Positron Emission Tomography of Monoaminergic Vesicular Binding in Aging and Parkinson Disease

Subjects

Thirty-one subjects with PD (20 men, 11 women; age 64 ± 10 years (mean ± s.d.)—range 43 to 89 years; Table 1) and 75 normal subjects (40 men, 35 women; age 49 ± 15 years—range 20 to 79 years) underwent PET imaging with (+)-[¹¹C]DTBZ. The duration of symptomatic PD was 5 ± 3 years (range 1 to 15 years) at the time of imaging. Parkinson disease diagnosis was based on conventional criteria requiring the presence of at least two cardinal signs (resting tremor, bradykinesia, rigidity) and responsiveness to levodopa and or dopamine agonist treatment. No subjects had atypical features such as prominent cognitive impairment or postural instability within the first year or had supranuclear eye movement abnormalities (Gelb et al, 1999). Three subjects were untreated for PD at the time of PET imaging, but later showed symptomatic treatment responses. Of the remaining subjects, 25 were receiving carbidopa-levodopa, 5 dopamine agonists, 9 selegiline, 2 trihexyphenidyl, and 1 tolcapone. Parkinson disease subjects were examined clinically in the practically defined ‘off’ state (12 h after the last medication dosages) using the Unified Parkinson Disease Rating Scale (UPDRS; Fahn and Elton, 1987), the modified Hoehn and Yahr Scale (HY; Hoehn and Yahr, 1967), and the Schwab and England (1969) Activities of Daily Living Scale (SE). Patients were allowed to resume medication after clinical ratings and before PET scans. All patients had mild-to-moderate PD (HY 1 to 3: stage 1: n = 8; stage 1.5: n = 8; stage 2: n = 7, stage 2.5: n = 6; and stage 3: n = 2 patients).

OPEN IN VIEWER

Table 1

Parkinson disease subject characteristics

Age	Gender	Duration	HY	SE	UPDRSIII	T	R	B	Medications a
43	M	3	1.0	0.9	6	+	+	+	L—240
44	F	5	1.0	1.0	4	+	+	+	L—400; D—10
53	F	5	1.0	0.9	8	+	+	+	L—300
57	M	4	1.0	0.9	16	+	+	+	L—300
58	M	1	1.0	0.9	8	+	+	+	L—400; D—10
63	F	3	1.0	1.0	10	+	−	+	D—5
67	M	2	1.0	0.8	17	+	+	+	None
71	F	1	1.0	1.0	10	+	+	+	None
48	F	1	1.5	0.8	9	+	+	+	P—3
59	M	7	1.5	0.9	16	−	+	+	L—100; P—6; D—5
61	M	1	1.5	0.9	8	+	+	+	None
64	M	3	1.5	0.9	9	+	−	+	P—1.5
68	M	2	1.5	0.8	10	+	+	+	L—600
73	M	8	1.5	0.9	7	+	+	+	L—480; D—5
74	M	7	1.5	0.8	10	+	+	+	L—1000; Br—22.5; D—10
80	M	3	1.5	0.9	7	−	+	+	L—500
54	M	8	2.0	0.8	23	+	+	+	L—580
60	M	12	2.0	0.8	27	+	+	+	L—980; D—5
64	M	6	2.0	0.8	27	+	+	+	L—960; Tr—8
68	F	8	2.0	0.9	10	+	+	+	L—750
68	F	4	2.0	0.8	13	+	+	+	L—300
75	F	5	2.0	0.8	13	+	+	+	L—420; D—5
76	M	22	2.0	1.0	20	−	+	+	L—400
58	M	5	2.5	0.8	26	+	+	+	L—140; Tr—8; D—5
60	M	3	2.5	0.9	23	+	+	+	L—300; To—600
63	M	5	2.5	0.8	19	+	+	+	L—630
64	F	10	2.5	0.8	40	+	+	+	L—560; Br—15; D—5
79	M	6	2.5	0.7	34	+	+	+	L—800
89	F	1	2.5	0.9	20	+	+	+	L—280
68	M	8	3.0	0.7	33	+	+	+	L—630; D—5
69	F	15	3.0	0.8	30	+	+	+	L—800

a

Values reflect the total daily medication dosages. Sustained-release levodopa was scaled by 0.75, reflecting its incomplete bioavailability. B = bradykinesia; Br = bromocriptine; D = deprenyl; L = levodopa; P = pergolide; R = rigidity; T = tremor; To = tolcapone; Tr = trihexyphenidyl.

None of the normal subjects were taking psychotropic drugs or had a history of neurological, psychiatric, or major medical diseases, and all had a normal neurologic examination. The experimental protocol was approved by The University of Michigan committees on the use of human subjects in research and on the human use of radioisotopes, and written informed consent was obtained before all procedures.

