\
EPA
OSWER Directive #9200.0-76
June 2010
RELATIVE BIO AVAILABILITY OF ARSENIC
IN SOILS AT 11 HAZARDOUS WASTE SITES
USING AN IN VIVO JUVENILE SWINE
METHOD
Bioavailability Subcommittee of the Technical Review Workgroup
Office of Solid Waste and Emergency Response
U.S. Environmental Protection Agency
Washington, DC 20408
-------
ACKNOWLEDGMENTS
The work described in this report is the product of a team effort involving a large number of people. In
particular, the following individuals contributed significantly to the findings reported here:
Most of the in vivo studies described in this report were planned and sponsored by USEPA, Region 8.
Technical and managerial direction for this project was provided mainly by Christopher P. Weis, PhD,
DABT, and Gerry M. Henningsen, DVM, PhD, DABT/DABVT.
All of the in vivo studies described in this report were performed by Stan W. Casteel, DVM, PhD,
DABVT, at the Veterinary Medical Diagnostic Laboratory, College of Veterinary Medicine, University of
Missouri, Columbia, Missouri. Dr. Casteel was supported by Larry D. Brown, DVM, MPH, Ross P.
Cowart, DVM, MS, DACVIM, James R. Turk, DVM, PhD, DACVP, John T. Payne, DVM, MS,
DACVS, Steven L. Stockham, DVM, MS, DACVP, Margaret E. Dunsmore, BS, and Roberto E. Guzman,
DVM, MS. Analysis of urine and other biological samples was performed by Dr. Edward Hindenberger,
of L.E.T., Inc., Columbia, Missouri.
Technical support in planning and evaluating the animal studies was provided by staff from Roy F.
Weston, Inc.; ISSI Consulting Group; and SRC.
Dr. Drexler, University of Colorado, Boulder, performed all of the electron microprobe and particle size
analyses of the test materials evaluated in these studies.
Dr. Timothy Barry, USEPA National Center for Environmental Economics, provided on-going support in
the development, testing, and evaluation of the statistical methods used in dose response curve fitting and
data reduction.
Dr. William Brattin, SRC, provided assistance in the design of the swine bioassay and data reduction
procedures, and authorship of this report.
-------
1 EXECUTIVE SUMMARY
2 INTRODUCTION
3 Reliable analysis of the potential hazard to humans from ingestion of a chemical depends upon
4 accurate information on a number of key parameters, including the concentration of the chemical in
5 environmental media (e.g., soil, dust, water, food, air, paint), intake rates of each medium, and the rate
6 and extent of absorption ("bioavailability") of the chemical by the body from each ingested medium.
7 Knowledge of bioavailability is important because the amount of a chemical (e.g., arsenic) that actually
8 enters the body from an ingested medium depends on the physical-chemical properties of the chemical
9 and of the medium. Accurate assessment of the human health risks resulting from oral exposure to
10 arsenic requires knowledge of the amount of arsenic absorbed from the gastrointestinal tract into the
11 body. When reliable data are available on the relative bioavailability (RBA) of a chemical in a site
12 medium (e.g., soil), this information can be used to improve the accuracy of exposure and risk
13 calculations at that site. Available RBA data can be used to adjust default oral toxicity values (reference
14 dose and slope factor) to account for differences in absorption between the chemical ingested in water and
15 the chemical ingested in site media, assuming the toxicity factors are based on a readily soluble form of
16 the chemical.
17 This document summarizes a number of in vivo studies that have been performed in young swine
18 to investigate the RBA of arsenic in different environmental media.
19 METHODS
20 Basic In Vivo Experimental Design
21 All in vivo studies were performed using young swine. Swine were selected for use because
22 available physiological data indicate that young swine are a good model for the human gastrointestinal
23 system. Groups of animals (usually 5 per dose group) were exposed to test material or reference material
24 for 12-15 days. Dosing was usually oral, although some groups were exposed to sodium arsenate by
25 gavage or by intravenous injection.
26 Samples of urine were collected from each animal on several different days during the study (the
27 exact days varied from study to study). Prior to analysis, samples of urine were digested using one of two
28 alternative methods. Studies that used the first digestion method are referred to as Phase II, and studies
29 that used the second digestion method are referred to as Phase III. After digestion, all samples were
30 analyzed for arsenic using the hydride method.
in
-------
1 Basic Method for Estimating RBA
2 Arsenic that is absorbed into the body from the gastrointestinal tract is excreted in the urine
3 within 1-2 days (see Table 2-1). Based on this, the RBA of a test material may be estimated by
4 measuring the urinary excretion fraction (UEF) of arsenic administered in test material and in reference
5 material (sodium arsenate), and calculating the ratio of the two UEF values:
6 RBAftest material) = UEF (test material) / UEE(sodium arsenate)
7 The UEF for each material (test soil, sodium arsenate) is estimated by plotting the mass of arsenic
8 excreted by each animal as a function of the dose administered, and then fitting the data for the two test
9 materials to a simultaneous weighted regression model. The slopes estimated for each test material are
10 direct estimates of the UEF. The RBA is estimated as the ratio of the slopes (slope test material/slope
11 sodium arsenate); the regression model also provides estimates of the uncertainty in the slope estimates.
12 A complete description of the regression model is included in Appendix A of the report.
13 RESULTS
14
15
16
17
18
In total, 29 test materials were investigated using the in vivo swine bioassay (two in duplicate). In
three cases, the amount of arsenic administered was too low to allow reliable measurement of RBA, and
the results for these samples are not considered to be meaningful. Values for the remaining all 29 test
materials are shown below.
Summary of RBA Estimates for Phase II and Phase III Test Materials
Phase
Phase II
Experiment
2
4
5
6
7
8
9
10
11
15
Sample
Bingham Creek Channel Soil
Murray Smelter Slag
Jasper County High Lead Millb
Aspen Bermb
Aspen Residentialb
Butte Soil
Midvale Slag
California Gulch Phase I Residential
Soil
California Gulch Fe/Mn PbO
California Gulch AV Slag
Palmerton Location 2
Palmerton Location 4
California Gulch AV Slag
Murray Smelter Soil
Clark Fork Tailings
Arsenic
Concentration3
(ppm)
149
695
16.4
66.9
16.7
234
591
203
110
1050
110
134
1050
310
181
RBA ± SEM
3 9% ±8%
55% ±10%
327% ±105%
100% ±46%
128% ±52%
9% ±3%
23% ±4%
8% ±3%
57% ±12%
13% ±4%
49% ±10%
61% ±11%
18% ±2%
33% ±5%
51% ±6%
-------
Summary of RBA Estimates for Phase II and Phase III Test Materials
Phase
Phase III
Experiment
1
2
3
4
5
6
7
Sample
VBI70 TM1
VBI70 TM2
VBI70 TM3
VBI70 TM4
VBI70 TM5
VBI70 TM6
Butte TM1
Butte TM2
Aberjona River TM1
Aberjona River TM2
El Paso TM1
El Paso TM2
ACC Utility Pole Soil
ACC Dislodgeable Arsenic
Arsenic
Concentration3
(ppm)
312
983
390
813
368
516
234
367
676.3
312.8
74
73
320
3500
RBA ± SEM
40% ±4%
42% ±4%
37% ±3%
24% ±2%
21% ±2%
24% ±3%
18% ±3%
24% ±2%
3 8% ±2%
52% ±2%
44% ±3%
37% ±3%
47% ±3%
26% ±1%
SEM = Standard error of the mean, an indicator of the relative uncertainty around the RBA estimate (see Appendix A)
aSame sample as evaluated in Phase II
bThe amount of arsenic administered was too low to allow reliable measurement of RBA, and the results for these samples are
not considered to be meaningful
2 As seen, using sodium arsenate as a relative frame of reference, estimated RBA values range
3 from less than 10% to more than 60% (excluding the 3 values considered to be unreliable). This wide
4 variability supports the conclusion that there can be important differences in RBA between different types
5 of samples, and that use of a site-specific RBA value is likely to increase the accuracy of risk estimates
6 for arsenic. This conclusion is also consistent with the similarity between the coefficient of variability of
7 the dose-UEF slope for test materials (0.38) and the coefficient of variability of estimated RBAs for the
8 same test materials (0.32).
9 Correlation of RBA with Arsenic Geochemistry
10 One objective of this project was to obtain preliminary information on which chemical forms or
11 mineral associations of arsenic tend to have high bioavailability and which tend to have low
12 bioavailability. Geochemical speciation data were obtained for 20 different test materials using electron
13 microprobe analysis. A total of 28 different arsenic phases were represented in the test materials; some
14 test materials contained more than one arsenic phase. In order to derive quantitative estimates of phase-
15 specific RBA values, a multivariate linear regression approach was used. Because the total number of
16 phases (28) was larger than the number of RBA measurements (20), the existing data are not sufficient to
17 perform a robust regression analysis based on individual phases. A screening-level analysis was
18 performed by grouping the 28 different phases into broader categories based on professional judgment
-------
1 regarding the expected degree of similarity between members of a group. Only the arsenic mass in
2 partially or entirely liberated particles (arsenic-bearing grains that are partially or entirely exposed on
3 their outer surfaces) was included in this analysis. Based on this analysis, it is possible to assign tentative
4 qualitative estimates of bioavailability, as follows:
6
7
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Low Bioavailability
As2O3
Sulfosalts
Medium Bioavailability
As Phosphate
FeAs Oxide
PbAs Oxide
MnAs Oxide
Fe and Zn sulfates
High Bioavailability
FeAsO
CONCLUSION
The data from the investigations performed under this program support the following main
conclusions:
1. Juvenile swine constitute a useful and stable animal model for measuring the relative
bioavailability of arsenic in a variety of soil or soil-like test materials. The Phase III protocol
described in this report is the recommended standard operating procedure (SOP) for the juvenile
swine RBA assay.
2. There are clear differences in the in vivo RBA of arsenic between different types of test materials,
ranging from less than 10% to more than 60%. Thus, knowledge of the RBA value for different
types of test materials at a site can be important for improving arsenic risk assessments at a site.
3. Available data are not yet sufficient to allow reliable quantitative calculation of the RBA for a test
material based only on knowledge of the relative amounts of arsenic mineral phases present.
However, tentative qualitative estimates of low, medium, or high bioavailability have been made
based on the major phase type of the arsenic containing waste material.
4. Additional extraction steps were identified and necessary to convert urinary organoarsenic
metabolites to inorganic arsenic for analysis of total arsenic in urine.
5. Due to limitations in detection limits for measurement of arsenic in urine, a minimum arsenic
dose of 25 (ig/kg bw-day is recommended for the juvenile swine RBA assay, so that the amount
of arsenic excreted in urine reaches a measurable quantity.
