"luxury perfusion") and to increases relative to metabolic demand in others (relative "luxury perfu sion"). These phenomena were observed mainly between the ...
Journal of Cerebral Blood Flow and Metabolism
2:321-335 © 1982 Raven Press, New York
Cerebral Oxygen Metabolism and Blood Flow in Human Cerebral Ischemic Infarction
O. L. Lenzi, R. S. J. Frackowiak, and T. Jones qf Medicine (NeuroloRY), Royal Postgraduate Medical School, Hammersmith Hospital, London, EnRland
Medical Research Council Cyclotron Unit and Department
Summary: Fifteen patients with acute cerebral hemispheric infarcts have been studied with positron emission tomography and the oxygen-I5 steady-state inhalation technique. Thirteen follow-up studies were also performed. The values of cerebral oxygen metabolism (CMR02), cerebral blood flow (CBF), and oxygen extraction ratio (OER) have been calculated for the infarcted re gions, their borders, the symmetrical regions in contralateral cerebral hemi spheres, and the cerebellar hemispheres. This study demonstrates that in the completed stroke there are thresholds for regional CMR02 and regional CBF below which the general clinical outcome of the patients is usually poor. The ischaemic lesions invariably produce an uncoupling between the greatly de creased metabolic demand and the less affected blood supply, with very fre quent instances of relative hyperperfusion. Remote effects of the hemispheric infarcts have been demonstrated, such as crossed cerebellar diaschisis and contralateral transhemispheric depression. The level of consciousness corre lates with oxygen uptake and blood flow both in the posterior fossa and in the contralateral cerebral hemispheres. The follow-up studies of individual patients underline the high variability of metabolism-to-flow balance during the acute phase of the illness, and stress the need for more studies focused on repeated assessments of homogeneous patient populations. Key Words: Cerebral blood flow-Cerebral metabolism-Positron emission tomography-Stroke.
Clinicians observe the end result, that is, neurological symptoms and signs; however, the steps leading to this final clinical picture are largely unknown. Experimental research on animal stroke models and the clinical application of techniques for the assessment of regional cerebral perfusion have made little improvement to our knowledge of these events, nor suggested any new therapeutic rationale. The recent introduction of positron emission to mography (PET) has offered a unique potential for measuring cerebral blood flow and metabolism in humans, in vivo and noninvasively. The hypothesis is that a better understanding of the impairment of the balance between cerebral metabolism and blood flow in cerebral ischemia should offer new ideas concerning treatment and allow its objective monitoring. Hence, institutes with PET facilities
The physiology of the nervous system is strictly dependent on a close relationship between perfu sion, metabolism, and local function. Any impair ment of one of these variables induces a chain of events that leads to the derangement of the others and to the further secondary aggravation of the ini tial change. In stroke, the aetiopathological event is a temporary or permanent reduction in perfusion beyond the capacity of the cerebral tissue to com pensate and maintain adequate metabolism, and, therefore, function. Addre s s corre spondence and reprint requests to Dr. Lenzi at Department of Neurology, IlIa Cattedra, University of Rome, Viale dell' Universita' 30, 00184 Rome, Italy. Abbreviations used: PET, Positron emission tomography;
rCBF, regional cerebral blood flow; rCMR02, regional cerebral oxygen metabolism; rO E R, regional oxygen extraction.
321
322
G. L. LENZI ET AL.
have begun to study and report on these patients (Ackermann et aI., 1978; Baron et aI., 1978; Kuhl et aI., 1980; Baron et aI., 1981; Lenzi et aI., 1981). This paper presents the first quantitative report of rCMR02 and rCBF in patients with stroke, studied in the acute phase of the disease with PET and with the steady-state oxygen-15 inhalation technique (Jones et aI., 1976; Frackowiak et aI., 1980). MATERIALS
Patients with acute, completed, ischemic cere bral infarcts were studied prospectively. Fifteen patients were successfully investigated at intervals from 1 to 34 days after the onset of symptoms. A second follow-up measurement was made in four patients. Three patients were studied three times and one patient on four occasions. A total of 30 measurements were performed, of which 28 are reported in this paper. Two studies were discarded from the analysis because the physiological defect corresponding to a small cap sular infarct was unresolved by the positron tomo graph. Fifteen studies were omitted from this series for one or more of the following reasons: (a) patient movement during the measurement of the tracer's cerebral distribution-this invalidates quantifica tion; (2) morphological evidence of infarction in the
posterior fossa; (3) initial measurement performed more than 5 weeks after the ictus; and (4) predomi nantly haemorrhagic infarction shown by x-ray CT scan. These restrictions were imposed a priori in order to limit the study to as reasonably homogene ous a pathology as practically possible. Twelve infarcts were left-sided and three were right-sided. Ten patients suffered a major, three a moderate, and two a mild functional deficit. Ten patients were alert at their first examination, four stuporous, and one comatose. Six patients had pre viously documented hypertension and three pa tients had previous presumed cerebral ischemic episodes (e.g., reversible ischemic attacks, as de fined by Loeb (1978». One patient was diabetic. Six patients had known ischemic heart disease. The age range was from 27 to 88 years (mean 62.5 ± 16.1). Eleven were male and four female. The evolution of signs and symptoms was fol lowed by clinical examination at presentation and then again within 24 h of positron scanning. The clinical picture at scanning and at follow-up for each patient is described in Table 1. The heterogeneity of the patient population in terms of age at onset, risk factors, presumed antecedent physiological history, and site and extent of infarction, precluded any but the most general statement concerning clinical out come. All the focal functional defects were com-
TABLE 1. S ummary of the patient population
Patient
Sex
Side and
Days from onset of
Level of
Age
vascular
symptoms to
conscious-
Clinical
(yr)
territory
first scana
ness"
e volutiond
Severity"
1
F
87
R, M CA - PC A
19
L
A
Good
2
72 68
L, MCA R, MCA
4
3
M M
4
S S
C S
Poor Poor
4
M
63
L, M C A
21
L
A
Good
5
M
61
L, MCA
2
S
S
Good
6
M
88
L, M C A
6
S
A
Poor
7
M
68
L, M C A
1
S
A
Poor
8
M
64
L, MCA
4
S
S
Good
9
F
66
L , MCA
8
S
A
Good
10
F
36
L, M C A
10
S
A
Poor
11
M
69
L, M C A
21
L
A
Good
12
M
53
R, MCA
26
S
A
Poor
13
M
51
L , MCA
4
S
A
Poor
14
F
27
L, IC
34
M
A
Good
15
M
64
L, M C A
32
M
S
Good
a Days from onset of symptoms: time of the PET scan in relation to the occurrence of the stroke. " Severity of neurological impairment: L, light; M, moderate; S, severe. C Level of consciousness at the time of the first PET scan: A, alert; S, stuporose; C , comatose. d Clinical e volution of the illness: goo d , improvement of the general clinical condition; poor, static or deteriorating general clinical
condition. M C A , middle cerebral artery; P C A , posterior cerebral artery; I C , internal carotid.
