AD-
NEUROBEHAVIORAL EFFECTS OF CARBON MONOXIDE (CO)
EXPOSURE IN HUMANS:
ELEVATED CARBOXYHEMOGLOBIN (COHb) AND
CEREBROVASCULAR RESPONSES
FINAL REPORT
Vernon A. Benignus-'-, Matthew L. Petrovick^
and James D. Prah-'-
May 19, 1989
•'•U.S. Environmental Protection Agency
Human Studies Division
Clinical Research Branch
Research Triangle Park, NC 27711
and
Department of Psychology
University of North Carolina
Chapel Hill, NC 27599
^u.S. Environmental Protection Agency
Neurotoxicology Division
Systems Development Branch
Research Triangle Park, NC 27711
Supported by
U.S. ARMY MEDICAL RESEARCH AND DEVELOPMENT COMMAND
Fort Detrick, Frederick, MD 21701-5012
Project Order 81PP1812
Contracting Officer's Representative
MAJ David L. Farmer
Health Effects Research Division
U.S. Army Biomedical Research and Development Laboratory
Approved for public release;
distribution unlimited
The findings in this report are not to be construed as an
official Department of the Army position unless
designated by other authorized documents.
-------
AD-
NEUROBEHAVIORAL EFFECTS OF CARBON MONOXIDE (CO)
EXPOSURE IN HUMANS:
ELEVATED CARBOXYHEMOGLOBIN (COHb) AND
CEREBROVASCULAR RESPONSES
FINAL REPORT
Vernon A. Benignus , Matthew L. Petrovick^
and James D. Prah-'-
May 19, 1989
1-U.S. Environmental Protection Agency
Human Studies Division
Clinical Research Branch
Research Triangle Park, NC 27711
and
Department of Psychology
University of North Carolina
Chapel Hill, NC 27599
^u.S. Environmental Protection Agency
Neurotoxicology Division
Systems Development Branch
Research Triangle Park, NC 27711
Supported by
U.S. ARMY MEDICAL RESEARCH AND DEVELOPMENT COMMAND
Fort Detrick, Frederick, MD 21701-5012
Project Order 81PP1812
Contracting Officer's Representative
MAJ David L. Farmer
Health Effects Research Division
U.S. Army Biomedical Research and Development Laboratory
Approved for public release;
distribution unlimited
The findings in this report are not to be construed as an
official Department of the Army position unless
designated by other authorized documents.
-------
sgcllAiTV CLASSIFICATION of THI$ PAGE
REPORT DOCUMENTATION PAGE
OM8 Ha. 0704-0188
la. REPORT SECURITY CLASSIFICATION
Unclassified
1b. RESTRICTIVE MARKINGS
2a. SECURITY CLASSIFICATION AUTHORITY
2D. ^CLASSIFICATION/DOWNGRADING SCHEDULE
3. DISTRIBUTION /AVAILABILITY OF REPORT
Approved for public release: distribution
unlimited
PERFORMING ORGANIZATION REPORT NUMBER(S)
5. MONITORING ORGANIZATION REPORT NUMBER(S)
M. NAME OF PERFORMING ORGANIZATION
U.S. Enviroamental Protection
Agency
66. OFFICE SYMBOL
(If appticabto)
7a. NAME OF MONITORING ORGANIZATION
U.S. Army Biomedical Research
Development Laboratory
and
6c ADDRESS (City, SU», and ZIFCodt)
Human Studies Division
Clinical Research Branch
Research Triangle Park, NC 27711
7b. ADDRESS (Crty, StaM, and ZIPCodt)
Fort Detrick
Frederick, MD 21701-5010
la. NAME OF FUNDING /SPONSORING
ORGANIZATION U.S. Army Medical
Research & Development Command
8b. OFFICE SYMBOL
(If appticaofe)
9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER
Project Order B1PP1812
8c ADDRESS (City. Stat», and ZIP Code)
Fort Detrick
Frederick, MD 21701-5012
10. SOURCE OF FUNDING NUMBERS
PROGRAM
ELEMENT NO.
62777A
BH&62
777A878
TASK
NO.
CA
WORK UNIT
ACCESSION NO.
287
11. TITLE (todud* Security OuiMIutb.^ NEUROBEHAVIORAL EFFECTS OF CARBON MONOXIDE (CO) EXPOSURE
IN HUMANS: ELEVATED CARBOXYHEMOGLOBIN (COHb) AND CEREBROVASCULAR RESPONSES
12. PERSONAL AUTHOR(S)
Vernon A. Benignus, Matthew L. Petrovick and James D. Prah
13a. TYPE OF REPORT
Final Report
1136. TIME COVERED
FROM 1985 TO 1983
14. DATE OF REPORT (»ar, A«ontn,Oay>
1 qqp Mav 23
15. PAGtCOUNT
58
16. SUPPLEMENTARY NOTATION
17.
COSATI COOES
FIELD
06
GROUP
u/
14
SUB-GROUP
18. SUBJECT TERMS (Conttnu* on rtwnv if rwctnan/ and Mtntrfy by otodr nwnter)
Carbon Monoxide, CO, Carboxyhemoglobin (COHb),
Brain Blood Flow, Cerebrovascular
19. ABSTRACT (Continue on rtwrw H mnsuty and Mtnmy by otodc numter)
A two-channel cranial impedance plethysmograph (CIP) was
designed and constructed as a noninvasive measure of brain blood
flow (BBF) in man. The instrument was designed to reduce some of
the problems with instability and difficulty of use found in
earlier commercially-available models. The CIP has been
previously validated against other measures of BBF.
During carboxyhemoglobin (COHb) formation, BBF is known to
increase. When BBF increases it compensates for the reduced
ability of the blood to carry oxygen in the presence of COHb.
Fifteen men breathed carbon monoxide (CO) to produce increases in
COHb values ranging from endogenous 18.4%. Increased COHb was
significantly related to a relative increase in BBF (p < 0.019).
Data from a similar experiment on dogs was obtained from Richard
continued, next page
20. DISTRIBUTION /AVAILABILITY OF ABSTRACT
GO UNCLASSIFIED/UNLIMITED El SAME AS RPT. Q OTIC USERS
'2* NAME OF RESPONSIBLE INDIVIDUAL
' " Mary Frances Bostian
21. ABSTRACT SE.QJRITY CLASSIFICATION
Unclassified
22b. TELEPHONE f/ndud* Ana Cod*)
301/663-7325
^mr*
O Form 1473, JUN 86
SECURITY CLASSIFICATION Of THIS PAGE
-------
ABSTRACT Ctd.
Traystman for reanalysis and comparison to human results. Human
and dog data did not differ significantly in the relation to
increased COHb. Various tests demonstrated that the CIP is a
reliable device. It was also shown that the information in the
two CIP channels (left and right sides of the head) is redundant
with respect to increased COHb. There was a substantial amount of
scatter about the line of best fit. Some subjects (both humans
and dogs) did not show increased BBF with high-level COHb. It
was hypothesized that subjects who sufficiently increased BBF
would not be behaviorally affected by COHb. Subjects whose BBF
did not increase after exposure would not, hypothetically, have
compensated and would therefore show behavioral impairment. The
importance of testing the hypothesis with future work was
emphasized.
111
-------
FOREWORD
Opinions, interpretations, conclusions and recommendations are
those of the authors and are not necessarily endorsed by the U.S
Army.
X Where copyrighted material is quoted, permission has been
obtained to use such material.
Were material from documents designated for limited
distribution is quoted, premission has been obtained to use the
material.
X Citations of commercial organizations and trade names in
this report do not constitute an official Department of Army
endorsement or approval of the products or services of these
organizations.
In conducting research using animals, the investigator(s)
adhered to the "Guide for the Care and Use of Laboratory
Animals", prepared by the Committtee on Care and Use of
Laboratory Animals of the Institute of Laboratory Resources,
National Research Council (NIH Publication No. 86-23, Revised
1985)
X For the protection of human subjects, the investigator(s)
adhered to policies of applicable Federal Law 45 CFR 46.
In conducting research utiliing recombinant DNA technology,
the Tnvestigaor(s) adhered to current guidelines promulgated by
the National Institute of Health.
The manuscript has been reviewed by the Health Effects
Research Laboratory, U.S. Environmental Protection Agency, and
approved for publication. Mention of trade names or commercial
products does not constitute endorsement or recommendation for
use.
