Quantitative in vivo autoradiography with positron emission tomography

47

Brain Research Reviews, 1 (1979) 47-68 0 ElsevierjNorth-Holland Biomedical Press

QUANTITATIVE IN VIYO A~URA~I~~RAP~Y EMISSION TOMOGRAPHY

Key words:

WITH POSITRON

position emission tomography-autoradio~aphy Bow - brain metabotism

-

cerebral

blood

CONTENTS I, Intruduction. ................................. 2. Radion~~~d~ and radiopharmaceuticals ..................... 3. Detection systems. ............................... 4. Tracertechniques ............................... 4.1. Blood volume ............................... 4.2. Metabolism ................................ 4.3. Tissue chemical composition ......................... .............................. 4.4.Bioadflow.. 5. Conclusions .................................. Acknowledglsments ................................ References ....................................

47 49 52 53 53 55 62 64 66 66 66

I. INTRODUCTION

An understanding of most processes occurring in the human brain in health and disease must ultimately be based on a precise knowledge of the dynamic underlying biochemical events that occur in vivo. Although biochemists have developed a very sophisticated ~ders~n~ng of chemical events occurring in brain tissue, their work has been hampered by difficulty in making direct measurements of dynamic events in living tissue, especially that of man. It is obvious that simplified test tube systems may not always accurately represent the situation in vivo. Progress toward the development of a satisfactory means of obtaining dynamic biochemical informaLon on the human brain began with the pioneering work of Kety and Schmidt in 1948. Their introduction of the nitrous oxide technique for the measurement of cerebral blood fiowl” when combined with arteriovenous measurements of various substrates and metabolites provided our first clear information on the relationship between substrate delivery and utilization in man. Although this tech-

48 nique was widely utilized and provided much valuable information, it suffered from the obvious flaw of not providing regional information. Dynamic regional differences in both circulation and metabolism were obscured, if not lost completely, by this approach. The introduction of a method for the measurement of regional cerebra1 blood flow based on the external detection of the clearance of freely diffusible radioactive gases such as xenon-l 33 from the brainlo provided much additional information on regional hemodynamics in the human brain, and demonstrated for the first time the dynamic interplay between localized functional events in the brain and its circuIation18. However, this technique precluded the measurement of regional metabolism, thus significantly reducing its capacity to provide the kind of information necessary. The introduction into the medical environment of cyclotron-produced, positronemitting radiopharmaceuticals for the measurement of brain hemodynamics and metabolism partially fulfilled this need for regional metabolic and biochemical information. In vivo techniques were developed for the measurement not only of regional blood Ilowa7, glucosez7, and oxygen utilization36 in humans, but also regional cerebral blood volume5 and vascular permeability 24.These techniques, however, have received limited application because they require the intracarotid injection of the radiopharmaceutical as well as the presence of a cyclotron in the immediate medical environment. Thus, until recently we have been unable to obtain more than a quasiregional assessment of cerebral blood flow in the human brain and, when circumstances permit, only limited metabolic information. Three significant developments move us closer to the capability of safely acquiring regional in vivo biochemical information in the human brain. First, the appearance within the medical environment of apparatus for nuclear bombardment, such as cyclotrons and linear accelerators, coupled with ingenious techniques for rapid synthesis of radiopharmaceuticals, has provided many radiopharmaceuticals suitable for in vivo regional hemodynamic and metabolic studie@. Second, the parallel development of appropriate mathematical models has provided the basis for practical algorithms that allow parameters of biochemical and physiological significance to be estimated from the data. Third, the more recent development of detection systems employing the concept of positron emission tomography (PET) permits us to monitor the fate of these radiopharmaceuticals in vivo in a truly regional and quantitative manner. Emission tomography is a visualization technique in nuclear medicine that yields an image of the distribution of a previously administered radionu~lide in any desired transverse section of the body. Positron emission tomography utilizes the unique properties of the annihilation radiation generated when positrons are absorbed in matter. It is characterized by the fact that an image reconstructed from the radioactive counting data is an accurate and quantitative representation of the spatial distribution of a radionuclide in the chosen section. This approach is analogous to quantitative autoradiography but has the added advantage of allowing in viva studies. In a larger sense PET represents a complex closely interactive team approach to clinical investigation, Among the several individuals whose diverse talents form an integral part of this team effort are physicists and engineers, radiochemists, applied mathematicians,

49 biochemists, and physiologists. It is the purpose of this review to present an introduction to the concept of PET and the radionuclides, radiopharmaceuticals, detection systems and tracer techniques involved in the actual measurement of tissue hemodynamics, metabolism, and biochemistry. 2. RADXONUCLIDES

AND RADIOPHARMACEUTICALS

A small number of radionu~l~des, especiallyls0, lsN, llC, and 18F, possess chemical and physical properties which make them uniquely suitable for obtaining quantitative, regional, in vivo, biochemical, and physiological information. First, they decay by positron emission (Fig. 1). Positrons, which are positively charged electrons, are emitted from the nucleus of these radionuclides, which are unstable because they have a deficiency of neutrons with respect to the stable state. Positrons lose their kinetic energy in matter in a manner similar to that of electrons. When positrons are brought to rest, they interact with an electron. The two particles undergo annihilation and the mass of the two particles is converted into two annihilation photons traveling 180” from each other with an energy of 511 keV. This is known as annihilation radiation. The high energy of this annihilation radiation (511 keV) gives it greater tissue penetration and, thus, better det~tability than, for example, the 140 keV gamma-ray photons of technetium-99m more commonly used in nuclear medicine imaging. The annihiIation radiation can be uniquely detected by two radiation detectors placed well outside a region of interest and connected to a coincidence circuit (Fig. 2) which records an event from the tissue only if both detectors sense an annihilation photon simultaneously. This coincidence detection of annihilation radiation forms the basis for detection systems used in PET (see below).

onnihilation

photon

Fig. 1. Radionuc~idesemployed in positron emission tomography (PET) decay by the emission of positrons (/I+) or positive electrons from a nucleus unstable because of a deficiency of neutrons. Positrons lose their kinetic energy in matter after traveling a finite distance (-1-6 mm) and, when they are brought to rest, interact with electrons @->. The two particles undergo annihilation and their mass is converted to two annihilation photons traveling at 180” from each other with an energy of 511 keV. It is these annihilation photons that are detected by the imaging device.

Fig. 2. A schematic representation of a radiation detector arrangement for PET. Left: because each radioactive decay of a positron-emitting radionuclide results in two annihilation photons traveling at 180” from each other (Fig. l), coincidence detection is employed to localize the event in the tissue. A coincidence circuit between pairs of detectors arrayed about the imaged object records an event only if both detectors in the circuit record an event simultaneously. Right: in order to increase the number of coincidence lines through an imaged object and, hence, the information gathering abilities of the imaging device, each radiation detector is in coincidence with multiple opposing radiation detectors as shown in this schematic diagram.