J Cereb Blood Flow Metabol, Vol. 2, No. 3, 1982
323
CMR02 AND CBF IN STROKE
plete, and a moderate improvement in aphasia or power in the limbs was the maximal focal functional change observed. Patients were, however, graded into one group that showed a significant and marked improvement in general physical status between presentation and scanning, and a second group that remained static or declined. These two groups are named "good-" and "poor-outcome," respec tively. The criteria used in this assessment included the level of consciousness, the severity of the neurological deficit, and the final degree of inde pendence achieved following the ictus. This clas sification was used to assess the remote or generalized clinical effects of a focal cerebral in farct. The control group consisted of 14 normal volun teers, aged 49-74 years (mean age, 61 ± 8), exam ined in our Center as part of a study of the normal population (Frackowiak et aI., unpublished obser vations). METHODS
Quantification of regional cerebral oxygen me tabolism (rCMR02), regional oxygen extraction (rOER), and regional cerebral blood flow (rCBF) with PET and with steady-state oxygen-15 inhala tion technique has been fully described for normal humans in our previous publications (Frackowiak et aI., 1980; Lammertsma et aI., 1981). The extension of the technique to stroke patients is neither simple nor linear, a problem shared by other approaches using tracer techniques, e.g., the F-18deoxyglucose model (Kuhl et aI., 1980). In addition to the prob lems arising from the extension of the model from healthy tissue to infarcted brain, the methods of data analysis and presentation potentially affect the results and the conclusions drawn. Ischemic infarcts are not uniform in size or site. They affect patients of different ages and are lo cated in different cerebral regions with different grey and white matter composition. Therefore, quantitative values per se do not necessarily linearly reflect the clinical severity of lesions. On the other hand, widespread and/or remote effects of regional lesions, which are often present, do not permit a reliable qualitative analysis of data on the basis of side-to-side comparisons. We have, there fore, chosen to analyze our data in a rigidly defined manner, being well aware that such a decision might decrease the potential sensitivity of the method but
would also minimize the subjective manipulation of data. The quantitative data have been obtained with the same procedures utilized in our previous work on normals (Frackowiak et aI., unpublished observa tions) and on patients with dementia (Frackowiak et aI., 1981), i.e., using cerebral regions of interest of 9 x 9 pixels (2.25 x 2.25 cm) that, therefore, in evitably contain mixtures of cortical grey and un derlying white matter. These regions of interest were located in the center and at the anterior and posterior borders of the infarct. Strictly homologous regions of interest were located in the contralateral, unaffected hemisphere. A similarly sized region of interest was selected from that part of the contra lateral hemisphere with maximal values for rCMR02 and rCBF. Finally, a region of interest was located in each cerebellar hemisphere. Thus, for each patient, nine regions of interest were analysed in which the mean values and stan dard deviations for rCMR02, rOER, and rCBF were calculated: (1) infarct; (2) contralateral mirror locus; (3-4) borders of the infarct; (5-6) contralateral mirror loci; (7) peak; (8-9) cerebellar hemispheres. Figure 1 illustrates this analysis. In cases in which there was a reduction of rCMR02 and/or rCBF val ues by 80% or more from the normal in the infarcted region, the rOER values become statistically un certain due to the low levels of activity resulting in poor count rates. Such values were, therefore, not incorporated in the group analysis (2 cases). The data are presented as follows: (1) quantitative val ues (mean and standard deviation) from the indi vidual regions of interest, and (2) relative values, obtained by comparing mirror regions from the two hemispheres. RESULTS Normal Distribution
In normal subjects, the cerebral activity distribu tions for rCMR02 and rCBF were the same and there was no significant difference between left and right hemispheres for each variable. Figure 2 shows a representative PET scan from our series of 14 normal, "elderly" volunteers aged 49-74 years (mean age 61 ± 8) (Frackowiak et aI., unpublished observations). Long-term reproducibility was studied in normals (2-10 months apart). Variability was up to ± 12% for rCMR02 and to ± 20% for rCBF. The mean re-
J Cereb Blood Flow Metabol, Vol. 2, No. 3, 1982
324
G. L. LENZI ET AL.
FIG.1. Method of analysing quantitative data using standard 5-cm2 regions of interest sited in relation to the focal infarc tion as described in the text. Nine regions were used for analysis: the infarct and the contralateral hemispheric mirror locus, anterior and posterior borders of the infarct and the contralateral mirror loci, the region with highest metabolic activity in the unaffected hemisphere, and two regions from the two cerebellar hemispheres.