-------
EXECUTIVE SUMMARY
The present document contains (a) a brief review of the
literature regarding increased brain blood flow (BBF) in response
to elevated carboxyhemoglobin (COHb), (b) a description of an
instrument to measure BBF in humans in a noninvasive manner and
(c) a report of an experiment in which BBF was measured in a
group of men with various COHb levels.
It is a well-established fact that, in dogs and sheep, BBF
increases as COHb is formed. This statement originates from
experiments involving invasive, direct measurement of blood flow
in the arterial blood supply to the brain. The increased BBF is
hypothesized to provide a compensatory mechanism for the
reduction in oxygen-carrying capacity of blood when COHb is
present. The BBF compensation is (in group averages) adequate to
compensate for both the lost oxygen-carrying capacity and a
shifted oxyhemoglobin dissociation curve.
If compensation were adequate in all subjects at all times
and for all parts of the brain, then there should be no
behavioral effects of COHb unless there is some other mechanism
than hypoxia by which COHb produces behavioral effects. Review
of behavioral experiments implies that below 20 - 30% COHb little
reliable effect on average performance is noted in healthy, young
men at rest, who are not simultaneously exposed to other
toxicants. It has also been shown that oxygen consumption in the
brain does not begin to decline until after about 20 - 30% COHb.
Thus, there is apparent agreement between the behavioral
literature and the compensatory BBF data.
-------
An instrument was designed and constructed to measure BBF in
man. The device is a cranial impedance plethysmograph (CIP).
This method of measurement of BBF had previously been
standardized against other, more traditional methods of BBF
quantification by other researchers. The instrument and data
reduction methods were tested in the present experiment. While
no measure of BBF in ml/min can be made with the CIP, it is
possible to calculate the relative change in BBF, R(BBF). This
is a dimensionless unit which is commonly used in the BBF
literature -
To test the method of BBF measurement, fifteen men breathed
air and carbon monoxide (CO) mixtures from a Douglas bag. CO
concentrations in parts per million (ppm) in the bag were
calculated and mixed to produce COHb levels ranging from
endogenous to 19%. The BBF was measured before and after
exposure and R(BBF) was computed for each subject. No behavior
was measured.
The R(BBF) increased as a function of COHb. The slope and
intercept of the regression function in humans were compared to a
function fitted to data from Richard Traystman in dogs. There
was no significant difference between the two functions. A
single function was fitted to the combined dog and human data.
A notable feature of the R(BBF) data was its wide scatter
about the line of best fit. Some subjects did not exhibit
compensatory increases while others appear to have
overcompensated. This was true of both humans and dogs. It was
hypothesized that those subjects not compensating for COHb might
-------
be behaviorally impaired by the COHb. The subjects who do
compensate would show reduced or no behavioral effects. If this
hypothesis were demonstrated, it could lead to the description of
a population of subjects who are especially sensitive to
behavioral impairment by COHb. No test of this hypothesis could
be performed with the present data set.
-------
TABLE OF CONTENTS
FOREWORD 1
EXECUTIVE SUMMARY 2
LIST OF FIGURES 6
TABLE OF ABBREVIATIONS 7
INTRODUCTION 8
BBF RESPONSE TO ELEVATED COHb 8
The Compensatory Mechanism Hypothesis 8
Implications of Compensatory mechanism 9
Assumption 1 10
Assumption 2 10
Assumption 3 11
Assumption 4 11
Assumption 5 11
BBF COMPENSATION AND BEHAVIORAL EFFECTS 11
PURPOSE OF THE PRESENT STUDY 13
METHODS 13
SUBJECTS 13
INSTRUMENTATION 14
Cranial Impedance Plethysmograph (CIP) 14
Display and Recording 18
CO EXPOSURE 18
PROCEDURE 19
CIP DATA REDUCTION 21
STATISTICS 23
RESULTS 23
R(BBF) AS A FUNCTION OF INCREASED COHb ( ACOHb) 24
EXPLORATORY EXPERIMENTS AND ANALYSES 24
Comparison of Human and Dog Data 24
Number of CIP Cycles Required 25
Differences Between Channels 26
DISCUSSION 27
R( BBF ) AS A FUNCTION OF ACOHB 27
IMPLICATION OF OBSERVATIONS ABOUT BBF ON BEHAVIOR 28
HUMAN VS . DOG R ( BBF ) 28
OTHER EXPLORATORY RESULTS 30
CONCLUSIONS 30
TABLE 1 32
GLOSSARY 33
FIGURES 34
LITERATURE CITED 39
APPENDIX - TECHNICAL DESCRIPTION OF THE CRANIAL
IMPEDANCE PLETHYSMOGRAPH (CIP) INSTRUMENT 41
INTRODUCTION 42
SYSTEM DESCRIPTION 43
IMPEDANCE DISPLAYS AND OUTPUT CONNECTORS 44
SUBJECT ELECTRICAL SAFETY 45
FRONT PANEL CONTROLS 45
REAR PANEL CONNECTIONS 47
INITIAL SETUP 47
OPERATION 48
SPECIFICATIONS 50
DISTRIBUTION LIST 53
-------
LIST OF FIGURES
Figure Figure Heading Pa9e
No.
1. Plots of the two-channel CIP and the ECG 33
2. Diagram of the CIP wave and its first
derivative, showing the measurements of A,
A', T and t. These measurements are used to
compute the derived measure, F, as defined
by Jacquy et al. (1974 ) 34
3. Scatter plot and fitted function for the values
of R(BBF) as a function of ACOHb for 14 men 35
4. Scatter plot and fitted function for the values
of R(BBF) as a function of ACOHb for 14 dogs 36
5. Scatter plot and fitted function for the values
of R(BBF) as a function of ACOHb for the pooled
data of 14 men and 14 dogs 37
-------
TABLE OF ABBREVIATIONS
BBF Brain blood flow. Also see R(BBF).
CIP Cranial impedance plethysmograph, an instrument or its
output which measures an analog of brain blood
flow by measuring the cranial electrical
bioimpedance changes.
CO Carbon monoxide, an invisible, odorless gas which is
the product of incomplete combustion.
COHb Carboxyhemoglobin, a measure of the level of carbon
monoxide in the blood.
p Probability of an event or outcome of a test of
statistical significance.
R(BBF) Relative change in brain blood flow.
ppm Parts per million.
r Correlation coefficient (Pearson product-moment).
-------
INTRODUCTION
The present document contains (a) a discussion of the data
regarding increased brain blood flow (BBF) in response to
elevated carboxyhemoglobin (COHb), (b) a description of an
instrument to measure an analogue of BBF in humans in a
noninvasive manner and (c) a report of an experiment in which an
analogue of BBF was measured in a group of men with various
levels of COHb.
BBF RESPONSE TO ELEVATED COHb
The Compensatory Mechanism Hypothesis. The most definitive
and quantitative information about the BBF response to COHb is
contained in a review and summary of a series of experiments
performed in Richard Traystman's laboratory (Jones & Traystman,
1984). From this work, involving invasive, direct measurement of
flow in the brain arterial supply in dogs and sheep, it appears
that as COHb increases there is a corresponding proportional
increase in BBF. The rise in BBF is apparently due to brain
vascular dilation.
As COHb increases, the blood's ability to carry oxygen (©2)
is diminished in direct proportion. The effect of the increased
BBF is to tend to compensate for the reduced 02~carrying capacity
of the blood. Thus, the BBF response to increased COHb can be
teleologically regarded as a compensatory mechanism.
When COHb increases, not only is the 02~carrying capacity of
the blood reduced, but the ©2 dissociation curve is also shifted
to the left (Lambertsen, 1980). The effect of the shift in the
dissociation characteristic is to decrease the amount of 02 which
-------
is unloaded from the blood into the tissue (Lambertsen, 1980).
If a compensatory mechanism is to be adequate to maintain an
unchanged 02 delivery to the tissues, then BBF must rise by more
than the decreased (^-carrying capacity due to COHb because there
must be sufficient additional 02 delivered to also compensate for
the more difficult 02 unloading.
The issue of adequacy of the compensatory BBF mechanism has
also been addressed by Jones and Traystman (1984). Two lines of
argument are offered as evidence that the compensation is
sufficient to maintain unchanged 02 supply to tissue even in the
face of 20 - 30% COHb. First, the mean increase in BBF which was
observed in subjects was larger than needed to compensate for the
reduced 02~carrying capacity, thus possibly reflecting a change
great enough to also compensate for the increased difficulty in
unloading 02 from blood to tissue. Second, the mean ©2
consumption of the whole brain was not observed to fall for
elevated COHb values in the above range. Jones and Traystman
therefore hypothesized that the brain vasodilation is the
effector part of a closed-loop 02 regulation system.
