32
33
34 Default Initial Parameter Values
35 alpha = 15.6395
36 rho = 0 Specified
37 beta_0 = 20.22
38 beta_l = 0.101
39
40
41
42 Parameter Estimates
43
44 95.0% Wald Confidence
45 Interval
46 Variable Estimate Std. Err. Lower Conf. Limit
47 Upper Conf. Limit
48 alpha 29.028 7.49498 14.3381
49 43.7179
50 beta_0 20.22 1.43994 17.3978
51 23.0422
52 beta_l 0.101 0.0157738 0.070084
53 0.131916
54
55
56 Asymptotic Correlation Matrix of Parameter Estimates
57
58 alpha beta_0 beta_l
59 alpha 1 6.2e-015 1.5e-015
60 beta_0 6.2e-015 1 -0.73
61 beta_l 1.5e-015 -0.73 1
62
63
64 Table of Data and Estimated Values of Interest
A-75 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Dose
Res.
0
50
150
Obs Mean
Obs Std Dev Est Mean
Est Std Dev
10
10
10
16.6
30.6
33.6
4.48
5.05
1.16
20.2
25.3
35.4
5.39
5.39
5.39
-2.1
3.15
-1.05
Model Descriptions for likelihoods calculated
Model Al: Yij
Var{e(ij) }
Model A2: Yij
Var{e(ij)
= Mu (i) + e(ij)
Mu(i) + e(ij )
Sigma(i)^2
Model R: Yi = Mu + e(i)
Var{e(i) } =
Likelihoods of Interest
Model Log(likelihood) DF
Al -54.666589 4
A2 -46.093905 6
fitted -65.523922 2
R -78.954450 2
AIC
117 .333177
104.187810
135.047844
161.908899
Test 1:
levels
Test 2:
Test 3 :
Does response and/or variances differ among dose
(A2 vs. R)
Are Variances Homogeneous (Al vs A2)
Does the Model for the Mean Fit (Al vs. fitted)
Test
Test 1
Test 2
Test 3
Tests of Interest
-2*log(Likelihood Ratio) Test df
65.7211
17.1454
21 .7147
4
2
1
p-value
<.0001
0.0001892
<.0001
The p-value for Test 1 is less than .05. There appears
to be a
difference between response and/or variances among the
dose levels.