gional values and standard deviations in these el derly normals was 3.75 ± 0.60 ml O2/100 mllmin for rCMR02; 39.3 ± 5.8 mlilOO mllmin for rCBF; and 0.56 ± 0.06 for rOER (cerebral blood volume uncor rected). In the 12 stroke patients, 3 weeks after infarction, the mean regional values in the contralateral hemi sphere was 3.45 ± 0.85 ml O2/100 mllmin for rCMR02; 31.6 ± 5.81 mlilOO mllmin for rCBF; and 0.56 ± 0.10 for rOER, thus indicating a restitution of energy balance, but at slightly lower mean oxy gen consumption (-9%). rCMR02 in the Core of the Infarct
Ischemic infarction leading to a completed stroke results in a decrease of rCMR02 to values ranging from 0.5 to 2.5 ml O2/100 mllmin. In only one case
J Cereb Blood Flow Me/abol, Vol. 2, No.3, 1982
was rCMR02 found to be in the normal range. No patient in the group studied showed full focal func tional recovery. Figure 3A shows the quantitative rCMR02 values for the infarcted region plotted against the interval from the onset of symptoms. Figure 3B shows the rCMR02 changes relative to the rCMR02 in the contralateral mirror focus for the same patients. This qualitative analysis is shown in an attempt to compensate for the different premor bid metabolic activities in the different patients. However, this type of comparative data is affected by global and/or remote changes secondary to the stroke, clinically reflected, for example, by the level of consciousness. The data suggest that in the early phase (first 2 weeks) after the infarction, patients with a good final clinical outcome (as defined above) had a rCMR02 in the infarcted region that never fell below 1.25 ml O2/100 mllmin, or below 40% of the contralateral value. Some infarcted regions, when the size of the infarct was small, showed later meta bolic recovery. In the "stabilized" phase, that is, beyond 2 weeks following infarction, no rCMR02 separation between good- and poor-outcome groups is shown. This is paralleled by the lack of functional recovery observed clinically and reflects the degree of local brain death in the infarcted area. rCBF in the Core of the Infarct
The pattern of rCBF change is less predictable than that for rCMR02 due to instances of both ab solute increases of rCBF in some patients (absolute "luxury perfusion" ) and to increases relative to metabolic demand in others (relative "luxury perfu sion"). These phenomena were observed mainly between the 6th and 21st day after infarction. In the early phase, an absolute decrease of rCBF below 20 mlllOO milmin seems to correlate with poor clinical outcome. All the patients with a rCBF above 20 mlllOO mllmin showed some clinical im provement, except in one case with very high ab solute lUxury perfusion in which later further de terioration was observed, and one other case in which at follow-up, rCBF recovered to a nearly normal value. In the cases where no absolute luxury perfusion was seen within the lesion, the average rCBF in the core of the infarct was 47% of normal, a less pro nounced depression than that of rCMR02 (36% of normal). Figure 4A shows the absolute values of
325
CMR02 AND CBF IN STROKE
FIG. 2. Functional images of C MRO" CBF, and OER from a normal elderly volunteer. Ab solute values were as follows for CMRO, (ml 0,1100 ml/min) and for CBF (ml/100 ml/min), respectively: frontal cortex, 3.49 and 36.4; parietal cortex, 3.11 and 35.1; insula, 3.66 and 40.5; occipital cortex, 4.58 and 42.5; thalamus, 3.94 and 41.5; centrum semiovale, 1.90 and 22.7.
FIG. 3. rCMRO, in the core of the in farct. Absolute (A) and relative (8) rCMRO, values obtained with a 9 x 9-pixel region of interest placed in the grey matter at the center of the in farcted area. Relative rCMRO, values
A
are percent of the value in the symmet rical regions of interest in the con tralateral cerebral hemispheres. The
B
o
0000 OUTCOME
•
BAD
OUTCOME
time of the PET scan in relation to the onset of the neurological symp tomatology is on the x-axis. Patients have been divided, according to their clinical evolution into "good" (open circles) and "poor" (filled circles) out
60
come. The shaded area in A indicates the mean rCMRO, for the normal el derly population ± 1 SO. In B, the lines interconnecting different studies indi cate individual patients who had follow-up PET scans. From A it follows that in the first days after a stroke, 6 of 9 patients with a poor clinical outcome had an rCMRO, below 1.25 ml 0,/100 ml/min. In the stabilized phase, a re lationship between rCMRO, and clini cal outcome is not identifiable. B shows that all the patients with a poor final outcome have an rCMRO, that does not attain 50% of the value in the
'·0
i
'0
20
DAYS
30
I
'0
20
DAYS
30
i
40
I
contralateral mirror locus. It follows that the absolute depression o f rCMRO, i n the core o f the lesion bears a relationship to the general outcome of the illness.
J Cereb Blood Flow Metabol, Vol. 2, No. 3, 1982
326
G. L. LENZI ET AL.
B
A o
GOOD OUTCOME BAD
OUTCOME
eo
%150 10
FIG.4. rCBF in the core of the infarct.
(A) absolute and (B) relative rCBF val ues were obtained and plotted as in
Z E
Fig. 3. The figure indicates that in the early phase after a stroke the rCBF
60
VI --ISO E C> C> � �
falls in the majority of cases, with a threshold at 20 ml/100 ml/min that separates the "good outcome" from the "poor outcome" patients. This fig ure underlines the occurrence of in stances of absolute lUxury perfusion.
100
VI --I E lO
50
10
20
DAYS
rCBF in the infarcts. It indicates that rCBF remains well below normal both acutely and in the stabilized phase of the disease. Figure 4B illustrates the changes in rCBF relative to the contralateral side. This figure underlines the larger variability of rCBF. Comparing Fig. 4B and Fig. 3B, it appears that in the first few days after infarction, rCBF may be decreased (5 of 9 cases), normal (2 cases), or in creased (2 cases), whereas in all these cases rCMR02 was decreased.
20
DAYS
30
rOER in the Core of the Infarct
The rOER in the early phase invariably indicates uncoupling between flow and oxygen metabolism, that is, between energy supply and demand. The four cases with rOER values near to or within the normal range had similar decreases of both rCBF and rCMR02• In the stabilized phase, rOER tends to return to ward normal values, but remains, in general, lower
A o
10
30
B
GOOD OUTCOME BAD
OUTCOME
% 150
0·1
::�0
FIG. 5. rOER in the core of the infarct.
(A) absolute and (B)
0·3 0·2 0·1
10
20 DAYS
i
30
I
J Cereb Blood Flow Metabol, Vol. 2, No.3, 1982
relative rOER val
ues were obtained and plotted as in Figs. 3 and 4. The figure indicates the high frequency of uncoupling between rCMRO, and rCBF after stroke.
