SUMMARY. To investigate further the topographical, clinical and temporal correlates of crossed cerebellar diaschisis (CCD) after supratentorial stroke, ...
Brain (1986), 109, 677-694
CROSSED CEREBELLAR DIASCHISIS FURTHER STUDIES by P . P A N T A N O , 1 * J. C. B A R O N , 1 ' 2 Y. SAMSON, 1 M. G. BOUSSER, 2 C. DEROUESNE 2 and D. COMAR 1 {From the ' Service Hospitalier Frideric Joliot, CEA, Departement de Biologie, Orsay, and 2Clinique des Maladies du Systime Nerveux, Hopital de la Salpitriire, Paris, France) SUMMARY To investigate further the topographical, clinical and temporal correlates of crossed cerebellar diaschisis (CCD) after supratentorial stroke, 55 patients suffering from a single unilateral ischaemic stroke in the carotid artery territory were studied with the quantitative oxygen-15 steady-state technique and positron tomography. Fourteen patients had one or more follow-up studies, contributing a total of 72 studies. The phenomenon of CCD, defined by depressed oxygen consumption in the contralateral cerebellum, was statistically significant in 58 % of the studies. It was more prominent when the supratentorial infarct involved the internal capsule or the cortical mantle extensively, consistent with the hypothesis that it results from destruction of the corticopontocerebellar fibres. Although CCD was associated with the presence of hemiparesis, it also occurred in patients without hemiparesis and was not seen in all those with hemiparesis, suggesting that destruction of the pyramidal tract is neither necessary nor sufficient to induce CCD. Finally, CCD tended to persist over long periods of time after a stroke, pointing towards a transneuronal degeneration possibly akin to crossed cerebellar atrophy as a likely explanation for CCD. Nevertheless, CCD could be seen within hours of a stroke and sometimes disappeared within a few days, suggesting a temporal continuum between early, potentially reversible functional hypometabolism (diaschisis) and irreversible degeneration. INTRODUCTION Quantitative mapping of cerebral blood flow and metabolism by positron emission tomography (PET) has revealed remote effects of focal cerebral lesions in man (Kuhl et al., 1980; Baron et al., 1981a; Metter et al., 1981; Phelps et al., 1981; Lenzi et al., 1982). Metabolic depression in the cerebellar hemisphere contralateral to supratentorial infarction was first described by Baron and coworkers (Baron et al., 1980a, b, 19816) using PET and the steady-state oxygen-15 technique. This phenomenon, termed 'crossed cerebellar diaschisis' (CCD), was interpreted as a functional deactivation, presumably caused by loss of excitatory afferent inputs. Among the cerebrocerebellar pathways possibly involved, the corticopontocerebellar • On leave from III Cattedra Neurologica, Roma, 00184, Italy. Correspondence to: Dr J. C. Baron, Service Hospitalier Frederic Joliot, CEA, Departement de Biologie, 91406 Orsay, France. © Oxford University Press 1986
678
P. PANTANO AND OTHERS
system was initially favoured (Baron et al., 1980a, b). The hypothesis advanced made CCD an example of transneuronal metabolic depression. Subsequent PET studies showed that CCD affects both oxygen consumption and glucose utilization (Lenzi et al., 1982; Heiss et al., 1983; Martin and Raichle, 1983; Baron et al., 1984; Kushner et al, 1984; Patronas et al., 1984). Likewise, it was observed in studies of cerebral perfusion by single-photon emission tomography (Biersack et al., 1984; Meneghetti et al., 1984; Cesaro et al., 1985). These studies, however, used a numerically limited series of patients. Correlations of CCD both with the topographic location of supratentorial damage and the clinical features, especially the occurrence of hemiparesis, remain unclear. Moreover, the temporal evolution of the phenomenon is still a matter of debate. We have therefore examined retrospectively a large series of patients with unilateral supratentorial infarction in order to analyse these in greater detail. PATIENTS AND METHODS Patients A total of 55 patients with a single supratentorial infarct in the internal carotid artery territory were studied. All subjects were selected retrospectively from a large series of patients in which regional cerebral blood flow (rCBF) and oxygen metabolism (CMRO2) had been measured with PET and the 15O2 method. Criteria for selection were as follows (1) technically