Implications of Compensatory Mechanism. It is potentially
important to study BBF compensation for COHb elevation in humans.
If whole-brain ©2 delivery is regulated at the tissue level in
the face of increasing COHb, brain function should not be
impaired. The preceding statement rests upon a number of
assumptions which are explicated and discussed below, but which
have not been tested.
-------
Assumption 1: Regulation of 02 supply at the tissue level
is homogenous across all regions of the brain. This is not
likely to be an exactly accurate assumption. Even if the
assumption were not valid in detail, but deviations in regulation
were small, the function of brain regions might not be disturbed
because of redundancies within the regions.
Differences in vasodilatory responses to COHb across brain
regions were reported in sheep (Koehler et al., 1984) and in cats
(Okeda et al., 1987). It is possible that the differences in
regional BBF response to COHb were proportional to 02 utilization
rate and therefore appropriate to the demand. However, the BBF
in one area of the brain, the neurohypophysis, has been shown to
be unresponsive to COHb (Hanley et al., 1986; Wilson et al.,
1987). Too little data are available to assess the implications
of the findings or the effect of deviations from the assumption.
To the extent that the above assumption is violated to an
important degree, behavioral effects of carbon monoxide (CO)
could occur despite compensatory action.
Assumption 2; Regulation remains effective over the entire
exposure duration. A number of other compensatory mechanisms do
not behave in a the above manner, e.g. the brain vasoresponse to
hypoxic hypoxia (Krasney et al., 1984; Manohar et al., 1984) and
for reduced carbon dioxide (C02), (Albrecht et al., 1987;
Raichele et al. , 1970). The BBF response in the above cases
declined over the course of hours or days, depending upon the
conditions. It is not, however, clear that the decline of the
BBF response represents a failure of compensatory mechanisms so
10
-------
much as it represents some other, more long-term adaptive
mechanism coming into play. No data on the duration of the BBF
response to COHb elevation are available.
Assumption 3: Regulation is equally effective for all
subjects and all occasions of measurement. Again, there are no
relevant data for CO in the peer-reviewed literature.
Assumption 4: There would be a behavioral decrement if there
were a decrease in 02 supply. It is not certain that a decrement
in brain function would occur if the 02 supply were slightly
decremented. Brain tissue could increase its 02 extraction to
further compensate. No data are available on this possibility.
Assumption 5; The neurotoxic effect of CO exposure is
entirely due to the hypoxic consequences of COHb formation.
There is evidence that CO hypoxia is not the only mechanism by
which CO produces effects (Piantadosi et al., 1987). The latter
is apparently important only at high CO concentrations, however.
If other toxic mechanisms played an important role, the BBF
compensatory mechanism might be only partly effective. Current
evidence seems to indicate, however, that other mechanisms have
only very small effects under conditions of exposure of interest
in this document.
BBF COMPENSATION AND BEHAVIORAL EFFECTS.
Benignus et al. (1989) reanalyzed reports of behavioral
effects of COHb elevation. They concluded that the effects of
COHb less than ca. 20% on brain function were small or absent.
The conclusion was limited to normal, healthy, young males, who
were not simultaneously exposed to other toxicants, under minimal
11
-------
physical work conditions. This is the conclusion which would be
expected from the compensatory mechanism hypothesis and data as
outlined above (if all assumptions are valid) because brain ©2
supply is held constant by the compensatory mechanism and the ©2
consumption does not fall below 20 - 30% COHb.
The conclusion of Benignus et al. (1989) was tempered by
some less formal observations. It appears that in nearly all of
the reports of behavioral effects of elevated COHb, there were
slight (but not statistically significant) elevations of mean
behavioral error scores for COHb values less than 20%. It also
appears that if data from individual subjects are examined, some
subjects demonstrated decrements in performance while others did
not or did so to a lesser extent (authors' own observations and
those of Steven Horvath, 1989). The latter observation could
explain the slight elevations in mean behavioral error scores and
produce a high variance among subjects. If the above less formal
observations were true, the conclusions of Benignus et al. (1989)
should be re-stated. It is possible that some subjects respond
to COHb elevation while others do not, or respond some of the
time and not other times. Thus, it is possible that in the
individual, if not in the group means, the effects of COHb
elevation could be large and important.
The compensatory vasodilation hypothesis (above) is based
upon data from group averages, as are behavioral conclusions.
Assumptions made about the compensatory process regarding equal
effects for all subjects and constant effects over time
(assumptions number 2 and 3) may not hold. If either or both of
12
-------
the assumptions do not hold, some subjects may not fully
compensate at all times and thereby show behavioral effects of
COHb elevation. No data are available to assess the validity of
the various assumptions and it was therefore considered important
to devise and test a method to collect such data in a noninvasive
way in humans.
PURPOSE OF THE PRESENT STUDY
The compensatory vasodilation hypothesis may well be
important to the issue of the mechanism of COHb effects (or non-
effects) on brain function. No recent data are available on
brain vasodilation due to COHb elevation in humans. Thus, it is
not known if humans exhibit the same extent of compensatory
response as do dogs and sheep.
The present study was designed to (a) devise a modern, non-
invasive method of estimating human BBF (b) expose humans to CO
while measuring BBF and (c) compare results to data in dog. Even
though the eventual application of BBF data would be to the
prediction of behavioral effects, no behavioral data were
collected in this study. It was not deemed prudent to obtain
behavioral data until the newly-designed cranial impedance
plethysmograqph) (CIP) instrument and methods had been
demonstrated to yield reliable and plausible results.
METHODS
SUBJECTS
Subjects were 15 men, aged 18.8 - 33.6 yrs (mean = 25.1,
SD = 4.52). Subjects weighed from 68.0 - 92.5 kg (mean = 80.6,
SD = 7.50), and were from 170.2 - 189.2 cm tall (mean = 179.3,
13
-------
SD = 5.70). Recruitment for participation was by public
advertisement. Each subject was paid $36 plus travel costs for
participation. Informed consent was obtained by written
statement and by oral exchanges. For the protection of human
subjects, the investigators have adhered to the policies of
applicable Federal Law 45CFR46. The protocol was approved by
Human Use Review Office of the Department of the Army, Office of
the Surgeon General, by the Committee on the Protection of the
Rights of Human Subjects of the School of Medicine, University of
North Carolina at Chapel Hill as well as by the Research Ethics
Committee of the Department of Psychology of the University of
North Carolina at Chapel Hill.
INSTRUMENTATION
Cranial Impedance Plethysmograph (CIP). The instrument
which was used to estimate human BBF used the impedance
plethysmograph principle. Earlier versions of such instruments,
also called rheoencephalographs (Jenkner, 1962), were unstable
and difficult to use. The present CIP instrument was developed
jointly by the second author of the present report and personnel
at the Research Triangle Institute in Research Triangle Park, NC.
Modern solid state technology was used and, as a result, the
device is extremely stable, reliable and simple to use. A
detailed description and evaluation of the CIP is given in the
Appendix. The principles of operation of the CIP are discussed
in the following paragraphs.
The impedance of the cranium fluctuates in a pulsatile
manner, synchronized with the cardiac cycle. Figure 1 is a plot
14
-------
of the left and right CIP wave along with the electrocardiogram
(ECG). Quantitative measures derived from the CIP have been
developed by various investigators and related to BBF by
comparison with other, more standardized, measures.
Quantification is usually preceded by ensemble averaging of the
CIP waves to reduce the influence of artifacts and presumably
random deviations which occur from wave to wave. Measures are
then performed on the ensemble-averaged CIP wave. The following
is a brief review of the relevant literature.
Jacquy et al. (1974) defined a derived CIP measure and
compared it to measures of BBF made simultaneously on the same
human subjects via the xenon clearance method. The derived
measure (F), a descriptor of brain blood flow, was made in 37
persons whose BBF was manipulated with injections of papaverine
and by CC>2 inhalation. F was calculated from measures made on
the ensemble-averaged CIP waveform and its first derivative as
shown in Figure 2. The CIP was averaged over 30-100 cycles. The
equation for F is given as follows, without the calibration
coefficients, which were eliminated for clarity.