(+)-[¹¹C]Dihydrotetrabenazine Positron Emission Tomography Scans

No-carrier-added (+)-[¹¹C]DTBZ (250 to 1000 Ci/mmol at the time of injection) was prepared as reported previously (Jewett et al, 1997). Positron emission tomography image data were acquired in three-dimensional mode (septa retracted) on a Siemens/CTI ECAT EXACT-47 scanner (Knoxville, TN, USA), which images a 15.8 cm axial field-of-view and permits reconstruction of 47 contiguous 3.375-mm-thick tissue slices. All subjects were studied supine, with eyes and ears unoccluded, resting quietly in a dimly lit room. Subjects were injected with 10 to 18 mCi (370 to 666 MBq) of (+)-[¹¹C]DTBZ containing less than 50 μg DTBZ mass. At this dosage, less than 10% of striatal VMAT2 binding sites could be occupied (less than 10 nmol/L [¹¹C]DTBZ peak striatal concentration and approximately 100 nmol/L striatal VMAT2 concentration as reported by Scherman et al, 1989). In all, 55% of the dose was administered by intravenous bolus injection over the first 15 to 30secs of the study, while the remaining 45% of the dose was continuously infused over the next 60 mins, resulting in stable arterial tracer levels and equilibrium with brain tracer levels after 30 mins (Koeppe et al, 1997). A series of 4 PET image frames were acquired: 0 to 4 mins; 30 to 40 mins; 40 to 50 mins; and 50 to 60 mins. Each frame was reconstructed using a Hanning filter with a cutoff of 0.5 cycles/projection ray, achieving a reconstructed resolution of approximately 9 mm full-width at half-maximum. The measured attenuation correction was applied to all image frames based on a 10- to 12-min 2-D transmission scan performed immediately before DTBZ injection. Images were corrected for scattered events by the method of Ollinger and Johns (1993) and Ollinger (1995) as implemented by CTI, Inc. All image frames were spatially coregistered within subjects with a rigid-body transformation to reduce the effects of subject motion during the imaging session (Minoshima et al, 1993b), followed by linear transformation to a reference atlas anatomical orientation (Talairach and Tournoux, 1988) and scale (Minoshima et al, 1993a, 1994).

Blood samples were withdrawn from a radial artery catheter at 10-sec intervals for the first 2 mins, and then at 2.5, 3, 4, 30, 40, 50, and 60mins after DTBZ injection. Plasma was separated from red blood cells by centrifugation and was assayed for ¹¹C activity in a NaI well counter. Aliquots of plasma obtained at 1, 2, and 3 mins and thereafter were also processed for determination of radiolabeled metabolites using a rapid Sep-Pak C₁₈ chromatographic technique (Frey et al, 1992). Plasma (0.5 mL) was added to phosphate-buffered saline (PBS, 0.5 mL) containing approximately 0.05 μCi of authentic [³H]DTBZ and the sample was applied to a Sep-Pak C₁₈ chromatography column. The eluates: (1) from initial column loading combined with a subsequent wash with 9 mL of PBS:ethanol, 65:35, and (2) from a subsequent wash with 5 mL of absolute ethanol, were each assayed for ¹¹C activity in the well counter. After decay of the ¹¹C, aliquots of the two eluates were assayed for tritium content by liquid scintillation spectrometry. The distribution of [³H]DTBZ between the two eluates was used to correct for losses (approximately 20%) of unmetabolized (+)-[¹¹C]DTBZ eluted with the metabolites in the first column wash.

Parametric Dihydrotetrabenazine Transport and Type-2 Vesicular Monoamine Transporter Binding Maps

Equilibrium modeling was used to estimate the total tissue-to-plasma distribution volume (DV_tot) in each imaged voxel (Koeppe et al, 1997). The average concentration of (+)-[¹¹C]DTBZ in tissue measured with PET (C_PET) in the three 10-min duration images between 30 and 60 mins after beginning infusion of DTBZ was divided by the metabolite-corrected arterial plasma concentration (C_P) averaged over the 30-, 40-, 50-, and 60-min samples:

D V tot = C PET / C P

(1)

DV_tot contains contributions of free and nonspecifically bound ligand (DV_ns) in addition to that of specifically bound ligand (DV_sp). Measured plasma and brain regional concentrations varied on average by 2% to 4% during the 30- to 60-min equilibrium period. We estimated specific DTBZ binding by subtraction of the occipital cortex value (DV_ctx), a reference region very low in VMAT2 binding sites, with the assumption that the free and nonspecific distribution is uniform across the brain (DV_ns= DV_ctx)at equilibrium:

D V sp = D V tot - D V ns

(2)

Expression of DV_sp relative to DV_ns results in a binding potential (BP; Mintun et al, 1984; Lammertsma et al, 1996), which is proportional to the density of tissuebinding sites (B_max) divided by the equilibrium ligand dissociation constant (K_d):

B P = D V sp / D V ns ≈ B max / K d

(3)

(the Lammertsma et al (1996) definition of BP). The assumption of minimal specific (+)-[¹¹C]DTBZ binding in the occipital cortex is supported by in vitro VMAT2 binding results in rodent and human brain, which indicate cortical binding levels between 5% and 8% of those in the caudate nucleus (CD) and putamen (Vander Borght et al, 1995a; Scherman et al, 1988) and by prior human PET imaging studies comparing the distributions of the active (+)- and inactive (—)-[¹¹C]DTBZ enantiomers (Koeppe et al, 1999). The PET studies using the inactive enantiomer also showed uniform cerebral distribution of free plus nonspecific uptake of (+)-[¹¹C]DTBZ (Koeppe et al, 1999).

In addition to VMAT2 binding of DTBZ, we estimated the blood-to-brain transport of DTBZ (K₁) in each imaged voxel from the initial 4-min scan frame (C_PET(0–4)) and the arterial plasma time course using the following equation:

C PET ( 0 - 4 ) = ∫ 0 4 K 1 C P ( t ) ⊗ e - ( K 1 / D V tot ) t

(4)

where ⊗ represents the mathematical operation of convolution. No correction for tracer arrival time disparity between radial and carotid arteries was made. Procedurally, after calculation of the voxelwise total distribution volume (DV_tot), the blood to brain transport rate constant (K₁) was estimated voxel by voxel using an ‘autoradiographic’ approach proposed originally for estimation of blood flow (Herscovitch et al, 1983).