VI
-------
TABLE OF CONTENTS
ACKNOWLEDGMENTS ii
EXECUTIVE SUMMARY iii
LIST OF TABLES viii
LIST OF FIGURES viii
LIST OF APPENDICES ix
ACRONYMS AND ABBREVIATIONS x
1.0 INTRODUCTION 1
1.1 Overview 1
1.2 Using Relative Bioavailability Data to Improve Risk Calculations for Arsenic 1
2.0 EXPERIMENTAL METHODS FOR ESTIMATING ARSENIC RBA BY IN VIVO STUDIES ..2
2.1 Basic Approach for Measuring RBA/« Vivo 2
2.2 Experimental Methods 4
2.2.1 Study Designs 4
2.2.2 Experimental Animals 5
2.2.3 Diet 5
2.2.4 Dosing 6
2.2.5 Collection and Preservation of Urine 6
2.2.6 Arsenic Analysis 7
2.2.6.1 Sample Digestion 7
2.2.6.2 Arsenic Analysis by Hydride Generation 8
2.2.7 Quality Assurance 8
2.2.8 Test Material Characterization 12
2.3 Results 14
2.3.1 RBA Estimates 14
2.3.2 Effect of Low Analytical Recovery on Phase II RBA Values 15
2.3.3 Effect of Food on Arsenic Absorption 16
2.4 Correlation of RBA with Arsenic Geochemistry 16
2.5 Discussion of In Vivo Results 18
3.0 CONCLUSIONS 19
4.0 REFERENCES 20
VII
-------
LIST OF TABLES
Table 2-1. Summary of Arsenic Excretion Studies in Humans and Animals Exposed to Soluble Arsenic
Compounds in Water 22
Table 2-2. Typical Swine Feed Composition 23
Table 2-3. Description of Test Materials 24
Table 2-4. Relative Mass of Arsenic By Mineral Phase in Test Materials 28
Table 2-5. Size Distributions of Arsenic Particles 31
Table 2-6. Matrix Associations of Arsenic Particles 32
Table 2-7. RBA Estimates for Arsenic in Test Materials 33
Table 2-8. Summary Statistics for Dose-UEF Slopes and RBA Estimates for Phase III RBA Assays 35
Table 2-9. Consolidated Arsenic Phases 36
Table 2-10. Relative Arsenic Mass for Consolidated Phase Groupings 38
Table 2-11. Estimated Group-Specific RBA Values for Liberated Particles 39
LIST OF FIGURES
Figure 2-1. Excretion of Soluble As in Humans and Animals3 40
Figure 2-2. Conceptual Model for Arsenic Absorption and Excretion 41
Figure 2-3. Quality Assurance Data from Phase II Pilot Studies3 42
Figure 2-4. Phase III Performance Evaluation Samples3 43
Figure 2-5. Phase III Blind Duplicate Samples 3 44
Figure 2-6. Phase III Inter-Laboratory Comparison3 45
Figure 2-7. Uncertainty in RBA Values3 46
Vlll
-------
LIST OF APPENDICES
APPENDIX A: DETAILED DATA REDUCTION PROCEDURE
APPENDIX B: STUDY DESIGNS
APPENDIX C: TEST MATERIAL CHARACTERIZATION
APPENDIX D: DETAILED RAW DATA FILES (see electronic files on attached CD)
APPENDIX E: DETAILED DATA FITTING AND RBA CALCULATIONS
IX
-------
ACRONYMS AND ABBREVIATIONS
AAS Atomic absorption spectrometer
ABA Absolute bioavailability
AFo Oral absorption fraction
bw Body weight
°C Degrees Celsius
CV Coefficient of variation (SD/mean)
DMA Dimethylarsinic acid
EDS Energy dispersive spectrometer
EMPA Electron Microprobe Analysis
ERA Environmental Resource Associates
GLP Good Laboratory Practices
HC1 Hydrochloric acid
HNO3 Nitric acid
ICP-AES Inductively Coupled Plasma-Atomic Emission Spectrometry
IRIS Integrated Risk Information System
kg Kilogram
KI Potassium iodide
L Liter
mg Milligram
mL Milliliter
MMA Monomethylarsonic acid
NaAs Sodium arsenate
NIST National Institute of Standards and Testing
NRCC National Resource Council Canada (Institute for National Measurement Standards)
ORD USEPA Office of Research and Development
oRfD Oral reference dose
oSF Oral slope factor
PE Performance Evaluation
ppm Parts per million
QA Quality assurance
RBA Relative bioavailability
RME Reasonable maximum exposure
SD Standard deviation
SEM Standard error of the mean
SOP Standard operating procedure
TAL Target Analyte List
UEF Urinary excretion fraction
ug Microgram
urn Micrometer
U.S. EPA U.S. Environmental Protection Agency
WDS Wavelength dispersive spectrometers
XAS X-ray absorption spectroscopy
-------
1 1.0 INTRODUCTION
2 1.1 Overview
3 Accurate assessment of the human health risks resulting from oral exposure to arsenic requires
4 knowledge of the amount of arsenic absorbed from the gastrointestinal tract into the body. This
5 information on gastrointestinal absorption may be described either in absolute or relative terms:
6 Absolute Bioavailability (ABA) is the ratio of the amount of arsenic absorbed to the amount ingested:
7 ABA = (Absorbed Dose) / (Ingested Dose)
8 This ratio is also referred to as the oral absorption fraction (AFo).
9 Relative Bioavailability (RBA) is the ratio of the absolute bioavailability of arsenic present in some test
10 material to the absolute bioavailability of arsenic in some appropriate reference material:
11 RBA = ABA (test) /ABA (reference)
12 Usually the form of arsenic used as the reference material is an arsenic compound dissolved in
13 water or a readily soluble form (e.g., sodium arsenate) that is expected to completely dissolve when
14 ingested.
15 For example, if 100 (ig of arsenic dissolved in drinking water were ingested and a total of 90 (ig
16 were absorbed into the body, the ABA would be 0.90 (90%). Likewise, if 100 (ig of arsenic contained in
17 soil were ingested and 30 (ig were absorbed into the body, the ABA for soil would be 0.30 (30%). If the
18 arsenic dissolved in water was used as the frame of reference for describing the relative amount of arsenic
19 absorbed from soil, the RBA would be 0.30/0.90, or 0.33 (33%).
20 When reliable data are available on the RBA of a chemical (e.g., arsenic) in a site medium (e.g.,
21 soil), this information can be used to improve the accuracy of exposure and risk calculations at that site.
22 Available RBA data can be used to adjust default oral toxicity values (reference dose and slope factor) to
23 account for differences in absorption between the chemical ingested in water and the chemical ingested in
24 site media, assuming the toxicity factors are based on a readily soluble form of the chemical.
25 1.2 Using Relative Bioavailability Data to Improve Risk Calculations for Arsenic
26 The Risk Assessment Guidance for Superfund (RAGS) Part A (U.S. EPA, 1989) and Guidance
27 for Evaluating the Bioavailability of Metals in Soils for Use in Human Health Risk Assessment (U.S.
28 EPA, 2007) discuss making adjustments to exposure estimates in Superfund site-specific risk assessments
29 when the medium of exposure in the exposure assessment differs from the medium of exposure assumed
30 by the toxicity value (cancer slope factor, reference dose value, etc.) based upon site-specific
-------
1 bioavailability data. When a reliable RBA value is available for a particular site medium (e.g., soil), the
2 RBA can be used to adjust estimate of the daily intake (DI) as follows:
4 2.0 EXPERIMENTAL METHODS FOR ESTIMATING ARSENIC RBA BY IN VIVO
5 STUDIES
6 All in vivo studies were performed according to the spirit and guidelines of Good Laboratory
7 Practices (GLP: 40 CFR 792). Standard Operating Procedures (SOPs) that included detailed methods for
8 all of the components of each study were prepared, approved, and distributed to all team members prior to
9 all studies.
10 2.1 Basic Approach for Measuring RBA In Vivo
11 Summary of Arsenic Toxicokinetics
12 Available data from studies on the absorption and excretion of soluble arsenic compounds in
13 humans and animals are summarized in Table 2-1. Based on the fecal excretion data, absorption of
14 soluble arsenic compounds (sodium arsenate and sodium arsenite) typically appears to be at least 90% in
15 both humans and animals.
16 Estimates of biliary excretion are available from studies in which soluble arsenic compounds
17 have been given by intravenous injection. Results from studies by Johnson and Farmer (1991) and
18 Freeman et al. (1994) indicate biliary excretion is probably about 4-8% of the absorbed dose. Correction
19 of fecal excretion data by subtraction of 8% to account for biliary excretion suggests that absorption of
20 soluble arsenic is probably close to 100% in most cases.
21 Figure 2-1 plots the urinary excretion data from Table 2-1. It is apparent that typical urinary
22 recovery of soluble arsenic in humans (top panel) is dose-independent, and averages about 67% (range =
23 45 to 85%). Urinary recovery of arsenic in rodents (Figure 2-1, lower panel) is similar, with an average
24 value of 70% (range = 36 to 94%). Often the sum of arsenic recovery in urine plus feces is slightly less
25 than 100%. This could be partly due to experimental error, but is more likely due to retention of some
26 arsenic in tissues such as skin and hair.
27 Conceptual Model
28 Based on the human and animal data above, it appears that both absorption and excretion are
29 likely to be linear (i.e., dose independent) processes at dose levels well above those expected from
-------
1 exposure to arsenic in soil (e.g., 1000 ppm x 100 mg/day = 100 (ig/day). Figure 2-2 shows a conceptual
2 model for the toxicokinetic fate of ingested arsenic that is based on concept that absorption and excretion
3 are linear. Key points of the model are as follows:
4 If 100% of all absorbed arsenic were excreted in the urine, the UEF would be equal to the oral
5 absorption fraction or ABA. However, some absorbed arsenic is excreted in the feces via the bile
6 and some absorbed arsenic enters tissue compartments (e.g., skin, hair) from which it is cleared
7 very slowly or not at all. Thus, the urinary excretion fraction should not be equated with the
8 absolute absorption fraction.
9 The RBA of two orally administered materials (e.g., a test soil and sodium arsenate) can be
10 calculated from the ratio of the urinary excretion fraction of the two materials. This calculation is
1 1 independent of the extent of tissue binding or biliary excretion, because the fraction of absorbed
12 arsenic that is excreted in urine (Ku), which does depend on tissue binding and biliary excretion,
13 cancels in the calculation:
., , UEF(x) AF,(x)-K. AFa(x)
^"^-U^W-TfM^-TfM
15 where:
16 RBA(x vs. y) is the relative bioavailability of As in test material (r) vs. sodium arsenate
17 (y);
1 8 UEF is the urinary excretion fraction of the dose excreted in urine;
19 AF0 is the absorption fraction, which is the fraction of the dose absorbed following oral
20 administration; and
21 Ku is the fraction of the absorbed dose excreted in urine.
22
23 Thus, measurement of the urinary excretion fraction ((ig/day excreted in urine per (ig/day
24 administered) of test material and reference material (sodium arsenate) is the key experimental goal in
25 these arsenic RBA studies.
26 Estimation of UEF
27 The amount of arsenic excreted in urine ((ig/day) is calculated as the product of urinary
28 concentration ((ig/L) and urinary volume (L/day). The UEF is the rate of As excreted in urine (mL/day)
29 divided by the dose (mg/day). Conceptually, the UEF could be estimated for each animal on each day
30 that data are collected, and the UEF estimates for a particular dose material could then be averaged across
3 1 different animals, dose levels, and days. However, this approach does not account for baseline intake and
32 excretion of arsenic in the control group (unexposed animals), and tends to overemphasize UEF values at
-------
1 the low end of the dose range where the estimate of urinary excretion is most uncertain. A more robust
2 approach, used in this evaluation, is to plot the mass excreted by each animal as a function of the dose
3 administered to each animal, and then fit a linear regression line to the combined data. The slope of this
4 line is a direct estimate of the UEF ((ig/day excreted per (ig/day ingested). This approach automatically
5 accounts for baseline arsenic ingestion and excretion in control (unexposed) animals, and is not
6 disproportionately influenced by measurement error at the low end of the dose curve.
7 The process of deriving the best fit linear regression lines through the data is complicated by the
8 fact that the equations for each dose material in a study must have the same intercept, and because the
9 variability in the data tend to increase as the dose increases (this is referred to as heteroscedasticity). In
10 order to address these issues, the data from each study were fit using simultaneous weighted linear
11 regression, as detailed in Appendix A.
12 2.2 Experimental Methods
13 2.2.1 Study Designs
14 Phase II Study Designs
15 Measurement of arsenic bioavailability in most Phase II studies was performed in parallel with
16 studies designed to estimate lead bioavailability (U.S. EPA, 2007). Groups of animals (typically 4 or 5
17 per dose group) were given oral doses of a test material (e.g., soil, tailings, slag, sediment) twice daily for
18 15 days, and 24-hour urine samples were collected several times during the study (typically on days 7 and
19 14). Because the main focus of these studies was on lead RBA, these early studies did not include groups
20 of animals that were exposed to an arsenic reference material. Thus, these studies, taken alone, were not
21 sufficient to allow for an estimation of the arsenic RBA of the test materials.
22 In order to address this data gap and provide data on the urinary excretion fraction of a suitable
23 reference material, two "pilot studies" (Phase II, Experiments 10 and 15) were performed to establish the
24 urinary excretion fraction for sodium arsenate administered by three different routes: orally with a small
25 amount of food, orally by gavage (no food), and by intravenous injection.
26 Appendix B1 provides the detailed study designs for each Phase II study, and Appendix B2
27 provides the detailed designs for the two pilot studies.
28 Phase III Study Designs
29 After the completion of the Phase II studies, a modified study design was developed that was
30 specifically optimized for evaluation of arsenic RBA, rather than lead RBA. In this design, each study
31 includes a set of animals exposed to the reference material (sodium arsenate) and one to three different
-------
1 test materials, each at two or three different dose levels. In some cases, the doses of arsenic (expressed as
2 (ig/day) were held constant over time, rather than being adjusted to account for changing body weight.
3 This is because the basic computational approach used to estimate RBA (described above) compares the
4 mass of arsenic excreted in urine ((ig/day) to the mass of arsenic ingested ((ig/day), so body weight
5 adjustments are not needed.