0·4
10
20
DAYS
30
,
I
327
CMR02 AND CBF IN STROKE
in patients with a poor final clinical outcome and a larger area of infarction. The absolute values of rOER are shown in Fig. SA, the relative values in Fig. SB. This figure indi cates two major trends during the early phase: either a profound depression, to values below SO% of normal, or an increase with respect to the con tralateral side. "Critical perfusion," that is, an in crease in rOER, was detected in one case. Luxury perfusion, defined by a decreased rOER, was the most common pattern of uncoupling seen in these patients (S out of 9) in the early phase. The occur rence and severity of uncoupling is more frequent in the earliest post-ictal than in later studies. rCMR02, rCBF and rOER in the Borders of Infarct
In general, the pathophysiology of the tissue bor dering the infarct was similar to that observed in the infarct itself. The depression of rCMR02 and rCBF was, however, smaller. There was no correlation between rCMR02, rCBF, and rOER values and the general clinical outcome. While rCMR02 values were invariably decreased, rCBF values showed in creases in absolute, and, more frequently, in rela tive terms. Uncoupling of energy supply and de mand was present in the first 2 weeks in 11 of 13 studies, with critical perfusion in 2, and luxury perfusion in 9 cases. In the stabilized phase, rOER values were found to lie close to or within normal range, both in absolute and relative terms. Figure 6 shows values for rCMR02, rCBF, and rOER, absolute (A, C, and E) and relative (B, D, and F), all plotted against the time interval from the onset of symptoms. Remote Effects of Cerebral Hemispheric Infarction
Crossed Cerebellar Diaschisis
Cerebral hemispheric infarcts produced a parallel decrease in rCMR02 and rCBF in the cerebellar hemispheres contralateral to the side of the lesion. Comparing the affected side to the contralateral, the depression was 16 ± 22% for rCMR02 and 19 ± 27% for rCBF in the early phase of the infarction. The depression increased in the stabilized phase, reaching 2S ± 20% for rCMR02 and 27 ± 23% for rCBF. The side-to-side variability in normal volun teers never exceeded ± IS% for the cerebellar hemi-
spheres. There was no correlation between the de gree of crossed cerebellar diaschisis and the general clinical outcome of the patients. The crossed cere bellar diaschisis was not evident in cases in which the dimensions of the infarct were small. In three cases, there was a small depression of rCMR02 and rCBF in the ipsilateral cerebellar hemisphere. Two of these patients had had previous infarcts in the other cerebral hemisphere and the third had an infarct confined to the anterior tem poral lobe. When the infarct predominantly in volved the parietal lobe, the crossed cerebellar diaschisis was SO% greater than in cases where the lesion affected anterior regions only. Figure 7 shows the changes in cerebellar rCMR02 and rCBF as percent ipsilateral to contralateral from the cere bral infarct. Posterior Fossa Effects
In the posterior fossa, a correlation between level of consciousness at scanning, metabolic activity, and flow was observed. Patients with impaired con sciousness at the time of the study had lower mean values for rCMROz and rCBF (Table 2) than did patients with normal consciousness and normal volunteers. Due to parallel decreases in rCMR02 and rCBF, the rOER remained normal. In patients with preserved consciousness at scanning, rOER was significantly lower than in the normal group. These data may reflect the decreased functional needs of the comatose-stuporose patients. Since rOER was within normal limits, they do not support a primary ischemic mechanism due to mechanical (increased intracranial pressure) obstruction for the observed metabolic depression. Transhemispheric Depression (Diffuse Diaschisis)
Patients with impaired consciousness showed lower mean rCMR02 and rCBF in the cerebral hemisphere contralateral to the infarct than did pa tients who were alert at the time of the study. Pa tients with unimpaired consciousness showed a modest reduction in rCMR02, while rCBF remained within normal limits (Table 3). Thus, for these pa tients, the rOER decreased, signalling a relative luxury perfusion situation presumably due to the decreased metabolic needs. That the effect is generalized is indicated by com paring the "peak" regions of interest in the early and in the stabilized phases. In the early phase (14
J Cereb Blood FloII' Metabol, Vol. 2, No.3, 1982
328
G. L. LENZI ET AL. B
A o
GOOO OU1t:OHE
•
8AO
% 150
OUTCOME
6-0 � z
E
III
-'
E 0 0
3-0
..
50
�1O -'
E
1-0
i 10
i 20
30
0
GOOO OUTCOME
.
BAD
DAYS
.
IIJ
I
10
20
i
IIJ
30
DAYS
(
I
0 OUTCOME
7.0
FIG. 6. rCMRO" rCBF, and rOER in the borders of the infarct. The values have been obtained from 9 x 9-pixel
bO
regions of interest located at the borders between the infarcted and the
% 150
normal tissue. The graphs
z
i:
III
-'
E 0 0
.0
III
30
�
-' E
10
� 0
. .
.
0 .
\,
0---'""""
early phase after the stroke when compared with the values in the core
'0
of the infarct. In the stabilized phase, there is a re-coupling at lower absolute levels for both rCMRO, and rCBF. i
20 DAY S
10
30
10
IIJ
20
E 0
.
lO
40
30
IIJ
DAYS F
GOOD OUTCOME BAD
OUTCOME
0/015 0.7
11-5
and E
hyperperfusion, and a conse quent high occurrence of uncoupling (increased or decreased rOER) in the
50
0 0 0
.
A, C,
(left) are absolute values; the graphs B, 0, and F (right) are relative values. The figure shows a lesser impairment of rCMRO" an increased frequency of
100
50
�
100
0-4 0-3 50 11-2 IH
10
20
30
I
40
I
DAYS
J Cereb Blood Flow Metabol. Vol. 2. No.3. 1982
I
10
20
DAYS
I 50
329
CMR02 AND CBF IN STROKE
o
GOOD OUTCOME BAD
OUTCOME
0/0 160
FIG. 7. Crossed cerebellar diaschisis. Comparison of absolute values for rCMRO, and rCBF between the cere bellar hemispheres ipsilateral to the infarct and the contralateral symmet rical regions of interest. The shaded zone of the graphs indicates the side-to-side variability for the normal
140
�
r..o 120 0 cr :E 100
BO
elderly population. The phenomenon of crossed cerebellar diaschisis was
60
present in
14
of 25 cases.
