Local Cerebral Blood Volume Determined by ThreeDimensional Reconstruction of Radionuclide
Scan Data
By David E. Kuhl, Martin Reivich, Abass Alavi, Istvan Nyary, and Muni M. Staum
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
ABSTRACT
We developed a method to determine in man absolute values of local cerebral blood
volume (LCBV) localized throughout the brain in three dimensions and presented in a
cross-sectional picture format. Previously, absolute values of LCBV have been determined
in vivo by stimulated X-ray fluorescence, but these determinations have been limited to
one point in the brain at a time. All other previous estimates of LCBV by external
emission counting have been contaminated by the significant contribution of blood in the
overlying scalp and cranium. In our method, a transverse section scan is made after the
injection of 99mTc-labeled red blood cells into a peripheral vein. Data processing then gives
a point-to-point estimate of absolute radionuclide concentration analogous to an
autoradiograph. After the concentration of blood activity is determined, counting data are
converted to a two-dimensional map of LCBV representing a cross section
at a known level of the brain. In a series of five baboons, the following equation was
obtained for the regression plane that relates LCBV in the center of the brain to arterial
carbon dioxide tension (Paco2) and mean arterial blood pressure (MABP):
LCBV = 2.88 + 0.049Paco2 - 0.013MABP.
In patients, LCBV values ranged from 2 to 4 ml/100 g depending on location; higher values
corresponded to regions of cerebral cortex. Differences in blood volumes of focal brain
lesions were also quantified.
99m
KEY WORDS
Tc-labeled red blood cells
autoregulation of blood
flow
CO2 response
• Knowledge of local cerebral blood volume
(LCBV) in the living patient would be useful for
assessing cerebral hemodynamics. Changes in regional cerebral blood volume have been shown to
occur in physiological states such as mentation (1),
sleep (2), and alterations in arterial carbon dioxide
tension (3) as well as in pathological states such as
increased intracranial pressure (4). However, the
accurate measurement of LCBV in vivo is difficult.
From the Departments of Radiology (Nuclear Medicine) and
Neurology, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania 19104.
This investigation was supported in part by U. S. Public
Health Service Grants 2R01 GM-16248 and 5 P01 NS-10939
from the National Institutes of Health and by U. S. Atomic
Energy Commission Contract AT(ll-l)-3399.
Dr. Reivich is a recipient of U. S. Public Health Service Research Career Development Award 5K03-HL-11,896; Dr. Alavi
was supported by U. S. Public Health Service Training Grant 3
T01 GM01762. Dr. Nyary was on leave from the Experimental
Research Department, University Semmelweiss Medical
School, Budapest, Hungary.
Please address reprint requests to: David E. Kuhl, M.D.,
Nuclear Medicine Division, Department of Radiology, Hospital
of the University of Pennsylvania, Philadelphia, Pennsylvania
19104.
Received June 3, 1974. Accepted for publication February 21,
1975.
610
baboons
man
section scanning
Previous estimates by external radionuclide counting have been complicated by the significant contribution of blood in the overlying scalp and
cranium (5). To avoid this limitation, we have
developed a relatively noninvasive method for the
absolute measurement of LCBV localized in three
dimensions throughout the brain: the point-topoint concentration of radioactive blood in a transverse section picture of the brain is determined by
scanning and compared with the concentration in
the peripheral blood.
Recently, TerPogossian and his co-workers (6-9)
have described their results with a new method for
measuring LCBV with three-dimensional resolution; they determined LCBV for a single point in
the brain by measuring the stimulated X-ray
fluorescence of an iodinated contrast material.
Although their method gives accurate data with
three-dimensional localization, which previous
techniques have lacked, only information from one
point in the brain at a time is obtained. Also, the
choice of indicators with their method is more
limited than it is when radionuclides are used, a
feature that may become more important with
further extension of the method which we propose.
Circulation Research, Vol. 36, May 1975
611
LOCAL CEREBRAL BLOOD VOLUME
Methods
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
Our method required an injection of labeled (99mTc)
red blood cells into a peripheral vein followed by transverse section scanning of the brain. During the scan, a
blood sample was taken, and the activity in both red
blood cells and plasma was determined. Counting data
from scanning were reconstructed and corrected to represent the point-to-point concentration of radioactivity in
a cross section of the brain. The measured concentration
of blood radioactivity was used to convert the brain
activity to LCBV expressed in milliliters of blood per 100
g of brain tissue. The result was a two-dimensional map
of LCBV that represented a cross section of the brain at a
known level.
Scan Procedure.—We used a transverse section scanner that allows visualization of the radioactivity in body
organs in the form of a cross-sectional picture (10-13). In
such an instrument, radiation detectors are moved in a
sequence of tangential passes at regular angular intervals
around the head so that any radioactive structure in the
selected cross section is viewed from many different
directions (Fig. 1). A digital computer then reconstructs
these counting data into a cross-sectional picture (14).