It seems appropriate to model the data
The p-value for Test 2 is less than .05. Consider
running a
non-homogeneous variance model
The p-value for Test 3 is less than .05. You may want
to try a
different model
Benchmark Dose Computation
Specified effect =
A-76
DRAFT: DO NOT CITE OR QUOTE
-------
1 Risk Type = Estimated standard deviations from the control mean
2
o
J
4 Confidence level = 0.95
5
6 BMD = 53.3442
7
8
9 BMDL = 39.9199
A-77 DRAFT: DO NOT CITE OR QUOTE
-------
1 ====================================================================
2 Polynomial Model. Revision: 2.2 Date: 9/12/2002
3 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY DOCUMENTS\_CYANO
4 TOX REV\MODELING\HEINZE_ENZYMES_AND_LIVER_WT.(d)
5 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
6 DOCUMENTS\_CYANO TOX REV\MODELING\HEINZE_ENZYMES_AND_LIVER_WT.pit
7 Thu May 18 09:42:43 2006
8 ====================================================================
9
10 BMDS MODEL RUN
12
13 The form of the response function is:
14
15 Y[dose] = beta_0 + beta_l*dose + beta_2*dose^2 + ...
16
17
18 Dependent variable = MEAN
19 Independent variable = COLUMN1
20 The polynomial coefficients are restricted to be positive
21 The variance is to be modeled as Var(i) = alpha*mean(i)^rho
22
23 Total number of dose groups = 3
24 Total number of records with missing values = 0
25 Maximum number of iterations = 250
26 Relative Function Convergence has been set to: le-008
27 Parameter Convergence has been set to: le-008
28
29
30
31 Default Initial Parameter Values
32 alpha = 15.6395
33 rho = 0
34 beta_0 = 20.22
35 beta_l = 0.101
36
37
38
39 Parameter Estimates
40
41 95.0% Wald Confidence
42 Interval
43 Variable Estimate Std. Err. Lower Conf. Limit
44 Upper Conf. Limit
45 alpha 2.44639e+006 4.30425e+006 -5.98979e+006
46 1.08826e+007
47 rho -3.55294 0.556161 -4.643
48 -2.46289
49 beta_0 21.3722 1.64609 18.146
50 24.5985
51 beta_l 0.0863674 0.0134905 0.0599266
52 0.112808
53
54
55 Asymptotic Correlation Matrix of Parameter Estimates
56
57 alpha rho beta_0 beta_l
58 alpha 1 -0.99 0.19 -0.23
59 rho -0.99 1 -0.21 0.25
60 beta_0 0.19 -0.21 1 -0.89
61 beta_l -0.23 0.25 -0.89 1
62
63
64 Table of Data and Estimated Values of Interest
A-78 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Dose
Res.
0
50
150
Obs Mean
Obs Std Dev Est Mean
Est Std Dev
10
10
10
16.6
30.6
33.6
4.48
5.05
1.16
21.4
25.7
34.3
6.79
4.9
2.93
-2.2
3.2
-0.808
Model Descriptions for likelihoods calculated
Model Al: Yij
Var{e(ij) }
Model A2: Yij
Var{e(ij) }
Model A3: Yij
Var{e(ij) }
Model R: Yi
Var{e(i) }
Model
Al
A2
A3
fitted
R
= Mu (i) + e(ij)
Mu(i) + e(ij )
Sigma(i)^2
Mu(i) + e(ij )
alpha* (Mud) )
Mu + e(i)
Sigma^2
Likelihoods of Interest
Log(likelihood) DF
-54.666589 4
-46.093905 6
-54.022155 5
-61.092641 4
-78.954450 2
AIC
117.333177
104.187810
118 . 044309
130.185281
161.908899
Test 1:
levels?
Test 2 :
Test 3:
Test 4:
Explanation of Tests
Does response and/or variances differ among Dose
(A2 vs. R)
Are Variances Homogeneous? (Al vs A2)
Are variances adequately modeled? (A2 vs. A3)
Does the Model for the Mean Fit? (A3 vs. fitted)