6 Appendix B3 provides the detailed study designs for each Phase III study.
7 2.2.2 Experimental Animals
8 Juvenile swine were selected for use in these studies because their gastrointestinal physiology is
9 more similar to humans than most other animal models (Weis and LaVelle, 1991). All animals were
10 young males of the Pig Improvement Corporation genetically defined Line 26, and were purchased from
11 Chinn Farms, Clarence, MO. All studies used intact animals, except for one (the second VBI70 study),
12 which used castrated animals. The number of animals purchased for each study was typically 6-8 more
13 than required by the protocol. These animals were usually purchased at age 4-5 weeks (weaning occurs
14 at age 3 weeks), and they were then held under quarantine for one week to observe their health before
15 beginning exposure to test materials. Any animals that appeared to be in poor health during this
16 quarantine period were excluded. To minimize weight variations between animals and groups, extra
17 animals most different in body weight (either heavier or lighter) four days prior to exposure (day-4) were
18 also excluded from the study. The remaining animals were assigned to dose groups at random. When
19 exposure began (day zero), the animals were about 5-6 weeks old and weighed an average of about 7-
20 12 kg.
21 All animals were housed in individual stainless steel cages. Each animal was examined by a
22 certified veterinary clinician (swine specialist) prior to being placed on study, and all animals were
23 examined daily by an attending veterinarian while on study. There were no instances where animals that
24 became ill could not be promptly restored to good health by appropriate treatment, so no animals were
25 removed from the studies.
26 2.2.3 Diet
27 Animals provided by the supplier were weaned onto standard pig chow purchased from MFA
28 Inc., Columbia, MO. In order to minimize arsenic exposure from the diet, the animals were gradually
29 transitioned from the MFA feed to a special feed (Zeigler Brothers, Inc., Gardners, PA) over the time
30 interval from day -7 to day -3; this feed was then maintained for the duration of the study. The feed was
31 nutritionally complete and met all requirements of the National Institutes of Health-National Research
32 Council. The typical nutritional components and chemical analysis of the feed is presented in Table 2-2.
-------
1 Each day every animal was given an amount of feed equal to 5% (4% in the Aberjona River study) of the
2 mean body weight of all animals on study. Feed amounts were adjusted every three days, when pigs were
3 weighed. Feed was administered in two equal portions of 2.5% (2% in the Aberjona River study) of the
4 mean body weight at 11:00 AM and 5:00 PM daily. Periodic analysis of feed samples indicated that the
5 arsenic level was generally below the detection limit (0.1 ppm), which corresponds to a dose contribution
6 from food of less than 5 (ig/kg-day (less than 50 (ig/day for a 10 kg animal).
7 Drinking water was provided ad libitum via self-activated watering nozzles within each cage.
8 Periodic analysis of samples from randomly selected drinking water nozzles indicated the arsenic
9 concentration was less than the detection limit (about 1 (ig/L). Assuming water intake of about
10 0.1 L/kg-day, this corresponds to a dose contribution from water of less than 0.1 (ig/kg-day (1 (ig/day for
11 a 10 kg animal).
12 2.2.4 Dosing
13 Animals were exposed to sodium arsenate (abbreviated in this report as "NaAs") or a test material
14 for 12-15 days, with the dose for each day being administered in two equal portions given at 9:00 AM
15 and 3:00 PM (two hours before feeding). Animals were administered dose material when in a semi-fasted
16 state (i.e., two hours before feeding) to avoid the presence of food in the stomach, which is known to
17 reduce absorption of arsenic. In Phase II, doses were based on measured group mean body weights and
18 were adjusted every three days to account for animal growth. In most Phase III studies, doses were held
19 constant (independent of body weight).
20 Dose material was placed in the center of a small portion (about 5 grams) of moistened feed
21 (referred to as a "doughball"), which was administered to the animals by hand. In cases where the mass
22 of soil was too large to fit into one doughball, the test material was distributed among two or more
23 doughballs. Occasionally, some animals did not consume some or the entire dose (usually because the
24 dose dropped from their mouth while chewing). All missed doses were recorded and the time-weighted
25 average dose calculation for each animal was adjusted downward accordingly.
26 2.2.5 Collection and Preservation of Urine
27 Samples of urine were collected from each animal on several different days during the study (the
28 exact days varied from study to study). Collection began at about 8:00 AM and ended 24 hours later in
29 the Phase II studies and 48 hours later in most Phase III studies. The urine was collected in a stainless
30 steel pan placed beneath each cage, which drained into a plastic storage bottle. Each collection pan was
31 fitted with a nylon screen to minimize contamination with feces or spilled food. At the end of each
32 collection period, the urine volume was measured and two 60-mL portions were removed for analysis.
-------
1 Each 60 mL sample was preserved by addition of 0.6 mL of concentrated nitric acid. These samples were
2 refrigerated until sample analysis.
3 2.2.6 Arsenic Analysis
4 All samples were assigned random chain-of-custody tag numbers and submitted to the analytical
5 laboratory in a blind fashion. Arsenic concentrations in urine were measured using a hydride generation
6 approach. This method requires that all arsenic exist in the form of inorganic arsenic before hydride
7 generation. Because arsenic in urine can exist in organic forms (monomethylarsonic acid [MMA] and
8 dimethylarsinic acid [DMA]) as well as inorganic forms, digestion of the urine prior to analysis is
9 required.
10 2.2.6.1 Sample Digestion
11 Two different methods of arsenic digestion prior to analysis were employed during this project.
12 The first method was used during Phase II and a revised method was used for Phase III studies. As
13 discussed in greater detail below (see PE Samples and Blind Duplicates in Section 2.2.7), this change in
14 digestion method was adopted because recovery of total arsenic from urine and other biological samples
15 using the first method was limited by incomplete conversion of organic metabolites of arsenic (MMA and
16 DMA) to inorganic arsenic. The revised method produced improved recoveries of these metabolites and
17 of total arsenic.
18 Digestion Method 1
19 A 25 mL aliquot of acidified urine was removed and placed in a clean 100 mL glass beaker.
20 20 mL of concentrated nitric acid and 2.5 mL of concentrated perchloric acid were then added. The
21 beaker was covered with a watch glass and placed on a hot plate to reflux for 4-12 hours. After this
22 period, the heat was increased to drive off the nitric acid and to cause the perchloric acid to fume. After
23 about 10 minutes of fuming, the digestate was cooled slightly and diluted with 20 mL of distilled water.
24 This was heated until clear, and then cooled and diluted to 50 mL.
25 Digestion Method 2
26 A 25 mL aliquot of acidified urine was removed and placed in a clean 100 mL beaker. 3.0 mL of
27 methanol, 10.0 mL of 40% (w/v) magnesium nitrate hexahydrate, and 10.0 mL of concentrated trace
28 metal grade nitric acid (HNO3) were then added. The beaker was covered with a watch glass and placed
29 on a hot plate to reflux for 8-12 hours at 70-80°C. After this, the temperature was increased to 200°C,
30 and the watch glass was moved back to allow faster evaporation. The sample was then heated to
31 complete dryness (8-12 hours), covered with a watch glass, and allowed to cool. Dried samples were
-------
1 transferred to a cool muffle furnace which was heated at a rate of 1 degree/minute to a temperature of
2 500°C, and then held at 500°C for 3 hours before cooling. Ashed samples were dissolved by adding 5 mL
3 distilled water and 5 mL concentrated trace metal grade hydrochloric acid (HC1), and boiling gently until
4 the white residue was completely dissolved. After cooling, the dissolved sample was diluted with
5 distilled water to 50.0 mL and held until analysis.
6 2.2.6.2 Arsenic Analysis by Hydride Generation
7 Arsenic concentrations in urine were measured by hydride generation. Samples were prepared
8 for hydride generation by dilution with a solution of 10% HC1, 10% potassium iodide (KI), and 5%
9 ascorbic acid. The samples were diluted 1/10 or 1/5 (v/v), depending on the detection limit desired.
10 Samples were held in the diluting fluid at least 30 minutes before analysis, but overnight was preferred.
11 Analysis was performed on a Perkin-Elmer 3100 atomic absorption spectrometer (AAS) equipped with a
12 FIAS 200 flow injection system. Calibration standards were prepared in dilution fluid (10% HC1, 10%
13 KI, 5% ascorbic acid) at concentrations of 0.0, 0.2, 1.0, 5.0, 10.0, and 15.0 (ig/L.
14 The detection limit of the method was evaluated by performing 10 replicate analyses of a low
15 standard (about 1 (ig/L). The detection limit was defined as three times the standard deviation of these 10
16 analyses. A 1/10 dilution typically gave a detection limit of about 2 (ig/L, while a dilution of 1/5 typically
17 yielded a detection limit of about 1 (ig/L. All responses below the detection limit were evaluated at one-
18 half the detection limit.
19 2.2.7 Quality Assurance
20 A number of quality assurance (QA) steps were taken throughout the studies to assess and
21 document the quality of the data that were collected. These steps are summarized below.
22 Blanks
23 Blank samples analyzed with each batch of samples never yielded a measurable level of arsenic,
24 with all values being reported as less than 2.0 (ig/L of arsenic.
25 Spike Recovery
26 Randomly selected samples were spiked with known amounts of inorganic arsenic (5-20 (ig) and
27 the recovery of the added arsenic was measured. In Phase II, recovery of arsenic from spiked samples
28 typically ranged from 95 to 105%, with an average across all analyses of 99.8%. In Phase III, recovery of
29 arsenic from spiked samples typically ranged from 83 to 120%, with an average across all analyses of
30 103%.
-------
1 Laboratory Duplicates
2 Random urine samples were selected for duplicate analysis by the analyst. In Phase II, the
3 average absolute difference across all pairs of duplicates samples was 2.4 (ig/L (n = 58). In Phase III, the
4 average absolute difference across all samples was 2.3 (ig/L (n = 115).
5 Laboratory Control Standards
6 Samples of various reference materials were analyzed with each set up test samples. Results for
7 these standards are summarized below:
Reference
Material
Description
Certified
Value
Measured Results
Mean
(% Certified
Value)
Standard
Deviation
n
Phase n
ERA Potable WatR
#697 (Trace Metals,
Lot 3413)
NIST 2670 Elevated
Plain water spiked
with inorganic trace
metals
Normal human urine
spiked with inorganic
trace elements
68.8 ng/L
480 ± 100 ng/L
23.6 ng/L
(34.3%)
451 ng/L
(94%)
10.2 ng/L
12.8 ng/L
12
26
Phase HI
ERA Waste WatR
#500 (Trace Metals,
LotPOSl)
ERA Waste WatR
#500 (Trace
Metals, Lot 99 106)
ERA Waste WatR
#500 (Trace
Metals, Lot 9978)
NIST 2670 Elevated
NIST 1640
NRCC Dolt-2
NRCC Tort-2
NIST 1566b
Plain water spiked
with inorganic trace
metals
Plain water spiked
with inorganic trace
metals
Plain water spiked
with inorganic trace
metals
Normal human urine
spiked with inorganic
trace elements
Natural water
containing trace
elements (not spiked)
Dogfish liver (not
spiked)
Lobster
hepatopancreas (not
spiked)
Oyster tissue (not
spiked)
366 ng/L
347 ng/L
92.9 ng/L
480 ± 100 ng/L
0.0267 ± 0.0004
lig/g
16.6 ± 1.1 ng/g
dry wt
21.6 ± 1.8ng/g
dry wt
7.65 ± 0.65
Hg/g dry wt
361 ng/L
(98.6%)
328 ng/L
(95%)
96 ng/L
(103%)
544 ng/L
(113%)
0.027 ng/g
(99.4%)
14.7 ng/g dry wt
(88.6%)
21.3 ng/gdry wt
(98.8%)
7.6 ng/g dry wt
(99.9%)
7.2 ng/L
6.7 ng/L
1.7 ng/L
9.6 ng/L
0.001 ng/g
0.8 ng/g dry wt
1.2 ng/gdry wt
0.5 ng/g dry wt
220
38
90
7
2
10
12
13
ERA: Environmental Resource Associates
NIST: National Institute of Standards and Technology
NRCC: National Resource Council Canada (Institute for National Measurement Standards)
-------
4
5
6
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
As seen, results were good with the exception of one standard (ERA #697) in Phase II. The low
recovery from these samples is not understood.
PE Samples and Blind Duplicates
In addition to these laboratory-based (non-blind) QA procedures, a series of blind Performance
Evaluation (PE) samples (known concentrations of sodium arsenate in control urine) and blind duplicates
were submitted to the laboratory in a random fashion, commingled with normal test samples.
The combined results for samples evaluated during the Phase II pilot studies are shown in
Figure 2-3. As seen in Panel A, there was good accuracy on sodium arsenate PE samples (10, 30, and
1000 (ig/L) throughout the duration of each study. As shown in Panel B, there was also good
reproducibility between blind duplicate samples.