F = [A/t(A')]/T
Terms in the above equation are defined in Figure 2. The data
revealed that F was correlated with the xenon clearance measure
of gray-matter BBF (r = 0.95).
A simpler derived CIP measure was compared to BBF as
measured by a radioisotope venous dilution method using 20
patients of various cerebral ischemic disturbances (Hadjiev,
1968). The CIP quantification consisted of the time to peak
15
-------
divided by the total wave time as measured on the un-averaged CIP
wave. The ratio was averaged for five cycles of CIP waves for
each subject. The two measures of BBF were not conducted
simultaneously but in close temporal contiguity. The measures
correlated well (r = 0.81). Possibly the correlation was not as
high as that demonstrated by Jacquy et al. (1974) because of (a)
the fewer cycles averaged (b) the non-simultaneous measurement of
BBF by the two methods (c) the possibly lower range of BBF values
which occurred in the patients or (d) the different derived
measure of CIP which was used.
BBF as measured by the xenon clearance method was compared
to CIP quantified from the ensemble average of 16 waves (Colditz
et al., 1988). Subjects were nine infants on artificial
ventilation whose PaCC>2 altered their BBF. The amplitude of the
averaged CIP was used as the measure. A low correlation
(r = 0.67) was observed. Such a low correlation could have
resulted from (a) the small number of cycles averaged (b) the
small number of subjects studied (c) the possibly small range of
BBF which occurred or (d) the particular derived measure of CIP
may have been innapropriate.
No absolute value of BBF can be computed from the impedance
measure, but presumably relative changes are accurately
portrayed. Thus, if a measurement of CIP is taken as a baseline,
before administration of a substance which may be a cerebral
vasodilator (e.g. CO), and the measurement is repeated while the
subject is under influence of the vasodilator, then the relative
change in the CIP measure is the same as the relative
16
-------
vasodilation. The foregoing is true if the measure of CIP is a
linear function of BBF. The assumption of linear relationship to
BBF appears to be valid for the measures of CIP proposed by
Jacquy et al. (1974 ) .
From the positioning of the electrodes and knowledge of
cranial anatomy, most of the variation in the CIP may be assumed
to be due to BBF although contamination by extra-brain (scalp)
blood flow is possible. The effect of extra-brain blood flow
contamination of the impedance measurement depends upon the
relative volume of such flow and its variability with respect to
COHb. The following is a logical evaluation of the possible
effects of extra-brain blood flow on the CIP. If extra-brain
blood flow does not vary as a function of COHb, or varies in the
same way as BBF, then it will not affect the CIP measure. If it
varies inversely with BBF, then the measurement of impedance
would be attenuated, but proportional to BBF. If it varies non-
systematically with respect to COHb then the impedance measure
would be made more variable, but the mean CIP values would still
be well related to BBF. Thus, it may be argued on strictly
logical grounds that even if extra-brain blood flow contamination
were present in the impedance measure, the measure would still be
a good analogue of BBF in terms of the mean measurement.
The influence of scalp blood flow was tested in 15 subjects
by Jevning et al. (1989). They computed various measures from
the averaged CIP, among which was the F measure of Jacquy et al.
(1974). The CIP measures were taken with and without a
constrictive band around the head to temporarily stop scalp blood
17
-------
flow. Differences between conditions were extremely small,
indicating that scalp blood flow was a negligible component of F.
Display and Recording. The pulsatile signals of the CIP
were displayed along with the EGG and an eye movement channel, on
a Grass model 7 D polygraph with the low frequency time constant
set at 0.15 Hz (half amplitude point) and high frequency filter
set at 40 KHz. The high frequency cutoff is not critical beause
the output of the present CIP instrument contains so little
noise. The outputs of the two Grass amplifiers for the CIP were
digitized by a 12-bit analog-to-digital converter at the rate of
100 samples per sec. Digitized data were stored on disk in an
IBM PC/XT. Computation of F was performed by the IBM PC/XT via
an offline program written by the senior author of the present
document.
CO EXPOSURE
During the experiment each subject breathed sufficient CO
designed to produce one of three approximate COHb levels;
endogenous, 12.5 or 17.0%. Exposure to CO was achieved by having
subjects breathe air/CO mixtures from a Douglas bag. The
concentration in the bag was fixed for each of three groups at
either 0, 6,000 or 9,600 ppm. The total bag volume was 30 1. No
effort was made to achieve particular COHb levels for each
person by, e.g., using physiological variables to compute the
amount of CO required. It was considered desirable, for
regression analysis purposes, to have a more-or-less continuous
distribution of COHb than to achieve 3 distinct groups. Bags
were prepared by an experimenter who had no contact with the
18
-------
subject. Neither the subject nor those experimenters in contact
with the subject were informed of the bag contents until after
the experiment.
PROCEDURE
Before being accepted into the study, potential subjects
were given routine physical examinations with special attention
to cardiac health. Following this, if no problems were detected,
they were given a twelve-lead resting ECG. If no problems were
noted, they were given a standard Bruce exercise ECG. Only
persons who showed no signs of abnormality of any kind were
accepted for study.
Subjects arrived in the laboratory between 0830 and 0930 and
were given a brief physical examination by a physician.
Following informed consent, pre-exposure blood was drawn and
Beckman silver-silver chloride (1 cm diameter) electrodes were
attached to the subject's head using EC-2 electrode paste.
Excitation signal (100 kHz, 4 ma) was applied between two
electrodes, one attached immediately above inion, the other
attached to the center forehead, 5 cm above nasion. Reference
leads for the two impedance measurement channels were attached to
each mastoid process. The two inputs of the impedance
measurement channels were attached to the forehead, 5 cm above
nasion and 4 cm on each side of midline.
After electrode preparation, a subject was seated in a
double walled audiometric testing chamber approximately 3x3x3
m in size. To avoid movement-induced artifacts in the CIP, he
was trained to relax during the measurement periods. The
19
-------
relaxation training consisted of instructing him to perform a
continuous mental inventory of all muscles including the tongue,
ears, scalp, temples, jaw and eyes to assure that each of the
muscles was relaxed and not moving. The subject was to allow his
head to drop forward until his chin came to rest on his chest.
When the subject became relaxed, a pre-exposure baseline
period of CIP was recorded for 2.5 min. Following the baseline
recording the subject was instructed to breathe the contents of a
60 1 Douglas bag which had been filled with 30 1 of either air or
one of two air/CO mixtures. The bag breathing required 3.0 - 6.2
min (mean - 4.2, SD » 1.1). The subject was then removed from
the chamber and blood was drawn 2 min after the end of bag
breathing. The subject then returned to the chamber for a post-
exposure CIP measurement. The time from the end of bag breathing
until the beginning of post-exposure CIP measurement was 4.55 -
8.47 min (mean = 6.15, SD = 1.0). Note that no information can
be gleaned from the present design regarding the persistence of
the compensatory BBF response because only short term CO exposure
was used.
The subject then exited the chamber and electrodes were
removed. If his COHb was above 10%, he was requested to breathe
normobaric 02 for up to 1 hr until his COHb was reduced to less
than 10%. Blood samples were drawn as appropriate to assure COHb
reduction. After reduction of COHb to less than 10% the subject
was released with the admonition to avoid even moderate exercise
and activities which required alertness or fine motor
performance.
20
-------
CIP DATA REDUCTION.
The value F of Jacquy et al. (1974) was computed on the
ensemble-averaged CIP data from the present experiment. The
equation for F and its interpretation is given above in the
paragraph entitled "INSTRUMENTATION", "Cranial Impedance
Plethysmograph (CIP)". One value of F was computed for each
subject and each condition in the experiment.
Ensemble averaging was performed off-line to allow adequate
artifact rejection. During averaging each cycle of the two-
channel CIP was displayed on a CRT screen and the operator was
then required to determine whether the displayed waves were
contaminated by artifact or distorted in a number of ways (see
below). If no problems were found, the data were passed to the
routine for computation of the ensemble averages. Before
inclusion into the ensemble average, a straight line was fitted
between the beginning and end points of the CIP wave. The
straight line was then subtracted from the wave to eliminate
baseline drift. When the number of required cycles was reached,
the averaged cycles were quantified by computation of F.
The criteria for acceptance of the displayed waves were (a)
the dicrotic notch (see Figure 1) had to be present (b) no
truncation or clipping could be present (c) the falling portion
after the dichrotic notch had to be relatively linear and (d)
there could be only moderate baseline drift (maximum of
approximately 30 degree baseline angle with respect to
horizontal). The baseline drift was assessed by connecting an
imaginary line between the beginning and end points of the wave.