51

Second, the chemical nature of 150, lsN, and 1% so resembles the normal constituents of molecules which compose living matter that they can be incorporated in substances that will be included in most metabolic processes. Third, these radio~uclides decay with short half-lives (ls0 = 2.05 min; ‘sN = 9.96 min; W = 20.34 min; rsF = 110 min). Initially these short half-lives might appear to be a major disadvantage because they not only require a dedicated cyclotron in the immediate environment and rapid radiopharmaceutical synthesis, but also metabolic processes sufbciently short to be studied. Indeed, the need for a dedicated cyclotron has, in the past, limited the use of these radionuclides to a few major medical centers worldwide. This will continue to be a requirement for this type of research, but the number of medical cyclotrons is gradually increasing and their operation simplified. Further, many ingenious, fast labeling techniques have been developed (e.g. see Fig. 3) which have already provided a surprisingly large number of metabolic substrates and other molecules of potential importance in the study of the brain*a. An increased awareness of this area of research, and stimulation by interested neuroscientists, will undoubtedly lead to larger numbers of highly specific radiopharmaceuticals. Finaily, experience gathered over the past two decades with these radionuclides indicates that the overwhelming majority of metabolic processes in the brain are in fact relatively fast with respect to the useful life of these radionuclides. A number of studies of importance to our understanding of brain biochemistry and physiology have been carried out with molecules labeled with these radionuclides. These include studies in

PRODUCTION

OF 11C-GLUCOSE

POW Time]

DO Minutd “C-Glucose

‘koz

CT

l

K-

?zschord

Cydotron

1

f’C- fructose _ ‘Y-sucrose

Starved

-i!

Starch

*Phosphates

Alcohol Extraction

“C-Glucose

‘K-fructose + Phosphates ll.- c _ _._ 7 1T.-

a”crose

J

1 Acid

“C-Glucose [75 Minutes]

Hish Pressure liiuid Chromatography

“C-Glucose

Neutralize

“C-Glucose

on Co* Fo;m of Ion EXC~OIUW Resin

“C-Fructose +

+ Concentrate

ttC- Fktose

II

Fig. 3. The synthesis of radiopharmaceuticals for PET often requires ingenious techniques devised by radiochemists to overcome the short half-life of the radionuclide and deliver a high specific activity tracer. This Sgure illustrates such a technique for the synthesis oflW]gIucose beginning with Wlabeled carbon dioxides*, The half-life of carbon-l 1 is 20.34 min.

52 man and animals of brain oxygen consumption; brain glucose transport, kinetics, and metabolism; brain permeability studies; and studies of brain blood volume. Two additional advantages accrue from the short half-life of these radionuclides. First, because of their short half-life, the dose of radiation to the subject as well as those handling the radionuclides is significantly reduced. Second, the short physical half-life of these radionuclides, particularly of 150, permits repeated studies in the same subject, not only because of the low radiation dose but also because the amount of background activity from one study to the next remains minimal to non-existent. This minimizes the need for cumbersome schemes that would otherwise be necessary to account for complex decaying background radioactivity. 3. DETECTION SYSTEMS

Annihilation radiation, because each event consists of two annihilation photons traveling at 180” from each other (Fig. l), can be detected by two radiation detectors connected to a coincidence circuit (Fig. 2). This circuit records an event only if both detectors sense annihilation photons simultaneously. This coincidence detection of the annihilation radiation provides a method of ‘electronic’ collimation since the two detectors can record coincidence events only in a volume of space established by straight lines joining the two detectors. Thus, coincidence detection of annihilation radiation provides a nearly uniform field of view (or sensitivity) in the region between the two detectors. Coincidence detection of the annihilation radiation also permits an easy, accurate correction for the attenuation of the radiation in the tissuezl. Thus it can be shown that the attenuation suffered by the annihilation radiation detected by coincidence is practically independent of the position of the source of the positrons within the tissue between the two detectors21. A number of detection systems have been built and are currently in use1.3p 1sJs-41. In its simplest form such a detection system consists of a pair of scintillation detectors scanning the imaged object at different angles. However, most operational systems incorporate multiple detectors placed around the imaged object, either arranged in banks to form a hexagon, or in a full circle, to improve the efficiency of the radiation collection. Each detector is then operated in coincidence with multiple opposing detectors creating many coincidence lines through the imaged object (Fig. 2). In addition, the detectors are usually moved about the imaged object by rotation, translation or more complex movements. The image is reconstructed by means of a computer-applied convolution-type algorithm. It is important to note that the computer reconstruction of the distribution of a radionuclide in a section of the imaged object is a quantitative representation of its spatial distribution. Actual calibration of the instrument is achieved by imaging appropriately designed phantoms containing known amounts of radioactivity4. This is an especially important feature of these detection systems because it forms the basis of the quantitative tracer methods employed (see below). The resolution of these detection systems has steadily improved. Early instruments had resolution elements approximately 1.5-2.0 cm on a side. This has now been

53

reduced to less than 1.Ocm. In considering ultimate resolution capabilities of PET it is important to remember that the positron itself travels a finite distance in tissue before the annihilation event occurs (Fig. 1)20. This distance varies from 1 to 6 mm, depending in part on the energy of the particular positronso. This distance obviously limits the ultimate resolution of detection devices designed to reconstruct the distribution of a radionuclide by externally detecting armihilation photons. The proper imaging of an organ such as the brain obviously requires several tomographic sections. Detection systems capable of yielding only one section at a time must be operated sequentially with movement of the imaged object with respect to the tomograph between measurements. Refinements in technology have now led to detection systems that are capable of obtaining multiple tomographic sections simultaneouslys*Js. The ability to obtain multiple tomographic sections simultaneously is particularly advantageous when dynamic tracer studies are planned (see below). With single slice instruments such studies must, of necessity, be restricted to a single tomographic section. Further, the ability to obtain multiple tomographic sections simultaneously makes it possible to reconstruct the data in sagittal and coronal slices in addition to the standard horizontal slices. 4. TRACER

TECHNIQUES

4.1. Blood volume

The measurement BLOOD A. PET

B.

RewIutIon

Peripheral

C

VOLUME

of regional cerebral blood volume (rCBV) in transverse (V,,)

Element

Vein

[Counts Set-‘(ml

Tissue)-‘] x 100

V, =. [Counts

Sec?g

Blood)“]x

p x sx f

Fig. 4. Regional cerebral blood volume (CBV) is determined from data obtained (A) from the equilibrium tomographic image of the vascular tracer [Wlcarboxyhemoglobin in the brain (counts/ set/ml tissue) and (B) venous blood samples (counts/&g blood) obtained during the PET scan. Regional CBV is then calculated as shown in (C) where ,gis the density of blood (N 1.05 g/ml), 6 is the tissue density (-1.05 g/ml) and f is the ratio of the mean tissue hematocrit to the large vessel hematocrit (-0.85).

