The dementia associated with progressive supranuclear palsy (PSP) is ... 18F-fluoro-2-deoxyglucose18FDG in 6 patients presumed to have PSP and was ...
Brain (1985), 108,785-799
SUBCORTICAL DEMENTIA FRONTAL CORTEX HYPOMETABOLISM DETECTED BY POSITRON TOMOGRAPHY IN PATIENTS WITH PROGRESSIVE SUPRANUCLEAR PALSY by R. D ' A N T O N A , 1
J. C. B A R O N , 1 . 2 4
F. VIADER,
Y. SAMSON, 1
M. S E R D A R U , 3
3
Y. AGID and J. CAMBIER4
{From the 'Service Hospitalier Frederic Joliot, CEA, Departement de Biologie, 91406 Orsay, the 2 Clinique des Maladies du Systeme Nerveux, La Salpetriere, 75013 Paris, the 3Clinique de Neurologieet de Neuropsychologie, La Salpetriere, 75013 Paris, and the 4Clinique Neurologique de la Faculte Bichat, Hopital Beaujon, 100 Boulevard du General Leclerc, 92110 Clichy, France)
SUMMARY The dementia associated with progressive supranuclear palsy (PSP) is considered to be subcortical because the cerebral cortex, unlike the subcortical structures, is usually free from major neuropathological lesions; the characteristic symptoms point to a dysfunction of the prefrontal lobe. The regional cerebral metabolic rate of glucose (rCMR Glu) was studied by positron emission tomography and 18 F-fluoro-2-deoxyglucose18FDG in 6 patients presumed to have PSP and was compared with values found in 8 control subjects of similar age. The results obtained showed a highly significant rCMR Glu decrease in the prefrontal cortex of our patients. The loss of several subcortical afferents to prefrontal cortex may be responsible for the frontal cortical hypometabolism present in PSP.
INTRODUCTION
Despite several positron emission tomography (PET) studies of regional cerebral metabolism in Alzheimer's type and multi-infarct dementia (Lenzi and Jones, 1980; Frackowiack et al., 1981; Benson et al., 1983; de Leon et al., 1983; Foster et al., 1983; Friedland et al., 1983), subcortical dementia has remained largely unexplored using this new research tool. Subcortical dementia is characterized by (1) symptoms suggesting dysfunction of the prefrontal cortex and (2) neuropathological involvement of subcortical structures without major cortical lesions (Albert et al., 1974; Cummings and Benson, 1984; Cambier et al., 1985). According to Albert et al. (1974), loss of activating afferences to the cortex resulting from subcortical lesions presumably causes this type of dementia. Recently (Mayeux et al., 1983), the respective contributions of cortical and subcortical involvement in the dementias of Alzheimer's, Parkinson's, and Huntington's diseases have been questioned on 1
Correspondence to: Dr J. C. Baron.
786
R. D'ANTONA AND OTHERS
neuropsychological, neuropathological, and neurochemical grounds. The intellectual decline that occurs in progressive supranuclear palsy (PSP, SteeleRichardson-Olszewski syndrome) has, however, remained the most characteristic subcortical dementia, both clinically and pathologically (Cummings and Benson, 1984). Our aim in studying with PET the regional cerebral glucose utilization of patients with suspected PSP was to assess whether the clinical signs suggesting prefrontal lobe dysfunction were attended by corresponding hypometabolism. Since regional glucose consumption is closely linked to neuronal activity (SokolofT, 1982), this finding would be a marker of cortical neuronal dysfunction, presumably resulting from subcortical lesions, and hence strengthen the concept of subcortical dementia. PATIENTS AND METHODS Patients Six patients (mean age 63.3 + 4.5 years) with probable PSP were studied (Table 1). In addition to falls, nuchal rigidity, bradykinesia and supranuclear impairment of vertical gaze, they all exhibited symptoms and signs of prefrontal lobe dysfunction (Table 2). Cases 1 to 3 also exhibited imitation and utilization behaviour (Lhermitte, 1983). Informed consent was obtained. For comparison, 8 subjects of similar age (59.0+ 10.0 years) without known brain disorder, intellectual