Striatal Volume-of-Interest Type-2 Vesicular Monoamine Transporter Binding Analyses

Peak BP activity was determined in striatal volumes of interest (VOI) representing the head of the CD, the anterior putamen (AP), and the posterior putamen (PP) using rectangular sampling regions measuring 4 (antero-posterior) × 2 (medio-lateral) × 1 (dorso-ventral) voxels (1 voxel = 2.25 mm³). Data were extracted from images after linear transformation to Talairach space. Operationally, the transaxial image level best showing the head of the CD and the putamen was identified on the basis of the K₁ parametric images (approximately 0 to 5 mm above the AC–PC plane according to the atlas of Talairach and Tournoux, 1988). At this level, a 4-voxel-deep (antero-posterior) mediolateral sampling profile was centered over the head of the CD in the corresponding parametric DV_tot image. The 2-pixel-wide mediolateral maximum of the profile was then extracted from each hemisphere. Subsequently, the antero-posterior sampling profile was moved 4 voxels posteriorly to estimate DV_tot in AP, and then moved posteriorly again by 4 voxels to measure PP DV_tot. The DV_tot of the occipital cortex reference region was determined in a manually configured VOI applied at a transaxial level guided again by the K₁ parametric map.

For each striatal subregion, BP was calculated as described above (see equation (3)). The left and right sides were analyzed separately, then averaged within each subject to provide a BP for each subregion in each subject. Total putamen BP was computed by averaging AP and PP values. To compute total striatal DTBZ BP, all three striatal subregions were averaged.

Anatomic Standardization and Voxel-Wise Analyses of Dihydrotetrabenazine K₁ and Tissue-to-Plasma Tracer Distribution Volume

Parametric K₁ and DV images were further anatomically standardized by nonlinear deformation (‘warping’) to the stereotaxic coordinate system of Talairach and Tournoux (1988) on the basis of the K₁ images, as described previously (Minoshima et al, 1994; Frey et al, 1996b). Volume of interest data were extracted from the standardized parametric K₁ images, using a predefined set of brain regions determined from the Talairach and Tourneaux atlas. Predefined volumes included all Brodmann areas and the major subcortical regions for both left and right hemispheres. In addition to these VOI analyses, we performed voxel-by-voxel analyses on the K₁ and DV ratio (DVR) images (DVR = DV_tot/DV_ns = 1+ BP).

Statistics

Age-related changes in normal subjects K₁ and BP were examined by linear and exponential regression analyses of parametric values with SYSTAT (Systat Software, Richmond, CA, USA), and the significance of intercept and slope parameter estimates were assessed by analysis of variance (ANOVA).

Regressions of K₁ and BP versus age were each performed in the normal subjects to assess regionally specific effects of aging. Striatal ROI DTBZ BP was assessed in PD subjects, employing age covariation as determined from the normal subjects. All comparisons of PD with normal subjects employed a subset of the normals with age similar to the PD subjects (n = 51; mean age 58 years; range 45 to 79 years; see Figure 1). Voxelwise analyses were performed on K₁ and DVR maps, unadjusted for effects of age. Significance of differences was assessed with unpaired, unequal variance, Student's t-tests.

OPEN IN VIEWER

Figure 1

Posterior putamen DTBZ BP versus age in normal (filled circles = men; open circles = women) and PD subjects (filled squares = men, open squares = women). The line is the regression of PP DTBZ BP versus age in normal subjects using a declining exponential model (see Table 2). The shaded region indicates the subset of normal subjects used for group comparisons with PD. Within the normal group, there is a significant age-related decline in DTBZ BP. There is complete separation of normal and PD subject groups. There are no significant gender differences in normal or PD subjects.

Correlations between DTBZ BP and clinical status of PD subjects were investigated employing the Pearson correlation model with post-hoc significance assessed with the Bartlett χ² test as implemented in SYSTAT. Comparison of side-to-side clinical and striatal DTBZ binding asymmetries was assessed with relative asymmetry measures of UPDRS_III and striatal ROI bindings. Asymmetry scores of UPDRS_III questions with right and left components (items #20 to 26) were calculated by subtracting the left summed scores from right summed scores and normalizing by dividing with the aggregate sum of the left and right scores of these questions (final range of values = −1.0 to 1.0). A similar transformation was used for assessing DTBZ binding asymmetry. Left total striatal DTBZ binding deficit (difference between subject- and age-adjusted predicted normal total striatal DTBZ binding) was subtracted from the right total striatal DTBZ binding deficit and divided by the summed right and left striatal deficits (range of final values = −1.0 to 1.0).

Blood-Brain Dihydrotetrabenazine Transport in Normal Aging and Parkinson Disease

Regression analyses of (+)-[¹¹C]DTBZ K₁ in normal subjects revealed an age-related global decline in ligand transport, averaging a 4.4% ± 1.4% loss per decade of life, consistent with the effect of the known normal age-related decline in cerebral blood flow (Leenders et al, 1990a). After removal of this global effect, no region changed by more than 2% per decade. Quantitative DTBZ K₁ estimates were not significantly different between PD and age-comparable normal subject groups either globally or in any of the atlas-based VOIs, though PD subjects averaged approximately 4% lower in cortical and 1% lower in subcortical structures than normal subjects. Voxel-by-voxel comparison of K₁ between PD and controls revealed significant reduction in parietal cortical K₁. Parametric K₁ images were also analyzed after normalization to each subject's global mean. Global mean K₁ was defined as the average of those voxels falling within 80% of the brain peak value. Statistical significance of the parietal cortex reduction in PD was enhanced in this analysis owing to reduced intersubject variance after removal of individual subject global K₁ effects. Volume of interest analysis of globally normalized K₁ values revealed significant, approximately 3% decreases in three regions (all in the parietal cortex); Brodmann areas 7, 39 and 40, with P-values less than 0.001, 0.004, and 0.001, respectively. No other VOI or voxels revealed significant differences in PD.