Initially, these QA results were interpreted to indicate that the analytical procedure was operating
correctly. However, the low recovery of arsenic for the ERA standard, as well as the observation that the
recovery of arsenic from the urine of animals administered sodium arsenate was lower than expected,
suggested that a problem did exist. In order to investigate this, a series of PE samples were prepared by
addition of three different concentrations of each of the four major urinary arsenic metabolites to control
urine, and each was analyzed in triplicate. The results are summarized below:
Urinary
Metabolite
Arsenate
Arsenite
MMA
DMA
Average Recovery
(Method 1)
101±2%
93±2%
73±3%
15±4%
As seen, recovery of inorganic forms of arsenic were within reasonable bounds, but recovery of
MMA was somewhat decreased and recovery of DMA was very poor. Based on the expectation that this
low recovery was based on incomplete conversion of MMA and DMA to inorganic arsenic prior to
hydride generation, a more vigorous digestion method was developed (see Digestion Method 2 in
Section 2.2.6). Recovery of each urinary metabolite using this new digestion method is summarized
below:
10
-------
Urinary
Metabolite
Arsenate
Arsenite
MMA
DMA
Average Recovery
(Method 2)
106±2%
106±7%
107±3%
113±3%
2 As seen, the revised digestion method yielded good recovery of all metabolites, including both
3 MMA and DMA. On this basis, the revised digestion method was used on all arsenic RBA studies
4 following the completion of Phase II.
5 The results for the Phase III PE samples are shown in Figure 2-4. As seen, the PE samples
6 included several different concentrations each of four different types of arsenic (As+3, As+5, MMA, and
7 DMA). With the exception of one unexplained outlier, there was good recovery of the arsenic from all
8 four types of PE sample.
9 The results for the blind duplicates from Phase III are shown in Figure 2-5. As seen, there was
10 good agreement between results for duplicate pairs, with an average absolute difference between pairs of
11 about 6.0 (ig/L and an average relative percent difference of about 1.5%.
12 Inter-laboratory Comparison
13 In two Phase III studies (Experiments 1 and 2), a series of samples was submitted to a second
14 laboratory for inter-laboratory comparison of results. This included investigative samples (urine samples
15 collected from study animals) as well as several PE samples. The results are shown in Figure 2-6. As
16 seen, there is generally good agreement between the two laboratories, with somewhat better
17 reproducibility for the Phase III studies.
18 Conclusion
19 Based on the results of all of the quality assurance samples and steps described above, it is
20 concluded that the analytical results for samples of urine are generally of high quality and are suitable for
21 derivation of reliable estimates of arsenic absorption from test materials. The only potential limitation is
22 that recovery of organic arsenic (especially DMA) is low in Phase II studies, which will tend to result in
23 an underestimate of UEF values. However, since RBA calculations are based on the ratio of two UEFs, if
24 both UEFs are underestimated by the same amount, then the resultant RBA may still be reliable (see
25 Section 2.3.2, below).
11
-------
1 2.2.8 Test Material Characterization
2 Table 2-3 describes the test materials for which RBA was measured in this program and provides
3 the analytical results for arsenic. Data on other Target Analyte List (TAL) metals, if available, are
4 provided in Appendix C. As seen, 27 different test materials were investigated (two in duplicate). In all
5 cases, these samples were sieved prior to analysis and dosing, and only materials which passed through a
6 60-mesh screen (corresponding to particles smaller than about 250 (im) were used. This is because it is
7 believed that soil particles less than about 250 (im are most likely to adhere to the hands and be ingested
8 by hand-to-mouth contact, especially in young children.
9 Many of the test materials1 were characterized with regard to arsenic mineral phase, particle size
10 distribution, and matrix association using electron microprobe analysis (BMPA). In this procedure, an
11 electron microprobe with combined energy dispersive spectrometer (EDS) and multiple wavelength
12 dispersive spectrometers (WDS) was used to evaluate the elemental composition of arsenic-bearing
13 particles. A 1 to 2 gram split of dried sample was placed in a 2.5 cm plastic mold and impregnated with
14 epoxy. Once the sample was hardened, it was polished and carbon coated for EMPA. TheEMPAwas
15 operated at 15 kV accelerating voltage, with a 20 nA current and a 1 micron focused beam. Instrument
16 response was calibrated using certified mineral or pure metal standards and counting times were chosen to
17 provide 3-sigma detection limits of between 100-200 ppm. Elemental concentrations were corrected
18 using ZAP factors and concentration errors were generally less than 5% relative. For a more detailed
19 explanation of the EMPA method of analyses see Birks (1971) or Heinrich (1981).
20 Although the electron microprobe is capable of determining the precise stoichiometry of the
21 elements in any given particle, this was not attempted in this project. This is mainly because investing
22 time in obtaining precise stoichiometry decreases the number of different particles that can be examined.
23 In addition, many arsenic-bearing particles are not composed of a pure mineral phase with an exact
24 stoichiometry, but are characterized by arsenic that is either adsorbed onto other mineral particles, or is a
25 mixture of phases that are undergoing transition from one phase to another. For this reason, particles
26 were classified into "phases" that may not be purely stoichiometric and may contain a mixture of similar
27 chemical phases. The first step used in the assignment of a phase designation was to determine if the
1 Arsenic was not speciated in three Phase II samples (Aspen Berm, Aspen Residential, and Jasper County High
Lead Mill) because the concentration of arsenic in each material was too low (17 ppm, 67 ppm, and 16 ppm, respectively)
to allow reliable evaluation In addition, speciation data were unavailable for four Phase III samples (El Paso TM1, El
Paso TM2, ACC Utility Pole Soil, and ACC Dislodgeable Arsenic).
12
-------
1 phase was an oxide, carbonate, sulfide, sulfate, or phosphate. Secondly, with the exception of the
2 "phosphates," the major cation associated with the phase was identified. Therefore, phases such as
3 Fe-sulfate, FeOOH, MnOOH, PbMO, AsMO, or PbMSO4 were identified (where M represents "metal").
4 Some of these phases could represent a stoichiometric mineral form, but most are likely to be metastable
5 and/or amorphous and have some quantity of arsenic sorbed to their surface.
6 The "phosphate" group is even more generic in that the only common dominant ion is PO4.
7 Although arsenic and phosphorous are both oxy-anions, a number of particles that contain both arsenic
8 and phosphate have been identified. As above, these might include minerals that contain mixtures of
9 phosphate and arsenate such as walentaite (Ca,Mn,Fe)Fe3(AsO4,PO4)4-7H2O, morelandite (Ba,Ca,Pb)5
10 Cl[AsO4,PO4]3, or turneaureite Ca5(Cl)[(AsO4, PO4)3], but more likely represent arsenic adsorbed onto
11 other phosphate-containing particles.
12 Detailed EMPA results are presented in Appendix C and the results, expressed as relative arsenic
13 mass, are summarized in Table 2-4. The relative arsenic mass for a particular phase is the estimated
14 percentage of the total arsenic in a sample that is present in that phase. Of the 28 different phases
15 detected in one or more samples, 14 are relatively minor, with relative arsenic mass values less than 5%.
16 However, the remaining 14 phases occur at concentrations that could contribute significantly to the
17 bioavailability of the sample.
18 Table 2-5 summarizes data on the size distribution of arsenic-containing particles (measured as
19 the longest dimension) in each sample. As seen, most samples contain a range of particle sizes, with the
20 majority of particles being less than 50 (im in diameter.
21 Table 2-6 summarizes information on the degree to which arsenic-bearing grains in each sample
22 are partially or entirely exposed on their outer surfaces (liberated), or are entirely enclosed within a larger
23 particle of rock or slag (included). Data are presented both on a simple particle frequency basis and on
24 the basis of relative arsenic mass. As seen, the majority of arsenic-bearing particles in all samples are
25 partly or entirely liberated.
26 In interpreting the results of the particle speciation studies, it is important to understand that, on a
27 mass basis, only a tiny fraction of the total sample is evaluated by electron microprobe and, hence, there
28 is moderate uncertainty as to whether the results for the grains examined are truly representative of the
29 sample as a whole.
30 It is also worth noting that other speciation methods are available to determine the chemical
31 forms of metals in soil systems. Each method has distinct advantages and disadvantages; and some
32 methods provide more robust data than others (see D'Amore et al. 2005). One such technique is X-ray
13
-------
1 absorption spectroscopy (XAS) for which USEPA Office of Research and Development (ORD) has
2 resident experts to conduct studies and the service is available to support Regional research efforts. XAS
3 probes the sub-atomic structure of elements to distinguish specific bonding mechanisms which leads to
4 precise determination of metal speciation. An example for As is differentiation of As sorbed to an iron
5 oxide versus As present as the mineral scorodite (FeAsO4) for which XAS can easily identify the different
6 phases that have vastly different bioavailability behaviors whereas EMPA will identify both phases as
7 containing As, Fe, and O.
8 2.3 Results
9 2.3.1 RBA Estimates
10 Detailed raw data for each study are provided in Appendix D. Results of simultaneous weighted
11 linear regression fitting and RBA calculations are presented in Appendix E. The results are summarized
12 below.
13 The upper portion of Table 2-7 summarizes the RBA results for all Phase II studies, and the lower
14 portion summarizes the results for materials studied during Phase III. As seen, using sodium arsenate as a
15 relative frame of reference, estimated RBA values range from 8% to more than 100%. This wide
16 variability supports the conclusion that there can be important differences in RBA between different types
17 of samples and that use of a site-specific RBA value is likely to increase the accuracy of risk estimates for
18 arsenic. Available data do not include replicate estimates of RBA of the same test materials; therefore,
19 there is no empirical basis for estimating variability in the RBA estimates that might be attributable to
20 within-test material variability as opposed to between-test material variability. Although ABA of As is
21 not estimated in the data reduction procedure for the swine assays, RBA is estimated as the ratio of the
22 slopes of the dose-UEF relationships for sodium arsenate and the test material. Table 2-8 provides
23 summary statistics for the dose-UEF slopes for sodium arsenate and all test materials assayed in the
24 Region 8 Phase III studies. The coefficient of variation (SD/mean) for the sodium arsenate slopes is
25 approximately 0.13 (N=7). This variability reflects an unknown combination of biological variability in
26 As bioavailability and other assay variables that contribute to variability in the measurement of the dose-
27 UEF slope. The coefficient of variability for the dose-UEF slopes for the test materials is 0.38 (N=14),
28 and is greater than that for sodium arsenate by a factor of approximately 3. The difference in the two
29 estimates reflects, at least in part, the additional variability introduced into the dose-UEF slope estimates
30 contributed by differences in bioavailability of the test materials. This outcome suggests that test material
31 characteristics contribute substantially to the observed variability in RBA estimates. This conclusion is
32 also consistent with the similarity between the coefficient of variability of the dose-UEF slope for test
33 materials (0.38) and the estimated RBAs for the same test materials (0.32).
14
-------
1 Figure 2-7 shows that the uncertainty in the RBA value for a test material (as reflected by the
2 difference between the upper bound and the lower bound) depends on the dose of arsenic administered in
3 the study. As seen, three of the test materials (Aspen Berm, Aspen Residential, and Jasper County High
4 Lead Mill) were administered only at low dose levels (less than 20 (ig/kg bw-day) and have extremely
5 wide uncertainty bounds around the RBA estimates. This is due mainly to the fact that the concentrations
6 of arsenic in the urine were very low and, hence, were difficult to quantify with good accuracy and also
7 difficult to distinguish from baseline. Because of the high uncertainty in these results, the data from these
8 three test materials are not considered further. Thus, based on these results, a minimum daily As dose of
9 25 (ig/kg-bw/day is recommended to ensure the amount if excreted in urine reaches a measurable quantity
10 and, that is to minimize uncertainty in RBA estimates.
11 2.3.2 Effect of Low Analytical Recovery on Phase II RBA Values
12 As noted above, all of the calculations of arsenic RBA performed during Phase II are based on
13 data obtained using an analytical method that had low recovery of organic metabolites of arsenic, which
14 raises a concern over the accuracy of the results. However, the low recovery of arsenic is not necessarily
15 a basis for complete distrust of the results. This is because the RBA is a ratio of two measured values,
16 and if the degree of error (underestimation) is the same in both the numerator and denominator, then the
17 error will cancel and the resulting ratio will be correct. However, the degree of error in each
18 measurement depends on the relative concentration of the metabolites in the urine: if the level of MMA
19 and DMA is low, the error will be smaller than if the levels of MMA and DMA are high. Thus, the key
20 question is whether or not the ratio of the urinary metabolites tends to be relatively constant as a function
21 of dose and dose material, at least over the range of exposures investigated in the Phase II studies.
22 The most direct approach for testing this question is to measure the relative concentration of each
23 metabolite (As+3, As+5, MMA, DMA) in urine from a number of animals exposed to a series of different
24 dose levels and dose materials. This approach was attempted, but the results for quality control samples
25 indicated that the results were not reliable, presumably due to the technical difficulty of performing the
26 separation and quantification of the individual metabolites. Therefore, this approach was not pursued
27 further.