impairment or vascular risk factors were studied using the same method. Methods Regional cerebral glucose utilization (rCMR Glu) was measured using the l8F-fluoro-2-deoxyD-glucose (• 8 FDG) method (Phelps et al., 1979; Reivich et al., 1979) as applied to PET (Phelps et al., 1979; Huang et al., 1980). The principles and validation of the method, which is the in vivo application to man of the autoradiographic technique of SokololT using 14C-2-deoxy-D-glucose (SokolofT et al., 1977), have been described in detail elsewhere, as well as the mathematical model, the procedure and the normal values (Phelps et al., 1979; Reivich el al., 1979; Huang et al., 1980), and will therefore not be reviewed here. Detailed accounts of the procedure as applied in our centre have also been published (Baron et al., 1982, 1984; Rougemont et al., 1984). In brief, the examination is performed under resting conditions with eyes closed, and external stimulation is kept to a minimum. A bolus of 8 to 10 mCi of 1 8 FDG is injected into one antecubital vein, and arterial blood samples are obtained from a radial artery catheter at predetermined intervals from time of injection until the end of scanning. Three planes parallel to the orbitomeatal (OM) line, located respectively 1.5 cm, 3.5 cm and 5.5 cm above this line, are studied using the ECAT II (ORTEC, EG and G) single slice positron tomograph (lateral resolution is about 16 mm and slice thickness about 19 mm). The acquisition time for each plane is about 10 min, with midscan times ranging from 55 to 77 min after injection of I 8 FDG. The images are then reconstructed by computer using measured attenuation correction (obtained by means of a prior 6 8 Ge- 6 8 Ga transmission scan) and system calibration data. These images, which represent absolute 1 8 F radioactive concentration, are then transformed, pixel by pixel, into functional images representing CMR Glu in mg/100 g/min. This is accomplished using the operational equation proposed by Phelps et al. (1979), which contains the four FDG rate constants (including the dephosphorylation constant kj) measured by these authors in 13 young healthy volunteers, the 'lumped constant' estimated at 0.42 by the same authors, the pixel 1 8 F concentration at time of study (decay-corrected), the time course of arterial plasma 18 F concentration from injection time to acquisition time, and the arterial plasma glucose content during the study (mean of 5 determinations).
TABLE 1. NEUROLOGICAL FEATURES OF THE 6 PATIENTS WITH PSP Case no. age/sex 1/62/M
Disease duration (yrs) 5
Ocular motility Upgaze palsy
Postural hypertonia ++
Limb rigidity ++
Bradykinesia ++
+++
+++
+++
+++
+++
+++
+
0
+
+
Functional Treatment disability (effect) ++ Bromocriptine (no effect) +++ L-DOPA, bromocriptine (no effect) +++ L-DOPA (no effect)
2/62/M
7
3/74/M
8
4/63/F
2
Vertical gaze palsy Vertical gaze palsy Upgaze palsy
5/65/F
5
Downgaze palsy
++
++
++
+++
6/57/M
1
Mildly impaired vertical gaze
+
0
+
+
L-DOPA, bromocriptine (no effect) Bromocriptine (no effect) Bromocriptine (slight improvement)
CTscan Brainstem atrophy Mild ventricular enlargement Mild cortical atrophy and ventricular enlargement Normal Moderate brainstem and cortical atrophy Normal
•n
O
z H >
o 2 w o r
0 = no alteration; + = alteration, mild; + + = moderate, + + + = severe.
2 TABLE 2. FRONTAL LOBE SIGNS
Case no. 1
2 3 4 5 6
Bradyphrcnia
Execution of orders
Apathy, loss of initiative
Impaired judgement
Loss of abstracting ability
Personality and behavioural Speech changes Perseverations alterations Jocular Present Dysarthria, verbal paraphasias * Severe dysarthria * Severe dysarthria Mild dysarthria Mild depression, Absent irritability Slowing Personality Present of fluency changed Dysphonia Episodic Absent irritability
c w o Other Grasping Grasping Grasping — Grasping —
0 = no alteration; + = alteration, mild; + + = moderate; + + + = severe; * = impossible to evaluate because of severity of illness.