aT 10
20
30
i
I
10
, 40
20
I
DAYS
DAYS
studies), peak rCMROz was 3.96 ± 1.21 ml 02/100 ml/min, peak rCBF was 43.4 ± 25.2 ml/l00 ml/min, and peak rOER was 0.50 ± 0.12. These three vari ables in the stabilized phase (15 studies) were, re spectively, 5.02 ± 1.14, 43.6 ± 10.6, and 0.59 ± 0.08. These data indicate that the main effect of the acute infarct is on oxygen uptake, with an uncou pling from flow that results in a relative over perfusion of the brain. After 21 days, the tendency is for re-coupling of metabolism to flow. Further more, only for rCMR02 is the difference from early to stabilized phase statistically significant (p < 0.05). Mirror Diaschisis
In addition to this generalized effect over the contralateral hemisphere, regional variations ap peared. In particular, a decrease in oxygen metabo lism and blood flow was detected in the "mirror to
the infarct" regions of the contralateral hemisphere, in which no morphological abnormalities were seen. This was present in the first 2 weeks (5 of 14 cases) but not later (1 of 15 cases). Both rCMR02 and rCBF were 70% or less than the values in the adja cent "border" regions. In the early phase after in farction, this focal depression was often superim posed on the previously described generalized de pression of oxygen consumption and flow. Longitudinal Studies
Due to its essential noninvasiveness, the steady state oxygen- IS inhalation technique is suitable for repeated studies of patients during the course of disease. Follow-up studies were performed in 9 of the 15 patients in the present series, and they are identified by data-linking in Figs. 3-7B. These follow-up studies show the natural history of the mismatching between flow and metabolism in indi-
TABLE 2. Mean rCMR02, rCBF, and rOER absolute v alues and s tandard dev iations in the posterior fos s a (ipsilateral cerebellar hemis pheres ) in the normal elderly population, in patients without impairment of consciousness (alert) at the time of PET s can, and in patients with impairment of consciousness (stupor-coma)
Level of Group studied
N
Value
Elderly normal
rOER
CBP
r C M R02
significance
Level of Value
Level of
significance
Value
significance
14
3.S0 ± 0.50
Nsa
39.6 ± 5.0
Nsa
0.56 ± O.OS
O.loa
Alert
9
4.21 ± 1.05
0.10"
45.0 ± 21.0
0.0005"
0.50 ± 0.13
NS"
Stupor-coma
5
3.43 ± 0.32
O.loe
30.9 ± 4.9
0.10"
0.5S ± 0.10
NSc
a Level of significance (one-tailed t test) for "elderly normal" group co mpared to "alert" group. o Level of significance (one-tailed t test) for "elderly normal" group compared to "stupor - coma" group.
e Level of significance (one-tailed t test) for "alert" group compared to "stupor-coma" group.
J Cereb Blood Flow Metabol, Vol. 2. No.3, 1982
330
G. L. LENZI ET AL.
TABLE 3. Mean rCMR02, rCBF, and rOER absolute v alues and s tandard dev iations in the three s elected regions of interes t in the cerebral hemispheres contralateral to the infarct
N
Value
Elderly normal
significance
Level o f
Level o f
Level o f Group studied
rOER
rCBF
r C M R02
Value
significance
Value
significance
Nsa
14
3.80 ± 0.50
O.Ola
Alert
9
3.13 ± 1.01
0.0005"
39.6 ± 5.0 39.7 ± 16.8
0.0005b
0.55 ± 0.08 0.44 ± 0.07
0.005a NSb
Stupor�coma
5
2.43 ± 0.44
0.10"
23.9 ± 6.10
0.025'
0.54 ± 0.09
0.025"
a Level of significance (one-tailed t test) for "elderly normal" group co mpared to "alert" group. b Level of s ignificance (one-tailed t test) for "elderly normal" group compared to "stupor�coma" group. " Level of s ignificance (one-tailed t test) for "alert" group compared to "stupor�coma" group.
vidual patients. However, general trends were not evident when a statistical analysis of the data was performed. The detailed results of two individual cases are presented. They serve to illustrate the two opposite patterns of uncoupling seen in ischemic cerebral infarction. Figure 8 shows three studies performed on pa tient no. 8. (See Table 1 for clinical details.) There is uncoupling between rCBF and rCMR02 in both the infarct and its borders, with absolute luxury perfu sion that lasted at least 10 days in the core of the lesion. Figure 9 shows three studies performed on pa tient no. 7. (See Table 1 for clinical details.) In this patient, the uncoupling between rCBF and rCMR02 was opposite to that in the previous case, with an
increase in rOER in the lesion and its borders during the early phase. By the 29th day after infarction, a reversal had occurred with relative luxury perfusion in the core of the lesion. Figure 10 shows, in these two cases, how rCMR02, rCBF, and rOER varied with time in the center of the lesion (top) and in its borders (bottom). The data are expressed in a qualitative way by com paring the infarcted regions to the contralateral, "mirror" foci. In both cases, the final pattern was a relative luxury perfusion within the infarct, and re coupling in its borders. DISCUSSION
This paper presents the first quantitative regional study of CMR02, CBF, and OER in patients with
FIG. 8. Functional images from the OM +
4
plane in patient no. 8. The
pattern of mismatch found in this patient should be compared to that in Fig. 9. The study demonstrates the defect in metabolism in the left parietal region, but in this case the oxygen extraction is elevated, im plying critical perfusion in the center of the infarct. With time, this pattern gives way to one of lUXUry perfusion, presumably a reflection of the much smaller oxygen de mand of the established infarct.