54

sections of the human brain was one of the first measurements accomplished with PET. The early introduction of this measurement was, in part, the result of its simplicity and accuracy. The most satisfactory way to measure rCBV using PET is to label the circulating blood pool of the body with the vascular tracer 1%Iabeled carboxyhemoglobin. This is achieved by causing the subject to inhale air which contains trace amounts of carbon monoxide labeled with carbon-l 1. The regional cerebral blood volume is then determined from data obtained from the equilibrium tomographic image of ~llC]carboxyhemoglobin in the brain (counts/set/ml tissue) and venous blood samples (counts/set/g blood) obtained during the time of the emission scan, as illustrated in Fig. 4. Measurements of rCBV in man by PET not only reveal the expected regional differences in vascular density between gray and white matter, but also provide a clear delineation of the major vascular structures, primarily venous, surrounding the brain

Fig, 5. A PET image of the equilibrium distribution of the vascular tracer [rrC]carboxyhemogiobin in a normal adult human showing larger vascular structures, primarily veins and venous sinuses, and the difference between gray and white matter vascular density. The individual scans, oriented in the horizontal plane, were taken at the levels indicated on the diagram of the head.

55 (Fig. 5). Furthermore, the responsiveness of the cerebral blood volume to changes in arterial carbon dioxide tension observed with emission tomography compare favorably with a variety of other such measurements of cerebral blood volumea. The ability to measure regional cerebral blood volume by PET is important for three reasons. First, it provides in a simple and accurate manner an assessment of regional perfusion within the brain. Changes in regional cerebral blood volume have been correlated with changes in regional cerebral blood flow’, and as such provide a potential index of blood flow. However, it must be remembered that blood volume is a relatively insensitive index of blood flow, changing little in relationship to large changes in blood flowr, and further, that blood volume may at times change quite independent of blood flow due to the phenomenon of blood flow autoregulation*. Second, the technique employed for the measurement of rCBV, when combined with other radiopharmaceuticals, can provide a variety of measurements in addition to that of blood volume alone. For example, a measurement of the equilibrium distribution of the plasma tracer W-labeled methyl albumin when combined sequentially with a measurement of erythrocyte tracer W-labeled carboxyhemoglobin, allows the measurement of the tissue hematocrit. Finally, the measurement of rCBV assumes major importance in the other measurements possible with PET. It provides the only means of determining the concentration of a radiopharmaceutical in the vascular compartment as compared to the extravascular compartment of tissue. The ability to make this distinction accurately is a unique feature of PET and of importance in the measurement of metabolism and tissue chemical composition. 4.2. Metabolism Quantitative measurement of regional metabolism in vivo is well within the capability of PET, but is a measurement of somewhat greater conceptual complexity than the measurement of regional blood volume. Such measurements are exemplified by the quantitative studies of cerebral glucose metabolism using W-labeled glucosezg and isF-labeled 2-deoxy-D-glucoses1 and illustrated by images such as that shown in Fig. 6. In order to understand the tracer strategy employed in these measurements of metabolism by PET, it is first important to understand the behavior of a labeled substrate such as [Wlglucose or a labeled substrate analogue such as [lsF]2-deoxy-pglucose. These are illustrated in Fig. 7. Because of the complex behavior of the tracers employed in metabolic studies with PET, rather sophisticated mathematical models and data acquisition schemes have been developed to extract metabolic rates from the data provided by the quantitative PET images and samples of arterial blood specific activity. These are illustrated in Fig. 8. From the material presented in Figs. 7 and 8, it should be apparent that several important assumptions must be addressed in the application of these techniques. First, we assume that the tracer is transported and metabolized in the same manner and at the same rate as the compound being traced (i.e. tracee). This requirement is clearly met when a substrate such as glucose is labeled with W. [W]Glucose is biologically

Fig. 6. A PET image (center) of a normal human brain following the intravenous administration of llC-labeled glucose. The distribution of activity in the images reflects the local glucose metabolic rate. A standard X-ray ~mputerized tomographic image and a brain slice taken at approximately the same Ievel are shown for comparison.

indistinguishable from unlabeled o-glucose. This requirement is not met in the case of substrate analogue tracers such as [lsF]2-deoxy-D-glucose, thus requiring appropriate correction factors (Fig. 8C). Second, we assume that the metabolized tracer is retained within the area of interest during the measurement period. In the case of a radiolabeled substrate such as [Wlglucose, achievement of this requirement is based on the following reasoning. Immediately upon entry into brain cells [Wlglucose is phosphorylated to [WIglucos~6-phospha~. In this form, it is only released by brain cells when it has been metabolized, as it is assumed that brain does not contain sufiicient quantities of the enzyme necessary for dephosphorylation (glucose-6-phosphatase) and brain cells are impermeable to glucose-6-phosphate. As [Wlglucose is metabolized, it enters several large metabolic pools, the specific activity of the labeled metabolites remains low, and

57 A. Radiolabeled

Subsirols

PET Retolution r__________--

(*XCnHn) Elsmrnf _ __-

- _)

Afteri infl

8.

“OIlI

outflow

Radiolabeled

Substrate

PET Resolution ~__--------_--___~ Arterial

inflpw

’

Anoiogue

(*XV)

Element 1

ox.,

_

VEIJOUS oulflow

Exlracsllular

Fig. 7. A diagrammatic representation of the fate of a radiolabeled substrate (*XCaH,) such as [rtC]glucose or radiolabeled substrate analogue (*XY) such as [i6F]2-deoxy-D-glucose. Both are introduced into the vascular compart~nt from a peripheral site such as a vein. They traverse the vasculature in the region of interest, usually entering brain tissue by a highly specific mechanism of carrier-mediated, facilitated diffusion. Because a significant pool of the substance being traced (e.g. glucose) usually exists in the tissue the radiolabeled tracer moves back and forth across the blood-brain barrier in response to local concentration gradients. Within the cell the tracer must be transformed, through metabolism, to a chemical species that does not readily leave the cell in order to permit satisfactory measurement of its metabolism. In the case of radiolabeled glucose (A) or a glucose analogue (Bj ph~pho~lation through the action of hexokinase irreversibly tr~sforms the tracer into an impermeable compound. Analogues (B) are so designed to remain irreversibly trapped by resisting further metabolic breakdown whereas a tracer such as radiolabeled glucose (A) will eventually be metabolized to labeled substances such as carbon dioxide which readily passes from the cell and the tissue. Mathematical models and data acquisition schemes must account for these complexities in order to achieve an accurate measurement of metabolism.