21
-------
The criteria are qualitative and therefore difficult to specify
exactly. The reliability of the selection procedure was tested
by having all records analyzed by two independent observers who
were blind to the exposure conditions. The correlation between
their results was r = 0.95. Thus, the procedures were
demonstrably reliable.
For each subject and both conditions, the ensemble average
contained 50 cycles of data. Derived measures (F) were computed
for the left and right channels and were considered to be
measures of BBF for the left and right sides of the head. The
final measures of vasodilation were computed as the relative
increase of F from pre- to post-exposure for left and right
channels. The relative increase in F will be called the relative
increase in brain blood flow, R(BBF). This procedure avoids
absolute units of flow and is the same as that used in the
literature in dogs and sheep (Jones & Traystman, 1984) except
that the units in the literature are usually expressed as
percentages.
Computational steps involving CIP quantification can be
summarized as follows:
(a) Select undistorted CIP cycles.
(b) Subtract TbajEkeJ.ine drift from each acceptable
pair of CI? waves (left and right channels).
(b) Compute ensemble average CIP cycle for both
channels..
(c) Compute F for each 50-cycle ensemble averaged
CIP wave.
22
-------
(d) Average F across both channels to produce one
measure per subject, per condition. Fewer measures
were desired to simplify the hypothesis testing.
Individual channels were later analyzed in an
exploratory manner.
(e) Compute R(BBF) from pre to post exposure.
After hypothesis tests were made, changes were made
in the above procedure and the data were reanalyzed
on an exploratory basis. The purpose of the
reanalyses was to explore the sensitivity of the
results to the data reduction methods.
STATISTICS.
A straight line was fitted to R(BBF) as a function of COHb
using BMDP statistical software for microcomputers (Dixon, 1988),
program 1R. The significance of the fit was evaluated by the F
test as computed by program Ir. The alpha level was 0.05.
During exploratory analyses, regression solutions were
compared for significant differences using methods from
Kleinbaum, Kupper and Muller (1988, p 266-269). Reliability of
R(BBF) estimation methods was analyzed as a function of the
number of CIP cycles included in the ensemble averages by use of
the Spearman-Brown equation (Ghiselli, 1964).
RESULTS
A total of seven statistical analyses were conducted, each
of them to test a particular hypothesis. One of the hypotheses
was formed on an a-priori basis, the others were exploratory.
Table 1 is a list of the hypotheses and the results of their
23
-------
tests. In the following text the hypothesis numbers will be used
for reference to Table 1.
R(BBF) AS A FUNCTION OF INCREASED COHb (ACOHb).
Figure 3 is a plot of R(BBF) for each subject as a function
of that subject's ACOHb. The line in Figure 3 is a regression
line the particulars of which are listed in Table 1, hypothesis
1A. The fitted line accounted for a significant amount of
variance (r = 0.62, p = 0.019).
EXPLORATORY EXPERIMENTS AND ANALYSES.
Comparison of Human and Dog Data. To compare the human
results with those from experiments on dogs, raw data were
obtained from Richard Traystman. Figure 4 is a plot of dog
R(BBF) as a function of ACOHb. Exploratory regression analysis
(see Table 1, Hypothesis IE) yielded r = 0.47, p = 0.087. The
experimental design employed by Traystman was to use each dog as
his own control for 1 or 2 elevated COHb measures. Consequently,
there were no independent pre-exposure baseline data as there
were for the human data. It is probable that the non-significant
result from hypothesis IE was due to the fact that the pre-
exposure data were not available thus reducing the range of
ACOHb. For exploratory purposes, it was decided to use the
fitted function given in Table 1.
It was desired to explore the possibility that the functions
fitted to the human data and the dog data did not differ
significantly. Slopes and intercepts of the dog and human data
were tested for significant differences (Kleinbaum et al., 1988,
pp 266-269). The two intercepts were not significantly
24
-------
different, p > 0.5 (see Table 1, hypothesis 2E). The two slopes
did not differ, p > 0.1 (see Table 1, hypothesis 3E). Thus
there was no evidence that the lines of best fit for human and
dog R(BBF) were different. Qualitatively, the intercepts were
nearly the same. The slopes had quite different values (0.0403
vs. 0.0135). The lack of a significant difference in the slopes,
however, implied that there was a large uncertainty about one or
both of the estimates.
Because the two lines cannot be said to differ, a single
function was fitted to the pooled data from the human and dog
experiments (Figure 5). The regression analysis yielded
r = 0.69, p < 0.0001 (see Table I, hypothesis 4E). The slope and
intercepts of the fitted function most closely approximated that
of the dog data.
Number of CIP Cycles Required. In the above analyses of the
human CIP, the ensemble average contained 50 CIP cycles. From an
efficiency standpoint it would be desirable to use as few cycles
as possible. Having to average fewer cycles would also mean that
a shorter length of raw CIP data would be needed and therefore
more temporal resolution in an ongoing record of BBF could be
obtained. Too few cycles, however, might yield unreliable
results.
To determine the loss of stability (reliability) as a
function of the number of cycles averaged, the entire data set
was analyzed twice, using only 10 CIP cycles in the ensemble
average (10 CIP cycles in each channel and the results from the
two channels averaged). The first 10 cycles of CIP came from the
25
-------
early part of the 2.5 min block of data and the second 10 cycles
came from a later part. Pre- and postexposure records were
analyzed to produce early and late estimates of F and from these
the R(BBF) was computed for both early and late parts of the
record. Correlation between the early and late estimates of area
(the reliability coefficient) was 0.71, p < 0.007 (see Table 1,
hypothesis 5E).
The Spearman-Brown formula (Ghiselli, 1964) was used to
estimate the reliability for various numbers of CIP cycles in the
averaging. Figure 6 is a plot of the estimated reliability of
the R(BBF) as a function of number of CIP cycles, from 10 to 150
cycles. As can be seen, the reliability of the 50-cycle data as
used in the analysis of the above experiment, was estimated to be
0.92. Fewer CIP cycles in the average would be expected to yield
lower reliabilities and therefore greater variance in estimates
of area across conditions or individuals.
The intercept for the fitted line to relate the two 10-cycle
estimates of R(BBF) was near zero (0.078). The slope of the line
was 1.0294, (nearly unity). It appears that the two estimates
not only correlated well but were nearly identical.
Differences between channels. To compare information
between left and right channels, the R(BBF) was not averaged
across the two channels. When data from the 50-CIP-cycle
analysis was used, the correlation between left and right
channel R(BBF) was r = 0.90 (see Table 1, hypothesis 6E).
Apparently, there is no variation due to ACOHb in one channel
that is not found in the other.
26
-------
DISCUSSION
R(BBF) AS A FUNCTION OF ACOHb
The R(BBF) as computed in the present study is significantly
related to ACOHb. The correlation is approximately 0.62.
Observations about R(BBF), based upon the present results, are
potentially important. From Figure 3, it may be seen that in the
unexposed control group, the BBF strongly tended to decline from
the baseline to the "post-exposure" measurement even though there
was no change in COHb. This change might reflect a decreased
brain activation due to reduced anxiety about the exposure or the
experimental procedures in general or a reduced level of stress.
No data are available from which to deduce the cause of the
decreased BBF. The amount of BBF increase for the CO-exposed
subjects must, therefore, be compared to the predicted value for
the unexposed controls at an equivalent time in the experiment
since without COHb, they too would presumably have exhibited
vasoconstriction. The best measure of effect is the difference
between the control and the experimental groups, not the absolute
vasodilation of the experimental group, since the latter is the
due to the contribution of the independent variable plus whatever
other variables are also acting on the control group.
From Figure 5, it is apparent that many of the subjects
(both humans and dogs) do not appreciably increase BBF over
control values after elevated COHb. The regression line slope is
significantly different from zero because most of the subjects
respond. Apparently, some subjects do not exhibit appreciable
brain vasodilation in the presence of COHb. The variation in BBF
27
-------
response may be due to individual differences, or to some other
unknown factor. Since there is high variation in BBF responses
Assumption 3 in the Introduction may not hold, i.e. there may be
variation which is related to the individual or to the occasion
of measurement. The apparent nonresponders and (for that matter)
the apparent over responders may, of course, be entirely due to
random variation in the measurement. No conclusion can be drawn
without further data.