The MARK III scanner used in this project was
designed and constructed in this laboratory (11). Each of
the four scintillation detectors' traveled back and forth
on one side of a square frame that was positioned to
coincide with the level of interest in the brain. The lines
of view of the four detectors moved simultaneously
through the plane of interest as each detector scanned in
a path tangential to the head. The scan cycle was a
sequence of six detector passes along the sides of the
detector square, each made with the frame rotated 15°
from the previous setting. A section scan took 2.5-20
minutes for completion, depending on the counting rates
available and the precision required. Counting data were
summed and recorded from each detector for each 0.8 cm
of detector travel. The resulting 768 values and their
corresponding position codes were recorded on perforated
paper tape for further processing.
Reconstruction of Scan Data.—The counting data,
gathered from known directions around the cross section
of the head, were used to reconstruct the original
distribution of radioactivity by an orthogonal tangent
correction (OTC) method (14). The calculations were
performed by a minicomputer.2 The OTC method is one
of a family of tangent correction processes for transverse
section reconstruction from data representing multiple
views or projections (15, 16). The reconstruction in the
OTC method is accomplished by first using counting
data from a single orthogonal pair of tangent passes and
forming a matrix distribution that reflects the proportionality that exists between them. The matrix data are
then redistributed to conform to the proportionality of
each succeeding orthogonal pair of tangent lines.
Our final matrix was a 64 x 64 array of picture
elements, which represented accurately the relative activity distribution in the cross section of the head. The
entire matrix could be multiplied by an experimentally
1
Each detector had a 2 x 0.5-inch Nal (Tl) crystal and a
focused collimator.
2
Varian 620/L which has 20K words of core memory with an
added 1.17 million words of disk memory.
Circulation Research, Vol. 36, May 1975
FIGURE 1
Transverse section scanning. The MARK III scanner had four
collimated scintillation detectors. The scan cycle was a series of
detector movements tangential to the plane of interest, and the
angular displacement between scan paths was 15°. Counting
data were reconstructed by computer to produce a cross-sectional picture of brain radioactivity. With knowledge of the
concentration of "mTc-labeled red blood cells in the peripheral
venous blood, it was possible to convert the brain radioactivity
into a two-dimensional map of LCBV that represented a cross
section of the brain at a known level. A = anterior, P =
posterior, L = left, and R = right.
determined calibration factor to convert the counts in
each picture element to absolute values of radioactivity
in units of microcuries per 100 g of brain tissue. In like
manner, an additional multiplier, dependent on the
measured activity of the blood sample, could be used to
convert the matrix values to LCBV in units of milliliters
of blood per 100 g of brain tissue.
For accurate estimates of brain activity by this
method, the volume of tissue must fill the field of view of
the detector. With the MARK III scanner, the diameter
of the field of view was approximately 2 cm. Therefore,
even though smaller spatial detail was resolved in these
section pictures, the activity estimate for each point or
picture element represented a cube of tissue measuring
approximately 2 x 2 x 2 cm.
Red Cell Labeling.—Ow method for labeling baboon
erythrocytes was similar to that of Atkins et al. (17).
After centrifuging 10 ml of a mixture (4:1) of whole blood
in a special ACD solution (Squibb), we removed 2.0 ml of
packed red blood cells. We added the desired amount of
99m
Tc in a volume not exceeding 1.0 ml of eluate from a
99m
Tc generator (New England Nuclear). After 5 minutes
of gentle agitation, we added 0.1 ml of a solution of
stannous chloride in ACD solution equivalent to not
more than 50 /xg of stannous ion. The stannous chloride
had been previously filtered through a 0.22^ membrane
filter. After another 5 minutes of gentle agitation, we
washed the red blood cells successively with two 10-ml
portions of saline by centrifuging and decanting the
supernatant solution each time to remove excess stan-
KUHL. REIVICH. ALAVI. NYARY. STAUM
612
nous ions and radioactivity not firmly bound to the cells.
Subsequent washings removed less than 2% of the
activity per wash. The overall yield was approximately
65%. We resuspended the labeled red blood cells in saline
and injected an accurately measured volume containing
the desired amount of radioactivity into the subject. A
sample of the injectate was retained as a reference
standard.
Calculation of LCBV.—LCBV was calculated as the
ratio of the 99mTc activity per weight of brain tissue,
determined by scan, to the 99mTc activity per milliliter of
cerebral whole blood, predicted by assay of a peripheral
venous blood sample. We assumed that the brain was
normal and that the minimum volume of brain tissue
quantified by this method ( 2 x 2 x 2 cm) was sufficiently
large that its average hematocrit could be considered
equal to the average vessel hematocrit for the whole
brain (0.85 x peripheral venous hematocrit [7]). Thus,
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
LCBV
x 100,
0.85Hct Crbc + (1 - 0.85Hct) Cplas
(1)
where Cbrain is the activity per gram of brain determined
by scan, 0.85 is the cerebral hematocrit correction, Hct
is the peripheral venous hematocrit, Cr4c and Cp(os are
the activities per milliliter of red blood cells and plasma,
respectively, and 100 converts LCBV to standard units of
milliliters per 100 g.