the egress of tracer from the brain is minimal for a period of about 5 min27. Thus, satisfactory measurement of regional cerebral glucose metabolism can be achieved using [Wlglucose when the measurement is performed within the first few minutes, or when a satisfactory correction can be achieved for the egress of the metabolized tracer (see below). The requirement for metabolic trapping within the tissue is more easily achieved when analogues such as lsF-labeled Zdeoxy-D-glucose are employed. fn the case of such a tracer, metabolism proceeds only through the phosphorylation stepss. Further metabolism is blocked because of the structure of the molecule. Thus, l*Flabeled 2-deoxy-D-glucose-6-phosphate remains irreversibly trapped within the cells of the central nervous system. Third, we assume that the amount of tracer not metabolized, i.e. free tracer in blood and extracellular fluid, is either negligible or accounted for at the time of the measurement. This requirement is most easily met when radiolabeled analogues are employed. Because they are irreversibly trapped within the cells of the region of

58

Time 8. Radiolabeled

Substrates

C

Substrate

Radialabeied

After Tracer

Administration

+

Analogues

Fig. 8. A general approach to data acquisition and analysis in the measurement of metabolism by PET. A: the data resulting from the intravenous administration of a tracer such as [W]glucose or [18F]2-deoxy-n-glucose consists of time varying blood specific activity cab(t) ) measured by repeated sampling of blood from a peripheral artery and local tissue activity (q(T) ) determined at discrete points in time by PET after tracer administration. At early times after tracer administration (T <= 5 min) activity in blood contributes significantly to the PET image whereas at late times (T’ > 30 min) blood activity contributes little to the image. The basic step in determining the metabolic rate from these data generated either by a radiolabeled substrate or a substrate analogue is to divide the tissue activity by the integral of the blood specific activity. B: in the case of a radiolabeled substrate such as [t*C]glucose the data are analyzed at early times to avoid egress of metabolized tracer from the tissue. As a result a correction must be made for unmetabolized tracer in the blood as well as the tissue. These corrections require a prior knowledge of the blood volume (Vb), which can be measured by PET using [W]carboxyhemoglobin and the concentration of tracee in blood (Cb). Additionally, two parameters, a and x, must be determined by separate experimentation. a denotes the ratio of forward to reverse unidirectional systemic tracer fluxes across the blood-brain barrier, and x the sum of the two compartmental rate constants for egress of tracee from tissue to blood and into the metabolic pathways. Details of the derivation of this model are given in refs. 27 and 29. C: in the case of a radiolabeled substrate analogue such as [18F]2-deoxy-n-glucose the data are analyzed at late times (T’ > 30 min) when blood acttnty is negligible. Corrections for the presence and behavior of unmetabolized tracer in the tissue must still be made and are based upon a prior knowledge of x, the compartmental rate constant for tracee movement from blood to tissue (KI*) and the concentration of tracer in plasma at the time of the measurement (C,*). The other correction accounts for the difference. between tracer and tracee that occurs when radiolabeled analogues are used. Here I equals the ratio of the distribution volume of tracer to that of tracee.; 0 is the fraction of tracee, in this case glucose, that once phosphorylated continues down the glycolytic pathways; and km*, V*max, , km and Vmax are the familiar Michaelis-Menton kinetic constants for tracer and tracee. Further details of this model are found in the work of Sokoloff et a1.33.

59 interests3 it is possible to delay the actual measurement until free tracer in tissue and blood has fallen to insignificant levels, thus minimizing the errors in estimating the free tracer in the tissue. In the case of glucose analogues, this usually requires 30-60 min33. Such a delay is not possible when radiolabeled substrates such as [llC]glucose are employed because, as pointed out above, significant egress of radiolabeled metabolites from the tissue begins within 5 min of the time of injectionz7. Thus, in the case of radiolabeled substrates, the tracer strategy employed must explicitly account for the free tracer in blood and tissue. This is accomplished, in part, by measuring the regional cerebral blood volume in association with the measurement of regional metabolism. By knowing the regional cerebral blood volume, it is then possible to compute not only the free tracer in blood, but also the free tracer within the tissue itselfz9. It should be clear from the foregoing that tracers in the form of radiolabeled substrates (e.g. [ilC]glucose) and radiolabeled substrate analogues (e.g. [1sF]2-deoxyD-glucose) each have their special advantages and disadvantages. It is important that these should be specifically noted. There are several advantages to the use of a radiolabeled substrate such as [llC]glucose. First, the tracer is biochemically identical to the compound being traced. Thus, corrections need not be made in the tracer model for differences in transport properties and enzyme affinities which vary among species (Fig. 8C)33. Such corrections may present an added difficulty when the organ of interest is diseased. Second, the measurement requires, by necessity, a relatively short time. This permits repeated measurements to be made during the course of one experiment should they be required for the evaluation of transient phenomena. Finally, there are currently available a large number of substrates labeled with positron-emitting radioisotope+, thus affording one the opportunity of making a wide variety of metabolic measurements. These advantages of radiolabeled substrates such as [ilC]glucose are tempered by several clear disadvantages. First, these tracers are not efficiently trapped as radiolabeled metabolites in the tissue 27,29.Thus the measurement strategy must be designed to permit measurement within the first few minutes following injection, or properly account for tracer egress from the tissue. One possible and rather simple means of accounting for this egress of metabolized tracer from the tissue in each subject would be to repeat the measurement of metabolism several times within the first 5 min following the injection of the tracer. The computed metabolic rate from these data would obviously fall as a function of time. The rate of its decline could be obtained from the data. Knowledge of this rate of decline as a function of time after injection should then allow an estimate of the true metabolic rate prior to egress of metabolized tracer. Preliminary data in our laboratory (unpublished) suggests that this is a very workable solution to this problem. We tested this approach with both [Vlglucose and 150-labeled oxygen, the latter being a most severe test of this idea because of the rapid movement of [150]water of metabolism from the tissue. In both cases our ability to determine the true utilization rate was excellent. In the case of cerebral oxygen utilization the correlation between the estimated and actual oxygen utilization, based on this strategy, was particularly gratifying (correlation coefficient = 0.90; number = 6). It should be noted that this strategy obviously restricts one to a single tissue slice, thus