n D m
2 m
z
788
R. D'ANTONA AND OTHERS
Data Analysis Regional grey matter CMR Glu values were calculated by a standard procedure calling on 4 cm 2 circular regions of interest (Lebrun-Grandie et al., 1983). Several Regions of Interest were placed visually on the video-display by means of a manual joystick in predetermined grey matter areas on the left hemisphere and were then mirror-copied automatically over the right hemisphere with respect to a vertical axis (Lebrun-Grandie et al., 1983). For each subject, Regionsof Interest were positioned in the following grey matter areas (Rougemont et al., 1984): cerebellum on plane I (OM + 1.5 cm); medial frontal, lateral frontal, temporal, temporo-occipital and occipital cortex on plane II (OM + 3.5 cm); and medial frontal, lateral frontal, parietal, parieto-occipital and occipital cortex on plane III (OM + 5.5 cm). It must be pointed out that in only 4 of the 6 PSP patients were reliable images of planes I and III obtained because of positioning difficulties related to neck rigidity. In addition to the absolute regional CM R Glu values so obtained, the following were also calculated for each patient: (1) the mean cortical CMR Glu, derived by averaging all the regional data for planes II and III separately; (2) normalized regional values, expressed as a percentage of the mean, obtained by dividing each regional CMR Glu by the mean cortical CMR Glu; and (3) a regional asymmetry index, expressed by the percentage right/left ratio for each pair of homologous regions. Statistical analysis of the results was performed by means of standard t test and analysis of variance procedures. RESULTS
Visual inspection of the CMR Glu images of the 6 PSP patients disclosed a striking decrease for CMR Glu in the cortex of both frontal lobes. This was somewhat variable from one patient to another, but was present in all (fig. 1).
FIG. 1. PET images representing rCMR Glu at brain level II (orbitomeatal line +3.5 cm) obtained in 6 patients with presumed PSP (images 1 -6) and 1 control subject (image 7). In these images, anterior is up and left is to the left of the reader; whiter shades of grey indicate higher metabolism. These images are not cross-scaled, so that differences in mean metabolic rate among patients are not apparent; this mode of display, however, allows visual assessment of the regional metabolic pattern. Compared with image 7, which is typically normal, every patient with PSP has an altered metabolic pattern, essentially a marked bifrontal hypometabolism, while posterior (occipital) regions appear falsely hypermetabolic (see Tables 1 and 2 for clinical details of the patients).
FRONTAL HYPOMETABOLISM IN SUBCORTICAL DEMENTIA
789
Level II CMR Glu 14 oc 12
8 10
1 8 1 6
rh ill | mF.
IF
TO
O
| Controls
K^3
PSP
Level III CMR Glu " 12 00
g 10
r+i
I 6 r+i I 4 2 0
mF
IF
I
PO
rh
P i
O
FIG. 2. Regional values of CMR Glu at brain levels II (OM + 3.5 cm) and III (OM + 5.5 cm) of 8 controls and 6 patients with PSP (data of level III are from 4 patients only—see Methods). Vertical bars represent mean rCMR Glu + 1 SD. The stars indicate statistically significant differences between patients and controls (P < 0.05). mF = medial frontal; IF = lateral frontal; T = temporal; TO = temporo-occipital; P = parietal; PO = parietooccipital; O = occipital.
Compared with control values, mean cortical and cerebellar glucose utilization was lower in PSP patients, but the difference did not reach statistical significance. Regional CMR Glu values of PSP patients were also lower than control values for all regions studied, but the difference was significant only in the frontal regions of plane II (P < 0.05) (Table 3,fig.2). Analysis of the relative (percentage) CMR Glu values revealed a highly significant reduction of this index in the frontal regions of both plane II and plane III (P < 0.01), and an increase in the occipital regions ( P < 0.01) (Table 4,fig.3). We found no significant right-left asymmetry of CMR Glu for any of the cortical regions studied (Table 5). Comparison between the severity of frontal lobe symptoms and the intensity of the frontal hypometabolism was limited by the small patient sample. Table 6 summarizes the data. There is no clear correlation between frontal signs and relative frontal hypometabolism (percentage values), but some trend of correlation is apparent when absolute rCMR Glu values are considered.
790
R. D'ANTONA AND OTHERS
TABLE 3. MEAN (+SD) VALUES OF CEREBRAL GLUCOSE UTILIZATION (CMR GLU IN MG/MIN/lOOg) AT BRAIN LEVELS II AND III
Level II Patients (6) Controls (8)
Level III Patients (4) Controls (8)
Medial frontal 4.86*
Lateral frontal
Temporal
4.74*
Temporooccipital
Occipital
Mean cortical
5.47
5.17
7.02
5.51
±1.81
±2.00
±2.09
±1.77
±1.50
±1.36
6.87 ±1.13
6.48
6.90
6.43
7.67
6.86
±1.15
±1.29
±1.30
±1.63
±1.24
Medial frontal
Lateral frontal
Parietal
Parietooccipital
Occipital
Mean cortical
4.71
5.12
5.64
6.08
7.43
5.92
±1.99
±2.13
±2.57
±2.81
±3.25
±2.61
6.70
6.54
6.68
6.68
7.99
6.93
±0.91
±1.02
±1.08
±1.18
±1.32
±1.06
* P< 0.05 with respect to controls.