IMPLICATION OF OBSERVATIONS ABOUT BBF ON BEHAVIOR
Unpublished observations (see Introduction) imply that only
some subjects' behavior may be affected by COHb. The present
results imply that the same may hold for the BBF compensatory
response. It seems to be a reasonable hypothesis that the
behavioral effect of COHb is inversely related to the BBF
compensatory response. The subjects who exhibit adequate BBF
compensation for COHb may show no behavioral effects while those
subjects that do not adequately vasodilate may be behaviorally
impaired. If the above hypothesis is valid, much of the variance
in the behavioral effects of CO exposure may be accounted for by
BBF variation. It could also be true that there exists a
population at increased risk of behavioral impairment by COHb due
to a failure to compensate by vasodilation. Such information
could be quite important if, indeed, the observed variation in
R(BBF) is other than random error of measurement.
HUMAN VS. DOG R(BBF)
It appears that the measure of R(BBF) used in humans in the
present study yields data which are similar to the R(BBF) data
28
-------
reported by Jones and Traystman (1984) in dogs. The slope of the
function derived from humans does not differ significantly from
that of dogs even though the numeric value of the slopes differed
markedly (human slope greater than dogs'). The numeric
difference between slopes could easily be due to the small range
of COHb values in the two studies, which could have resulted in
erroneous estimates. The slope of the human data could have also
been over estimated because of a single subject whose R(BBF) was
2.28 for a ACOHb of only 9.8% (a possible outlier). Conversely,
the slope of the dog data could have been erroneously estimated
because of the absence of independent baseline data thus leading
to the absence of control data at near-zero ACOHb. When the two
data sets were combined, however, the function fitted for the
pooled data was much more similar to the dog than the human
function. While no unique conclusion can be reached, it appears
that the slope of the human function was overestimated due to a
possible outlier. If the variation across subjects is not due to
random error (see above) the slope of the human function may not
be important anyway, because in the face of, e.g. real individual
differences, a regression analysis would be less appropriate.
Qualitatively, the scatter of the dog data is similar to the
scatter of the human data. The similarity of the two data bases,
the well-understood principles of impedance plethysmography and
the standardization of CIP against other measures of BBF (Jacquy
et al., 1974), lend credence to the use of CIP as a measure of
relative BBF.
29
-------
OTHER EXPLORATORY RESULTS
There seems to be no difference between the measured
response in left and right channels of the CIP to ACOHb. This is
due either to the lack of lateral asymmetry in the brain
circulatory system or the lack of the ability of the device to
resolve lateral differences. From the present data base, there
is no way to decide between the above alternatives.
The number of cycles of CIP over which averaging is carried
out is important to the reliability and stability of the
estimates of R(BBF). Figure 6 indicates that approximately 40
cycles must be averaged to obtain a reliability coefficient of
0.90. This figure is very useful for experimental design
purposes. By use of Figure 6, estimates of the duration of
measurements (via number of CIP cycles to average) in a
particular condition can be made, reliabilities can be estimated
and one aspect of the power of a test can be determined.
CONCLUSIONS
(1) The reliability of the CIP, using the methods of Jacquy et
al. (1974) to measure BBF as an analog of cranial impedance, is
high when sufficient cardiac cycles of data are included in the
ensemble average.
(2) The BBF as measured by the CIP in man compares well
with other measures of BBF in man and with BBF measures
made by other methods in dogs.
(3) The CIP did not show lateralization in the BBF response
to COHb.
30
-------
(4) Average BBF increases with COHb in dogs, sheep and
humans. There is, however, a large amount of variability across
individuals and/or occasion of measurement.
31
-------
TABLE 1. SUMMARY OF THE HYPOTHESES TESTED WITH
TEST STATISTICS
TEST
H*1 TEST STAT. df p r int slope
1A R(BBF) - f(ACOHb) F=7.34 1,12 0.019 0.62 0.893 0.0403
(HUMANS)
IE R(BBF) - F(ACOHb) F=3.48 1,12 0.087 0.47 0.987 0.0135
(DOGS)
2E HUMAN AND DOG INT- t-0.28 24 >0.50 n/a n/a n/a
ERCEPTS ARE SAME
3E HUMAN AND DOG SLOPES t=l.59 24 >0.10 n/a n/a n/a
ARE SAME
4E R(BBF) - F(ACOHb) F=23.2 1,26 <0.001 0.69 0.985 0.0139
MAN, DOG DATA POOLED
5E REPEAT ANALYSIS FOR F=10.9 1,12 0.007 0.71 0.078 1.0294
RELIABILITY (TEST 1)
6E DIFFERENCES BETWEEN F=49.9 1,12 <0.001 0.90 0.246 0.7903
LEFT/RIGHT R(BBF)
^•Hypothesis numbered 'A' was a priori. Those numbered 'E' were
exploratory.
32
-------
GLOSSARY
Blind An experimental strategy in which a participant
is kept uninformed about a condition. Also see
"Double Blind".
Digitization Conversion of a continuous signal of function
to a discrete number series.
Double blind An experimental strategy in which neither the
subject nor the experimenter in contact with
the subject is informed of the conditions of
exposure. The strategy is used to minimize
bias due to expectations.
Plethysmograph An instrument used to infer changes in blood
flow in an intact organ of the body from
changes in e.g. volume, electrical impedance,
optical or acoustic properties of the organ.
33
-------
RIGHT
CIP
LEFT
CIP
ECG
TIME
Figure 1. Plots of the two-channel CIP and the ECG. The
left and right CIP refer to the left and right sides of the
head. On each CIP cycle, points typical of point A
correspond to the opening of the aortic valve while poits
typical of point B are the dichrotic notch, corresponding to
the closing of the aortic valve.
-------
FIRST
DERIVATIVE
OF CIP
CIP
TIME
Figure 2. Diagram of the CIP wave and its first derivative,
showing the measurements of A, A', T and t. These
measurements are used to compute the derived measure, F, as
defined by Jacquy et al. (1974). The equation for F
(without calibration coefficients) is F = (A/t(A')]/T.
35
-------
2.2H
2.M
1.8-1
1.'
i.oH
3.4-
10
I
15
Figure 3. Scatter plot and fitted function for the values
of'R(BBF) as a function of ACOKb for 14 men. The regression
line was R(BBF) = 0 . 0403(ACOHb) + 0.89. r = 0.62,
p < 0-019 .
36
-------
2.4—
2.2 —
2.0—
1.8 —
1.6-
1.2 —
1.0 —
0.8 —
0.6 —
0.4
I
20
I
40
50
COHb (?)
Figure 4. Scatter plot and fitted function for the values
of R(BBF) as a function of ACOHb for 14 dogs. The
regression line was R(BBF) = 0.0135(ACOHb) + 0.99.
r = 0.47, p < 0.087 .
37
-------
2.4—
2.2-
2.0-
1.8-
1.6-
1.2-
1.0 —
0.3-1
0.6 —
0.4
20
I
40
I
60
COHb
Figure 5. Scatter plot and fitted function for the values
of R(BBF) as a function of ACOHb for the pooled data of 14
men and 14 dogs. Open circles are dog data, filled circles
are human data. The regression line was
R(BBF) = 0.0139(ACOHb) + 0.99. r = 0.69, p < 0.0001.
38
-------
LITERATURE CITED
Albrecht, R.F., Miletich, D.J., and Ruttle, M. 1987. Cerebral
effects of extended hyperventilation in unanesthetized goats.
Stroke. 18:649-655.
Benignus, V.A., Muller, K.E., and Malott, C.M. 1989. Dose-
effects functions for carboxyhemoglobin and behavior. Neurotox. &
Teratol. In Press, 1990.
Colditz, P., Greisen, G., and Pryds, 0. 1988. Comparison of
electrical impedance and 133-xenon clearance for the assessment
of cerebral blood flow in the newborn infant. Pediat. Res.
24:461-464.
Dixon, W.J. (ed.) 1988. BMDP Statistical Software Manual.
University of California Press, Berkely.
Ghiselli, E.E. 1964. Theory of Psychological Measurement.
McGraw-Hill, New York.
Hadjiev, D. 1968. A new method for quantitative evaluation of
cerebral blood flow by rheoencephalography. Brain Res. 8:213-
215.
Hanley, D.F., Wilson, D.A., Traystman, R.J. 1986. Effect of
hypoxia and hypercapnia on neurohypophyseal blood flow. Am. J.