Results
STUDIES IN BABOONS
Validity of Label.—To have a short scanning
time and an acceptable radiation exposure, we
chose short-lived 99mTc (half-life 6 hours) as a label
rather than the more well-established red blood cell
label 61Cr (half-life 28 days). Although the physical
characteristics of 99mTc are superior to those of 51Cr
for this purpose, red blood cell labeling methods for
99m
Tc are not as efficient and are still being
improved (17, 18).
When baboon red blood cells were labeled with
M
Cr, an insignificant amount of radioactivity appeared in the plasma during the several hours after
injection. However, when baboon red blood cells
were labeled with 99mTc by the method described
previously (17), approximately 14% of the red blood
cell activity appeared in the plasma almost immediately after injection, even though substantially
all of the unbound activity had been removed from
the sample prior to injection. After this, the red
blood cell concentration of 99mTc declined at a rate
of about 5%/hour. During the first hour, the plasma
concentration of 99mTc declined by 30% and then
fell at a rate of 5%/hour. The composition of this
unwanted plasma activity was not determined;
only a portion could be accounted for later in the
liver and the spleen, presumably sequestered there
as red cell fragments.
For a valid prediction of LCBV by Eq. 1, the
brain activity concentration (C6rain) can represent
activity that is in plasma as well as red blood cells,
but it must not include extravascular activity. If
the assumed relation between local cerebral hematocrit and peripheral venous hematocrit (0.85:1) is
correct and if Crbc and Cptas are known, LCBV will
be predicted correctly by Eq. 1. Any extravascular
99m
Tc will falsely increase the calculated value of
LCBV.
To estimate the magnitude of such an error, we
compared LCBV determinations in baboons when
99m
Tc and 61Cr red blood cell labels were used
simultaneously. Separate red cell samples from six
baboons were labeled with 99mTc and 51Cr and
reinjected into a peripheral vein when the animal's
arterial carbon dioxide tension (Pco2) was approximately 37 mm Hg. In the six baboons, four brain
biopsies were made less than 1 hour after the
injection and four were made 4-5 hours after the
injection under conditions of extremely high and
low values of LCBV. The concentration of 99mTc
and 51Cr was measured in the brain and blood
samples. Separate values of LCBV were calculated
for each label (ratio of activity per weight of tissue
to activity per milliliter of blood).
Figure 2 shows the correlation of in vitro determinations of LCBV by the99mTc-labeled red blood cell
method and the 51Cr-labeled red blood cell method.
9m T ( .
label
a,
N
I
Slope = 1.02 - . 0 2
Y Intercept = 2 2 ± . 0 7
5'
Cr
label
LCBV ml blood/IOO g
FIGURE 2
Correlation of in vitro determinations of LCBV in the baboon
brain by the MmTc-labeled red blood cell method and the
"Cr-labeled red blood cell method. Brain samples were taken
less than 1 hour after injection (open circles) or 4-5 hours after
injection (solid circles). The solid line represents the calculated
linear regression line through these data and the broken lines are
the 95% confidence limits. Extrauascular leak of "mTc was
considered responsible for the difference of 0.22 ml blood/100 g,
which was constant throughout the range of LCBV values.
Circulation Research, Vol. 36, May 1975
613
LOCAL CEREBRAL BLOOD VOLUME
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
The regression line and the 95% confidence intervals are shown. Note that the slope is 1.02 ± 0.02,
but the y intercept is 0.22 ± 0.07 ml blood/100 g and
not zero.
We interpreted these data as indicating an error
of +0.22 ml blood/100 g when the preparation of
99m
Tc-labeled red blood cells was used as an
indicator for LCBV determinations in baboons.
This error was constant throughout the range of
LCBV values. These results could be explained by
a small consistent percent (approximately 7%) of
labeled red blood cell activity moving into the
extravascular space very quickly after injection.
This extravascular concentration then remains
constant throughout the subsequent 4-5 hours and
is independent of any subsequent changes in
LCBV. Because of these results, in all subsequent
studies in normal baboons reported in this paper,
LCBV values determined by 99mTc-labeled red
blood cell counting were corrected by subtracting
0.22 ml blood/100 g from all measured values.
Accuracy and Reproducibility of Measuring
LCBV by Section Scan.—In a series of four baboons, approximately 15 me of 99mTc-labeled red
blood cells was injected into each animal. Four
hours later, peripheral venous blood samples were
obtained, and the baboons were killed. Section
scans were then made of the frozen intact heads.