TABLE 4. REGIONAL RELATIVE (PERCENTAGE) CMR GLU VALUES AT BRAIN LEVELS II AND III
Level II Patients (6) Controls (8)
Level III Patients (4) Controls (8)
Medial frontal
Lateral frontal
Temporal
Temporooccipital
-11.52* ± 3.77
-13.10*
-0.45
-8.45
± 4.11
±8.48
+ 0.77 ± 6.99
- 5.39 ± 4.04
±5.63 +0.57 ±5.33
Medial frontal
Lateral frontal
Parietal
Parietooccipital
-18.71* ± 6.85
-12.51* ± 2.99
-4.77
+ 0.95
±3.04
±7.62
+ 25.75* ± 5.05
- 2.95 ± 7.47
-
5.77
-3.70
-3.78
+ 7.99
± 1.41
±3.60
±6.29
± 1.32
-8.06
±3.10
< 0.01 with respect to controls.
Occipital + 27.88* ± 5.07 + 11.60 ± 7.89
Occipital
FRONTAL HYPOMETABOLISM IN SUBCORTICAL DEMENTIA
791
Level II
« 120
i 100 u
•
50
OS
mF
• • -+H
l
TO
IF
1 O
Controls PSP
Level III
g, 120 | 100
I 50 4)
1 1 mF
PO
IF
O
FIG. 3. Relative (percentage) regional CM R Glu values obtained by dividing each individual absolute rCM R Glu value by the corresponding mean cortical value (see Methods). Other comments are the same as for fig. 2. The results shown are those of Table 5, but now expressed relative to 100 per cent to allow easier visual assessment of the metabolic pattern (e.g. + 5 % is now shown as 105 %). *P < 0.01. Abbreviations as for fig. 2.
TABLE 5. MEAN RIGHT/LEFT RATIOS OF REGIONAL CMR GLU VALUES AT BRAIN LEVELS II AND III
Level II Patients (6)
Level III Patients (6)
Frontal
Temporal
Temporooccipital
1.034
1.026 ±0.034
1.041
1.058
±0.055
±0.058
±0.075
1.020 ±0.117
Frontal
Parietal
Temporooccipital
Occipital
Mean grey matter
1.033 ±0.064
±0.041
1.028 + 0.050
1.050 + 0.093
1.027 + 0.092
1.035
Occipital
No significant asymmetry compared with controls.
Mean grey matter
792
R. D'ANTONA AND OTHERS
DISCUSSION Our study establishes that cortical hypometabolism affecting both frontal lobes occurs in most, and possibly all, patients suffering from progressive supranuclear palsy. Thisfindingwas already obvious by visual inspection of the functional CMR Glu images (fig. 1), which show that this hypometabolism affects most of the frontal lobes in a distribution roughly corresponding to the prefrontal cortex (Fuster, 1980). Quantitative analysis of the regional data further confirmed reduced CMR Glu values in the frontal regions of PSP patients when compared with control values (Table 3, fig. 2). A more widespread and more significant frontal relative hypometabolism was found by analysis of the percentage CMR Glu values (Table 4, fig. 3), because this method compensates for intersubject variability in absolute CMR Glu by using each patient as its own control. This method also disclosed a significant occipital cortex hypermetabolism which, however, is only relative and due entirely to the reduction in absolute CMR Glu in the frontal cortex; as shown in Table 3 and fig. 2, occipital CMR Glu was in fact slightly lower than control values. Because the CMR Glu images shown in fig. 1 are also normalized, they too falsely suggest the presence of occipital hypermetabolism. Several questions may be raised regarding the methodology used. First, although the diagnosis of PSP can only be established post-mortem, the clinical features of our patients, together with the CT scan data and the lack of response to dopaminergic agonists, leave little doubt in this respect (Table 1). Secondly, objections may be raised against the use of the in vivo autoradiographic paradigm to measure CMR Glu in disease states, and particularly in situations of reduced metabolism (Huang et al., 1980; Hawkins et al., 1981; Baron et al., 1984). To overcome this potential problem, we obtained more reliable CMR Glu values by means of the 'kinetic' method, whereby the FDG kinetic rates are simultaneously measured regionally on each patient (Huang et ai, 1980; Baron et al., 1984; Rougemont et al., 1984). For various technical reasons, the 'kinetic' data of only 3 PSP patients and 5 control subjects could be analysed further; despite the small sample, this method confirmed the frontal hypometabolism found using the autoradiographic approach (mean relative medial and lateral frontal values: —7.7% and — 10.5% for patients, and +1.6% and —4.5% for control subjects, respectively). Thirdly, the measured CMR Glu values directly depend on the 'lumped constant' for FDG, but significant alterations in this variable occur only in situations of marked mismatching between glucose supply and demand (Pardridge et al., 1982) that are most unlikely to prevail in degenerative disorders of the brain. Finally, preferential cortical atrophy in the frontal regions, through the partial volume effect, may have produced the spurious appearance of frontal hypometabolism (Kuhl et al., 1984), but the fact that cortical atrophy was absent or only mild in our patients (Table 1) makes this possibility unlikely. The reduction in prefrontal cortex glucose utilization and, hence, neuronal function (Sokoloff, 1982), is in striking correspondence with the pattern of