Test
Test 1
Test 2
Test 3
Test 4
Tests of Interest
-2*log(Likelihood Ratio) Test df
65.7211
17 . 1454
15.8565
14.141
4
2
1
1
p-value
<.0001
0.0001892
<.0001
0.0001696
The p-value for Test 1 is less than .05. There appears
to be a
difference between response and/or variances among the
dose levels
It seems appropriate to model the data
The p-value for Test 2 is less than .05. A
non-homogeneous variance
model appears to be appropriate
A-79
DRAFT: DO NOT CITE OR QUOTE
-------
1
2 The p-value for Test 3 is less than .05. You may want
3 to consider a
4 different variance model
5
6 The p-value for Test 4 is less than .05. You may want
7 to try a different
8 model
9
10
11 Benchmark Dose Computation
12 Specified effect = 1
13
14 Risk Type = Estimated standard deviations from the control mean
15
16
17 Confidence level = 0.95
18
19 BMD = 78.6091
20
21
22 BMDL = 63.5158
23
24
25
26
27
A-80 DRAFT: DO NOT CITE OR QUOTE
-------
1 Heinze, 1999 Alkaline Phosphatase Changes
2
o
J
4 ====================================================================
5 Polynomial Model. Revision: 2.2 Date: 9/12/2002
6 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY DOCUMENTS\_CYANO
7 TOX REV\MODELING\HEINZE_ENZYMES_AND_LIVER_WT.(d)
8 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
9 DOCUMENTS\_CYANO TOX REV\MODELING\HEINZE_ENZYMES_AND_LIVER_WT.pit
10 Thu May 18 09:46:03 2006
11 ====================================================================
12
13 BMDS MODEL RUN
14
15
16 The form of the response function is:
17
18 Y[dose] = beta_0 + beta_l*dose + beta_2*dose^2 + ...
19
20
21 Dependent variable = MEAN
22 Independent variable = COLUMN1
23 rho is set to 0
24 The polynomial coefficients are restricted to be positive
25 A constant variance model is fit
26
27 Total number of dose groups = 3
28 Total number of records with missing values = 0
29 Maximum number of iterations = 250
30 Relative Function Convergence has been set to: le-008
31 Parameter Convergence has been set to: le-008
32
33
34
35 Default Initial Parameter Values
36 alpha = 7.59287
37 rho = 0 Specified
38 beta_0 = 10.6414
39 beta_l = 0
40
41
42
43 Parameter Estimates
44
45 95.0% Wald Confidence
46 Interval
47 Variable Estimate Std. Err. Lower Conf. Limit
48 Upper Conf. Limit
49 alpha 7.93453 2.04869 3.91918
50 11.9499
51 beta_0 10.6414 0.75283 9.16591
52 12.117
53 beta_l 0.0180286 0.00824683 0.00186506
54 0.0341921
55
56
57 Asymptotic Correlation Matrix of Parameter Estimates
58
59 alpha beta_0 beta_l
60 alpha 1 l.le-006 -4.7e-007
61 beta_0 l.le-006 1 -0.73
62 beta_l -4.7e-007 -0.73 1
63
64
A-81 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
J
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
Dose
Res.
Table of Data and Estimated Values of Interest
N Obs Mean Obs Std Dev Est Mean
Est Std Dev
0 10
50 10
150 10
9.67
13
12.9
2.2
3.81
1.85
10.6
11.5
13.3
2.82
2.82
2.82
Model Descriptions for likelihoods calculated
Model Al: Yij
Var{e(ij) }
Model A2: Yij
Var{e(ij) }
Model R: Yi
Var{e(i) }
Model
Al
A2
fitted
R
= Mu (i) + e (ij)
= Mu (i) + e (ij)
= Sigma(i)*2
= Mu + e(i)
Likelihoods of Interest
Log(likelihood) DF
-43.827730 4
-40.832314 6
-46.068366 2
-48.794188 2
AIC
95.655461
93.664628
96.136733
101.588375
Test 1:
levels
Test 2:
Test 3 :
Does response and/or variances differ among dose
(A2 vs. R)
Are Variances Homogeneous (Al vs A2)
Does the Model for the Mean Fit (Al vs. fitted)
Test
Test 1
Test 2
Test 3
Tests of Interest
-2*log(Likelihood Ratio) Test df
15.9237
5.99083
4 .48127
4
2
1
p-value
0.0003485
0.05002
0.03427
The p-value for Test 1 is less than .05. There appears
to be a
difference between response and/or variances among the
dose levels.