60 making it advantageous to employ a PET imaging device capable of rapidly scanning multiple slices simultaneouslys*~ss. Second, because the measurement must be made shortly after the injection of the tracer, it is necessary to couple the measurement of metabolism with the measurement of regional cerebral blood volume, thus adding to the logistical and computational complexity of the measurement. Finally, specific descriptors (a and x; Fig. 8B)ss of the behavior of free tracer in the tissue must be predetermined in order to compute the amount of tracer within the tissue which remains unmetabolized. With the advent of rapid, multislice PET imaging devicess8q3s*41it should be possible to determine such parameters on-line by obtaining, rapidly, sequential PET images of regional tissue activity as a function of time during the course of a slow intravenous infusion of tracer. These data, coupled with a knowledge of the time activity history of tracer in arterial blood, should allow the application of statistical procedures previously described and validated27 to estimate these parameters, as well as the metabolic rate. Successful implementation of such a strategy would not only circumvent the risk of assuming arbitrary values for such parameters, but would also greatly expand the information available from such a study. Thus, in addition to a measurement of the local utilization rate of glucose, one should be able to obtain such important measurements as the brain-to-blood glucose concentration ratio, relative free-glucose space in tissue, brain free-glucose concentration, and brain free-glucose turnover times7 in humans, for the first time. Radiolabeled substrate analogues have a number of unique advantages. First, these compounds are so designed as to remain irreversibly trapped within the tissue once they have been metabolized 33. Thus one is not concerned with the egress of metabolites when employing them to measure regional metabolic rates. Second, because these compounds remain irreversibly trapped within tissue, it is possible to minimize corrections for free tracer in tissue and blood such as is done when radiolabeled substrates are employed. By reducing or eliminating the necessity for making such corrections, one obviously enhances the accuracy of the measurement of metabolism. Radiolabeled substrate analogues do, however, have a number of noteworthy disadvantages. First and most importantly, they are not biochemically identical to the compound being traced. When radiolabeled substrate analogues are employed, it is necessary to correct for differences in transport properties and enzyme affinities which have been shown to vary among species33. Such corrections may present an added difficulty when the organ of interest is diseased. Second, because the strategy employed relies on the time (~30-60 min) for unmetabolized tracer to clear from blood and tissue, measurement time tends to preclude the examination of transient phenomena, and repeat measurements. This time requirement also necessitates the administration of somewhat larger quantities of the radionuclide and/or the use of radionuclides with longer half-lives (e.g. 1sF rather than 1X) in order to achieve adequate counting statistics in the PET image. Third, substrate analogues are capable of disrupting normal metabolism if they enter the cell in sufficiently high concentration. Although this is not a problem when tracer quantitites of radiolabeled substrate analogues are employed, it is important

61 that synthesis techniques employed in the production of such tracers assure very low concentrations of unlabeled carrier. The strategy outlined above for the measurement of metabolism, using as tracers radiolabeled substrates or substrate analogues, does not permit, in the present form, the measurement of tissue oxygen co~umptio~. This is not a problem of tracer labeling as the positron-emitting radionuclide 150 is readily available from most cyclotrons in the form of a gas43 which can then be used to label the subject’s blood as [lsO]oxyhemoglobin. Rather it is the prompt conversion of W-labeled oxygen to IsOlabeled water of metabolism in brain and its rapid egress from the tissue that precludes the direct application of the methods outlined above. Quite simply, the egress of metabolized tracer from the tissue occurs immediately and in significant quantitites. In response to the clear need for a means of measuring tissue oxygen consumption, Jones and his colleagueslr~r~, as well as Subramanyam et al.35 have proposed a technique for the measurement of the regional fractional extraction of oxygen. This measurement is a~omplished by the sequential inhalation of trace amounts of isO-labeled oxygen and isO-labeled carbon dioxide (Fig. 9). The inhalation of isO-labeled oxygen to equilibrium produces as brain image that results from the accumulation of molecular oxygen and water of metabolism within brain tissue. In order to obtain the fractional extraction of oxygen by the brain, it is necessary to correct this image for the egress of water of metabolism labeled with 150 from the INHALATION

INHALATION

of O’s0

of C’aOa

\ \ 0%Hgb I

A

H,'sOme+. =

i

c.ca

’ \

I

Co*t (&&i (Jonas et al. 1976)

i \ \

B

\

\

.I$

Hf %irc.

\

H,‘60circ

*

= fF) fF/V+kl

’c (Lenzl

at al,

1979)

Fig. 9. The tracer strategy employed in the measurement of the brain fractional extraction of oxygen. A: the concentration of water of metabolism (HP O,,t) in brain resulting from the inhalation of Or5 0 is dependent upon the blood concentration of radioactive oxygen (Ca+), the brain fractional extraction of oxygen ((Ca - CJCa), the blood flow (F) to a region of volume(V), and the physical decay constant of 150 (1; 0.335 min-r). B: the concentration of circulating water (HrPOerre) in brain resulting from the inhalation of Cr50e is only dependent upon flow (F) to a volume (V), 1 and c, the arterial concentration of HP0 drc. The ratio of Ha150,,t to Hst50ctre normalized by the ratio Ca*/c yields the fractional extraction of oxygen (see text).

62

brain and its recirculation back to the brain. In order to do this, tsO-labeled carbon dioxide is inhaled to equilibrium. Inhalation of 150-labeled carbon dioxide immediately labels the circulating water pool in blood because of the presence of carbonic anhydrase in the red cell 44. The resulting brain image simulates the recirculation of metabolic water to the brain. The regional oxygen fractional extraction can then be computed by dividing the equilibrium image of metabolic water (Hat50,& by the image of circulating water (HslsOcrrJ. Thus Hz

150,,t

H2 150circ

(Ca - Cv) Ca* .---____ (Ca) ’ C

where Ca and Cv represent the arterial and cerebral venous oxygen contents respectively and (Ca-Cv)/(Ca), the oxygen fractional extraction; Ca* represents the arterial concentration of radioactive oxygen and c the arterial concentration of circulating Hsl50. The ratio Ca*/c will be a constant for all regions of the tissue and hence this ‘normalization’15 procedure will produce a distribution of activity which represents the local fractional extraction of oxygen. Employing this technique, these investigators have clearly demonstrated regional differences in the fractional extraction of oxygen in the brain of man under normal circumstances and in a number of disease states. Their data clearly demonstrate an apparent potential usefulness of this technique in assessing regional metabolism in the brain. However, it must always be borne in mind that the fractional extraction of oxygen is not only influenced by the local oxygen utilization rate in brain, but also by the regional blood flow as can be seen from the following expression: fractional extraction of 02 =

oxygen consumption blood flow x arterial 02 content ’

Thus where oxygenutilization and blood flow move in parallel, for example during functional activation25 or during a seizuress, one would not observe a significant change in the local fractional extraction of oxygen. Thus a significant change in local oxygen would not be appreciated with this tracer technique. On the other hand, where blood flow is altered independently of metabolism, for example during hyperventilations*, a significant change in the local fractional extraction of oxygen would merely reflect the change in blood flow. Without a knowledge of the latter, it would not be possible to arrive at a correct interpretation of the change in the regional fractional extraction of oxygen. Clearly, this technique has promise, but only if it can be employed with an accurate measurement of regional blood flow. 4.3. Tissue chemical composition PET can be used to measure quantitatively the chemical composition of tissues of the body by using appropriately selected compounds labeled with positron-emitting radioisotopes that rapidly distribute between the blood and the tissue of interest. The measurement is based on the determination of the tissue-to-blood partition coefficient of the tracer at equilibrium. Knowledge of the tissue-to-blood partition coefficient plus the concentration of the compound of interest in the vascular compartment, permits the calculation of the actual tissue concentration as illustrated in Fig. 10.