Type-2 Vesicular Monoamine Transporter Binding Sites in Normal Aging

Regression analyses of striatal (+)-[¹¹C]DTBZ VOI BP in normal subjects revealed age-related decline in VMAT2 binding sites in all striatal subregions (Figure 1, Table 2). Both linear and exponential models gave indistinguishable fits, revealing approximately 0.5% decline per year in DTBZ binding. Rates of decline were equivalent in all striatal subregions (Table 2). There was no effect of gender on DTBZ binding in any striatal subregion (P > 0.09).

OPEN IN VIEWER

Table 2

Effect of normal aging on striatal VOI DTBZ binding

Striatal region	Intercept	Exponent	F	P	% loss/year
CD	1.89 ± 0.17	0.0054 ± 0.0011	25.14	0.000004	0.54
AP	2.21 ± 0.15	0.0048 ± 0.0010	20.84	0.000020	0.48
PP	2.23 ± 0.16	0.0059 ± 0.0011	26.92	0.000002	0.59
Striatum	2.11 ± 0.15	0.0053 ± 0.0010	26.16	0.000002	0.53

Values represent constants from regression analyses based on an exponential model (BP = Ae_-Bt) using data from 75 normal subjects between the ages of 20 and 79 years. AP = anterior putamen; CD = caudate nucleus; PP = posterior putamen.

Striatal Dihydrotetrabenazine Reductions in Parkinson Disease

There was no significant difference in occipital cortex DTBZ DV between PD and normal subjects. In PD subjects, the occipital cortex DV averaged 4.47 ± 0.92, while the DV of the PD-age-comparable normal subgroup was 4.84 ± 0.88 (P = 0.08 versus PD). It should be noted that the effect of the small reduction in the PD occipital cortex DV would be to relatively increase BP estimates (see equations (2) and (3)), thus potentially reducing, not enhancing, apparent deficits in PD striatal VMAT2 binding.

(+)-[¹¹C]Dihydrotetrabenazine BP was reduced significantly in all striatal VOI subregions of PD subjects (Table 3, Figure 1). There was complete separation between PD and normal subjects in (+)[¹¹C]DTBZ BP in AP, PP, putamen, and in total striatum measures. In CD, there was one mildly affected (HY 1.0) PD subject whose BP values overlapped those of the normal subjects.

OPEN IN VIEWER

Table 3

Striatal VOI DTBZ binding deficits in Parkinson disease

Striatal region	Parkinson disease	Normal	P	Z-score
Caudate nucleus, contralateral	0.610 ± 0.240	0.001 ± 0.169	1.35 × 10−16	3.59
Caudate nucleus, ipsilateral	0.518 ± 0.244	-0.040 ± 0.183	5.44 × 10−15	3.05
Anterior putamen, contralateral	1.140 ± 0.206	0.008 ± 0.198	2.45 × 10−34	5.71
Anterior putamen, ipsilateral	1.029 ± 0.200	-0.051 ± 0.198	2.73 × 10−29	5.45
Posterior putamen, contralateral	1.214 ± 0.173	0.007 ± 0.214	7.14 × 10−41	5.64
Posterior putamen, ipsilateral	1.111 ± 0.192	-0.060 ± 0.211	7.11 × 10−37	5.55
Putamen, contralateral	2.354 ± 0.357	0.016 ± 0.399	6.71 × 10−39	5.86
Putamen, ipsilateral	2.140 ± 0.405	-0.111 ± 0.399	7.16 × 10−34	5.65
Striatum, contralateral	0.989 ± 0.187	0.007 ± 0.176	3.15 × 10−32	5.57
Striatum, ipsilateral	0.887 ± 0.202	-0.049 ± 0.186	3.80 × 10−29	5.04

Values represent the means and s.d. of age-adjusted reductions in striatal DTBZ BP from Parkinson disease (n = 31) and normal subjects (n = 51) of comparable age ranges. The entire cohort of subjects was employed in the covariate adjustments, see Table 2. Deficits in the normal subjects should average ‘zero’ if the data from the broader age range are completely descriptive of the older normal subset. Indeed, the values are negligible in comparison to their variabilities and relative to the PD changes, confirming the validity of the covariation employed. Contralateral = striatum opposite the clinically most affected limbs. Ipsilateral = striatum ipsilateral to clinically most affected limbs. Z-score = [(PD mean-Normal mean)/Normal s.d.].

Voxel-by-voxel analysis of DTBZ DVR confirmed significant declines in binding throughout the PD striatum (Figure 2A), and additionally within the midbrain (Figure 2B). In the ventral midbrain, corresponding to the stereotaxic location of substantia nigra pars compacta (mediolateral: ± 6.5 mm; rostrocaudal: −18 mm; dorsoventral: −8 mm, in the reference space of Talairach and Tournoux, 1988), there were reductions approximating 50% of normal DTBZ binding, greatest on the brain side contralateral to the clinically most affected limbs. Ventral midbrain DTBZ binding correlates with ipsilateral CD significantly (most affected side, r = 0.412; P < 0.05, less affected side, r = 0.524; P < 0.005). There were no significant correlations with putaminal binding, probably due to floor effect in putaminal DVRs. In the dorsal midbrain, corresponding probably to the serotonergic raphe nuclei, there was no significant difference in DTBZ binding.