28 An alternative approach is to measure the UEF and RBA of several test materials using both
29 analytical methods, and to compare the results. This approach was implemented for two different test
30 materials (Butte TM1 and Butte TM2), and the results are shown below:
31
15
-------
Substance
Administered
Sodium Arsenate
Butte TM1
Butte TM2
Digestion Method 1
UEF
0.238
0.047
0.056
RBA
[1.00]
0.20
0.23
Digestion Method 2
UEF
0.890
0.158
0.210
RBA
[1.00]
0.18
0.24
2 As seen, the measured UEF for sodium arsenate based on Digestion Method 1 (24%) is much
3 lower than the UEF based on Digestion Method 2 (89%). However, the UEF of each of two different soil
4 test materials was also lower by approximately the same relative amount when measured by Digestion
5 Method 1 compared to Digestion Method 2, so the ratio (the RBA) was approximately constant when
6 calculated for each method. These results indicate that, even though the low recovery of arsenic in Phase
7 II studies is a basis for uncertainty in the RBA estimates derived during Phase II, the error due to low
8 recovery of organic metabolites of arsenic is likely to approximately cancel, and the final RBA estimates
9 are likely to be approximately correct. For this reason, the Phase II data were included in the overall
10 estimates of As RBA.
11 2.3.3 Effect of Food on Arsenic Absorption
12
13
14
15
16
17
18
19
20
21
22
23
In Phase II Pilot Study 2 (Experiment 15), some animals were dosed with NaAs via gavage in
order to compare the results with NaAs given in orally in doughballs. These results are shown below:
Substance
Administered
NaAs - Gavage
NaAs - Doughball
UEF
Slope
0.189
0.177
SEM
0.014
0.014
N
31
31
As seen, the UEF for sodium arsenate administered orally in a doughball is only slightly lower
than the UEF for sodium arsenate administered by gavage, indicating that the amount of feed (about 5
grams) used to administer the arsenic doses does not significantly affect arsenic absorption.
2.4 Correlation of RBA with Arsenic Geochemistry
One objective of this project was to obtain preliminary information on which mineral and
chemical forms of arsenic tend to have high bioavailability and which tend to have low bioavailability.
As noted above, data on chemical form or mineral association were obtained using EMPA. Detailed data
are presented in Appendix C and results are summarized in Section 2.2.8 and in Tables 2-4 to 2-6.
16
-------
1 In order to derive quantitative estimates of phase-specific RBA values, a multivariate linear
2 regression approach was used, employing the following basic model:
3 RBA=lu(f-RBAi}
4 where:
5 fi = Fraction of total arsenic present in phase "/'"
6 RBAi = Inherent RBA of phase"/"
7
8 However, because a total of 28 different phases were identified and reliable RBA results were
9 obtained for only 20 different samples, it is clear that the existing data are not sufficient to perform a
10 robust regression analysis. Instead, a screening-level analysis was performed, as follows. First, in order
11 to reduce the number of independent variables, the 28 different phases were grouped into 9 categories as
12 described in Table 2-9. These categories were based on professional judgment regarding the expected
13 degree of similarity between members of a group, along with information on the relative abundance of
14 each phase (see Table 2-4). Phases with low relative arsenic mass (maximum relative mass in any test
15 material less than 15%) were grouped together under "Minor Constituents;" these phases included AsMO,
16 AsMSO4, Clays, Paint, Pb Solder, Pb-As Vanidate, PbAsMO, PbAsSbCuO, PbCrO4, PbMO, PbMS,
17 PbMSO4, Pyrite, TiO2, and ZnSiO4. Next, the fraction of arsenic present in each group was calculated by
18 summing the relative arsenic mass for each phase in the group. Based on the expectation that particles
19 that are totally included (fully enclosed or encased in mineral or vitreous matrices) are not likely to
20 contribute significantly to the observed RBA value of a sample, only the relative arsenic mass in partially
21 or entirely liberated particles (partially or entirely exposed on their outer surfaces) was included in the
22 sum. The results are shown in Table 2-10.
23 Group-specific RBA values were then estimated by fitting the grouped data to the model using
24 minimization of square errors. Two different options were employed. In the first option, each fitting
25 parameter (group-specific RBA) was fully constrained to be between zero and one, inclusive. In the
26 second option, all parameters were unconstrained. Because the minor constituents do not contribute
27 significantly to the total arsenic mass in any of the tested materials, a reasonable estimate of their specific
28 RBA cannot be obtained. Therefore, an arbitrary coefficient of 0.5 was assumed for this group and the
29 coefficient was not treated as a fitting parameter. The resulting estimates of the group-specific average
30 RBA values for the remaining groups are shown in Table 2-11 (these values apply only to liberated
31 particles).
17
-------
1 As seen, there is a wide range of group-specific RBA values, with the precise values depending
2 on the method used to constrain the parameters. It is important to stress that these group-specific RBA
3 estimates are derived from a very limited data set, so the group-specific RBA estimates are inherently
4 very uncertain. In addition, both the measured sample RBA values and the relative arsenic mass in each
5 phase are subject to additional uncertainty. Therefore, the group-specific RBA estimates should not be
6 considered to be highly precise, and calculation of a quantitative sample-specific RBA value from these
7 estimates is not appropriate. Rather, it is more appropriate to consider the results of this study as
8 sufficient to support only a qualitative classification of phase-specific RBA values, as follows:
Low Unavailability
As2O3
Sulfosalts
Medium Unavailability
As Phosphate
FeAs Oxide
PbAs Oxide
MnAs Oxide
Fe and Zn Sulfates
High Unavailability
FeAsO
10
11 2.5 Discussion of In Vivo Results
12 The results of this investigation indicate that juvenile swine are a useful model for quantifying
13 gastrointestinal absorption of arsenic from different test materials, using urinary arsenic excretion as the
14 measurement endpoint. In addition, this experimental protocol can be used to estimate lead and arsenic
15 RBA in the same animals. Because of the size of juvenile swine (about 10 kg at the beginning of the
16 study), it is usually possible to administer doses of test soils that are relatively close to the range thought
17 to be of concern to humans. For example, in Pilot Study 1 (Phase II, Experiment 10), the low dose of slag
18 administered averaged about 260 mg/day, only slightly higher than the reasonable maximum exposure
19 (RME) value of 200 mg/day assumed for human children (U.S. EPA, 1991). Thus, most measurements
20 are obtained in a portion of the dose-response curve that is more relevant to humans than is achieved in
21 most other animal models.
22 Most studies of arsenic absorption employ a single dose protocol and measure urinary excretion
23 for 2-3 days. In contrast, these studies employed a repeated dosing protocol, with repeated 24- or
24 48-hour urine collections. An advantage of this protocol is that it reflects a more realistic human
25 exposure scenario than does a single dose protocol. Further, multiple measurements can be made from
26 the same animal on different days. In essence, data from different days allow multiple independent
27 estimates of the UEF, and these data can be combined (once steady state has been achieved) to provide a
28 robust estimate of the excretion fraction.
18
-------
1 The RBA results for different test materials investigated strongly support the view that absorption
2 of arsenic from soils and mine wastes is highly variable, and generally is not as well absorbed as soluble
3 arsenic. The detailed chemical mechanism accounting for this variable and reduced bioavailability of
4 arsenic in soil-like media is not known, but almost certainly is related to the chemical form of arsenic in
5 the sample.
6 Because arsenic in most test materials is absorbed less-extensively than soluble forms of arsenic,
7 and because soluble forms of arsenic are the basis of the oral RfD and oral slope factor for arsenic, the use
8 of the unadjusted toxicity factors for assessing human health risk from soil ingestion will usually lead to
9 an overestimate of risk. Consequently, measurement and application of site-specific RBA values to adjust
10 the toxicity factors to account for the lower level of absorption is expected to increase the accuracy and
11 decrease the uncertainty in human health risk assessments for arsenic in soil.
12 3.0 CONCLUSIONS
13 The data from the investigations performed under this program support the following main
14 conclusions:
15 1. Juvenile swine constitute a useful and stable animal model for measuring the relative
16 bioavailability of arsenic in a variety of soil or soil-like test materials. The Phase III protocol
17 described in this report is the recommended SOP for the juvenile swine RBA assay.
18 2. There are clear differences in the in vivo RBA of arsenic between different test materials, ranging
19 from less than 10% to more than 60%. Thus, knowledge of the RBA value for different materials
20 at a site can be very important for improving arsenic risk assessments at a site.
21 3. Available data are not yet sufficient to allow reliable calculation of the RBA for a test material
22 based only on knowledge of the relative amounts of the arsenic mineral phases present.
23 However, tentative qualitative estimates of low, medium, or high bioavailability have been made
24 based on the major phase type of the arsenic containing waste material.
25 4. For analysis of total arsenic in urine, additional extraction steps were identified and necessary to
26 convert urinary organoarsenic metabolites to inorganic arsenic.
27 5. Due to limitations in detection limits for measurement of arsenic in urine, a minimum arsenic
28 dose of 25 (ig/kg bw-day is recommended for the juvenile swine RBA assay, so that the amount
29 of arsenic excreted in urine reaches a measurable quantity.
19
-------
1 4.0 REFERENCES
2 Bettley, F.R. and O'Shea, J.A. 1975. The absorption of arsenic and its relation to carcinoma. Br. J.
3 Dermatol. 92:563-568.
4 Birks, L.S. 1971. Electron Probe Microanalysis, 2nd ed. New York: Wiley-Interscience.
5 Buchet, J.P., Lauwerys, R., and Roels, H. 1981a. Comparison of the urinary excretion of arsenic
6 metabolites after a single oral dose of sodium arsenite, monomethyl arsonate or dimethyl arsinate in man.
7 Int. Arch. Occup. Environ. Health 48:71-79.
8 Buchet, J.P., Lauwerys, R., and Roels, H. 1981b. Urinary excretion of inorganic arsenic and its
9 metabolites after repeated ingestion of sodium meta arsenite by volunteers. Int. Arch. Occup. Environ.
10 Health 48:111-118.
11 Charbonneau, S.M., Spencer, K., Bryce, F., and Sandi, E. 1978. Arsenic excretion by monkeys dosed
12 with arsenic-containing fish or with inorganic arsenic. Bull. Environ. Contam. Toxicol. 20:470-477.
13 Coulson, E.J., Remington, R.E., and Lynch, K.M. 1935. Metabolism in the rat of the naturally occurring
14 arsenic of shrimp as compared with arsenic trioxide. J. Nutrition 10:255-270.
15 Crecelius, E.A. 1977. Changes in the chemical speciation of arsenic following ingestion by man. Environ.
16 Health Perspect. 19:147-150.
17 D'Amore, J.M., Al-Abed, S.R., Scheckel, K.G. and Ryan, J.A. 2005. Methods for Speciation of Metals in
18 Soils: A Review. J. Environ. Qual. 34: 1707-1745.
19 Freeman, G.B., Johnson, J.D., Liao, S.C., Feder, P.I., Davis, A.O., Ruby, M.V., Schoof, R.A., Chaney,
20 R.L., and Bergstrom, P.D. 1994. Absolute bioavailability of lead acetate and mining waste lead in rats.
21 Toxicology 91:151-163.
22 Heinrich, K.F.J. 1981. Electron Beam X-ray Microanalysis. New York: Van Nostrand.
23 Johnson, L.R. and Farmer, J.G. 1991. Use of human metabolic studies and urinary arsenic speciation in
24 assessing arsenic exposure. Bull. Environ. Contam. Toxicol. 46:53-61.
25 Mappes, R. 1977. Experiments on excretion of arsenic in urine. Int. Arch. Occup. Environ. Health
26 40:267-272.
27 Marafante, E. and Vahter, M. 1987. Solubility, retention and metabolism of intratracheally and orally
28 administered inorganic arsenic compounds in the hamster. Environ. Res. 42:72-82.
29 Roberts, S.M., Weimar, W.R., Vinson, J.R., Munson, J.W., and Bergeron, R.J. 2002. Measurement of
30 arsenic bioavailability in soil using a primate model. Toxicol. Sci. 67(2): 303-310.
31 Roberts, S.M., Munson, J.W., Lowney, Y.W., and Ruby, M.V. 2007. Relative oral bioavailability of
32 arsenic from contaminated soils measured in the cynomolgus monkey. Toxicol. Sci. 95(1): 281-288.
33 Tarn, G.K.H., Charbonneau, S.M., Bryce, F., Pomroy, C., and Sandi, E. 1979. Metabolism of inorganic
34 arsenic (74As) in humans following oral ingestion. Toxicol. Appl. Pharmacol. 50:319-322.