63 TISSUE:BLOOD pi. PET Resolution

8.

Peripheral

PARTITION

COEFFICIENT

(X)

Element

Artery

or

Vein

C.

[Counts See-‘(g

Tissue)eL]-[Counts

Sece’(g

Bloodlm’]xVb

X= [Counts Set-‘(g

Blood)-‘]

Fig. 10. The tracer strategy employed in the measurement of the tissue to blood partition coefficient of a substance (1) is shown. At tracer equilibrium the PET image (A) measures activity in blood and tissue (counts/set/g tissue). Activity in blood (counts/%/g blood) is measured separately in a sample of blood from a peripheral artery or vein. f. is calculatedas shownin (C), wherethe volumeof blood in the region of interest (Vb) is determinedas previouslydescribed(see Fig. 4).

We have tested this concept by measuring brain carbon dioxide content using W-labeled carbon dioxide as our tracer 26. These measurements were performed in adult rhesus monkeys at different levels of arterial carbon dioxide tension, which were adjusted by varying the partial pressure of carbon dioxide in the inspired air. The results of this study are shown in Fig. 11. These data compare favorably with direct measurements of tissue carbon dioxide content and show the expected relationship between arterial carbon dioxide tension and tissue CO2 content. These data illustrate the unique potential of PET specifically to explore, in vivo and quantitatively, brain tissue acid-base chemistry. The ability to obtain this information in humans should provide valuable new insights with regard to the regulation of acid-base chemistry in normal and diseased states of the brain. It is important to emphasize the general applicability of this type of measurement. It is not restricted to the measurement of tissue carbon dioxide content and acid-base status with WOs. The approach may be employed with a variety of radiopharmaceuticals to explore the chemical properties of the brain or any other organ of the body. One of the most obvious applications of this method is its potential to measure tissue drug levels. The study of tissue drug levels with PET might not only refine our appraisal of the therapeutic efficacy of specific drugs, i.e. anticonvulsants, but also would permit, for the first time, a study of specific brain receptor, in vivo and regionally, in humans when employed with drugs with specific receptor-binding characteristics. This approach to the study of specific brain receptors is analogous in all respects to studies in laboratory animal@ except that it can be conducted in vivo

64 30

-

,25

.

Y.

0.23X

+ 5.75

rao.95

/

ii E20. ; 0 z 0

15-

0” 0

IO -

: z I-

s-

01

0

_ (Metsetsr 8

IO

20

30

pace*

40

60

Sisrjo,1971)

70

80

tmrn~~

Fig. 11. The relationship between brain tissue carbon dioxide content measured with [W]carbon dioxide and the arterial carbon dioxide tension (PaCOg in adult rhesus monkeys, The solid line is the regression line computed for the data. The broken line was developed from the data of Messenter and Siesjoz7 for comparison.

and, thus, in humans. The ability to make such measurements in humans may assume special importance in the study of certain conditions for which a satisfactory animal model does not exist (e.g. schizophrenia, Parkinson’s disease). 4.4. Hood&w The development of a completely satisfactory measurement of regional cerebral blood flow (rCBF) employing PET has lagged behind developments in other areas of PET tracer methodology. This is somewhat ironic in that prior to the advent of PET, blood flow was the primary parameter available to the clinical investigator interested in hemod~amic and metabofic data on the human brain. All other measurements, including blood volume, metabolism, and tissue chemical composition, were obtained, at best, with great difficulty. The reason for this state of affairs is related to the fact that techniques for the measurement of blood flow have, to date, depended upon the analysis of dynamic tracer data. Such analysis requires data sampling times of 1 set or less. PET units currently in operation have not been able to collect sufficient data for an adequate r~const~ction of a tissue slice in less than l-2 min, although units currently being tested****1 may offer the temporal resolution necessary for truly dynamic tracer studies in the measurement of rCBF. As a result, the development of methods for the measurement of blood flow have focused on techniques that require equilibrium imaging. One approach, illustrated in Fig. 9B, relates an equilibrium image of 150-labeled water, obtained by the continous inhalation of 150-labeled carbon dioxide, to rCBF. As Lenzi et al.15 and others have shown 35, the amount of the tracer [rsO]water in a region of the brain or other organ under these circumstances is proportional to rCBF.

65 Thus, the equation in Fig. 9B can be presented in the following manner to permit culation of blood flow (F) to a region of volume (V): F/V =

Cal-

1 c/CT-P--’

where A is the physical decay constant of l&O (~0.335 min-l),

P the tissue-to-blood equilibrium partition coefficient for water (gray matter N 1.04; white matter 1: 0.88; whole brain = 0.95)Q’ and c and CT are the arterial blood and tissue concentration of circulating HQrsO, respectively. Although superficially appealing, this approach has two significant problems. First, the relationship between blood flow (F/V) and the measured parameters c and CT is non-linear 1s. For example, it can easily be shown that starting at a F/V of 50 ml/100 g/min, a normal resting blood flow in man3’, a 40 % increase in blood flow is accompanied by only a 12 % increase in the ratio ~/CT, whereas a 40 oAdecrease in blood flow is accompanied by a 22 ‘A decrease in C/CT. Thus as blood flow increases the measured parameters become less sensitive to change, whereas with a decrease in blood flow they become more sensitive. Sceond, the estimate of blood flow by this method is very sensitive to errors in the measurement of c and Cr. For example, at a blood flow of 50 ml/100 g/min, letting 3,be 0.35 min-l and P be 0.95 (ref. 37), it can be shown that measurement errors leading to a 10 % overestimation of C/CT produce a 20% underestimation of blood flow. As blood flow increases, error magnification becomes greater because the value of C/CT approaches the value of l/P. The converse is also true. In considering the likelihood of errors in the determination of C/CT it should be noted that this measurement actually involves several measurements, including (i) the response of the PET imaging system to radioactivity in the region of interest; (ii) arterial blood radioactivity, and (iii) the relationship between measurement (i) and (ii) as determined by PET imaging of appropriately designed and calibrated phantoms4. Thus the opportunity for errors to occur is high. A second approach to the estimation of rCBF with PET is analogous to the radiolabeled microsphere method employed in laboratory animalsis. While the microsphere method employed in animals depends upon mechanical trapping of the tracer in the microcirculation of the organ of interest, PET studies have employed the concept of metabolic trapping. Specifically, rQN-labeled ammonia has been employed as a tracer to map perfusion in the brain22 and other organs9 because of its ability to be efficiently trapped, presumably by immediate incorporation into amino acids. Although conceptually sound, this approach is not likely to succeed in the case of rQNammonia because this tracer is not freely permeable 22, due to the fact that some of the tracer in blood always exists in the impermeable form of ammonium ion (1QNH4+). Because of this it can be shown that the amount deposited in the tissue as a function of blood flow will progressively deviate from the expected. Thus, as blood flow increases, estimated flow based on 13N-ammonia deposition will be increasingly underestimated. The particular case of lQN-ammonia is further complicated by the fact that the ratio of the diffusible species (1sNHQ)to the non-diffusible (iQNH4+) in blood is a function of blood PH. Therefore, PET images of rQN-ammonia purporting to depict a map of brain perfusion or blood flow must be viewed with caution. Clearly what is