OPEN IN VIEWER

Figure 2

(A) Dihydrotetrabenazine binding in normal and PD subjects. Images depict group-average DTBZ DVR parametric maps in normal (n = 51, top row) and in PD subjects (n = 31, middle row). Individual images have been oriented before averaging so that the hemisphere contralateral to the most affected limbs (PD) or with the lowest striatal binding (normal subjects) is depicted on the image left. The pattern of significant reductions in PD is depicted superimposed on T1-weighted MRI for anatomic reference (bottom row). Images represent trans-axial levels from the vertex (left column) to the base (right column) of the brain at 10-mm intervals. The parametric images are presented in pseudocolor according to the scale at the bottom right (DVR: 0 to 3; t-statistic: 0 to 6). Striatal DTBZ binding is markedly decreased throughout the PD striatum, with the greatest change in the PP and least effect in the CD. The hemisphere contralateral to the clinically most affected body side has the greatest decrement in DTBZ binding. (B) Significant DTBZ DV reductions in the PD midbrain. Images show t-statistical maps superimposed on MRI for anatomical reference from upper (left image) to lower (right image) midbrain levels at contiguous 3.375-mm intervals. Differences correspond to the expected anatomic location of the substantia nigra, and are greatest contralateral (image left) to the most clinically affected body side.

Intrastriatal Dihydrotetrabenazine Deficit Patterns in Parkinson Disease

In normal subjects, there was no significant striatal DTBZ BP difference between cerebral hemispheres.

In contrast, PD subjects show significant asymmetries (Table 4, Figure 3). The ratios of subregional and total striatal DTBZ BP between the hemisphere contralateral to the clinically most affected limbs and the ipsilateral side were reduced significantly (Table 4). Analogous results were found in ratios of PP-to-CD DTBZ BP. These ratios were reduced significantly in PD subjects in contrast to normal subjects, where there was only a modest anterior-to-posterior increasing gradient (Table 4). Contralateral-to-ipsilateral and anterior-to-posterior (PP/CD) BP patterns differentiate PD and normal subject groups (Figure 3). The PP-to-CD ratio, in particular, exhibits complete separation of normal and PD subjects.

OPEN IN VIEWER

Table 4

Intrastriatal VOI DTBZ BP patterns in Parkinson disease

Striatal regional ratios	Parkinson disease	Normal	P	Z-score
CD C/I	0.90 ± 0.15	0.97 ± 0.05	1.05 × 10-02	-1.39
AP C/I	0.85 ± 0.21	0.97 ± 0.04	4.65 × 10-03	-2.56
PP C/I	0.80 ± 0.22	0.96 ± 0.06	5.36 × 10-04	-2.81
Putamen C/I	0.83 ± 0.20	0.96 ± 0.04	8.44 × 10-04	-3.41
Striatum C/I	0.86 ± 0.16	0.97 ± 0.02	6.99 × 10-04	-4.77
Ipsilateral PP/CD	0.45 ± 0.15	1.14 ± 0.13	1.75 × 10-29	-5.29
Contralateral PP/CD	0.51 ± 0.14	1.16 ± 0.11	6.52 × 10-28	-5.94
Bilateral PP/CD	0.48 ± 0.13	1.15 ± 0.11	3.11 × 10-30	-6.26

AP = anterior putamen; C/I = BP of the brain hemisphere contralateral to clinically most affected limbs divided by BP of the hemisphere ipsilateral to the most affected limbs; CD = caudate nucleus; contralateral = cerebral hemisphere contralateral to most affected limbs; ipsilateral = cerebral hemisphere ipsilateral to most affected limbs; PP = posterior putamen.

OPEN IN VIEWER

Figure 3

Intrastriatal DTBZ BP patterns in PD. Dihydrotetrabenazine BP ratios from striatal VOI are depicted for PD (squares) and normal (circles) subjects. Both the rostrocaudal and the side-to-side asymmetry patterns of BP differ significantly in PD. There is complete separation of PD subjects in the rostrocaudal pattern of BP loss.

Clinical Correlates of Striatal Dihydrotetrabenazine in Parkinson Disease

In PD subjects, there were significant correlations between (+)-[¹¹C]DTBZ BP binding deficits and some clinical measures of disease status (Table 5, Figure 4). In CD, AP, PP and total striatum, (+)-[¹¹C]DTBZ BP deficits correlated significantly with disease duration. In CD, AP, and total striatum, DTBZ BP binding deficits correlated with SE ratings. There was no correlation in any striatal subregion or in the total striatum between DTBZ BP binding deficits and UPDRS_III scores or HY stage.

OPEN IN VIEWER

Table 5

Clinical correlations of striatal VOI DTBZ binding deficits in Parkinson disease

Striatal region	Duration	Hoehn and Yahr	Schwab and England	UPDRSIII
CD	0.542**	0.335	-0.403*	0.326
AP	0.514**	0.131	-0.373*	0.203
PP	0.377*	-0.123	-0.232	0.030
Striatum	0.529**	0.144	-0.376*	0.212

Pearson correlation coefficients of bilaterally averaged, age-adjusted DTBZ BP reductions and clinical measures in 31 PD patients. AP = anterior putamen; CD = caudate nucleus; PP = posterior putamen; striatum = average of CD, AP, PP.

*

P < 0.05

**

P < 0.005.

OPEN IN VIEWER

Figure 4

Disease duration versus age-adjusted total striatal DTBZ BP binding deficit in PD. The duration of clinical symptoms correlates significantly with the total striatal BP reduction averaged over both hemispheres (BP reduction = 0.7923 + 0.0294 × year; r² = 0.2795).

Striata contralateral to the most clinically affected hemibody had the greatest decrements in DTBZ BP, and there was significant correlation between asymmetry indices of striatal DTBZ binding and UPDRS_III scores (r = 0.78, P < 0.0001; Figure 5). In HY 1.0 and 1.5 subjects, PP DTBZ BP contralateral to the affected hemibody was decreased by 81%, while ipsilateral (clinically normal) PP DTBZ BP was diminished by 73% compared with normal subjects.