reliable measurement of CBF and CMRO2; (2) satisfactory visualization of cerebellum on the lower cut of the PET study; (3) absence of clinical symptoms and/or CT scan findings suggesting ischaemic episodes in the vertebrobasilar territory; and (4) absence of gross morphological alterations in the cerebellum on routine CT scanning. There were 29 male and 26 females, aged 25 to 82 years (mean age 61 ±14). All patients were examined neurologically immediately before the PET study. The age of the infarct ranged from 2 days to 3 years. None of the patients included in this study had alterations of consciousness. The patients were divided into subgroups according to the following criteria (1) duration from onset of clinical symptoms to PET study; (2) anatomical localization and extent of cerebral infarct by CT scan; (3) severity of motor impairment, graded as 0 = no deficit, 1 = mild deficit but able to walk and use hand, 2 = able to raise arm and/or leg but hand useless, and 3 = unable to raise either arm or leg. Repeated CT scans were available in all patients. The PET study was performed once in 41 patients and twice in 12 patients. Two other patients were studied on three and four occasions, respectively. A total of 72 PET studies was therefore available. An age-matched control group of 11 subjects (mean age 57 + 7), without any history of brain disease or vascular risk factors, was selected from a series of 19 controls (age-range 19-66 yrs) in whom we had found an age-related decline in cerebellar perfusion (but not of cerebellar oxygen metabolism) that fell just short of statistical significance (r = 0.446, P < 0.06). Methods The oxygen-15 continuous inhalation method (T. Jones et al., 1976) was used. Its theoretical basis, limitations, statistical accuracy, and experimental validation, as well as rCBF and CMRO2 values obtained in normal subjects, have been published previously (Subramanyam et al., 1978; Frackowiak et al., 1980; Baron et al., 1981c; S. C. Jones et al., 1982; Lammertsma et al., 1982; Steinling and Baron, 1982; Lebrun-Grandie et al., 1983; Steinling et al., 1985). In this study, PET examinations
CROSSED CEREBELLAR DIASCHISIS
679
were performed in dim light; the patient lay supine, with eyes closed and ears unplugged. Further details have been reported earlier (Lebrun-Grandie et al., 1983). Three cerebral planes were scanned, at 2, 4 and 6 cm above and 5° rostral to the orbitomeatal (OM) line. The initial positioning and the absence of head tilt during PET scans were insured by crossed laser beams projected on ink marks drawn on the patient's forehead. Tissue radioactivity was monitored by an Ecat II (Ortec) single slice PET device, with a spatial resolution of 16 mm in the lateral and 19 mm in the axial plane (slice thickness). Correction for attenuation was carried out using a 48 Ge- 6S Ga transmission scanning performed on the same planes. C 1 ! O 2 and 1 ! O 2 were administered consecutively to the patient by continuous inhalation at tracer doses using a standard oxygen mask. End-tidal CO 2 content was monitored during the study by a CO 2 analyser (Beckman LB2). Head scanning was started only after equilibrium was reached (about 10 min). Approximately one million true coincidence events were collected for each level studied ( ~ 2-6 min scanning time per plane). During both the C 1 5 O 2 and the ' 3 O 2 scanning of the third plane, duplicate arterial blood sampling was carried out, either by direct puncture of the
FIG. 1. Standardized protocol using a video display for obtaining cerebellar data from circular regions of interest (ROIs) (area: 13.6 cm2). The original rCBF image at the level of the cerebellum (lower cut) is used (1). First, an automated isocontour program draws a line joining the picture elements having individual CBF values less than 30% of the image maximum, and the 'vertical' (anteroposterior, AP) axis is positioned (2). Then, one circular (13.6 cm2) ROI is placed tangentially to the AP axis over the cerebellar hemisphere ipsilateral to the supratentorial infarct so that its most posterior point overlaps the isocontour line (2). In the final step, the above defined ROI is moved up 5 pixels and automatically copied over the contralateral side with respect to the AP axis (3). To obtain CMROj and OEF values, the 2 rCBF ROIs were copied over the corresponding CMRO2 and OEF images.