The total counts of each scan averaged approximately 60,000-100,000. We then used the scan data
to calculate brain activity concentration (MC/100 g)
and LCBV (ml blood/100 g) for a 2 x 2-cm region in
the center of the baboon brain. From each frozen
head, we sawed a section corresponding to the scan
cross section, and we removed a brain tissue
sample from the center of each section. Blood
samples and brain tissue samples were weighed
and counted; brain activity concentration and
LCBV were determined. There was good agreement
between the section scan results and the results
from in vitro counting; the coefficient of variation
of the data was 6.7% for brain activity concentration and 8.1% for LCBV.
The reproducibility of the scan method over time
is illustrated in Figure 3, which is a plot of
scan-determined LCBV in a 2 x 2-cm section
centered in a baboon brain measured 60, 82, 108,
129, and 139 minutes after injection. At the times
of measurement, the baboon's arterial Pco 2 was
33.9 ± 0.7 mm Hg and his mean arterial blood pressure was 122 ± 4 mm Hg. Each scan required 7.5
minutes and contained 250,000 counts. The mean
of the scan-determined LCBV values was 3.22 ml
blood/100 g with a coefficient of variation of 6.5%.
Circulation Research, Vol. 36, May 1975
LCBV
(ml blood /lOOg)
4r-
60
80
IOO
120
140
Time after injection ( min)
FIGURE 3
Reproducibility of repeated scan determinations of LCBV in a 2
x 2-cm section centered in a baboon brain. The baboon was
maintained in a steady state with an arterial Pco2 of 33.9 ±0.7
mm Hg and a mean arterial blood pressure of 122 ± 4 mm Hg.
The mean value of LCBV was 3.22 ml blood/100 g with a
coefficient of variation of 6.5%.
Responses to Arterial CO2 Tension and Mean
Arterial Blood Pressure.—Five baboons were anesthetized with Sernylan (1 mg/kg), paralyzed with
Flaxedil (5 mg/kg), and artificially ventilated with
a Harvard large-animal respirator (inspired gas
was 30% O2, 70% N 2 ). End-tidal Pco 2 was monitored with a Beckman infrared CO2 analyzer.
Catheters were" placed in both femoral arteries for
blood sample collection and blood pressure recording. The core temperature of the baboons was
monitored and kept constant (37 ± 0.2°C) by an
infrared heating lamp.
After the completion of surgery, a red blood cell
sample from each baboon was labeled with approximately 15 me of 99mTc and injected intravenously.
Alterations in arterial Pco 2 were induced by both
hyperventilation and CO2 inhalation (8% CO2, 30%
O2, 62% N 2 ). When blood pressure and end-tidal
Pco2 had stabilized, arterial blood samples were
drawn for blood gas determinations. Repeated
transverse section scans were made through the
head over the next 4-5 hours under normocapnic,
hypocapnic, and hypercapnic conditions. Depending on time and dose, each scan required 2.5-20
minutes for collection of approximately 50,000
counts/picture. In most instances, after at least 15
minutes with the baboon under stable conditions, a
series of four section scans was performed with a
scan time of 2.5 minutes for each. From these data
KUHL. REIVICH, ALAVI. NYARY. STAUM
614
and the blood sample data, LCBV for each scan
study was then calculated for a 2 x 2-cm section
centered in the baboon's head.
The primary data were then reduced by averaging corresponding values of LCBV and mean arterial blood pressure for the same arterial Pco2 level
(Table 1). Since there was a wide range of mean
arterial blood pressures (69 to 153 mm Hg), we
performed a bivariate analysis of our data (arterial
Pco 2 and mean arterial blood pressure were the
TABLE 1
Experimental Conditions for Scan-Determined
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
Baboon
N
Range
LCBV
Arterial Pco 2
(mmHg)
MABP
(mmHg)
LCBV
(ml blood/
100 g)
42.4
37.8
27.2
27.8
70.5
69.9
71.1
66.6
45.1
94
97
84
87
131
131
149
130
81
4.23
4.08
3.53
3.28
5.49
5.26
5.22
5.05
4.54
38.9
38.3
64.1
67.1
31.3
27.1
116
111
125
129
119
95
2.62
2.63
2.72
3.11
2.25
2.35
32.7
33.9
34.4
47.0
53.6
35.2
33.9
114
134
119
124
121
88
69
3.34
3.17
3.20
4.66
4.46
3.14
3.16
39.1
62.6
24.4
34.6
138
132
142
116
2.40
3.82
2.43
2.21
39.9
40.9
24.7
21.4
67.9
73.8
42.6
135
140
130
137
142
153
121
3.00
3.28
2.49
2.27
4.10
3.85
3.17
33
21.4-73.8
33
69-153
33
2.21-5.49
Pco 2 = carbon dioxide tension, MABP = mean arterial blood
pressure, and LCBV = local cerebral blood volume, N = number
of measurements.
independent variables, and LCBV was the dependent variable). The equation of the calculated
regression plane was:
LCBV = 2.88 + 0.049 P a co 2 - 0.013MABP,
(2)
where P a co 2 = arterial Pco2 and MABP = mean
arterial blood pressure. The overall fit was significant at the P < 0.01 level (F ratio test). Also, both
slopes were significant: the arterial Pco 2 slope at
the P < 0.001 level and the mean arterial blood
pressure slope at the P < 0.05 level (£-test). These
data agree well with the few similar results known
from the literature (Table 2).