66

needed is a freely permeable radiopharmaceutical that is completely trapped within tissue during a single pass through the microcirculation. Although no such radiopharmaceutical labeled with a positron-emitting radionuclide has yet emerged, several candidates are being considered by a number of laboratories. Finally, it seems reasonable to assume that rCBF will eventually be measured by PET in a manner analogous to the tissue autoradiographic method originally proposed by Landau et al.14 and later refined by Reivich et al.30 and Sokoloff and co-workerssa. In this technique as originally applied to animals, a freely diffusible radiolabeled tracer is infused intravenously for 1 min followed immediateiy by decapitation. Quanti~tive autoradiograms of brain slices and the time activity curve of arterial blood radioactivity form the data from which rCBF is calculated. It is clear that a PET image could be substituted for the tissue autoradiogram provided that an image could be obtained in a relatively short period of time (~5-10 set). Recent developments in PET technologyaT3a,41 suggest that this will be achieved in the near future and a number of freely diffusible tracers labeled with positron-emitting radionuclides could serve satisfactorily as the tracer for such measurements43. 5. CONCLUSIONS

The emergence of PET in the biomedical environment provides a unique new tool for sophisticated clinical investigation. The potential of the technique is obvious. PET should provide unique new information on the biochemistry and physiology of the human central nervous system in health and disease. Such information can provide a valuable interface between basic science studies in simple systems and the in vivo condition in man, which most investigators, whatever their tools, are striving to understand. The requirements for the successful implementation of PET may not be so obvious. It may be clear to most that expensive and complicated machinery including cyclotrons, imaging devices and computers are a necessary foundation, but not so clear that a critical mass of diverse human talents must be closely coordinated in order to make the full potential of PET a reality in the biochemical environment. Successful implementation of PET, if achieved, will represent a truly collaborative effort among physicists, chemists, mathematicians, and neurobiologists who share in common the desire to develop and implement quantitative techniques for the study of brain biochemistry and physiology in man. ACKNOWLEDGEMENTS

This work was supported (NINCDS).

by USPHS

Grants

HL13851

and

NS06833

REFERENCES 1 Bohm, C., Eriksson, L., Bergstrom, M., Litton, J., Sundman, R. and Singh, M., A computer assisted ring detector positron camera system for reconstruction tomography of the brain, I.E.E. E. Trans., NS-25 (1978) 624-637.

67 2 Budinger, T. F., Derenzo, S. E., Greenberg, W. L., Gullberg, G. T. and Huesman, R. H., Quantitative potentials of dynamic emission computed tomography, .I. nucl. Med., 19 (1978) 309-315. 3 Burnham, C. A., Aronow, S. and Brownell, G. L., A hybrid positron scanner, Phys. med. Biol., 15 (1970) 517-528. 4 Eichling, J. O., Higgins, C. S. and Ter-Pogossian, M. M., Determination of radionuclide concentrations with positron CT scanning (PETT), J. nucl. Med., 18 (1977) 845-847. 5 Eichling, J. O., Raichle, M. E., Grubb, R. L., Jr., Larson, K. B. and Ter-Pogossian, M. M., In vivo determination of cerebral blood volume with radioactive oxygen-15 in the monkey, Circulat Res., 37 (1975) 707-714. 6 Greenberg, J. H. , Alavi, A., Reivich, M., Kuhl, D. and Uzzell, B., Local cerebral blood volume response to carbon dioxide in man, Circulat. Res., 43 (1978) 324330. 7 Grubb, R. L., Jr., Raichle, M. E., Eichling, J. 0. and Ter-Pogossian, M. M., The effects of changes in PaCOz on cerebral blood volume, blood flow and vascular mean transit time, Stroke, 5 (1974) 630-639. 8 Grubb, R. L., Jr., Raichle, M. E., Phelps, M. E. and Ratcheson, R. A., Effects of increased intracranial pressure on cerebral blood volume, blood flow and oxygen utilization in monkeys, J. Neurosurg., 43 (1975) 385-398. 9 Harper, P. V., Lathrop, K. A., Frizek, H., Lembarer, N., Stark, V. and Hoffer, P. B., Clinical feasibility of myocardial imaging with isNHs, J. nucl. Med., 13 (1972) 278-280. 10 Hoedt-Rasmussen, K., Sveinsdottir, E. and Lassen, N. A., Regional cerebral blood flow in man determined by intra-arterial injection of radioactive inert gas, Circular. Res., 18 (1966) 237-247. 11 Jones, T., Chesler, D. A. and Ter-Pogossian, M. M., The continuous inhalation of oxygen-15 for assessing regional oxygen extraction in the brain of man, Brit. J. Radiol., 49 (1976) 339-343. 12 Kety, S. S. and Schmidt, C. F., The nitrous oxide method for the quantitative determination of cerebral blood flow in man: theory, procedure and normal values, J. c&n. Invest., 27 (1948) 476-483. 13 Kuhar, M. J., Murrin, L. C., Malouf, A. T. and Klumm, N., Dopamine receptor binding in vivo: the feasibility of autoradiographic studies, Life Sci., 22 (1978) 203-210. 14 Landau, W. M., Freygang, W. H., Jr., Rowland, L. P., Sokoloff, L. and Kety, S. S., The local circulation of the living brain; values in the anesthetized and unanesthetized cat, Trans. Amer. Neural. Assoc., 80 (1955) 125-129. 15 Lenzi, G. L., Jones, T., McKenzie, C. G., Buckingham, P. D., Clark, J. C. and Moss, S., Study of regional cerebral metabolism and blood flow relationships in man using the method of continuously inhaling oxygen-15 and oxygen-15 labeled carbon dioxide, J. Neural. Neurosurg. Psychiat., 41 (1978) l-10. 16 Marcus, M. L., Heistad, D. D., Ehrhardt, J. C. and Abboud, F. M., Total and regional cerebral blood flow measurement with 7-, lo-, 15-, 25-, and 50-pm microspheres, J. appl. Physiol., 40 (1976) 501-507. 17 Messenter, K. and Siesjo, B. K., The intracellular pH in the brain in acute and sustained hypercapnia, Actaphysiol. stand., 83 (1971) 210-219. 18 Olesen, J., Contralateral focal increase in cerebral blood flow in man during arm work, Brain, 94 (1971) 635-646. 19 Phelps, M. E., Hoffman, E. J., Huang, S. C. and Kuhl, D. E., ECAT: a new computerized tomographic imaging system for positron-emitting radio-pharmaceuticals, J. nucl. Med., 19 (1978) 635-647. 20 Phelps, M. E., Hoffman, E. J., Huang, S. C. and Ter-Pogossian, M. M., Effect of positron range on spatial resolution, J. nucl. Med., 16 (1975) 649-652. 21 Phelps, M. E., Hoffman, E. J., Mullani, N. A. and Ter-Pogossian, M. M., Application of annihilation coincidence detection to transaxial reconstruction tomography, J. nucI. Med., 16 (1975) 210-224. 22 Phelps, M. E., Hoffman, E. J. and Raybaud, C., Factors which affect cerebral uptake and retention of 13NHs, Stroke, 8 (1977) 694702. 23 Plum, F., Posner, J. B. and Troy, B., Cerebral metabolic and circulatory responses to induced convulsions in animals, Arch. Neural. (Chic.), 18 (1968) 1-13. 24 Raichle, M. E., Eichling, J. O., Straatmann, M. G., Welch, M. J., Larson, K. B. and TerPogossian, M. M., Blood-brain barrier permeability of llC-labeled alcohols and 150-labeled water, Amer. J. Physiol., 230 (1976) 543-552. 25 Raichle, M. E., Grubb, R. L., Jr., Gado, M. H., Eichling, J. 0. and Ter-Pogossian, M. M., Correl-