Physiol. 250: H77-H15.
Horvath, S.M. 5/1989. Private communication, Durham, NC.
Jacquy, J., Dekoninck, W.J., Piraux, A., Calay, R., Bacq, J.,
Levy, D., and Noel, G. 1974. Cerebral blood flow and
quantitative rheoencephalography. Electroenceph. and Clin.
Neurophysiol. 36:507-511.
Jenkner, F.L. 1962. Rheoencephalography - A Method for the
Continuous Registration of Cerebrovascular Changes. Charles C.
Thomas, Springfield, IL, 1962.
Jevning, R., Fernando, G., and Wilson, A.F. 1989. Evaluation of
consistency among different electrical impedance indices of
relative cerebral blood flow in normal resting individuals. J of
Biomed. Eng. 1153-56.
Jones, M.D. Jr., and Traystman, R.J. 1984. Cerebral oxygenation
of the fetus, newborn, and adult. Semin. Perinatol. 8:205-216.
Kleinbaum, D.G., Kupper, L.L., and Muller, K.E. 1988. Applied
Regression Analysis and Other Multivariable Methods. PWS-Kent
Publishing Company, Boston.
39
-------
Koehler, R.C., Traystman, R.J., Zeger, S., Rogers, M.C., and
Jones, M.D. 1984. Comparison of cerebrovascular responses to
hypoxic and carbon monoxide hypoxia in newborn and adult sheep.
J. Cereb. Blood Flow Metab. 4:115-122.
Krasney, J.A., McDonald, B.W., and Matalon, S. 1984. Regional
circulatory responses to 96 hours of hypoxia in conscious sheep.
Respir. Physiol. 57:73-88.
Lambertsen, C.J. 1980. Effects of excessive pressures of oxygen,
nitrogen, helium, carbon dioxide, and carbon monoxide. In V-
Mountcastle (Ed.), Medical Physiology, vol. 2, pp. 1901-1944. St.
Louis: C.V. Mosby Company.
Manohar, M. , Parks, C.M., Busch, and Bisgard, G.E. 1984. Bovine
regional brain blood flow during sojourn at a simulated altitude
of 3500 m. Respir. Physiol. 58:111-122.
Okeda, R., Matsuo, T., Kuroiwa, T., Nakai, M., Tajima, T., and
Takahashi, H. 1987. Regional cerebral blood flow of acute carbon
monoxide poisoning in cats. Acta Neuropathol. 72:389-393.
Piantadosi, C.A., Sylvia, A.L. and Jobsis-Vandervliet, F.F. 1987.
Differences in brain cytochrome responses to carbon monoxide and
cyanide in vivo. J. Appl. Physiol. 62:1277-1284.
Raichle, M.E., Posner, J.B., and Plum, F. 1970. Cerebral blood
flow during and after hyperventilation. Arch. Neurol. 23:394-
403.
Wilson, D.A., Manley, D.F., Feldman, M.A., Traystman, R.J. 1987.
Influence of chemoreceptors on neurohypophyseal blood flow during
hypoxic hypoxia. Circ. Res. 61:1194-11101.
40
-------
APPENDIX
TECHNICAL DESCRIPTION OF THE
CRANIAL IMPEDANCE PLETHYSMOGRAPH (CIP!
INSTRUMENT
41
-------
APPENDIX
TECHNICAL DESCRIPTION OF THE CRANIAL
IMPEDANCE PLETHYSMOGRAPH (CIP) INSTRUMENT
INTRODUCTION
Noninvasive measurement of cerebral blood flow can be
accomplished by a variety of medical instrumentation, i.e.
nuclear magnetic resonance (NMR), radio labeled isotopes in the
blood, and doppler backscatter of red blood cells exposed to
ultrasonic signals. Each method has its advantages,
disadvantages, cost effectiveness and the type of information
obtained. Each method is unique and may be favored over other
methods depending on the desired application. i.e., clinical
diagnostic, research, mobility, and various environmental
exposure situations.
Of the popular methods, (NMR) requires large size buildings
and computer controlled data acquisition systems. Other systems
include mobile carts which are far too costly and not suitable
for laboratory experiments or field applications. Alternative
noninvasive measurements of cerebral blood circulation have been
considered which are more cost effective and physically portable.
For the latter, the level of measurement accuracy and validation
is not equivalent to the more sophisticated and costly methods.
For the purposes of behavioral experiments involving carbon
monoxide exposures with (COHb levels of 10-20%), the impedance
plethysmographic method was selected for measurement of cerebral
pulsatile blood volume- This method is also known as
rheoencephalography (Jenkner, 1962). Commercially available
instruments, however, were found to be inadequate for one or more
42
-------
of the following reasons: (a) poor sensitivity to cerebral
vessel vasomotor events, (b) poor signal to noise ratio, (c)
D.C. drift problems, (d) excessive artifact, (e) excessive
size, (f) poor electrical safety characteristics, and (g) poor
calibration and data display methods.
In order to overcome the above limitations, a dedicated two
channel cranial impedance plethysmograph (CIP) instrument was
designed, and fabricated. The CIP, as a result of its basic
design, measures the change of electrical impedance (z) of
pulsatile red cell masses as they pass between specific
excitation (current source electrodes) and detection electrodes
(Jenkner, 1962). Briefly, the measurement of z-changes between
electrodes represent changes of electrical conductivity in a 100
KHz electrical field. This measurement, therefore, is not a
direct measure of any fluid flow component such as blood.
Conversely, the measured signal amplitude is generally
proportional to the fluid bolus profile.
SYSTEM DESCRIPTION
The CIP system used in the present study is comprised of
(1) a two channel impedance plethysmograph instrument, (2) a two
channel electrode system, (3) an auxiliary ECG channel and (4) a
power source- The CIP electrodes are placed in proximity to the
area of observation (cerebral vascular tree). One pick-up
electrode ( + ) channel (1) is placed over the right eyebrow. A
second pick-up electrode (+) channel (2) is placed over the left
eyebrow. Each CIP channel is referenced (-) to each mastoid
process. A 100 KHz, 4 ma square-wave excitation signal from a
43
-------
constant-current source is applied, to the center of the
forehead. The excitation current source is referenced to a
common electrode on the inion (back of the skull).
This electrode configuration creates a 100 KHz square-wave
field within the frontal to occipital skull region. As pulsatile
blood flow occurs, the 100 KHz signal is modulated and picked-up
by the (+) recording electrodes. The modulated signal is
subsequently processed to its analog slow-wave component.
As cerebral vascular modulation takes place, this change of
impedance is amplified and sent to a synchronous detector
circuit. The synchronous detector removes the 100 KHz (carrier)
signal but retains its slow-wave envelope as an analog signal.
The analog signal reflects the pulsatile blood volume profile
including the opening of the aortic valve and the dicrotic notch
or closing of the aortic valve (see Figure 1).
The analog signal is fed to a delta z rebalancing circuit
which has a maximum D.C. limit in the event of artifact, i.e., if
a lead movement, scalp flexing signal exceeded the delta z
threshold, the circuit is automatically re-zeroed. This prevents
overdriving or saturation of any recording device or analog to
digital converters connected to the CIP output. Each of two CIP
channels are virtually identical.
IMPEDANCE DISPLAYS AND OUTPUT CONNECTORS
Each channel contains its own LCD display for calibration
and subject impedance measurements. Each channel has the
following analog signal outputs:
44
-------
1. Zo = base impedance
2. Z = subject impedance
3. dz/dt = 1st derivative impedance
4. ECG = electrocardiogram
The detected (Zo) signal is also sent to a D.C. 16 Hz low
pass filter and to an isolation amplifier. Additionally, the
output of the rebalancing (z) signal is fed to an isolation
amplifier. When velocity of flow profile information is desired,
the (z) is fed to a differentiator amplifier and then to an
isolation amplifier, resulting in a dz/dt signal.
In addition to the impedance signals, a three lead ECG
signal is provided to check for cardiac rhythm synchronization
with CIP signals. The output connectors provide signal levels of
+1.5 volt for use with an analog recorder, or an analog to
digital converter.
SUBJECT ELECTRICAL SAFETY
All bioimpedance signals and the ECG signal are fed to
output connectors via isolation amplifiers. The isolation
amplifier provides low leakage currents and reduce possible shock
hazards when the CIP is connected to the test subject or other
instruments.
FRONT PANEL CONTROLS
A. Power Switch - This switch turns the instrument off and
on. When power is on the red POWER light is lit.