J Cereb Blood Flow Metabol, Vol. 2, No.3. 1982
CMR02 AND CBF IN STROKE
331
FIG. 9. Functional images from the O M + 4 plane i n patient no. 7 , who shows a metabolic defect in the left parietal re gion. The perfusion in this area is high and shows the rOER pattern of luxury perfusion. The levels of flow at day 8 are still higher than that through the normal hemisphere; this represents late abso lute luxury perfusion.
cerebral ischemia leading to completed infarction. Therefore, general aspects of the study require dis cussion before proceeding to the results and their implications. Three major points need critical examination: (a) the extension of the steady-state oxygen- I5 inhalation technique to the study of ab normal cerebral tissue; (b) the method of data analysis; and (c) the characteristics of the patient population studied. We refer to our previous papers for a detailed exposition of the study of cerebral physiology in normal humans by the oxygen-I5 inhalation tech nique and PET (Frackowiak et ai., 1980; Lam mertsma et ai., 1981). As far as the extension of this technique to diseased tissue is concerned, we share the reservations of Kuhl et ai. (1980), namely, that "interpreting the quantitative relationships" of per centage differences existing between values of per fusion and glucose metabolism "for focal lesions will remain problematic until there is more certainty of the trapping conditions for" the utilized tracer "in all states of diseased tissue." However, the oxygen- I5 technique may, in some respects, be considered more reliable than the 18F_ deoxyglucose technique for the study of pathologi cal cerebral tissue. Molecular oxygen, at the sub cellular level, has only one metabolic pathway: coupling with H+ ions produced in the cytochrome
system to form "metabolic" water. Being freely diffusible, oxygen is not affected by alterations of cellular membrane permeability or specific trans port mechanisms. Thus, a decrease in rCMR02 re vealed by this technique indicates a true decrease in cytochrome function, which represents a funda mental step in oxidative metabolism-the only energetic pathway for neurons under normal condi tions. Obviously, if neuronal metabolism in particular conditions utilizes pathways or substrates that do not reach the final cytochrome oxygen-hydrogen coupling, the oxygen- I5 inhalation approach will show a decreased CMR02 but the neuronal function may be unimpaired. However, it seems logical to assume that any impairment of the cytochrome system reflects a true impairment of the correct metabolic function of the cerebral tissue. Lesser reservations are needed when considering that part of the oxygen-15 technique concerned with the quantification of rCBF. Labeled water may be considered as freely diffusible (Raichle et aI., 1974) in healthy and ischemic tissue for the values of flow observed in humans, in contrast to data obtained in the rhesus monkey (Eichling et ai., 1974). Our group (Lammertsma et aI., 1981) has shown that rCBF measurements in the grey matter are within 6% of their actual values.
J C"reb Blood Flow Metabol, Vol. 2, No.3, 1982
332
C. L. LENZI ET AL.
B
A .. _-
o
+40
-CMRO
• -
CBF
'" =OER ,
,
,
-40
- 80
\
"
•
�
x--
�//" ' �
-
�-
- ....
-��
"
--0
,,/
"
+40
-40
'",
'-
--,
0-
/:y--
-80
DAYS
20
DAYS
I 20
FIG. 10. Follow-up studies of patients no. 8 (A) and no. 7 (B). The rCMR02 (open circles), rCBF (filled circles), and rOER (triangles) values in the infarct (top) and in the borders of the infarct (bottom) are presented as percent of the values in the contralateral symmetrical regions of interest. Note the luxury perfusion in A and the critical perfusion in B. Both these phenomena were maximal in the early phase and vanished at the third follow-up study (4 weeks after the stroke).
The procedure used for the regional quantitative analysis of the data is similar to that described in our recent report on dementia (Frackowiak et al., 1981). The use of large regions of interest, and ac ceptance of the limitations on the absolute nature of the quantitative data that this imposes, is more than offset by increasing the unbiased nature of the mea surements and allowing for objective assessment of regional information. The use of a rigid system of siting regions of interest imposed other limitations. The definition and siting of the infarcted and border regions of interest was guided by conventional CT scanning, but a certain amount of sUbjectivity was unavoid able. The border regions of interest include a mix ture of healthy and ischemic tissue, since they com prise ill-defined regions of transition between dam aged and normal tissue. No attempt was made to site the infarct and border regions of interest ac cording to cerebral arterial territory.
J Cereb Blood Flow Metabol, Vol. 2. No.3, 1982
Besides the limitations imposed by their defini tion, both anatomical and physiopathological, the border regions of interest hold great interest with regard to the final clinical outcome, because these areas may contain regions in which the damage is still potentially reversible and in which, therefore, therapy could be effective. The quantitative values have been presented together with relative values comparing homologous contralateral regions. Both presentations contain intrinsic limitations. Quan titative values do not take into account the age of the patient, the previous condition of the brain, or the size, the position, and the grey-to-white matter composition of the infarcted region. On the other hand, relative values do not take into account dis tant effects of local lesions, and assume a compari son between diseased and normal tissue that is not necessarily true. In addition, they do not allow in terpretation of follow-up studies in the meaningful way that quantification allows. The observations of the present study are further limited by the inhomogeneity of the patient popula tion. The present patient series is homogeneous only in that all patients suffered an acute ischemic cerebral infarct resulting in a completed stroke. Age, aetiopathogenesis, risk factors, and previous illnesses were widely heterogeneous. We must ac cept that cerebral infarction is an exceedingly di verse disease, and analysis of the pathophysiology implies that great care must be taken in defining appropriate homogeneous patient populations to answer specific questions of the natural history, aetiopathogenesis, or effects of therapy. Keeping in mind these reservations and limitations, it is possi ble to discuss the results and the findings of the present study. In the infarcted regions, thresholds for rCMR02 and rCBF were identified below which the general clinical outcome of the patient (as defined) was poor. The correlation between low rCBF values and poor outcome has been described by many authors (Fieschi et aI., 1968; Wong et aI., 1973; Heiss, 1979). However, rCMR02 is a more reliable index of the clinical outcome. An rCMR02 decrease below 40% of the values in the contralateral hemi sphere invariably correlated with a poor outcome. The disturbance of the relationship between CBF and metabolism in the infarcted region was a con stant finding. Our data agree with the reports of Baron et aI. (1978) and of Ackerman et aI. (1981) on the high frequency of luxury perfusion, i.e., de creased OER, and on the low occurrence of critical