35 U.S. EPA (U.S. Environmental Protection Agency). 1991. Human Health Evaluation Manual,
36 Supplemental Guidance: Standard Default Exposure Factors. United States Environmental Protection
37 Agency, Office of Solid Waste and Emergency Response. Washington, DC. OSWER Directive 9285.6-
38 03. March 25, 1991. Available online at:
39 http://www.epa.gov/oswer/riskassessment/pdf/defaultExposureParams.pdf.
20
-------
1 U.S. EPA (U.S. Environmental Protection Agency). 2007. Estimation of Relative Bioavailability of Lead
2 in Soil and Soil-Like Materials by In Vivo and In Vitro Methods. United States Environmental Protection
3 Agency, Office of Solid Waste and Emergency Response. Washington, DC. OSWER 9285.7-77.
4 Available online at:
5 http://www.epa.gov/superfund/health/contaminants/bioavailability/lead tsdmain.pdf.
6 U.S. EPA (U.S. Environmental Protection Agency). 2009. Arsenic, inorganic. Integrated Risk
7 Information System (IRIS). U.S. Environmental Protection Agency. National Center for Environmental
8 Assessment. Washington, DC. Available online at: http://www.epa.gov/ncea/iris/subst/0278.htm.
9 Vahter, M. 1981. Biotransformation of trivalent and pentavalent inorganic arsenic in mice and rats.
10 Environ. Res. 25:286-293.
11 Vahter, M., and Norin, H. 1980. Metabolism of 74As-labeled trivalent and pentavalent inorganic arsenic in
12 mice. Environ. Res. 21:446-457.
13 Weis, C.P., and LaVelle, J.M. 1991. Characteristics to consider when choosing an animal model for the
14 study of lead bioavailability. In: Proceedings of the International Symposium on the Bioavailability and
15 Dietary Uptake of Lead. Sci. Technol. Let. 3:113-119.
16 Yamauchi, H. and Yamamura, Y. 1985. Metabolism and excretion of orally administrated arsenic trioxide
17 in the hamster. Toxicology 34:113-121.
21
-------
Table 2-1. Summary of Arsenic Excretion Studies in Humans and Animals Exposed to
Soluble Arsenic Compounds in Water
Species
Human
Human
Human
Human
Human
Human
Human
Human
Human
Human
Human
Human
Hamster
Mouse
Mouse
Mouse
Mouse
Mouse
Mouse
Monkey
Monkey
Monkey
Hamster
Hamster
Mouse
Mouse
Mouse
Mouse
Mouse
Mouse
Sex
M,F
M
M
M
M
M
NS
M
M
M
M
M
NS
M
M
M
M
M
M
F
M
M
M
NS
M
M
M
M
M
M
N
4
3
1
1
1
1
2
1
1
1
6
2
4
5
5
5
5
5
5
4
5
7
5
4
5
5
5
5
5
5
Chemical
Form
NS
NaAsO2
NaAsO2
NaAsO2
NaAsO2
NaAsO2
As2O3
As2O3
Mixture
Na2HAsO4
Na2HAsO4
Na2HAsO4
NaAsO2
NaAsO2
NaAsO2
NaAsO2
NaAsO2
NaAsO2
NaAsO2
As2O3
Na2HAsO4
Na2HAsO4
As203
Na2HAsO4
Na2HAsO4
Na2HAsO4
Na2HAsO4
Na2HAsO4
Na2HAsO4
Na2HAsO4
Dose
Jig/day
8520
500
125
250
500
1000
1000
760
63
200
0.01
220
2000
400
4000
40
400
2000
4000
1000
360
50-200
4500
2000
400
4000
40
400
2000
4000
Days
Exposed
1
1
5
5
5
5
1
5
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
Percent Recovered
Urine
NA
45
54
73
74
64
85
70
80
50
58
67
36
90
65
88
91
86
75
73
49
40
49
74
77
89
94
93
92
85
Feces
4
NA
NA
NA
NA
NA
1.4
NA
NA
NA
NA
NA
49
7
9
NA
NA
NA
NA
NA
2
42
11
12
8
6
NA
NA
NA
NA
Days
10
4
14
14
14
14
5
22
3
3
6
7
3
2
2
2
2
2
2
14
4
4
5
3
2
2
2
2
2
2
Reference
Bettley and O'Shea 1975
Buchetetal. 1981a
Buchetetal. 1981b
Coulsonetal. 1935
Mappes 1977
Crecelius 1977
Tametal. 1979
Johnson and Farmer
1991
Marafante and Vahter
1987
Vahter and Norm 1980
Vahter 1981
Charbonneau 1978
Roberts et al. 2002
Roberts et al. 2007
Yamauchi and
Yamamura 1985
Marafante and Vahter
1987
Vahter and Norm 1980
Vahter 1981
-------
Table 2-2. Typical Swine Feed Composition
Nutrient Name
Protein
Arginine
Lysine
Methionine
Met+Cys
Tryptophan
Histidine
Leucine
Isoleucine
Phenylalanine
Phe+Tyr
Threonine
Valine
Fat
Saturated Fat
Unsaturated Fat
Linoleic 18:2:6
Linoleic 18:3:3
Crude Fiber
Ash
Calcium
Phos Total
Available Phosphorous
Sodium
Potassium
Amount
20.10%
1.21%
1.47%
0.84%
0.59%
0.28%
0.56%
1.82%
1.13%
1.11%
2.05%
0.82%
1.19%
4.44%
0.56%
3.74%
1.94%
0.04%
3.80%
4.33%
0.87%
0.77%
0.70%
0.24%
0.37%
Nutrient Name
Chlorine
Magnesium
Sulfur
Manganese
Zinc
Iron
Copper
Cobalt
Iodine
Selenium
Nitrogen Free Extract
Vitamin A
Vitamin D3
Vitamin E
Vitamin K
Thiamine
Riboflavin
Niacin
Pantothenic Acid
Choline
Pyridoxine
Folacin
Biotin
Vitamin B 12
Amount
0.19%
0.05%
0.03%
20.4719 ppm
1 18.0608 ppm
135.3710 ppm
8. 1062 ppm
0.01 10 ppm
0.2075 ppm
0.3 196 ppm
60.23%
5.1892kIU/kg
0.6486 klU/kg
87.2080 lU/kg
0.9089 ppm
9. 1681 ppm
10.2290 ppm
30. 1147 ppm
19. 1250 ppm
1019.8600 ppm
8.2302 ppm
2.0476 ppm
0.2038 ppm
23. 44 16 ppm
Feed obtained from and nutritional values provided by Zeigler Bros., Inc
23
-------
Table 2-3. Description of Test Materials
Phase
II
Experiment
2
4
5
6
7
Sand
10 (Pilot 1)
Sample
Designation
Bingham Creek
Channel Soil
Jasper County High
Lead Mill
Murray Smelter Slag
Aspen Berm
Aspen Residential
Butte Soil
Midvale Slag
California Gulch
Phase I Residential
Soil
California Gulch
Fe/MnPbO
California Gulch AV
Slag
Site
Kennecott NPL Site,
Salt Lake City, Utah
Jasper County,
Missouri Superfund
Site
Murray Smelter
Superfund Site
Smuggler Mountain
NPL Site, Aspen,
Colorado
Smuggler Mountain
NPL Site, Aspen,
Colorado
Silver Bow
Creek/Butte Area NPL
Site, Butte, Montana
Midvale Slag NPL
Site, Midvale, Utah
California Gulch NPL
Site, Leadville,
Colorado
California Gulch NPL
Site, Leadville,
Colorado
California Gulch NPL
Site, Leadville,
Colorado
Sample Description
Soil composite of samples containing
3000 ppm or greater of lead; collected
from a residential area (Jordan View
Estates) located along Bingham Creek in
the community of West Jordan, Utah
Soil composite collected from an on-site
location
Composite of samples collected from
areas where exposed slag existed on site
Composite of samples collected from
the Racquet Club property (including a
parking lot and a vacant lot)
Composite of samples collected from
residential properties within the study
area
Soil composite collected from waste
rock dumps in Butte Priority Soils
Operable Unit (BPSOU)
Composite of samples collected from a
water-quenched slag pile in Midvale
Slag Operable Unit 2
Soil composite collected from
residential properties within Leadville
Soil composite collected from near the
Lake Fork Trailer Park located
southwest of Leadville near the
Arkansas River
Sample collected from a water-
quenched slag pile on the property of the
former Arkansas Valley (AV) Smelter,
located just west of Leadville
Arsenic
Concentration3
(ppm)
149
16
695
67
17
234
591
203
110
1050
Lead
Concentration3
(ppm)
6330
6940
11,700
14,200
3870
8530
8170
7510
4320
10,600
24
-------
Table 2-3. Description of Test Materials
Phase
III
Experiment
9
11
15 (Pilot 2)
1
2
3
Sample
Designation
Palmerton Location 2
Palmerton Location 4
Murray Smelter Soil
Clark Fork Tailings
VBI70 TM1
VBI70 TM2
VBI70 TM3
VBI70 TM4
VBI70 TM5
VBI70 TM6
Butte TM1
Site
New Jersey Zinc NPL
Site, Palmerton,
Pennsylvania
New Jersey Zinc NPL
Site, Palmerton,
Pennsylvania
Murray Smelter
Superfund Site
Milltown Reservoir
Sediments NPL Site,
Milltown, Montana
Vasquez Boulevard
and 1-70 NPL Site,
Denver, Colorado
Vasquez Boulevard
and 1-70 NPL Site,
Denver, Colorado
Vasquez Boulevard
and 1-70 NPL Site,
Denver, Colorado
Vasquez Boulevard
and 1-70 NPL Site,
Denver, Colorado
Vasquez Boulevard
and 1-70 NPL Site,
Denver, Colorado
Vasquez Boulevard
and 1-70 NPL Site,
Denver, Colorado
Silver Bow
Creek/Butte Area NPL
Site, Butte, Montana
Sample Description
Soil composite collected from on-site
Soil composite collected from on-site
Soil composite collected from on-site
Sample collected from a tailings deposit
along the banks of the Clark Fork River
on the property of the Grant-Kohrs
Ranch near Deer Lodge, Montana
Soil composite from impacted
residential property (Eastern
Swansea/Elyria neighborhood)
Soil composite from impacted
residential property (Western
Swansea/Elyria neighborhood)
Soil composite from impacted
residential property (Eastern Cole
neighborhood)
Soil composite from impacted
residential property (Western Cole
neighborhood)
Soil composite from impacted
residential property (Clayton
neighborhood)
Clean site soil (from the Swansea/Elyria
neighborhood) plus added PAX
pesticide
Soil composite collected from waste
rock dumps in Butte Priority Soils
Operable Unit (BPSOU)
Arsenic
Concentration3
(ppm)
110
134
310
181
312
983
390
813
368
516
234
Lead
Concentration3
(ppm)
3230
2150
3200
733
824
236
541
157
264
7980
25
-------
Table 2-3. Description of Test Materials
Phase
Experiment
4
5
6
Sample
Designation
Butte TM2
Aberjona River TM1
Aberjona River TM2
El Paso TM1
El Paso TM2
ACC Utility Pole Soil
Site
Silver Bow
Creek/Butte Area NPL
Site, Butte, Montana
Wells G & H
Superfund Site,
Woburn,
Massachusetts
Wells G & H
Superfund Site,
Woburn,
Massachusetts
El Paso/Dona Ana
County Metals Survey
site, El Paso County,
Texas, and Dona Ana
County, New Mexico
El Paso/Dona Ana
County Metals Survey
site, El Paso County,
Texas, and Dona Ana
County, New Mexico
- (Study sponsored by
American Chemistry
Council)
Sample Description
Soil composite collected from a
residential property located adjacent to a
railroad grade in Butte, Montana
Composite of sediment samples
containing arsenic concentrations
greater than 500 ppm, collected along
the Aberjona River, Massachusetts
Composite of sediment samples
containing arsenic concentrations from
180 to 460 ppm, collected along the
Aberjona River, Massachusetts
Soil sample collected approximately 1.5
miles east of the American Canal in El
Paso County, Texas
Soil sample collected approximately 1.5
miles east of the American Canal in El
Paso County, Texas
Soil affected by chromated copper
arsenate (CCA)-treated wood utility
poles from a test plot in Conley, Georgia
(soil was affected by being adjacent to
the poles for over ten years)
Arsenic
Concentration3
(ppm)
367
676
313
74
73
320
Lead
Concentration3
(ppm)
492
410
350
NM
NM
NM
26
-------
Table 2-3. Description of Test Materials
Phase
Experiment
7
Sample
Designation
ACC Dislodgeable
Arsenic
Site
- (Study sponsored by
American Chemistry
Council)
Sample Description
Dislodgeable material obtained from the
surface of chromated copper arsenate
(CCA)-treated wood (boards from in-
service residential decks, aged outdoors
for one to three years)
Arsenic
Concentration3
(ppm)
3500
Lead
Concentration3
(ppm)
NM
aValues are arithmetic means
All samples were analyzed by ICP/AES in accord with EPA Method 2007.