45
-------
B. ZO Meter - This meter indicates what has traditionally been
called the "base" impedance or the "static" impedance of the
CIP signal. The meter reads 0 to 70 ohms +1% of reading +0.1
ohm.
C. Operate-Calibrate Switch - This is a momentary-contact,
center-off switch which selects the operating mode of the
instrument. Momentarily pushing the switch in the Operate
direction connects the instrument to the front panel input
connectors during which the calibrate light is not lit.
Momentarily pushing the switch in the calibrate direction
connects the instrument to the internal calibrator signals
during which the calibrate light is lit.
D. Rebalance Switch - This switch has three positions AUTO, OFF,
and ZERO.
1. AUTO - In this position the instrument automatically
rebalances (re-zeros) the delta Z channel whenever the
threshold which has been set on the front panel
dial is exceeded. When rebalancing occurs, the rebalance
light on the front panel flashes and a 10 ms TTL pulse
appears at the rebalance output connector on the rear
panel.
2. OFF - This position sets the rebalance threshold to
+12V (+3 ohm). Under normal signal conditions
this effectively inhibits all rebalancing. (NOTE: Re-
balancing may still occur if the input is open circuited
and the delta Z channel saturates with noise.)
46
-------
3. ZERO (Momentary) - This position forces a rebalance to
occur by momentarily setting the rebalance threshold to
0 volts.
E. Rebalance Threshold Control - This ten-turn, continuaous dial
indicates the relative rebalance threshold on a scale of 0 to
10. This corresponds to an impedance range of 50 milliohms
(0) to 0.5 ohm (10.)
F. Z Input Connector - 9 pin input connector for the subject
impedance leads.
G. ECG Input Connector - 5 pin input connector for the subject
ECG leads.
REAR PANEL CONNECTIONS:
ECG Output - BNC
dZ/dt Output (CHI & CH2) - BNC
delta Z Output (CHl & CH2) - BNC
ZO Output (CHl & CH2) - BNC
Rebalance Pulse Output (CHl & CH2) - BNC
INITIAL SETUP
A. Connect the instrument to a standard 3-wire 120 VAC power
source
B. Turn the instrument on via the front panel power switch. The
red POWER light should light.
C. Switch both channels (1&2) to the calibrate mode by
momentarily pushing each operate-calibrate switch to the
calibrate position. Both red calibrate lights should light.
47
-------
D. Rebalance both channels (1&2) by momentarily depressing each
rebalance switch to zero position. If a rebalance operation
occurs the green rebalance light should flash.
E. Display the calibration signals. They should be as follows:
ECG - 2.0Vp-p Triangle Wave
dz/dt - 4.0VO-0 Square Wave
delta Z - 680mVp-p Triangle Wave
ZO - 0.0 volts
Rebalance Pulse - 0.0 volts except during rebalance
when each rebalance will generate
a 5V, 10 ms pulse.
F. Front Panel Meters - With the instrument in calibrate mode
both meters should indicate 35 ohm (+0.45 ohm).
OPERATION
A. Attach the CIP electrodes to the subject and connect them to
the Channel 1 input cable. The connections to this cable are
numbered as follows:
1 Current source
2 + Impedance Pickup
3 - Impedance Pickup
4 Current Return (common)
Connect the Channel 1 cable to the front panel of the
instrument. NOTE: If only one impedance channel is being used
it must be channel 1 because only channel 1 is connected to
the current source.
48
-------
B- If two impedance channels are being used, attach the second
set of CIP electrodes to the subject and connect them to the
Channel 2 input cable. The connections to this cable are
numbered:
1 OPEN
2 + Impedance Pickup
3 - Impedance Pickup
4 - OPEN
Connect the Channel 2 cable to the front panel of the
instrument. NOTE: Channel 2 does not contain a current
source. It must be used in conjunction with the Channel 1
current source to make a measurement. Similarly, if Channel
1 is switched into the calibrate mode the current source is
removed from the subject and Channel 2 will not record a
signal even though it remains in the operate mode.
C. Attach the ECG electrodes to the subject and connect them to
the input cable. Connect the ECG cable to the front panel of
the instrument.
D. Switch the impedance channels being used from the Calibrate
to the Operate mode. Again, if a single channel is being
used it must be Channel 1. Momentarily depress each
rebalance switch to the Zero position then set them to the
AUTO position. Set the rebalance threshold dial to 10. It
is best to leave the threshold set on 10 as this allows the
largest delta Z signal possible without saturating the dz/dt
ci rcuits.
49
-------
Connect an oscilloscope or strip chart recorder to the
outputs and observe subject waveforms.
SPECIFICATIONS
CURRENT SOURCE
Frequency
Output level
Effective output impedance
Dynamic range
Maximum open circuit output
voltage
II. Zo Channel
Output sensitivity
Bandwidth (-3dB)
Dynamic range
Noise
Output Impedance
III. delta Z Channel
Output sensitivity
Bandwidth (-3dB)
Dynamic range
Noise
Output Impedance
100 kHz, sinusoidal
4 mA rms
50 kohm
0 ohm to 2 kohm
+ 12V
200 mV/ohm
0 ohm = -7.0 V
DC to 0.16 Hz
0 ohm to 70 ohm
2mVp-p
50 ohm
8 V/ohm
DC to 100 Hz
+3 ohms (Rebalance off
4mVp-p
50 ohm
50
-------
IV. dz/dt Channel
Output sensitivity 2 V/(ohm/sec)
Bandwidth (-3dB) 2 Hz to 40 Hz
Dynamic range +Vp-p 0-40Hz (2.5 mohm/
sec equivalent)
Output Impedance 50 ohm
V. ECG Channel
Output sensitivity 1000 V/V
Bandwidth (-3dB) 0.07 Hz to 106 Hz
Dynamic range +12 mv input
Noise 4mVp-p output
Output Impedance 50 ohm
VI. Front Panel Meter
Range 0 to 70 ohms
Accuracy + 1% of reading +0.1 ohm
VII. Front Panel Threshold Potentiometer
Range Indicator Threshold Voltage Effective ohms
0 0.2 V +50 milliohms
10 2.2 V +550 milliohms
VIII. Rear Panel Rebalance Pulse
TTL compatible (0 to 5 V)
Pulse width 10 msec
51
-------
IX. Internal Circuits
Synchronous Demodulator - recovers only resistive (in-
phase) component
Digital Rebalance
Droop rate 0 V/sec
Resolution 16 bits (1 LSB = 763 uohm)
Dynamic range 0 to 10 V (0-50 ohm)
(Zo suppression resolution to within +6 mV at delta Z
output [+0.75 milliohms])
X. Isolation
Maximum leakage current, total instrument 14uA
Maximum lead leakage current to ground 6uA
Maximum isolated lead leakage current to
ground (120 VAC applied between lead
and AC power ground) 23uA
52
-------
DISTRIBUTION LIST
No. of Copies
15 Commander
U.S. Army Biomedical Research and
Development Laboratory
Attn: SGRD-UBZ-RA
Ft. Detrick, Frederick, MD 21701-5010
1 Commander
U.S. Army Medical Research and Development Command
Attn: SGRD-RMI-S
Fort Detrick, Frederick, MD 21701-5012
2 Defense Technical Information Center (DTIC)
Attn: DTIC-FDAC
Cameron Station
Alexandria, VA 22304-6145
1 Dean
School of Medicine
Uniformed Services University of the
Health Sciences
4301 Jones Bridge Road
Bethesda, MD 20014
1 Commander
U.S. Army Environmental Hygiene Agency
Attn: HSHB-0
Aberdeen Proving Grounds, MD 21010
1 Commander
U.S. Army Environmental Hygiene Agency
Attn: Librarian, HSHD-AD-L
Aberdeen Proving Ground, MD 21010
1 Walter Reed Army Institute of Research
Department of Respiratory Research
Attn: SGRD-UWH-E
Washington, D.C. 10307-5100
2 U.S. Army Research Institute of Environmental
Medicine
Attn: SGRD-UE-MEP
Natick, MA 01760-5007
1 U.S. Environmental Protection Agency
Human Studies Division
Research Triangle Park, NC 27711
1 U.S. Environmental Protection Agency
Library
Research Triangle Park, NC 27711
53
-------
Commandant
Academy of Health Sciences, U.S. Army
Attn: HSHA-CDB
Fort Sam Houston, TX 78234-6100
54
------- |