CMR02 AND CBF IN STROKE
perfusion, i.e., increased OER. Our subsequent ob servations strongly indicate that this is a time dependant phenomenon, the proportion of "critical perfusion" instances markedly increasing the closer to the ictus the PET study is performed (Wise et aI., unpublished observations). We emphasize that rOER values are reliable only if the reduction in rCBF and rCMR02 is not too profound. When the number of counts per pixel is less than approximately 18% of normal, the rOER value is greatly affected by statistical variations. In our series, we have rejected two rOER values be cause of this uncertainty. On the whole, the data from the infarcted area indicate no recovery of oxy gen metabolism if, 48 h after the onset of symptoms, the rCMR02 values are below 60% of their con tralateral mirror values. In the borders of the ischemic lesion, the changes in rCMR02 are similar to those observed in the core of the lesion, the impairment being of a lesser de gree. As previously found by Baron et al. (1980a), in no case was a hypermetabolic rim seen. This may complement the report (using the 18F-deoxyglucose technique) by Ginsberg et al. (1977) of a ring of glucose hypermetabolism that is probably reflecting increased anaerobic glycolysis. Increases in CBF were frequent, particularly in relation to the meta bolic demand, thus signaling uncoupling of aerobic glycolysis. Ischemic lesions affecting one cerebral hemi sphere are shown to produce generalized and re mote effects on other structures of the brain, e.g., crossed cerebellar diaschisis and transhemispheric depression, both diffuse and focal. The phenome non of crossed cerebellar atrophy was described by neurologists of the last century (see Dow and Moruzzi, 1958). Recently, Baron et al. (1980h) have described "a decrease in blood flow without change in OEF (OER), i.e., a matched decrease in CBF and oxygen metabolism in the cerebellar hemisphere contralateral to the supratentorial infarct" . These authors suggest a deactivation secondary to inter ruption of a cerebro-cerebellar loop. They did not find metabolic depression in the contralateral cere bellar hemisphere beyond 2 months after su pratentorial infarction. This is a curious finding if the crossed cerebellar atrophy is a morphological correlate of the physiological disturbance. Our data indicate that the phenomenon is observed up to the 50th day past infarction, being slightly greater in the later weeks than in the earlier weeks. In the six cases studied after the 50th day, crossed cerebellar
333
diaschisis was present and was greater than that seen at earlier stages (Lenzi et aI., 1981). The parietal cortex seems to play an important role in the development of crossed cerebellar dias chisis, in accord with the recent experimental data of Dauth et al. (1980) and with the as yet scanty knowledge of the cortico-ponto-cerebellar con nections (see Brodal, 1969). Another possible mechanism is that a relative deafferentation of the cerebellar hemisphere ipsilateral to the hemiplegic side occurs. The transhemispheric depression of rCMR02 and rCBF, as well as their depression in the posterior fossa, shows a clear-cut and easily understandable correlation with impairment of consciousness, as described by many authors (see Kety, 1956; Sokoloff, 1971). In addition, it is observed that pa tients with preserved consciousness nevertheless showed low rCMR02 values over the entire con tralateral hemisphere, with normal rCBF. In the early phase after infarction, a further local decrease of oxygen uptake and flow was detectable in nearly 50% of the cases in the contralateral locus from the infarct. These phenomena, contralateral diaschisis and mirror diaschisis, appear to reflect a primary metabolic impairment, as suggested also by the sig nificant decrease in rOER, which is a situation sim ilar to that reported in post-traumatic coma (Fieschi et aI., 1974). The definition of diaschisis rests on the lack of anatomical correlates of these physio logical patterns. They are typical of the immediate post-infarct period and do not have obvious func tional clinical correlates. The follow-up studies serve to emphasize the potentials as well as the limitations of the investiga tive technique. The measurement of cerebral oxygen consump tion and blood flow in a reproducible manner at a regional level, allowing ischemia to be quantified, represents a considerable methodological step in the study of the pathophysiology of cerebral infarc tion. The limitations arise more from the object of the study than from the utilized technique. That is, the cerebral infarcts disrupt the precise balance between energy supply and demand, and whereas a profound decrease in oxygen metabolism is the pathophysiological counterpart of late structural damage, rCBF and rOER behave less predictably and may vary widely in the same patient over short periods of time. This uniqueness of individual stroke leads to the following general remarks:
J Cereb Blood Flow Metabol, Vol, 2, No, 3, 1982
334
G. L. LENZI ET AL.