It seems appropriate to model the data
The p-value for Test 2 is greater than .05. A
homogeneous variance
model appears to be appropriate here
The p-value for Test 3 is less than .05. You may want
to try a
different model
Benchmark Dose Computation
-1.09
1.64
-0.545
A-82
DRAFT: DO NOT CITE OR QUOTE
-------
1 Specified effect = 1
2
3 Risk Type = Estimated standard deviations from the control mean
4
5
6 Confidence level = 0.95
7
8 BMD = 156.243
9
10
11 BMDL = 87.6609
12
13
14
15
16
A-83 DRAFT: DO NOT CITE OR QUOTE
-------
1 Fawell et al., 1999 Male Alanine Aminotransferase Changes
2
3 ====================================================================
4 Polynomial Model. Revision: 2.2 Date: 9/12/2002
5 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
6 DOCUMENTS\_CYANO TOX REV\MODELING\FAWELL_MALE_ALT.(d)
7 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
8 DOCUMENTS\_CYANO TOX REV\MODELING\FAWELL_MALE_ALT.pit
9 Tue May 09 12:59:55 2006
10 ====================================================================
11
12 HMDS MODEL RUN
13 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
14
15 The form of the response function is:
16
17 Y[dose] = beta_0 + beta_l*dose + beta_2*doseA2 + ...
18
19
20 Dependent variable = MEAN
21 Independent variable = COLUMN1
22 rho is set to 0
23 The polynomial coefficients are restricted to be positive
24 A constant variance model is fit
25
26 Total number of dose groups = 4
27 Total number of records with missing values = 0
28 Maximum number of iterations = 250
29 Relative Function Convergence has been set to: le-008
30 Parameter Convergence has been set to: le-008
31
32
33
34 Default Initial Parameter Values
35 alpha = 1
36 rho = 0 Specified
37 beta_0 = 30.4717
38 beta_l = 0.129124
39
40
41
42 Parameter Estimates
43
44 95.0% Wald
45 Confidence Interval
46 Variable Estimate Std. Err. Lower Conf. Limit
47 Upper Conf. Limit
48 alpha 1584.79 289.34 1017.69
49 2151.88
50 beta_0 30.4717 6.47011 17.7905
51 43.1529
52 beta_l 0.129124 0.0126792 0.104273
53 0.153974
54
55
56 Asymptotic Correlation Matrix of Parameter Estimates
57
A-84 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
6
4
5
6
7
9
10
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
alpha beta 0 beta 1
alpha 1 -6.1e-OC>8 9.2e-OC>7
beta 0 -6.1e-008 1 -0.61
beta 1 9.2e-007 -0.61 1
Table of Data and Estimated Values of Interest
Dose N Obs Mean Obs Std Dev Est Mean Est Std Dev
Res .
-
0 15 27 8 30.5 39.8
40 15 37 17.2 35.6 39.8
200 15 59 28 56.3 39.8
1000 15 159 75 160 39.8
Model Descriptions for likelihoods calculated
Model Al: Yij = Mu(i) + e(ij)
Var{e (ij ) } = SigmaA2
Model A2: Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma (i)A2
Model R: Yi = Mu + e(i)
Var{e(i) } = SigmaA2
Likelihoods of Interest
Model Log (likelihood) DF AIC
Al -250.943933 5 511.887866
A2 -216.540867 8 449.081734
fitted -251.046212 2 506.092423
R -281.663312 2 567.326624
Test 1: Does response and/or variances differ among dose
levels
(A2 vs. R)
Test 2: Are Variances Homogeneous (Al vs A2 )
Test 3: Does the Model for the Mean Fit (Al vs. fitted)
Tests of Interest
Test -2*log (Likelihood Ratio) Test df p-value
Test 1 130.245 6 <.0001
Test 2 68.8061 3 <.0001
Test 3 0.204557 2 0.9028
The p-value for Test 1 is less than .05. There appears
to be a
ChiA2
-0.338
0.133
0.263
-0.0579
A-85
DRAFT: DO NOT CITE OR QUOTE
-------
1 difference between response and/or variances among the
2 dose levels.
3 It seems appropriate to model the data
4
5 The p-value for Test 2 is less than .05. Consider
6 running a
7 non-homogeneous variance model
8
9 The p-value for Test 3 is greater than .05. The model
10 chosen appears
11 to adequately describe the data
12
13
14
15 Benchmark Dose Computation
16 Specified effect = 1
17
18 Risk Type = Estimated standard deviations from the control mean
19
20
21 Confidence level = 0.95
22
23 BMD = 308.304
24
25
26 BMDL = 252.245
27
A-86 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
6
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Polynomial Model. Revision: 2.2 Date: 9/12/2002
Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
DOCUMENTS\_CYANO TOX REV\MODELING\FAWELL_MALE_ALT.(d)
Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
DOCUMENTS\_CYANO TOX REV\MODELING\FAWELL_MALE_ALT.pit
Tue May 09 13:00:39 2006
HMDS MODEL RUN
The form of the response function is:
Y[dose] = beta 0 + beta l*dose + beta 2*doseA2
Dependent variable = MEAN
Independent variable = COLUMN1
The polynomial coefficients are restricted to be positive
The variance is to be modeled as Var(i) = alpha*mean(i)Arho
Total number of dose groups = 4
Total number of records with missing values = 0
Maximum number of iterations = 250
Relative Function Convergence has been set to: le-008
Parameter Convergence has been set to: le-008
Default Initial Parameter Values
alpha = 1
rho = 0
beta_0 = 30.4717
beta 1 = 0.129124
Confidence Interval
Variable Estimate
Upper Conf. Limit
alpha 0.098324
0.32663
rho 2.15231
2.72391
beta_0 28.6439
33.2276
beta_l 0.141692
0.177912
Parameter Estimates
Std. Err.