OPEN IN VIEWER

Figure 5

Relative clinical asymmetry (UPDRS_III) versus relative total striatal DTBZ BP deficit asymmetry in PD. The relative right–left hemisphere (BP deficit) and body motor impairment (UPDRS) values in each subject were expressed relative to the total (bilateral sum) deficits in each PD subject, yielding a relative estimate of side-to-side pathological asymmetry. There is significant correlation of DTBZ binding and clinical motor asymmetry in PD (DTBZ BP = 0.0173–0.0842 × UPDRS; r² = 0.6044).

Normal Nigrostriatal Aging

The magnitude of decline associated with normal age in striatal VMAT2 binding sites in the present study is similar to results obtained with a variety of other methods, including postmortem assays of nigral neuron loss with normal aging, postmortem in vitro studies of nigrostriatal terminal integrity, and prior in vivo imaging studies (McGeer et al, 1977; Zelnik et al, 1986; Scherman et al, 1989; Fearnley and Lees, 1991; DeKeyser et al, 1990; Volkow et al, 1994; van Dyck et al, 1995). Conversely, a recent postmortem study of age-related changes in human striatal dopaminergic markers found no evidence of age-related decline in striatal VMAT2 expression (Haycock et al, 2003), but this study had a significantly smaller number of subjects over the age of 20 than ours, and postmortem measurements might be confounded by agonal and postmortem changes in VMAT2 levels, increasing variation and reducing the chance of finding an age-related change.

Basal Ganglia Type-2 Vesicular Monoamine Transporter Binding Changes in Parkinson Disease

We found no overlap of DTBZ BP ranges in AP, PP or total striatal DTBZ binding sites between PD and normal subjects. Specific PP (+)-[¹¹C]DTBZ binding was most reduced and CD binding least affected in PD. Nevertheless, there was near complete separation in CD DTBZ binding between PD and normal subjects, with only one minimally symptomatic PD subject (HY stage 1) overlapping the normal range. We did not use anatomic MRI for anatomic adjustment or atrophy correction. While there is age-related change in regional brain volumes that can be depicted by brain MRI, there are no significant differences in striatal volumes between control and medication-responsive PD subjects. We conclude that the losses in DBTZ binding in PD subjects reflect nigrostriatal damage rather than atrophy (Good et al, 2001; Schulz et al, 1999; Burton et al, 2004; Summerfield et al, 2005).

Parkinson disease alters also the subregional pattern of striatal DTBZ binding. Dihydrotetrabenazine BP is normally symmetric between hemispheres, while in PD there is significant asymmetry of binding, with reduction greatest in the striatum contralateral to the clinically most affected limbs. The normal rostrocaudal gradient of increasing DTBZ binding is also altered in PD, where a marked decline in the ratio of PP-to-CD DTBZ binding is observed. Our results are consistent with findings in postmortem VMAT2 binding assays, showing near-complete losses of specific VMAT2 binding in the PP in more advanced PD subjects (Scherman et al, 1989; Lehéricy et al, 1994) and more prominent nigrostriatal projection lesions in the PP of less severely affected PD subjects (Kish et al, 1988; Fearnley and Lees, 1991; Damier et al, 1999; German et al, 1989; Gibb et al, 1990).

Striatal DTBZ binding changes are correlated with ventral midbrain reductions in PD subject binding, corresponding to the coordinates of the substantia nigra and reduced greatest contralateral to the most clinically affected limbs. This result is noteworthy since normal BPs in the midbrain are approximately 0.3, indicating that only ∼ 25% of total DTBZ uptake in this region is due to specific binding. Parkinson disease subject dorsal midbrain, probably corresponding to the serotonergic raphe nuclei (Zubieta et al, 2000), was not different statistically from age-matched normals. Although postmortem studies show loss of serotonergic neurons in PD (Hornykiewicz and Kish, 1987; Chinaglia et al, 1993), our result suggests that serotonergic neuron degeneration might be a later feature of PD.

Our results are in good agreement with postmortem studies, suggesting 70 to 80% losses of PP dopamine (Kish et al, 1988; Bernheimer et al, 1973) and perhaps somewhat lesser losses of nigral neurons (Fearnley and Lees, 1991; Damier et al, 1999) when symptoms of PD are initially manifest. We were able to assess the DTBZ deficit accompanying the clinical ‘threshold’ of PD by examination of binding contralateral to the clinically unaffected side in HY 1 and 1.5 subjects. In these subjects, we found 73% loss of PP DTBZ binding in the ‘unaffected’ PP. The DTBZ binding ‘threshold’ for development of clinical parkinsonism is therefore between this value and that of the ‘affected’ side (81% loss) in these most mildly affected subjects.

K₁/Regional Cerebral Blood Flow Changes

In contrast to DTBZ-binding data, there were no differences between PD subjects and age-comparable normals in striatal K₁ values. Voxel-by-voxel and VOI analysis, however, showed significant decreases in PD parietal cortex. Since DTBZ K 1 largely reflects regional cerebral blood flow, it may constitute an indirect measure of regional brain ‘functional’ and energy metabolic activity (Koeppe et al, 2005). Prior FDG-PET studies of regional cerebral metabolism in PD show hypometabolism in the cerebral cortex (Peppard et al, 1992; Eberling et al, 1994; Piert et al, 1996; Vander Borght et al, 1997; Bohnen et al, 1999). Prior studies have, however, usually shown more widespread reductions involving the occipital and posterior cingulate cortices in addition to the parietal association area (Peppard et al, 1992; Eberling et al, 1994; Piert et al, 1996; Vander Borght et al, 1997; Bohnen et al, 1999). We did find evidence of widespread reductions in regional cortical K₁, but these differences did not attain significance with the relatively stringent statistical thresholds we employed (data not shown). Our data are consistent also with the results of Eidelberg et al (1990, 1994) and Fukuda et al (2001), who have described a ‘network pattern’ of regional metabolic covariations in PD using a variation of principal component analysis. In this analysis, parietal association cortex is identified as an area of relative metabolic reduction covarying with increases in the striatum and pallidum and with decreases in the thalamus and prefrontal and occipital cortices.