68 ation between regional cerebral blood flow and oxidative metabolism,

Arch. Neural.

(Chic.),

33

(1976) 523-526. 26 Raichle, M. E., Grubb, R. L., Jr. and Higgins, C. S., Measurement of brain tissue carbon dioxide content in vivo by emission tomography, Bruin Research, 166 (1979) 413-417.

27 Raichle, M. E., Larson, K. B., Phelps, M. E., Grubb, R. L., Jr., Welch, M. J. and Ter-Pogossian, M. M., In vivo measurement of brain glucose transport and metabolism employing glucose -“C, Amer. J. Physiol., 228 (1975) 1936-1948. 28 Raichle, M. E., Posner, J. B. and Plum, F., Cerebral blood flow during and after hyferventilation, Arch. Neurol. (Chic.), 23 (1970) 394403. 29 Raichle, M. E., Welch, M. J., Grubb, R. L., Jr., Higgins, C. S., Ter-Pogossian, M. M. and Larson, K. B., Measurement of regional substrate utilization rates by emission tomography, Science, 199 (1978) 986-987. 30 Reivich, M., Jehle, J., Sokoloff, L. and Kety, S. S., Measurement of regional cerebral blood flow with antipyrine-14C in awake cats, J. appl. Physioi., 27 (1969) 296-300. 31 Reivich, M., Kuhl, D., Wolf, A., Greenberg, J., Phelps, M., Ido, T., Cosella, N., Fowler, J., Hoffman, E., Alavi, A., Som, P. and Sokoloff, L., The [i*F]fluorodeoxyglucose method for the measurement of local cerebral glucose metabolism in man, Circulut. Res., 44 (1979) 127-137. 32 Sakurada, O., Kennedy, C., Jehle, J., Brown, J. D., Carbin, G. L. and Sokoloff, L., Measurement of local cerebral blood flow with iodo[14C]antipyrine, Amer. J. Physioi., 234 (1978) H59-H66. 33 Sokoloff, L., Reivich, M., Kennedy, C., deRosiers, M. H., Pattak, C. S., Pettigrew, K. C., Sakurada, 0. and Shinohara, M., The [14C]deoxyglucose method for the measurement of local cerebral glucose utilization: theory, procedure, and normal values in the conscious and anesthetized albino rat, J. Neurochem., 28 (1977) 897-916. 34 Straatmann, M. G. and Welch, M. J., The liquid chromatographic purification of carbon-11 labeled glucose, ht. J. appl. Radiat. Isotopes, 24 (1973) 234-236. 35 Subramanyam, R., Alpert, N. M., Hoop, B., Jr., Brownell, G. L. and Taveras, J. M., A model for regional cerebral oxygen distribution during continuous inhalation of 1502, C’s0 and CrsOs, J. nucl. Med., 19 (1978) 48-53. 36 Ter-Pogossian, M. M., Eichling, J. O., Davis, D. O., and Welch, M. J., The measure in vivo of

regional cerebral oxygen utilization by means of oxyhenoglobin labeled with radioactive cxygen15, J. c/in. Invest., 49 (1970) 381-391. 37 Ter-Pogossian, M. M., Eichling, J. O., Davis, D. O., Welch, M. J. and Metzger, J. M., The determination of regional cerebral blood flow by means of water labeled with radioactive oxygen-15, Radiology, 93 (1969) 3140. 38 Ter-Pogossian, M. M., Mullani, N. A., Hood, J. T., Higgins, C. S. and Currie,

C. M., A multislice positron emission computed tomograph (PETT IV) yielding transverse and longitudinal images, Radiology, 128 (1978) 477-484. 39 Ter-Pogossian, M. M., Mullani, N. A., Hood, J. T., Higgins, C. S. and Ficke, D. C., Design considerations for a positron emission transverse tomograph (PETT V) for imaging of the brain, J. Comput. Assist. Tomogr., 2 (1978) 539-544. 40 Ter-Pogossian, M. M., Phelps, M. E., Hoffman,

41 42

43 44

E. J. and Mullani, N. A., A positron-emission transaxial tomograph for nuclear imaging (PETT), Radiology, 114 (1975) 89-98. Thompson, C. J., Yamamoto, Y. L. and Meyer, E., Positome II: a high efficiency PET device for dynamic studies, J. Comput. Assist. Tomogr., 2 (1978) 650-651. Welch, M. J. (Ed.), Radiopharmaceuticals and Other Compounds Labelled with Short-lived Rudionuclides, Pergamon Press, New York, 1977. Welch, M. J. and Ter-Pogossian, M. M., Preparation of short half-lived radioactive gases for medical studies, Rudiut. Res., 36 (1968) 580-587. West, J. B. and Dollery, C. T., Uptake of oxygen-15 labeled COz compared with carbon-11 labeled CO2 in the lung, J. appl. Physiol., 17 (1962) 9-13.