The equation of the normalized regression plane
for scan-determined LCBV is shown in Figure 4
(plane A). This equation was obtained by first
calculating the regression plane for LCBV in each
baboon and determining the value of LCBV under
normal conditions, i.e., arterial Pco2 = 37 mm Hg
and mean arterial blood pressure = 120 mm Hg.
This value was then made unity and all other
LCBV values for the baboon were related to it. The
data for all of the baboons were pooled and
regression plane A was calculated.
A normalized regression plane was calculated in
a similar manner from external head counting data
that were obtained without benefit of section
reconstruction (Fig. 4, plane B). These counting
data were recorded during the section scans in each
lateral pass, only when the collimated detector was
directed at the brain. They represent the counting
data that one would expect to obtain by applying a
collimated radiation detector to the head to estimate cerebral blood volume without use of our
section reconstruction method. Note that the
slopes of plane B do not agree with those of plane A.
In fact, the mean arterial blood pressure slope of
plane B is opposite in sign to what would be
TABLE 2
Comparison of the Alterations in LCBVin Response to Changes
in Arterial Pco2 and Mean Arterial Blood Pressure Found
in the Present Experiments and in Previously Published
Investigations
Animal
Goat
Rhesus monkey
Rhesus monkey
Baboon
Arterial
Pco2
slope
MABP
slope
Reference
-0.015
-0.013
3
7
8
Present paper
0.043
0.049
0.049
Pco 2 = carbon dioxide tension and MABP = mean arterial
blood pressure.
Circulation Research, Vol. 36, May 1975
615
LOCAL CEREBRAL BLOOD VOLUME
LCBV Qiormollztd]
IOO
ACOgQnmHq)
Plane A
LCBV • 0.667 + 0.011 P - 0.0009 MABP
normal
ACO2
Plane B
L C B V 0.751 + 0.0038 P „ +T 0.0010 MABP
»co 2
normal
MABP [mmHg]
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
FIGURE 4
Response of LCBV to arterial Pcot (PAco,) end mean arterial
blood pressure (MABP). Plane A = external counting data with
section reconstruction and plane B = external counting data
without section reconstruction. The slopes are different for the
two normalized regression planes. The mean arterial blood
pressure slope of plane B is opposite in sign to what would be
expected with autoregulation. External counting without section reconstruction is an insensitiue indicator of LCBV.
expected in the presence of autoregulation. External counting without section reconstruction is an
insensitive indicator of cerebral blood volume, as
would be expected, since there is no discrimination
against the effect on the data of extracerebral
circulation.
INITIAL STUDIES IN MAN
Transendothelial Passage of Label.—In extending these studies to man, we replaced the 99mTc
labeling method of Atkins et al. (17) with the
method of Cutkowski and Dworkin (18) in which
red blood cells are pretreated with stannous glucoheptenate before labeling. With this modification,
the procedure was simplified, labeling efficiency
was increased, and label loss in vivo was reduced.
About 7% of the red blood cell activity still appeared in the plasma at the end of 15 minutes. Over
the next hour, the plasma concentration declined
by 30%, but the red blood cell concentration
declined by less than 2%.
We have not yet quantified the effect of possible
transendothelial passage of label on the measurement of LCBV in the normal or the diseased brain
of man. If there is any undesired extravascular
activity in the normal brain, then an even greater
concentration may be present when lesions of the
blood-brain barrier occur. The effect would be (1)
to mask small reductions in LCBV that might
Circulation Research, Vol. 36, May 1975
coincide with lesions of the blood-brain barrier or
(2) to suggest LCBV increases in the damaged
brain when there are none.
Our more recent studies in man suggest that this
problem is not serious. For example, the scans
shown in Figure 7 are of a brain tumor that had a
low LCBV and a damaged blood-brain barrier. As
expected, 99mTc-pertechnetate scans showed a
high concentration of activity localized in the tumor
(over a 3:1 increase compared with normal brain
tissue). But, 99mTc-red cell scans showed almost
no activity localized in the tumor. This result would
not have occurred if a significant amount of activity
had entered the extravascular space. The unwanted plasma activity may largely be bound to albumin or to larger fragments of red blood cells, neither of which will readily pass the endothelium.
Regional Variations in Hematocrit.—Rosenblum
(19) has cautioned that, since measurement of
regional cerebral blood flow by the 133Xe method
depends on regional hematocrit, interpretation may
be incorrect if one assumes a constant regional
hematocrit in health and disease states. Because of
plasma skimming, regional cerebral hematocrits
may be altered in cerebral infarction or other
conditions characterized by endothelial damage
and during hyperviscosity states. This possibility
must also be considered in these measurements of
LCBV, since they also depend on regional hematocrit (Eq. 1).