Dihydrotetrabenazine-Positron Emission Tomography Versus Other Nigrostriatal Dopamine Terminal Markers

Estimates of DA nerve terminal loss in PD are dependent partly on the marker selected (Wilson et al, 1996b). Estimates of AADC activity with [¹⁸F]FDOPA suggest more modest nigrostriatal losses than do our VMAT2 results. [18F]6-fluorodopa uptake (K_i) values in mild PD often exceed 50% of normal putaminal values with significant overlap of values in normal subjects and PD patients (Sawle et al, 1994; Morrish et al, 1996b). Considerable experimental evidence suggests that AADC activity is regulated by extracellular dopamine levels via activation of dopamine D2 receptors (Hadjiconstantinou et al, 1993; Young et al, 1993; Cumming et al, 1995; Tedroff et al, 1999). Relative preservation of FDOPA K_i in PD striatum may result from upregulation of AADC activity in response to the loss of dopaminergic terminals (Kish et al, 1995; Wilson and Kish, 1996a; Tedroff et al, 1999; Lee et al, 2000). Although striatal AADC activity as measured by FDOPA K_i has good correlation with clinical measures of disease severity and with postmortem nigral neuron counts, it might be relatively insensitive to earlier changes in PD (Ishikawa et al, 1996) and be affected by dopaminergic drug treatment, reducing its usefulness as an objective measure of nigrostriatal terminal density.

Use of DAT ligands may have analogous pitfalls. In common with numerous other cell surface proteins, DAT molecules cycle between the cell surface membrane and endosomal compartments. Within endosomal compartments, DAT molecules may have reduced affinity for cocaine analogue tracers, leading to an apparent reduction in the number of binding sites (Zhu et al, 1997; Doolen and Zahniser, 2001). Dopamine receptor activation may regulate the distribution of DAT molecules between the cell membrane and endosomes (Batchelor and Schenk, 1998). Lee et al (2000) presented data that DAT binding sites are downregulated in PD striatum and Guttman et al (2001) presented data that dopaminomimetic agents downregulate striatal DAT binding in early PD subjects.

The potential regulation of FDOPA uptake and DAT binding by loss of striatal dopamine terminals and dopaminergic drugs requires cautious interpretation of FDOPA and DAT imaging data from neuroprotection trials. In the imaging substudy of the CALM-PD study comparing initial pramipexole with initial levodopa therapy, subjects receiving initial pramipexole therapy were reported to have relative preservation of striatal DAT binding when subjects were compared at 22, 34, and 46 months after randomization. Clinical measures of disease status, however, were identical at 46 months (Parkinson Study Group, 2002). The effect of pramipexole on DAT binding, moreover, appeared in the initial randomization to 22-month interval, and there was no evidence of a sustained neuroprotective effect, consistent with a pharmacologic effect on DAT expression after initiation of treatment (Albin et al, 2002; Albin and Frey, 2003; Ahlskog, 2003). Guttman et al (2001) suggested that levodopa had a greater effect on downregulating DAT binding than dopamine agonists, which would be consistent with the CALM-PD imaging results. Analogous results were obtained in the ELLDOPA study of possible levodopa toxicity. In ELLDOPA, a placebo-treated group was compared with two schedules of levodopa treatment for 40 weeks and a subset of subjects underwent single photon emission computed tomography (SPECT) imaging with the DAT ligand β-CIT at entry and conclusion of the trial (Parkinson Study Group, 2004). The greatest decrement in DAT binding was found in the highest levodopa treatment group, though this group had the best clinical ratings of disease severity, even when evaluated after a 2-week treatment washout. The discrepancy between imaging and clinical results is a possible reflection of levodopa treatment regulation of DAT binding and/or the long-duration clinical effects of levodopa treatment (Hauser et al, 2000). In a third recent neuroprotection study, REAL-PET, FDOPA imaging was used to assess the differences in disease status between initial ropinirole- and initial levodopa-treated subjects (Whone et al, 2003). As with the CALM-PD substudy, the initial dopamine agonist group had a significantly smaller decrement in FDOPA uptake than the initial levodopa treatment group. Comparison with clinical measures was not possible; subjects were not evaluated in the ‘off’ state at the conclusion of the trial. Unified Parkinson Disease Rating Scale ratings during the study, however, favored the initial levodopa-treated group, suggesting greater intensity of treatment in the initial levodopa-treated group. It is possible that the measured differences in AADC activity between the two arms of REAL-PET reflect a pharmacologic difference due to differential AADC regulation by differing treatment intensity rather than a difference in rate of nigrostriatal terminal degeneration.