680
P. PANTANO AND OTHERS
femoral artery or by radial artery catheterization, on which whole blood and plasma H 2 15 O and Hb I 5 O 2 concentrations, as well as PaCO 2 , PajO, pH, haematocrit and total oxygen content (Ca) were measured and averaged for further use. The reconstructed C 1 5 O 2 and 1 5 O 2 images were transformed pixel by pixel into CBF and oxygen extraction fraction (OEF) images using published equations (Frackowiak et al., 1980); the corresponding C M R 0 2 images were created according to CMRO 2 = rCBF x OEF x Ca. Data Analysis On the video display of the posterior fossa cut, one large circular region of interest (ROI) of 13.6 cm 2 in area was created over each cerebellar hemisphere according to a standardized protocol (fig. 1). Each study provided one set of cerebellar blood flow (CeBF) oxygen consumption (CeMRO 2 ) and oxygen extraction (CeOEF) data for the ipsilateral cerebellar hemisphere, and one set for the contralateral cerebellar hemisphere. From absolute CeBF and CeMRO 2 values obtained, two indexes were calculated (1) the asymmetry R —L index (AI) between right (R) and left (L) cerebellar hemispheres =——r~ x 200, expressed in percentK. -(- L age; and (2) the percentage difference between contralateral (CCH) and ipsilateral (ICH) cerebellar hemispheres (J%) = — x 100. The AI assesses the degree of cerebellar asymmetry irrespective of which side is abnormal, and was used to obtain control values and upper 95 per cent confidence limits from the normal subjects' studies. The A%, on the other hand, expresses the direction of the asymmetry (e.g., it will be negative if CCD is present), and was designed to explore the correlations between CCD and (1) time from onset, (2) infarct topography, and (3) motor impairment. The data were analysed by Student's t test and analysis of variance (Bonferroni method).
RESULTS
Neurological State, Cerebellar Blood Flow and Cerebellar Oxygen Consumption
The neurological findings are summarized in Table 1 and mean CeBF and CeMRO2 in the stroke studies and in the control group are shown in Table 2. Although CeBF and CeMRO2 of patients were depressed on both sides relative to control values, only the values contralateral to the infarct (CCH) differed significantly from controls (P < 0.05). In the stroke group, there was a significant increase {P < 0.001 by Student's t test) of the asymmetry index (AI) when compared with controls for both CeBF (13.8 ±1.1 and 4.8 ± 1.1, respectively, mean + SEM), and CeMRO2 (13.9±1.1 and 3.2±1.0, respectively). The percentage difference (A%) in CeBF and CeMRO2 was -10.4% ± 1.3 (SE) and -11.2% ±1.3 (SE), respectively. Both values are significantly different from 0 at P < 0.001. On the other hand, CeOEF did not differ significantly either between cerebellar hemispheres or between patients and controls (Table 2). The analysis of individual data showed that a significant cerebellar asymmetry (that is, individual value of the AI lying beyond the upper 95% confidence limits defined in controls) was present in 36/72 (50%) CeBF studies and in 42/72 (58%) CeMRO2 studies.
CROSSED CEREBELLAR DIASCHISIS
681
TABLE 1. NEUROLOGICAL FINDINGS
Hemiparesis Present (n = 47)
Other neurological signs None (n = 13) Aphasia (n = 14) Aphasia and homonymous hemianopia and/or hemianaesthesia (n = 10) Homonymous hemianopia and/or hemianaesthesia (n = 10)
Absent (n = 8)
Aphasia (n = 5) Aphasia and homonymous hemianopia
TABLE 2. MEAN ( ± S E M ) CEREBELLAR BLOOD FLOW (CeBF), OXYGEN METABOLISM (CeMRO 2 ), AND OXYGEN EXTRACTION FRACTION (CeOEF) IN IPSILATERAL AND CONTRALATERAL CEREBELLUM OF STROKE PATIENTS AND IN CONTROL GROUP
CeBF CeMRO2 (mil 100 mljmin) (mlO^lOO ml/min) CeOEF Mean SE Mean SE Mean SE
Group Patients Ipsilateral cerebellum Contralateral cerebellum Controls
72
46.4
2.11
4.43
0.19
0.55
0.01
72
41.6*
1.92
3.96*
0.15
0.56
0.01
11
54.1
3.38
4.99
0.32
0.56
0.02
*P < 0.05 vs controls by Bonferroni's test.