1. Generalized conclusions drawn from studies on cerebral metabolism and flow in stroke cannot be made if the patients are not homogeneous in terms of rCMR02-rCBF balance. 2. The natural history of stroke must be clarified in terms of rCMR02-rCBF balance in the very first hours as well as in the first days after the ictus. This requires a large body of well-documented, indi vidual case studies to build up the overall picture in this physiologically heterogeneous patient popula tion. 3. The interpretation of data gathered so far must be tempered by a careful consideration of uncer tainties regarding (a) the tracer behaviour in infarc ted tissues and (b) the biochemical natural history of infarcts. The protean, dynamic nature of post-ischemic pathophysiology is well-known to clinicians and has been hinted at by rCBF studies. Now PET has per mitted a direct demonstration of these changing patterns and of their evolution. We feel that PET may open the "black box" of cerebral infarction and offer the possibility of decoding the messages provided by the clinical signs and symptoms in stroke, thus suggesting more logical therapeutic rationales in this disease. Acknowledgment: Dr. G. L. Lenzi was partly sup ported by CNR Grants 78.02653.04 and 79.02161.04. Dr. R. S. J. Frackowiak was in receipt of an MRC Training Fellowship. Discussion with Dr. R. J. S. Wise is grate fully acknowledged. Thanks are offered to Mr. D. D. Vonberg and the staff of the MRC Cyclotron Unit for constant moral and technical support in the production of short-lived, positron-emitting radioisotopes. Finally, we would like to thank our clinician colleagues who made this study possible by allowing investigation of their pa
Human hemispheric infarction studied by positron emission tomography and 150 continuous inhalation technique. In: Computerized Tomogruphy (Caille JM, Salomon G , eds)
Berlin , Springer-Verlag, pp 231-237 Baron JC, Bous ser MG, Co mar D, Castaigne P (1980b) Cros sed cerebellar diaschisis in human supratentorial brain infarc tion. Ann Neurol 8:128 Baron JC, Bous ser M G , Comar D, Soussaline F, Castaigne P (1981) Noninvasive tomographic study of cerebral blood flow and oxygen metabolism in vivo: Potential s , limitations and clinical applications in cerebral ischemic disorders. Ellr NeuroI20:273-284
Brodal A (1969) Neurological Anatomy in Relation to Clinical Medicine. Oxford, Oxford University Press
Dauth G,
G i l m an
S,
Frey
K,
P e n n e y J (1980) (l4 C ) -2-
Deoxygluco se uptake in monkeys with hypotonic hemi plegia after precentral or postcentral lesions. Neurology 30:407 Dow RS, Moruzzi G (1958) The Physiology and Pathology of the Cerebellum. Minneapolis, University of Minnesota Press
Eichling JO, Raichle ME, Grubb R L Jr, Ter-Pogo s s ian MM (1974) Evidence of the li mitations of water as a freely dif
fu s ible tracer in brain of the rhe s u s monkey. Circ Res 35:358-364
Fieschi C , Agnoli A, B attistini N, Bozzao L, Prencipe M (1968) Derangement of regional cerebral blood flow and of its reg ulatory m e c h an i s m in acute cerebrova s c u l ar l e s i o n s . Neurology 18:1166- 1179
Fieschi C , B attistini N, Beduschi A, Ros selli L, Rossanda M, Pas sero S (1974) Regional cerebral blood flow and intraven tricular pressure in acute head injuries. J Neurol Neurosurg Psychiatry 37:1378-1388
Frackowiak RSJ, Lenzi G L , Jones T, Heather JD (1980) The quantitative measurement of regional cerebral blood flow and oxygen metabolism in man using oxygen-15 and posi tron emission tomography: Theory , procedure and normal values. J Comput Assist Tomogr 4:727-736 Frackowiak RSJ, Pozzilli C, Legg NJ , Du B oulay GH , Marshall J, Lenzi G L , Jones T (1981) Regional cerebral oxygen sup ply and utilisation in dementia: A clinical and physiological s t u d y with o x ygen-15 and p o s itron tomography. Brain 104:753-778
Ginsberg M D , Reivich M , Giandomenico A, Greenberg JH (1977) Local glucose utilisation in acute focal cerebral isch e m ia: Local d y s metaboli s m a n d diaschi s i s . Neurology 27:1042-1048
He i s s WD ( 1979) Relat i o n s h i p of cerebral b l o o d flow to neurological deficit and to long-term prognosis of stroke. In: 11 (Ziilen KI , Kaufmann W ,
tients.
Brain and Heart Infarct,
Hossmann KA, Hos smann V , eds), Berlin , Springer-Verlag,
REFERENCES
pp 280-292 Jones T, Chesler D A , Ter-Pogo ssian MM (1976) The continuous
Ackerman RH , Correia J A , Alpert N M , Gouliamos A, Ryder JW, Brownell G L , Taveras JM (1978) Positron imaging as a clinical tool for demonstrating di sturbed cerebral circulation and metabolism. J Comput Assist Tomogr 2:664
inhalation of oxygen-15 for assessing regional oxygen ex traction in the brain of man. Br J Radiol 49:339-343 Kety S S (1956) Human cerebral blood flow and oxygen con sumption as related to aging. Res Publ Assoc Res Nerv Ment Dis 35:31-45
Ackerman RH , Alpert N M , Correia J A , Finkelstein S , Buonanno F, Davis S M , Chang JY , Brownell GL, Taveras J M (1981)
Kuhl D E , Phelps ME , Kowell AP, Metter E J , Selin C, Winter J
Importance of monitoring metabolic function in assessing
(1980) Effects of stroke on local cerebral metabolism and
the severity of a stroke insult (CBF: An epiphenomenon?). J
perfusion: Mapping by emission computed tomography of 18FDG and 13NH3• Ann Neurol 8:47-60 Lammertsma A A , Jones T, Frackowiak RSJ , Lenzi GL (1981) A
Cereb Blood Flow Metabol I ( Suppl 1):S502-S503
Baron J C , Comar D, Bousser MG , Sous saline F, Crouzel C , Plummer D , Kellershohn C , Castaigne P (1978) Etude to
theoretical study of the steady state model for measuring
mographique, chez l'homme, du debit sanguin et de la con
regional cerebral blood flow and oxygen utilisation using
sumation d'oxygene d u cerveau par inhalation continue d'oxygene-15. Rev Neurol 134:546-556
Lenzi G L , Frackowiak R S J , Jones T, Heather J D , Lammertsma
Baron J C , B o u s ser M G , Co mar D,
Kellershohn C (1980a)
J Cereb Blood Flow Metabol, Vol. 2, No.3, 1982
oxygen-15. J Comput Assist Tomogr 5:544-550 A A , Rhodes CG , Pozzilli C (1981) CMR02 and CBF by
335
CMR02 AND CBF IN STROKE
oxygen-15 inhalation technique : Results in normal volun teers and cerebrovascular patients. ElI r Neuro l 20: 285 - 290 Lenzi G L , Frackowiak RSJ, Jones T (1981) Regional cerebral
Raichle M E , Eic hling J O , Grubb R L (1974) B rain permeability of water. Arch Neurol 30 : 319 - 321 Sokoloff L (1971) Neurophysiology and neurochemistry of coma.
blood flow ( C B F ) , oxygen utilisation ( C MRO,) and oxygen
I n : Neurochemis try of Hepatic Coma ( Polli E , ed) , Exp Bioi
e xtraction ratio (OER) in acute hemisphe ric stroke . J Cereb
Med 4 : 15 - 33 Basel , S Karger
Blo o d Flow Metabo l \ ( Suppl I ) : S504 - S505 Loeb C (1978) Clinical features and long-term follow up of pa
Wong E, B ull JWD, Du Boulay GR, Marshall J, Ross Russell
tients with reversible ischemic attacks (RIA). Acta Neuro l
pleted strokes and tran sient ischemic attacks : A clinical cor relation . Neuro logy 23: 949 - 952
S can d 57 : 47 \ - 480
R W , S y mon L (1973) Regional cerebral blood flow in com
J Cereb Blood Flow Me/abol, Vol . 2, No. 3 , 1982