The majority of available evidence indicates that striatal VMAT2 binding sites do not undergo up- or downregulation under conditions that alter the synthesis or turnover of dopamine in experimental animals (Naudon et al, 1994; Vander Borght et al, 1995b; Kilbourn et al, 1996; Wilson et al, 1996b; Zucker et al, 2001; Kemmerer et al, 2003). This is true in the normal striatum, the partially dopamine denervated striatum, and also in the massively dopamine denervated striatum. Similar results have been obtained in serotonergic pathways with manipulations of serotonergic neurotransmission (Vilpoux et al, 2000). Type-2 vesicular monoamine transporter transporter activity and its subcellular distribution within nerve terminals might, however, be regulated actively by monoaminergic neurotransmitter actions (Holtje et al, 2003; Fleckstein and Hanson, 2003; Truong et al, 2003; Ugarte et al, 2003). The changing distribution of VMAT2 among different compartments of striatal nerve terminals, however, does not alter total striatal VMAT2 binding as measured with radiolabeled tetrabenazine analogues (reviewed in Fleckstein and Hanson, 2003; Hogan et al, 2000). These preclinical studies suggest that DTBZ-PET provides an anatomic marker of nigrostriatal terminal density that is not readily influenced by alterations in dopaminergic neurotransmission. Human imaging experiments examining directly the possible regulation of DBTZ binding by dopaminergic agents remain to be performed. De La Fuente-Fernandez et al (2003) presented DTBZ-PET data indicating increased binding in the striata of individuals with dopa-responsive dystonia, which could indicate dopaminergic regulation of VMAT2 expression. An alternative explanation is that dopamine absence within synaptic vesicles exposes more tetrabenazine analogue binding sites due to lack of competition for the common binding site (De La Fuente-Fernandez et al, 2003). If lack of direct regulation by dopaminomimetic treatments is shown in PD, an additional practical advantage of in vivo VMAT2 assays is that measures can be performed without discontinuation of symptomatic dopaminergic therapy, as indicated by recent preclinical studies (Kilbourn et al, 1996).

Relationship of Nigrostriatal Imaging to Clinical Measures

Striatal FDOPA uptake is correlated with clinical measures of disease severity, particularly bradykinesia (Eidelberg et al, 1990; Antonini et al, 1995; Morrish et al, 1996a; Vingerhoets et al, 1997), and DAT losses as assessed by [¹²³I]β-CIT are correlated with the stage and severity of PD (Seibyl et al, 1995; Pirker, 2003). We also find correlations between striatal DTBZ binding and clinical measures of PD severity and asymmetry. We find that the duration of symptoms and SE ratings correlated with total striatal and subregional DTBZ binding. The highest correlations were with the DTBZ BP deficits in the CD and AP. The lack of correlations between clinical ratings and VMAT2 binding in the PP probably reflects a ‘floor’ effect because of the marked losses of PP DA innervation before the time of clinically manifest PD (Antonini et al, 1995; Seibyl et al, 1995; Vingerhoets et al, 1997). Our subject selection and limitations of clinical measures may contribute also to the absence of some correlations, notably the HY and UPDRSIII ratings. In studies reporting significant correlations between FDOPA uptake or DAT binding and clinical measures such as the HY scale and UPDRSIII, a wider range of subjects (HY 1 to 5) was studied (reviewed in Pirker, 2003). Our study of only HY 1 to 3 subjects limits the range available for correlation. If prior studies had been restricted to HY 1 to 3 subjects, it is possible that no significant correlations between tracer binding measures and clinical scales would have been observed. The [¹²³I]β-CIT study of Pirker (2003), for example, would probably not show a correlation between UPDRS scores and binding measures if HY stage 4 and 5 subjects were eliminated. In addition, the HY scale is a relatively crude disease index and distinction between HY 2 and HY 3 might be subtle. The most widely used quantitative clinical measure in PD is the UPDRS scale, and particularly the UPDRS_III subscale. This measure possesses considerable intrinsic variation and is altered by symptomatic treatment effects, even when used in the practically defined ‘off’ state (Hauser et al, 2000; Holloway and Dick, 2002). In particular, the variable long duration effect of levodopa and use of relatively long half-life dopamine agonists can confound interpretation of the ‘off’ UPDRS_III.

Clinical measures, such as time until need for levodopa treatment or change in UPDRS ratings, have been the primary tools for evaluating potential neuroprotective interventions (Holloway and Dick, 2002). Drawbacks of these measures include considerable variability, the confounding effects of dopaminomimetic treatments, and potential symptomatic effects of putatively neuroprotective treatments themselves. In addition, the relationship between clinical manifestations and underlying pathology is complex. Clinically manifest parkinsonism, for example, appears only when approximately 75% of PP nigrostriatal terminals are lost. Clinical measures reflect admixtures of the degree of underlying pathology, treatment effects, and the activity of poorly understood compensatory processes. In their cogent critique, Holloway and Dick (2002) point out that conventional clinical measures such as the UPDRS or time to some change in clinical status (e.g. need for dopaminomimetic treatment or emergence of motor fluctuations) are themselves putative surrogate biomarkers of disease progression rather than validated clinical endpoints.

A biomarker that reflects objectively the progression of pathology in PD would be an invaluable aid for evaluating potential neuroprotective interventions. Brooks et al (2003) presented explicit criteria for a suitable biomarker of disease progression. Molecular imaging methods, such as FDOPA-PET imaging, DAT-PET or -SPECT imaging, and VMAT2-PET imaging, satisfy almost all the criteria for a suitable biomarker. However, an important exception is the likely regulation of AADC activity and DAT distribution by partial striatal dopaminergic denervation and by dopaminomimetic agents, which potentially confound interpretation of FDOPA and β-CIT imaging studies. Preclinical studies of VMAT2 expression suggest that it is not subject to these regulatory limitations, and that (+)-[¹¹C]DTBZ-PET is a suitable candidate for the objective quantification of nigrostriatal integrity in PD. The present studies confirm and extend observations that VMAT2-PET imaging depicts the severity and distribution of nigrostriatal lesions in PD, in keeping with clinical and behavioral measures. Final evaluation of the suitability of (+)-[¹¹C]DTBZ-PET as a PD biomarker will require direct assessment of the effects of dopaminomimetic agents in previously unmedicated PD subjects.