We have not yet clarified the importance of
regional differences in brain hematocrit nor the
inconstancy of regional hematocrit in different
pathological states. The error introduced in the
LCBV measurement will depend on the magnitude
of the hematocrit difference and the volume of
brain tissues involved. With our present scanning
method, activity concentrations are quantified in
minimum volumes of brain tissue measuring 2 x 2
x 2 cm. The hematocrit of such a region is
probably not very different from the average cerebral hematocrit. A small error in LCBV due to
regional variations in hematocrit could become
larger, however, when improvements in technique
permit study of smaller tissue volumes.
The section method proposed in this paper may
be useful in clarifying this potential problem of
variations in regional hematocrit. For example, if
one uses a tracer that is strictly limited to red blood
cells (or plasma), the section method measures
local cerebral red blood cell volume (or local
cerebral plasma volume) without requiring assumptions concerning local cerebral hematocrit.
This fact suggests that section scans of one level in
616
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
the brain, after separate labeling of red cells and
plasma, should yield local cerebral hematocrit
values in the transverse section format. With these
kinds of data, one might assess the validity of
assuming a constant cerebral vessel hematocrit
under different conditions, as we have assumed in
this paper.
Initial Results.—Patient scans with the MARK
III scanner were begun 15 minutes after the injection of a 15-mc dose of 99mTc-labeled red blood cells
into an antecubital vein. This dose delivers approximately 200 mrad to the total body. The total
scanning time was 20 minutes. A blood sample was
taken from the antecubital vein in the arm opposite
to the injection site before and after scanning, and
the mean values of C rbc and Cpia9 from these
samples were used in Eq. 1. The counts from each
section scan totaled 25,000-50,000. In estimating
LCBV, no corrections were made for extravascular
label or possible variations in regional brain hematocrit.
In normal patients, LCBV values ranged from
about 2 to 4 ml/100 g, depending on the location
within the cross section. The higher blood volumes
coincided with regions containing cortex. The following two studies illustrate the potential usefulness of the method in evaluating the patient with a
brain tumor.
The first patient (E.K.) was a 72-year-old female
with a history of diabetes, hypertension, and a
clinical episode 3 months earlier which was thought
to be due to a cerebrovascular accident. Conventional scintillation camera imaging with 99mTc
pertechnetate (Fig. 5) demonstrated abnormal uptake in the right hemisphere. The appearance of
this abnormal scan suggested to us a cerebral
infarction in the distribution of the right middle
cerebral artery.
Figure 6 shows the transverse section scans done
first with 99mTc pertechnetate and, on the following
day, with 99mTc-labeled red blood cells. Like the
conventional scintillation camera results, the section scan of 99m Tc pertechnetate showed a local
alteration in the blood-brain barrier suggestive of
cerebral infarction (13). Unexpectedly, the red cell
or LCBV scan indicated that the lesion had a local
blood volume that was three times that of normal
brain tissue, a finding that would not be expected
in a 3-month-old cerebral infarction. Subsequently, a cerebral arteriogram showed an area of
markedly increased abnormal vascularity in the
region of high pertechnetate uptake that had been
noted on the radionuclide scan.
At autopsy a malignant glioma was found which
KUHL. REIVICH. ALAVI. NYARY, STAUM
B
Post
Ant Lt
Rt
FIGURE 5
Conventional scintillation camera views made with "mTc pertechnetate demonstrated an altered blood-brain barrier in the
distribution of the right middle cerebral artery of patient E.K.
Section studies of this patient are shown in Figure 6.
.••>-•
2
1
6 cm
FIGURE 6
Section scans of a malignant glioma in patient E.K. previously
thought to be a cerebral infarction. The section scan of 9amTc
pertechnetate (top left) demonstrates the local alteration in the
blood-brain barrier, and the section scan of °""Tc-labeled red
blood cells (bottom left) demonstrates the LCBV. Corresponding quantitative data are shown at the right. The lesion had
three times the normal blood volume. At autopsy, the tumor was
found to be much more highly vascular than was the adjacent
normal brain tissue.
involved the frontal, parietal, and temporal lobes
and had infiltrated the subarachnoid space. The
center of the lesion was necrotic, and it was not
possible to evaluate the number of blood vessels
there. However, in the more peripheral and viable
areas, especially in the subarachnoid space, the
vascular component of the neoplasm was extremely
prominent. The blood vessels (arteries and venCirculation Research, Vol. 36, May 1975
LOCAL CEREBRAL BLOOD VOLUME
617
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
ules) were extremely congested with red blood
cells, their lumens were dilated, and their number
appeared definitely increased compared with the
number in adjacent normal brain tissue.