Correlation Between CCD and Duration After Infarct Linear regression analysis showed no correlation (r = 0.12) between the A% CeMRO 2 and the duration after the stroke (fig. 2). Furthermore, there was no significant difference in mean A% CeMRO 2 between early ( < 1 month) and late ( > 1 month) studies of stroke patients (fig. 3). The A% CeMRO 2 of the 14 patients who had follow-up PET studies are given in Table 3. The magnitude of CCD remained approximately constant in 5 patients (Cases 281, 190, 258, 365, 216), diminished or disappeared in 3 (Cases 177, 294, 313) and increased in 4 (Cases
682
P. PANTANO AND OTHERS
+30 i
+20 +10 O
6 ~ -20 -30 -40
lw 2w 3w lm
2m
3m 4m 5m 6m 7m 8m 9m 10m l l m 12m > l y Time
FIG. 2. Plot of the 72 individual A% CeMROj (i.e., the percentage difference between contralateral and ipsilateral cerebellar hemispheres) against duration after stroke (w = week; m = month; y =• year).
1 m (n=30)
-5
6 -10 FIG. 3. Mean J % C e M R 0 2 ( ± 1 SEM) for the 'early study' group (less than 1 month since stroke onset, n = 42) and for the 'late study' group (more than 1 month, n •= 30). These values are not significantly different.
-15
345, 169, 233, 185). In the remaining 2 patients (Cases 177, 333) there was no significant metabolic depression in the contralateral cerebellar hemisphere in any of the PET studies. Illustrative examples are shown infigs4 and 5. Correlation Between CCD and Topography of Cerebral Infarct The frequency of significant CCD and the mean A% CeMRO2 according to the topography of the supratentorial infarct are shown in Table 4. Although the number of patients in four of the topographic subgroups (capsular, frontal, temporal, parietal) is small (less than 5), it appears that CCD is most frequent and
CROSSED CEREBELLAR DIASCHISIS
683
TABLE 3. PERCENTAGE ASYMMETRY IN CEREBELLAR OXYGEN CONSUMPTION (J% CeMROj) IN FOLLOW-UP STUDIES
Patients
Duration from onset
281
2d lid
A%CeMRO2 -18.2* -19.9*
177
3d lOd
- 9.6* + 3.1
294
3d lOd
-12.7* + 0.6
190
3d 17d
-18.5* -17.5*
216
3d 10 d 3m 6m
-17.0* -24.5* -21.6* -20.8*
345
4d 4m
-14.0* -23.5*
169
4d 9d
-13.0* -26.4*
233
5d 31 d
- 9.5* -16.8*
258
12d 38 d
-12.6* - 9.7*
365
12 d 4.5 m
-17.0* -17.4*
110
18 d 32 d
+ 9.4 + 14.7
313
lm 3m 8m
-36.8* -20.3* -22.0*
333
lm 5m
+ 4.9 + 9.8
185
2m 2.5 m
- 8.0 -16.3*
* Indicates significant CCD (based on AI values) with respect to 95% confidence limits of control subjects, d = days; m = months.
severe in deep middle cerebral artery territory infarction, followed by multilobar and capsular infarcts. Correlation Between CCD and Motor Deficit (Table 5) Mean A CeMRO 2 values were very similar among and highly significant within each hemiparesis grade subgroup, but frequency of CCD was highest in the most severely hemiparetic patients.