NM = Not Measured
27
-------
Table 2-4. Relative Mass of Arsenic By Mineral Phase in Test Materials
Ol
03
£
i
Experiment
2
4
6
7
8
11
"3.
03
to
Bingham
Creek
Channel
Soil
Murray
Smelter
Slag
Butte Soil"
Midvale
Slag
California
Gulch
Phase I
Residentia
ISoil
California
Gulch
Fe/Mn
PbO
California
Gulch AV
Slag
Palmerton
Location 2
Palmerton
Location 4
Murray
Ol
tj
e
03
Number of P
Counted
430
1108
636
1847
510
380
1472
111
105
355
Phase
£
03
O.
0
b£
-------
Table 2-4. Relative Mass of Arsenic By Mineral Phase in Test Materials
Ol
03
II
Experiment
15
1
2
3
"3.
03
to
Smelter
Soil
Clark
Fork
Tailings
VBI70
TM1
VBI70
TM2
VBI70
TM3
VBI70
TM4
VBI70
TM5
VBI70
TM6
Butte
TM2
Aberjona
River
TM1
Aberjona
River
TM2
<*l
(J
"5
Number of Pa
Counted
238
261
128
97
139
103
124
137
186
123
Phase
As Phosphate
16%
8%
4%
2%
<1%
<1%
0
54%
22%
80%
86%
97%
80%
O
<.
AsSbO
<,»
1%
Ol
03
2
'3
03
s
PbAsMO
PbAsSbCuO
AsMSO4
<.
£
1
<1%
03
u
,,,.
<,s
<«
,.
<,,.
,K,
Ol
2
40%
3%
3%
8%
»
3%
<,»
39%
69%
16%
FeAs Sulfate
24%
<,»
<,«
,,,.
<,«
18%
29%
27%
0
to
2%
55%
FeAsO
2%
Ol
a
6
-------
Table 2-4. Relative Mass of Arsenic By Mineral Phase in Test Materials
O)
03
£
^
Experimen
"3.
03
Ol
'£
03
a.
Number ol
Counted
Phase
£
0
0
O
AsSbO
Ol
03
a
=
03
£
PbAsMO
O
PbAsSbCu
AsMSO4
£
1
03
u
-S
~3
ft!
0
FeAsO
Ol
6
Ol
a
8
1
O
1
1
PbMS04
I1
_03
Sulfo salts
!
1
-
Ol
2
fi
9
u
fi
o
H
0
1 Same sample as evaluated in Phase III Experiment 3 (Butte TM2
30
-------
Table 2-5. Size Distributions of Arsenic Particles
Phase
II
III
Experiment
2
4
6
7
8
9
11
15
1
2
3
4
Sample
Bingham Creek Channel Soil
Murray Smelter Slag
Butte Soil3
Midvale Slag
California Gulch Phase I Residential Soil
California Gulch Fe/Mn PbO
California Gulch AV Slag
Palmerton Location 2
Palmerton Location 4
Murray Smelter Soil
Clark Fork Tailings
VBI70 TM1
VBI70 TM2
VBI70 TM3
VBI70 TM4
VBI70 TM5
VBI70 TM6
Butte TM2
Aberjona River TM1
Aberjona River TM2
Particle Size (jim)
0-5
71%
14%
21%
3%
22%
35%
21%
40%
21%
18%
34%
81%
59%
49%
45%
48%
63%
18%
33%
59%
6-10
14%
15%
9%
1%
16%
24%
9%
26%
28%
31%
20%
9%
20%
21%
32%
18%
23%
11%
34%
9%
11-20
6%
4%
16%
2%
14%
13%
2%
12%
18%
17%
17%
7%
10%
18%
13%
24%
6%
20%
6%
15%
21-50
6%
15%
26%
13%
22%
17%
11%
15%
19%
10%
21%
3%
9%
11%
9%
10%
6%
30%
13%
9%
51-100
3%
24%
17%
19%
16%
9%
12%
7%
13%
12%
7%
2%
1%
1%
2%
18%
6%
6%
101-150
<1%
23%
9%
40%
6%
2%
18%
<1%
7%
1%
<1%
4%
4%
2%
151-200
2%
1%
6%
1%
14%
3%
<1%
<1%
201-250
3%
<1%
14%
1%
7%
1%
2%
>250
<1%
<1%
<1%
<1%
6%
<1%
' Same sample as evaluated in Phase III Experiment 3 (Butte TM2).
31
-------
Table 2-6. Matrix Associations of Arsenic Particles
Phase
II
III
Experiment
2
4
6
7
8
9
11
15
1
2
o
J
4
Sample
Bingham Creek Channel Soil
Murray Smelter Slag
Butte Soil3
Midvale Slag
California Gulch Phase I
Residential Soil
California Gulch Fe/Mn PbO
California Gulch AV Slag
Palmerton Location 2
Palmerton Location 4
Murray Smelter Soil
Clark Fork Tailings
VBI70 TM1
VBI70 TM2
VBI70 TM3
VBI70 TM4
VBI70 TM5
VBI70 TM6
Butte TM2
Aberjona River TM1
Aberjona River TM2
Particle
Frequency
(Percent)
Liberated
100%
99%
92%
96%
88%
98%
85%
100%
84%
92%
99%
100%
99%
100%
100%
95%
100%
100%
100%
100%
Relative Arsenic Mass
(Percent)
Liberated
100%
95%
87%
78%
94%
100%
73%
100%
58%
79%
96%
100%
95%
100%
100%
100%
100%
100%
99%
100%
Included
0%
5%
13%
22%
6%
0%
27%
0%
42%
21%
4%
0%
5%
0%
0%
0%
0%
0%
1%
0%
1 Same sample as evaluated in Phase III Experiment 3 (Butte TM2).
32
-------
Table 2-7. RBA Estimates for Arsenic in Test Materials
Phase
Phase II
Phase
III
Experiment
2
4
5
6
7
8
9
10
11
15
1
2
3
4
5
6
Sample
Bingham Creek
Channel Soil
Murray Smelter
Slag
Jasper County
High Lead Mill
Aspen Berm
Aspen
Residential
Butte Soil
Midvale Slag
California Gulch
Phase I
Residential Soil
California Gulch
Fe/Mn PbO
California Gulch
AV Slag
Palmerton
Location 2
Palmerton
Location 4
California Gulch
AV Slag
Murray Smelter
Soil
Clark Fork
Tailings
VBI70 TM1
VBI70 TM2
VBI70 TM3
VBI70 TM4
VBI70 TM5
VBI70 TM6
Butte TM1
Butte TM2
Aberjona River
TM1
Aberjona River
TM2
El Paso TM1
El Paso TM2
ACC Utility Pole
Soil
Site
Bingham Creek
Murray Smelter
Region VII
Jasper County
Aspen
Aspen
Butte
Midvale
California
Gulch
California
Gulch
California
Gulch
Palmerton
Palmerton
California
Gulch
Murray Smelter
Clark Fork
VBI70
VBI70
VBI70
VBI70
VBI70
VBI70
Butte Arsenic
Butte Arsenic
Aberjona River
Aberjona River
El Paso
El Paso
ACC
Sample
Channel Soil
Slag Composite
High Lead
Smelter
Berm
Residential Soil
Composite
Soill
Slag Composite
Phase I
Residential Soil
Composite
FeMnPb Oxide
Soil
AV Smelter
Slag
Location 2
Location 4
AV Smelter
Slag
(reproducibility)
Soil Composite
Grant Kohrs
Tailings
TM1
TM2
TM3
TM4
TM5
TM6
Soil lb
Soil 2
River Sediment
- High Arsenic
River Sediment
- Low Arsenic
Soill
Soil 2
Soil Affected by
CCA-Treated
Wood Utility
Poles
Arsenic
Concentration3
(ppm)
149
695
16.4
66.9
16.7
234
591
203
110
1050
110
134
1050
310
181
312
983
390
813
368
516
234
367
676.3
312.8
74
73
320
RBA ± SEM
3 9% ±8%
55% ±10%
327% ±105%
100% ± 46%
128% ± 52%
9% ±3%
23% ± 4%
8% ±3%
57% ± 12%
13% ±4%
49% ± 10%
61%± 11%
18% ±2%
33% ±5%
51% ±6%
40% ±4%
42% ±4%
37% ±3%
24% ±2%
21% ±2%
24% ±3%
18% ±3%
24% ±2%
3 8% ±2%
52% ±2%
44% ±3%
37% ±3%
47% ±3%
33
-------
Table 2-7. RBA Estimates for Arsenic in Test Materials
Phase
Experiment
7
Sample
ACC
Dislodgeable
Arsenic
Site
ACC
Sample
Dislodgeable
Arsenic from
Weathered
CCA-Treated
Wood
Arsenic
Concentration3
(ppm)
3500
RBA ± SEM
26% ±1%
aValues are arithmetic means
b Same sample as evaluated in Phase II
SEM = Standard error of the mean, an indicator of the relative uncertainty around the RBA estimate (see Appendix A)
34
-------
Table 2-8. Summary Statistics for Dose-UEF Slopes and RBA Estimates for Phase III RBA
Assays
Parameter
N
Mean
SD
CV
Sodium Arsenate Slope
7
0.78
0.099
0.13
Test Material Slope
14
0.26
0.098
0.38
Test Material RBA
14
0.34
0.118
0.32
CV, coefficient of variation (SD/mean); RBA, relative bioavailability; SD, standard deviation; UEF, urinary excretion fraction
35
-------
Table 2-9. Consolidated Arsenic Phases
Phase
Grouping
As Phosphate
As203
FeAs Oxide
Fe & Zn Sulfates
FeAsO
MnAs Oxide
PbAs Oxide
Pyrite
Sulfosalts
Minor
Constituents
Phase
As Phosphate
As203
FeAs Oxide
FeAs Sulfate
ZnSO4
FeAsO
MnAs Oxide
PbAs Oxide
Pyrite
AgAsS
Sulfosalts
AsMO
AsMSO4
Other
Abbreviations Used
Phos, Phosphate
As
Fe, Fe Oxide, FeSi
Fe Sulfate, Sulf
FeAs
Mn, Mn Oxide
PbAsO
Py
Ags
Phase Description
Arsenic bearing phosphate: although naturally occurring forms are rare (arsenocrandallite-
CaAl3AsPO4-OH6), these may be metastable forms of phosphate with sorbed arsenic formed
by secondary soil processes.
Arsenic trioxide: a common pyrometallurgical-formed phase that is common to arsenic
kitchens or copper smelters. It can also be found as a product in old formulas for herbicides,
pesticides, and rodenticides.
Iron oxide (FeOOH) with sorbed arsenic and lead, probably from soil.
Iron-rich sulfates: probably related to jarosite (KFe3(OH)6(SO4)2) or plumbojarosite
(PbFe3(OH)6(SO4)2). Can form in oxide zone of hydrothermal deposits, but is also common
to baghouse dust associated with copper-lead smelters.
Zinc sulfates: recognized by an elemental composition dominated by zinc, sulfur, and
oxygen with minor quantities of lead, arsenic, and/or cadmium. Generally found as
inclusions in slag or in baghouse dust and sometimes used in commercial products.
Iron oxide (FeOOH) that is highly enriched with arsenic; probably a flue dust.
Arsenic sorbed to the surface of manganese oxide-containing particles in soil. Formed by
release of arsenic from soluble forms. Recognized by an elemental composition dominated
by manganese, arsenic, and oxygen.
A product released from smelter flues and sometimes used in commercial products.
Recognized by an elemental composition dominated by lead, arsenic, and oxygen.
Iron sulfide (FeS2): a gaunge mineral associated with base -metal ore deposits. Pyrite may
contain small quantities of arsenic or have arsenic sorbed to its oxidized surface.
Silver arsenic sulfides: a mineral form related to mining activity (from a class of minerals
referred to as Sulfosalts). These ores of silver may be in the chemical form of proustite
(Ag3AsS3), xanthoconite (Ag3AsS3), pearceite ((AgCu)2As2Sn), orpolybasite
((AgCu)16(Sb,As)2Sn).
A group consisting of more than 100 forms of unoxidized minerals composed of metal or
semimetals and sulfur, distinct from a sulfide. These include numerous arsenic -bearing
phases: tennantite (Cu12As4S13) and enargite (Cu3AsS4) are perhaps the most common.