The second patient (G.D.) was a 66-year-old
female with a small malignant glioma in the right
parietal lobe. An arteriogram revealed that peritumor edema had expanded the surrounding brain
and displaced the right anterior and posterior
cerebral artery to the left. Two months later, after
steroid therapy, the edema had subsided and
clinically the patient had improved. Initial and
subsequent section studies, first with 99mTc pertechnetate and then with 99mTc-labeled red blood
cells, are shown in Figure 7. The initial section scan
of 99mTc pertechnetate demonstrated the small
tumor by identifying its local alteration of the
blood-brain barrier. The section scan of 99mTclabeled red blood cells showed in the same location
a much larger region of reduced LCBV, probably a
result of compression of the local microcirculation
by peri tumor edema. Corresponding studies were
made 2 months later after steroid therapy. This
section scan of 99mTc pertechnetate showed a
-6-4-2
0
2
4
6D
reduction in tumor uptake of radioactivity, probably a result of steroid effect on tumor vessel
endothelium. As an incidental finding, this section
also showed an increased uptake in the adjacent
convexity due to a burr hole that had been required
in the interval for biopsy. However, the most
interesting feature of the 99mTc-red cell scan was a
more normal LCBV in the region of the tumor.
Presumably, the steroid treatment had diminished
the edema and relieved compression of the peritumor microcirculation.
Note in Figures 6 and 7 the correspondence of
normal LCBV values in the left hemisphere. Also
note the usefulness of the actual quantitative data
for side-to-side comparisons.
Discussion
Our knowledge of the status of cerebral hemodynamics in patients would be enhanced by a reliable
means of measuring absolute values of LCBV
without the need for intracarotid injection of indicator. Until now, the only method approaching
these requirements has been the X-ray fluorescence
technique of TerPogossian et al. (6-9). All previous
-6-4-2
0
2
4
6 cm
FIGURE 7
Section scans of a malignant glioma surrounded by edema in patient G.D. At the top of the figure,
section scans of "mTc pertechnetate show the alteration of the blood-brain barrier. At the bottom,
section scans of '""Tc-labeled red blood cells show LCBV. Initial studies are to the left, and studies
made 2 months later, after steroid therapy, are to the right. The section scans of '""Tc-labeled red
blood cells initially showed a large region of reduced LCBV due to compression of local
microcirculation by edema. After steroid therapy, peritumor edema decreased and peritumor LCBV
returned toward normal.
Circulation Research, Vol. 36, May 1975
618
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
in vivo methods are flawed by a lack of quantification and an inability to discriminate the effects of
extracerebral circulation when the indicator is
injected intravenously. With the method that we
have introduced in this paper, certain of the
restrictions of previous approaches are avoided.
Reconstruction of one or more transverse section
scans after intravenous injection of a radioactive
indicator permits an absolute measurement of
LCBV localized in three dimensions. With this
approach, each part of the brain can be studied
without regard to contaminating data from the
overlying scalp or skull or from other regions of the
brain. The method is an improvement over the
existing X-ray fluorescence technique in that the
quantitative data are presented as one or more
cross-sectional pictures, the attentuation effects by
the human skull on gamma emissions (140-500
kev) in radionuclide scanning are less severe than
those with the X-rays of lower energy which are
encountered in stimulated fluorescence (9), and the
use of a labeled radioactive indicator permits a
wider choice of alternative compounds which may
be administered to the patient in extremely small
amounts.
Several improvements in our method would lead
to better results. Red blood cell labeling should be
refined so that radioactivity appears nowhere but
in circulating red blood cells. We hope to realize
about a tenfold increase in detection sensitivity
with our new section scanner (MARK IV) which is
now under construction. This increase should permit faster scans with better counting statistics and
enable scans of multiple levels in the brain to be
made in a relatively brief period of time. Improvements in spatial resolution are also possible. The
collimated detectors used in this study view a slice
of brain about 2 cm thick, and the method quantifies activity within volumes that are about 2 cm in
diameter. Better spatial resolution in the detection
system would permit visualization and quantification of smaller structures.
As a general method, three-dimensional reconstruction of radionuclide scan data should become
more important in the study of the brain. Ideally,
such a technique would enable precise measurements of radionuclide concentration to be made
within functional structural units of the brain in
man; such measurements have not been possible
up until now. The same study principles might be
applied to other tracers enabling measurements of
local cerebral blood flow and local metabolic rates
for various metabolites within the brain in man.
Measurements of these kinds of regional functions
KUHL. REIVICH. ALAVI. NYARY. STAUM
with three-dimensional resolution throughout the
brain and without the necessity for intracarotid
injection of indicator would constitute a significant
advance over the presently available methods.
Acknowledgment
We thank Dr. Nicholas Gonatas, Dr. Robert Zimmerman,
Mr. Roy Q. Edwards, Mr. Anthony R. Ricci, Mr. Nelson L.
Marin, and Ms. Deborah A. Long for assistance in this project.