0.116485
0.291636
2.33865
0.0184799
95.0% Wald
Lower Conf. Limit
-0.129982
1.58072
24.0602
0.105472
Asymptotic Correlation Matrix of Parameter Estimates
A-87
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
8
9
10
11
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
alpha rho beta 0 beta 1
alpha 1 -0.99 O."06 -0.11
rho -0.99 1 -0.065 0.18
beta 0 0.06 -0.065 1 -0.39
beta 1 -0.17 0.18 -0.39 1
Table of Data and Estimated Values of Interest
Dose N Obs Mean Obs Std Dev Est Mean Est Std Dev
Res .
_
0 15 27 8 28.6 11.6
40 15 37 17.2 34.3 14.1
200 15 59 28 57 24.3
1000 15 159 75 170 79
Model Descriptions for likelihoods calculated
Model Al: Yij = Mu(i) + e(ij)
Var{e(ij) } = SigmaA2
Model A2 : Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma (i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = alpha* (Mu (i )) Arho
Model R: Yi = Mu + e(i)
Var{e(i)} = SigmaA2
Likelihoods of Interest
Model Log (likelihood) DF AIC
Al -250.943933 5 511.887866
A2 -216.540867 8 449.081734
A3 -217.537818 6 447.075636
fitted -219.838125 4 447.676251
R -281.663312 2 567.326624
Explanation of Tests
Test 1: Does response and/or variances differ among Dose
levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (Al vs A2 )
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
Tests of Interest
ChiA2
-0.549
0.739
0.321
-0.556
A-88
DRAFT: DO NOT CITE OR QUOTE
-------
1
2 Test -2*log(Likelihood Ratio) Test df p-value
o
J
4 Test 1 130.245 6 <.0001
5 Test 2 68.8061 3 <.0001
6 Test 3 1.9939 2 0.369
7 Test 4 4.60062 2 0.1002
8
9 The p-value for Test 1 is less than .05. There appears
10 to be a
11 difference between response and/or variances among the
12 dose levels
13 It seems appropriate to model the data
14
15 The p-value for Test 2 is less than .05. A
16 non-homogeneous variance
17 model appears to be appropriate
18
19 The p-value for Test 3 is greater than .05. The
20 modeled variance appears
21 to be appropriate here
22
23 The p-value for Test 4 is greater than .05. The model
24 chosen seems
25 to adequately describe the data
26
27
28 Benchmark Dose Computation
29 Specified effect = 1
30
31 Risk Type = Estimated standard deviations from the control mean
32
33
34 Confidence level = 0.95
35
36 BMD = 81.8426
37
38
39 BMDL = 58.3727
A-89 DRAFT: DO NOT CITE OR QUOTE
-------
Linear Model with 0.95 Confidence Level
1
2
o
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
I
o
8-
-------
1
2
o
6
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Default Initial Parameter Values
alpha =
rho =
beta_0 =
beta_l =
beta 2 =
1
0
28.5738
0.160635
0
Confidence Interval
Variable
Upper Conf. Limit
alpha
0.32663
rho
2.72391
beta_0
33.2276
beta_l
0.177912
beta 2
Estimate
0.0983241
2.15231
28.6439
0.141692
0
Parameter Estimates
Std. Err.