684
P. PANTANO AND OTHERS
OM+2 cm
OM+4 cm
FIG. 4. PET images showing CMRO2 at the level of the cerebellum (OM + 2 cm) (left side) and basal ganglia (OM+4 cm) (right side) on three different occasions in a 69-year-old woman (Case 294). The first study (upper panel) was performed several days after a single transient ischaemic attack involving therightcerebral hemisphere, and was unremarkable, as was the CT scan. The second study (middle panel), seven months later, was performed on the third day following the abrupt onset of left global hemiplegia, presumably due to embolism of the right carotid system from the heart (atrial fibrillation). There was a marked decrease in CMRO2 in the right lenticulocaudate-capsular area (straight arrow), corresponding to the CT appearance of a fresh infarct in the deep middle cerebral artery (MCA) territory. Cerebellar oxygen consumption was significantly less on the left than on the right side (3.71 and 4.25 ml/100 g/min, respectively) (ctirved arrow). The third study (lowermost panel) was performed after a further seven days, when the patient had notably improved clinically: the depression in CMRO2 in the right deep MCA territory had lessened (straight arrow), and cerebellar metabolism was now symmetric (3.72 and 3.70 ml/100 g/min on left and right side, respectively).
On the other hand, although the mean J% CeMRO2 of the patients without hemiparesis was not statistically significant, 3 of the 7 patients in this subgroup still had individually significant CCD; one of them (Case 216), whose CCD was significant at each of four follow-up studies, was unusual in that he had no
CROSSED CEREBELLAR DIASCHISIS
685
FIG. 5. CMRO2 images obtained in a 48-year-old woman (Case 313) who developed a right middle cerebral artery infarct (deep territory) due to embolism. She recovered well, and, at the first PET study a month later, had only a mild left hemiparesis. The study (top two images) showed a markedly depressed CMROj in the infareted area at level OM + 3.7 cm (straight arrow, top left image), and a contralateral cerebellar hypometabolism ( — 36.8% relative to right side, P < 0.05) at level OM+1.7 cm (top right image, curved arrow). When studied again at three months (bottom left image) and eight months (bottom right) after onset, the left cerebellar hemisphere was still hypometabolic (curved arrows), although numerically less than at first study (-20.3% and -22.0% relative to right side, respectively, P < 0.05).
hemiparesis whatsoever despite an extensive multilobar infarction; both of the remaining patients had occlusion of a major cerebral artery (internal carotid artery in one case, middle cerebral artery in the other). The 'no hemiparesis' subgroup also includes a patient (Case 110) with a crossed aphasia due to right temporoparietal infarction and a 'reversed' CCD (see Table 3). DISCUSSION
Before discussing the significance of our findings, several methodological issues need to be addressed. First, the presence of large, often asymmetric vascular spaces in the posterior fossa (mainly venous sinuses) might have caused errors in the
686
P. PANTANO AND OTHERS TABLE 4. FREQUENCY AND SEVERITY OF CCD WITH RESPECT TO TOPOGRAPHICAL LOCATION OF THE INFARCT
Location of infarct
Frontal Temporal Parietal Temporoparietal Frontoparietotemporal Deep MCA Capsular
No. of No. of patients I1% patients with significant CeMRO2 CCD (mean ;± .SEM) 4 1 - 3.6 ± 5.0 3 - 4.5 ± 2.9 0 1 3 - 6.9 ± 3.5 4 6 -11.5 ± 5.8 -10.1 ± 3.3* 13 6 -15.1 ± 2.6** 12 10 4 2 -10.6 ± 1.6*
* P < 0.01. ** P < 0.001 with resDect to zero. MCA = middle cerebral artery territory. TABLE 5. FREQUENCY AND MAGNITUDE OF CCD WITH RESPECT TO SEVERITY OF MOTOR IMPAIRMENT
Severity of motor No. of No. of patients A°/o impairment*' patients* with significant CeMRO2 CCD (mean ± SEM) 0 7 3 - 6.7 ± 4.1 (43%) 1 .9 . 5