References
1. RISBERG J, INGVAR DH: Regional changes in cerebral blood
volume during mental activity. Exp Brain Res 5:72-78,
1968
2. RISBERG J, GUSTAVSSON L, INGVAR DH: Regional cerebral
blood volume during paradoxical sleep. In Cerebral Blood
Flow: Clinical and Experimental Results, edited by M
Brock, C Fieschi, DH Ingvar, NA Lassen, and K
Schurmann. Berlin, Springer-Verlag, 1969, pp 101-103
3. SMITH AL, NEUFELD JR, OMINSKY AJ, WOLLMAN H: Effect of
arterial CO2 tension of cerebral blood flow, mean transit
time, and vascular volume. J Appl Physiol 31:701-707,
1971
4. LANGFITT TW, WEINSTEIN JD, HASSELL NF: Cerebral vaso:
motor paralysis produced by intracranial hypertension.
Neurology 15:622-641, 1965
5. OLDENDORF WH, LISAKA Y: Interference of scalp and skull
with external measurements of brain isotope content: I.
Isotope content of scalp and skull. J Nucl Med
10:177-183, 1969
6. TERPOGOSSIAN MM, PHELPS ME, GRUBB RL JR, GADO M:
Measure of regional cerebral blood volume in vivo by
means of stimulated X-ray fluorescent X-radiation, and
factors affecting this parameter. Eur Neurol 6:218-223,
1971-1972
7. PHELPS ME, GRUBB RL, TERPOGOSSIAN MM: Correlation
between PaCO2 and regional cerebral blood volume by
X-ray fluorescence. J Appl Physiol 35:274-280, 1973
8. GRUBB RL JR, PHELPS ME, RAICHLE ME, TERPOGOSSIAN
MM: Effects of arterial blood pressure on the regional
cerebral blood volume by X-ray fluorescence. Stroke
4:390-399, 1973
9. PHELPS ME, GRUBB RL JR, TERPOGOSSIAN MM: In vivo
regional cerebral blood volume by X-ray fluorescence:
Validation of method. J Appl Physiol 35:741-747, 1973
10. KUHL DE, EDWARDS RQ: Image separation radioisotope
scanning. Radiology 80:653-662, 1963
11. KUHL DE, EDWARDS RQ: The MARK III scanner: A compact
device for multiple-view and section scanning of the
brain. Radiology 96:553-570, 1970
12. KUHL DE, SANDERS TP: Characterizing brain lesions using
transverse section scanning. Radiology 98:317-328, 1971
13. MANLIO FL, MASLAND WS, KUHL DE, STAUM MM: Prognos-
tic significance of a deep-wedge pattern in transverse
section scanning of cerebral infarctions. Radiology
103:135-137, 1972
14. KUHL DE, EDWARDS RQ, RICCI AR, REIVICH M: Quantitative
section scanning using orthogonal tangent correction. J
Nucl Med 14:196-200, 1973
15. GORDON R, BENDER R, HERMAN GT: Algebraic reconstruc-
tion techniques (ART) for three-dimensional electron
microscopy and X-ray photography. J Theor Biol
29:471-481, 1970
Circulation Research, Vol. 36, May 1975
LOCAL CEREBRAL BLOOD VOLUME
16. GILBERT R: Iterative methods for the three-dimensional
reconstruction of an object from projections. J Theor Biol
36:105-117, 1972
17. ATKINS HL, ECKELMAN WC, KLOPPER JF, RICHARDS P:
Vascular imaging with 89mTc-red blood cells. Radiology
106:357-360, 1973
18. CUTKOWSKI RF, DWORKIN MJ: Kit-produced """Tc-labeled
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
Circulation Research, Vol. 36, May 1975
619
red cells for splenic imaging. J Nucl Med 15:1187-1191,
1974
19. ROSENBLUM WI: Can plasma skimming or inconstancy of
regional hematocrit introduce serious errors in regional
cerebral blood flow measurements or their interpretation? Stroke 3:248-254, 1972
Local cerebral blood volume determined by three-dimensional reconstruction of
radionuclide scan data.
D E Kuhl, M Reivich, A Alavi, I Nyary and M M Staum
Downloaded from http://circres.ahajournals.org/ by guest on June 17, 2017
Circ Res. 1975;36:610-619
doi: 10.1161/01.RES.36.5.610
Circulation Research is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231
Copyright © 1975 American Heart Association, Inc. All rights reserved.
Print ISSN: 0009-7330. Online ISSN: 1524-4571
The online version of this article, along with updated information and services, is located on the
World Wide Web at:
http://circres.ahajournals.org/content/36/5/610
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in
Circulation Research can be obtained via RightsLink, a service of the Copyright Clearance Center, not the
Editorial Office. Once the online version of the published article for which permission is being requested is
located, click Request Permissions in the middle column of the Web page under Services. Further information
about this process is available in the Permissions and Rights Question and Answer document.
Reprints: Information about reprints can be found online at:
http://www.lww.com/reprints
Subscriptions: Information about subscribing to Circulation Research is online at:
http://circres.ahajournals.org//subscriptions/