0.116485
0.291636
2.33865
0.0184799
NA
95.0% Wald
Lower Conf. Limit
-0.129982
1.58072
24.0602
0.105472
NA - Indicates that this parameter has hit a bound
implied by some inequality constraint and thus
has no standard error.
Asymptotic Correlation Matrix of Parameter Estimates
alpha
rho
beta_0
beta 1
alpha
1
-0.99
0.06
-0.17
rho
-0.99
1
-0.065
0.18
beta_0
0.06
-0.065
1
-0.39
beta_l
-0.17
0.18
-0.39
1
The following parameter(s) have been estimated at a
boundary
point or have been specified. Correlations are not
computed:
beta 2
Dose
Res .
Table of Data and Estimated Values of Interest
N Obs Mean Obs Std Dev Est Mean Est Std Dev ChiA2
0
40
15
15
27
37
17.2
28.6
34.3
11.6
14.1
-0.549
0.739
A-91
DRAFT: DO NOT CITE OR QUOTE
-------
1 200 15 59 28 57
2 1000 15 159 75 170
3
4
5
6 Model Descriptions for likelihoods calculated
7
9 Model Al: Yij = Mu(i) + e(ij)
10 Var{e(ij)} = SigmaA2
11
12 Model A2: Yij = Mu(i) + e(ij)
13 Var{e(ij)} = Sigma (i)A2
14
15 Model A3: Yij = Mu(i) + e(ij)
16 Var{e(ij)} = alpha* (Mu (i )) Arho
17
18 Model R: Yi = Mu + e(i)
19 Var{e(i) } = SigmaA2
20
21
22 Likelihoods of Interest
23
24 Model Log (likelihood) DF
25 Al -250.943933 5
26 A2 -216.540867 8
27 A3 -217.537818 6
28 fitted -219.838125 4
29 R -281.663312 2
30
31
32 Explanation of Tests
33
34 Test 1: Does response and/or variances differ
35 levels?
36 (A2 vs. R)
37 Test 2: Are Variances Homogeneous? (Al vs A2 )
38 Test 3: Are variances adequately modeled? (A2
24.3 0.321
79 -0.556
AIC
511.887866
449.081734
447.075636
447.676251
567.326624
among Dose
vs . A3 )
39 Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
40
41 Tests of Interest
42
43 Test -2*log (Likelihood Ratio) Test df
44
45 Test 1 130.245 6
46 Test 2 68.8061 3
47 Test 3 1.9939 2
48 Test 4 4.60062 1
49
50 The p-value for Test 1 is less than .05. There
51 to be a
p-value
<.0001
<.0001
0.369
0.03196
appears
52 difference between response and/or variances among the
53 dose levels
54 It seems appropriate to model the data
55
56 The p-value for Test 2 is less than .05. A
57 non-homogeneous variance
A-92
DRAFT: DO NOT CITE OR QUOTE
-------
1
2
o
6
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
model appears to be appropriate
The p-value for Test 3 is greater than .05.