Arsenic -metal oxides: these are arsenic -rich oxides formed from pyrometallurgical
processes. Common associated elements (M) include lead, antimony, copper, zinc, and/or
cadmium.
Arsenic -antimony oxide: this is a common pyrometalurgically formed phase that is common
to arsenic kitchens. Its occurrence is significant in "dirty" or "black" arsenic and is still
found in trace quantities in "white" arsenic.
36
-------
Table 2-9. Consolidated Arsenic Phases
Phase
Grouping
Phase
AsSbO
Barite
Clays
Paint
Pb Solder
Pb-As Vanidate
PbAsMO
PbAsSbCuO
PbCrO4
PbMO
PbMS
PbMSO4
Slag
Ti02
ZnSiO4
Other
Abbreviations Used
AISi
Pbsold
PbAsVo4
-
Ti
Phase Description
Arsenic -antimony oxide: this is a common pyrometalurgically formed phase that is common
to arsenic kitchens. Its occurrence is significant in "dirty" or "black" arsenic and is still
found in trace quantities in "white" arsenic.
Barium sulfate: common gaunge mineral with base metals. Will adsorb lead and arsenic
during smelting.
Arsenic sorbed to the surface of soil-forming clays (hydrated, Al-Mg silicates).
Arsenic may be present in some very old paint pigments or as a trace contaminant in lead,
copper, and antimony pigments.
Lead solder with trace levels of arsenic. Recognized by an elemental composition
dominated by lead and tin with minor base metals.
A phase probably associated with mining or smelting of copper-rich ores, not used in
commercial products. Recognized by an elemental composition dominated by lead, arsenic,
vanadium, and oxygen.
Lead-arsenic metal oxides: these are lead-arsenic rich oxides formed from pyrometallurgical
processes. Common associated elements (M) include antimony, copper, zinc, and/or
cadmium.
Lead-arsenic metal oxides: these are lead-arsenic rich oxides formed from pyrometallurgical
processes.
A common lead pigment in paint and a rare form of lead.
Lead-metal oxides: these are lead-rich oxides formed from pyrometallurgical processes.
Common associated elements (M) include arsenic, antimony, copper, zinc, and/or cadmium.
Lead-metal sulfides: these are lead-rich oxides formed from pyrometallurgical processes.
Common associated elements (M) include arsenic, antimony, copper, zinc, and/or cadmium.
Lead-metal sulfates: these are lead-rich oxides formed from pyrometallurgical processes.
Common associated elements (M) include arsenic, antimony, copper, zinc, and/or cadmium.
A waste by-product of pyrometallurgical activity. Recognized by an elemental composition
dominated by silica, calcium, iron, and oxygen with variable quantities of lead, arsenic,
copper, and/or zinc.
Rutile or anatase with surface sorbed arsenic in small quantities. Recognized by an
elemental composition dominated by titanium and oxygen.
Zinc silicate, recognized by an elemental composition dominated by zinc, silica, and oxygen
with minor quantities of lead, arsenic, and/or cadmium. Generally found as inclusions in
slag or in baghouse dust and sometimes used in commercial products.
37
-------
Table 2-10. Relative Arsenic Mass for Consolidated Phase Groupings
Phase
II
III
Experiment
2
4
6
7
8
9
11
15
1
2
3
4
Sample
Bingham Creek
Channel Soil
Murray Smelter
Slag
Butte Soil3
Midvale Slag
California
Gulch Phase I
Residential Soil
California
Gulch Fe/Mn
PbO
California
Gulch AV Slag
Palmerton
Location 2
Palmerton
Location 4
Murray Smelter
Soil
Clark Fork
Tailings
VBI70 TM1
VBI70TM2
VBI70 TM3
VBI70 TM4
VBI70 TM5
VBI70 TM6
Butte TM2
Aberjona River
TM1
Aberjona River
TM2
RBA
39.3%
55.1%
17.8%
22.9%
8.4%
56.6%
12.9%
49.2%
61.0%
33.0%
50.7%
40.3%
42.2%
36.7%
23.8%
21.2%
23.5%
23.6%
38.1%
52.4%
Arsenic
Cone.
(ppm)
149
695
234
591
203
110
1050
110
134
310
181
312
983
390
813
368
516
367
676
313
Number
of
Particles
Counted
430
1108
636
1847
510
380
1472
111
105
355
238
261
128
97
139
103
124
137
186
123
Phase (Liberated/Included)
As
Phosphate
8%
<1%
14%
5%
27%
<1%
16%
8%
4%
2%
<1%
<1%
<1%
7%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
As2O3
54%
17%
80%
86%
97%
80%
<1%
5%
<1%
<1%
<1%
<1%
FeAs
Oxide
11%
27%
18%
<1%
29%
23%
21%
5%
3%
40%
3%
3%
8%
2%
3%
<1%
39%
69%
16%
<1%
<1%
2%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
Fe&Zn
Sulfates
46%
10%
51%
<1%
11%
5%
<1%
<1%
6%
24%
<1%
<1%
<1%
<1%
18%
30%
82%
<1%
<1%
3%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
1%
<1%
FeAsO
38%
2%
<1%
<1%
MnAs
Oxide
<1%
<1%
16%
36%
66%
40%
10%
<1%
2%
<1%
5%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
PbAs
Oxide
34%
44%
65%
58%
<1%
66%
32%
70%
6%
10%
18%
<1%
5%
22%
26%
42%
21%
<1%
<1%
<1%
<1%
<1%
Sulfosalts
2%
1%
13%
42%
<1%
<1%
3%
<1%
Minor
Constits.
<1%
15%
<1%
11%
5%
<1%
16%
11%
5%
4%
<1%
<1%
<1%
<1%
<1%
<1%
1%
<1%
<1%
2%
<1%
<1%
<1%
<1%
5%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
<1%
1 Same sample as evaluated in Phase III Experiment 3 (Butte TM2).
38
-------
Table 2-11. Estimated Group-Specific RBA Values for Liberated Particles
Group Name
FeAsO
As Phosphate
FeAs Oxide
Fe & Zn Sulfates
PbAs Oxide
MnAs Oxide
As2O3
Sulfosalts
Estimated Grouj
Method 1
1.00
0.55
0.45
0.40
0.38
0.38
0.25
0.02
3-Specific RBA1
Method 2
1.42
0.59
0.44
0.40
0.38
0.35
0.25
0.01
RBA Category2
High
Medium
Medium
Medium
Medium
Medium
Low
Low
39
-------
Figure 2-1. Excretion of Soluble As in Humans and Animals"
4000
Human Data
y=0.6714x+14.194
R2 = 0.9905
1000 2000 3000 4000
Total Ingested Dose fug)
5000
6000
4000
3500
3000 -
^2500 H
0)
'§2000 -\
c
Animal Data
y=0.6972x + 61.986
R2 = 0.8831
1000
2000 3000
Total Ingested Dose(ug)
4000
5000
aSee Table 2-1 for literature sources of RBA estimates.
40
-------
Figure 2-2. Conceptual Model for Arsenic Absorption and Excretion
Absorbed
INGESTED DOSE (D)
-> Blood
AFn
l-AFn
Non-Absorbed
Tissue (t)
-> Urine (u)
-> Bile (b)
Hepatobilliary
circulation
-> Feces(F)
Where:
D = Ingested dose
AF0 = Oral Absorption Fraction
Kt= Fraction of absorbed arsenic which is retained in tissues
Ku = Fraction of absorbed arsenic which is excreted in urine
Kh = Fraction of absorbed arsenic which is excreted in the bile
BASIC EQUATIONS:
Amount Absorbed (|ig)
= D-AF0
Amount Excreted in Urine (|ig) = Amount absorbed Ku
= D-AF0-KU
Urinary Excretion Fraction (UEF) = Amount excreted /Amount Ingested
= (D-AF0-KH)/D
= AF0-KU
Relative Bioavailability (x vs. y) = UEF(x) / UEF(y)
= (AF0(x)-Ku)/(AF0(y)-Ku)
= AF0(x)/AF0(y)
41
-------
Figure 2-3. Quality Assurance Data from Phase II Pilot Studies"
Panel A: Blind Sodium Arsenate PE Samples
120
100
80
60
cc
&_
-I '
§
c
o
O
40
20
--A--
A
ominal
-5
5 10
Study Day
15
20
1000 -
D)
3 100 -
0)
"ro
to
o
1 10 -
c
00
t
1 -
Panel B: Blind Duplicate Analyses
X
X'*""
Line of Equality _X
\* .-'*
S '' '
y* *
s$
* ,x
*,,'-'
,'*''' *
10 100 1000
Original Value (ug/L)
aComparion of measured and actual (nominal) concentrations of performance evaluation (PE) samples for
urine (panel A), and between duplicate measurements on the same urine sample (panel B), for Phase II
studies. R2 for blind duplicates was 0.91 (n=30).
42
-------
Figure 2-4. Phase III Performance Evaluation Samples"
Sodium Arsen.ite (As+3)
600 ~
3 500 -
1=
'v 400 -
1
§ 300 -
oj
m
m 200 -
Ol
oj
S 100 -
0 *
:.,.,-
Line of Equality
\
\ /
\ '"
H f.-'
.'''
x"
_.''
J*_
0 100 200 300 400 500 600
Expected (ng/mL)
600 -
a soo -
"Bi
£=
a 400 -
o
f 300-
m
1 200 -
m
03
aj
s 100 -
0 4
(
MMA
Line of Equality _,''
\
\ ,-"
\/
.#'"
s
1 100 200 300 400 500 600
Expected (ng/niL)
600 -
CT500 -
c
o 400 -
o
Dl
g 300 -
05
m
a! 200 -
OJ
03
OJ
^ 100 -
0 4
[
Sodium Arsenite (As+5)
>''
Line of Equality ..' *
\
\
\ /
.,-'
,.,-*"
^
100 200 300 400 500 600
Expectedtng/mL)
DMA
600
3-500 -
E
c
^400 -
o
|300 -
cc
m
1 200 -
CO
CO
CD
^ 100 -
fl .
C
/
*'
x' ^
Line of Equality ,,''
\.''
/
/
,.,--'
^'
xf""
-''' Outlier
*'' - *
r '''»'
100 200 300 400 500 600
Expected(ng/mL)
aComparison of measured and actual concentrations of performance evaluation (PE) urine samples for Phase III studies. DMA, dimethylasinic
acid; MMA, monomethylarsonic acid. R2 values were <0.99 for the four analytes (N=35-37).
43
-------
Figure 2-5. Phase III Blind Duplicate Samples a
1400 -
1200 -
1000 -
E
S 800 -
CO
'co
'ro
ty 600 -
'a.
Q
400 -
200 -
0 4
C
_,,--'"'
./"""
Line of Equality
\
\x"
,,-''
,,-$'' 4
^,.- '
<*x '
yX *
200 400 600 800 1000 1200 1400
Primary Analysis (ng/mL)
aComparion between duplicate measurements on the same urine sample for Phase III studies. The R2 was
0.98 (n=72).
44
-------
Figure 2-6. Phase III Inter-Laboratory Comparison"
Phase III Experiment 1
(VBI70 Study 1)
D)
_Q
(0
"gioo -
o
o
CO
E
o
'
AX
A
A /
4^AA.""'
A ,X
^
A f* ^
*^&
+^S*
X''A
y'
x''
X*^v
x' Line of Equality
^x'
1 100 10000
Results from Primary Lab (ug/L)
10000 -
D)
ro
T3
§100 -
CO
E
o
M
1 -
1
Phase III Experiment 2
(VBI70 Study 2)
A Investigative Samples XX
»PE Samples
X
,'A
A''A
I"""
A^A
4'*
X
Line of Equality
100 10000
Results from Primary Lab (ug/L)
aComparison of interlaboratory results of analyses of arsenic in urine in two Phase III studies. Values for
R2 were 0.87 (n=24) for Experiment 1 and 1.0 (n=25) for Experiment 2. Samples included urines
collected during the RBA assay (investigative samples) and performance evaluation samples (PE).
45
-------
Figure 2-7. Uncertainty in RBA Values"
4 -
? 3.5 -
o
CQ
1 3 -
I 2.5 -
CQ
Q.
Q.
EL 2
Q)
O)
ro
9: 1.5 -
1^*
CO
1 1"
CQ
^ 0.5 -
0 -
1
*
* * * **
10 100 1000 10000
Mean Dose of Highest Dose Group (ug/kw-bwday)
aPlot of uncertainty range (90% confidence interval) against administered dose. The dose axis is the
group mean dose ((ig/kg-day) for the highest dosing group in each study. The confidence interval
increases substantially when the administered dose levels are less than 25 (ig/kg-day.
46
------- |