FRONTAL HYPOMETABOLISM IN SUBCORTICAL DEMENTIA
793
intellectual impairment observed in our patients (Table 2) and, more generally, in PSP, such as reduced capacity for critical and abstract thought, impaired attention, slowing of mental processes and perseverations, all symptoms that suggest dysfunctioning prefrontal cortex (Albert et al., 1974; Cummings and Benson, 1984; Cambier et al., 1985). In our small patient sample, there was no obvious correlation between the severity of the frontal lobe symptoms and the intensity of frontal hypometabolism (Table 6 and fig. 1); in fact, patients with early disease and moderate intellectual impairment (Cases 4 and 6) also had marked relative frontal hypometabolism, suggesting that the metabolic alteration may appear quite early in the course of the disease and, perhaps, even precede the occurrence of clinically overt frontal lobe symptoms. Further studies will be necessary to substantiate these points. The fact that frontal hypometabolism has not so far been a prominent feature in 18 FDG studies of other degenerative disorders of the brain such as Parkinson's disease (Kuhl et al., 1984; Rougemont et al., 1984), Alzheimer's type dementia (Benson et al., 1983; Foster et al., 1983; Friedland et al., 1983), Huntington's disease (Kuhl et al., 1982) or Wilson's disease (Hawkins et al., 1983), further suggests that it perhaps constitutes a useful marker of PSP. This would be concordant with the general observation that the most characteristic of the subcortical dementias is that found in PSP (Albert et al., 1974; Cummings and Benson, 1984). Two main hypotheses, not mutually exclusive, should be considered to explain this frontal neuronal hypometabolism: (1) cortical neuronal lesions, such as neuronal loss and neurofibrillary tangles, the hypometabolism observed being then a mere expression of decreased number of functional neurons; and (2) loss of stimulating afferences from subcortical structures involved in the disease process to (without loss of) functional neurons in the frontal cortex. Neuropathological studies have shown that PSP is characterized by neuronal loss, neurofibrillary tangles and gliosis affecting several subcortical structures, but with relative sparing of the cerebral cortex (Albert et al., 1974; Ishino et al., 1975; Steele, 1975). Sparse neurofibrillary tangles are uncommonly found in the cerebral
TABLE 6. CLINICOMETABOLIC CORRELATION
Case no.
Frontal lobe signs
CMR Glu (mg/lOO g\miri)
CMR Glu (%)
1 2 3 4 5 6
Moderate Severe Severe Mild Moderate Mild
4.02 2.59 5.05 5.15 5.07 6.92
-16.2 - 6.5 -11.5 -10.1 -13.6 -16.0
Values shown are mean of medial frontal and lateral frontal CMR Glu obtained at brain level II.
794
R. D'ANTONA AND OTHERS
cortex and they predominate in the hippocampus; when found, their density is generally of the order of what is seen in controls of similar age (for review see Ishino and Otsuki, 1976). Neuronal loss in the neocortex has not been reported, although, to our knowledge, morphometric studies are lacking. A recent post-mortem study, however, reported a 34 per cent reduction in dopaminergic receptor density in the frontal cortex of PSP patients (Ruberg et al., 1985), possibly indicating that some neuronal loss may occur in this location. Despite these reservations, the lack of gross morphological changes in the frontal cortex of PSP patients therefore makes unlikely the first hypothesis described above. Loss of activating subcortical afferents to the frontal cortex may therefore be considered the chief mechanism involved in frontal hypometabolism. Such a mechanism had been postulated earlier to explain the frontal lobe syndrome of PSP (Albert