modeled variance appears
to be appropriate here
The
The p-value for Test 4 is less than .05. You may want
to try a different
model
Benchmark Dose Computation
Specified effect = 1
Risk Type
Confidence level =
BMD =
BMDL =
Estimated standard deviations from the control mean
0.95
81.8426
58.3727
Polynomial Model with 0.95 Confidence Level
I
I
in
-------
1
2 ====================================================================
3 Power Model. $Revision: 2.1 $ $Date: 2000/10/11 20:57:36 $
4 Input Data File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
5 DOCUMENTS\_CYANO TOX REV\MODELING\FAWELL_MALE_ALT.(d)
6 Gnuplot Plotting File: C:\DOCUMENTS AND SETTINGS\HCLYNCH\MY
7 DOCUMENTS\_CYANO TOX REV\MODELING\FAWELL_MALE_ALT.pit
8 Tue May 09 13:03:23 2006
9 ====================================================================
10
11 HMDS MODEL RUN
17 ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
13
14 The form of the response function is:
15
16 Y[dose] = control + slope * doseApower
17
18
19 Dependent variable = MEAN
20 Independent variable = COLUMN1
21 The power is restricted to be greater than or equal to 1
22 The variance is to be modeled as Var(i) = alpha*mean(i)Arho
23
24 Total number of dose groups = 4
25 Total number of records with missing values = 0
26 Maximum number of iterations = 250
27 Relative Function Convergence has been set to: le-008
28 Parameter Convergence has been set to: le-008
29
30
31
32 Default Initial Parameter Values
33 alpha = 1692.21
34 rho = 0
35 control = 27
36 slope = 0.519763
37 power = 0.801589
38
39
40 Asymptotic Correlation Matrix of Parameter Estimates
41
42 alpha rho control slope power
43
44 alpha 1 -0.99 0.13 -0.46 0.48
45
46 rho -0.99 1 -0.12 0.41 -0.43
47
48 control 0.13 -0.12 1 -0.76 0.74
49
50 slope -0.46 0.41 -0.76 1 -1
51
52 power 0.48 -0.43 0.74 -1 1
53
54
55
56 Parameter Estimates
57
A-94 DRAFT: DO NOT CITE OR QUOTE
-------
1
2
3
4
5
6
7
9
10
11
12
13
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Variable Estimate Std. Err.
alpha 0.098324 0.141722
rho 2.15231 0.343016
control 28.6439 3.51452
slope 0.141692 0.218496
power 1 0.231677
Table of Data and Estimated Values of Interest
Dose N Obs Mean Obs Std Dev Est Mean Est Std Dev
Res .
_
0 15 27 8 28.6 11.6
40 15 37 17.2 34.3 14.1
200 15 59 28 57 24.3
1000 15 159 75 170 79
Model Descriptions for likelihoods calculated
Model Al: Yij = Mu(i) + e(ij)
Va r { e ( i j ) } = Si gma A 2
Model A2 : Yij = Mu(i) + e(ij)
Var{e(ij)} = Sigma (i)A2
Model A3: Yij = Mu(i) + e(ij)
Var{e(ij)} = alpha* (Mu (i )) Arho
Model R: Yi = Mu + e(i)
Var{e(i) } = SigmaA2
Likelihoods of Interest
Model Log (likelihood) DF AIC
Al -250.943933 5 511.887866
A2 -216.540867 8 449.081734
A3 -217.537818 6 447.075636
fitted -219.838125 5 449.676251
R -281.663312 2 567.326624
Explanation of Tests
Test 1: Does response and/or variances differ among Dose levels?
(A2 vs. R)
Test 2: Are Variances Homogeneous? (Al vs A2 )
Test 3: Are variances adequately modeled? (A2 vs. A3)
Test 4: Does the Model for the Mean Fit? (A3 vs. fitted)
ChiA2
-0.142
0.191
0.083
-0.144
A-95
DRAFT: DO NOT CITE OR QUOTE
-------
1 Tests of Interest
2
3 Test -2*log(Likelihood Ratio) d.f p-value
4
5 Test 1 130.245 6 <.00001
6 Test 2 68.8061 3 <.00001
7 Test 3 1.9939 2 0.369
8 Test 4 4.60062 1 0.03196
9
10 The p-value for Test 1 is less than .05. There appears to be a
11 difference between response and/or variances among the dose levels
12 It seems appropriate to model the data
13
14 The p-value for Test 2 is less than .05. A non-homogeneous variance
15 model appears to be appropriate
16
17 The p-value for Test 3 is greater than .05. The modeled variance appears
18 to be appropriate here
19
20 The p-value for Test 4 is less than .05. You may want to try a different
21 model
22
23
24 Benchmark Dose Computation
25 Specified effect = 1
26
27 Risk Type = Estimated standard deviations from the control mean
28
29 Confidence level = 0.95
30
31 BMD = 81.8426
32
33
34 BMDL = 58.3727
35
36
37
A-96 DRAFT: DO NOT CITE OR QUOTE
------- |