Extrastriatal 123I-FP-CIT SPECT impairment in Parkinson’s disease – the PPMI cohort

Neuropathological data and nuclear medicine imaging show extensive serotonergic impairment in Parkinson’s disease (PD). We undertook a case-controlled analysis of 123I-FP-CIT SPECT images to measure extrastriatal serotonergic transporters (SERT) in PD using the Parkinson’s Progression Markers Initiative (PPMI) cohort. We included all PD (n = 154) and Control subjects (n = 62) with available 123I-FP-CIT SPECT imaging and high-resolution T1-weighted MRI for coregistration (PD: mean age 61.6 years, 62% male, disease duration 26 months, MDS-UPDRS III score 22). 123I-FP-CIT SPECT images were processed with PETPVE12 using an exploratory voxel-wise analysis including partial-volume effect correction. Linear regressions were performed in the PD group to assess correlations between region of interest 123I-FP-CIT uptake and clinical motor and non-motor impairment. Compared to Controls, PD exhibited an uptake reduction in bilateral caudate nucleus, putamen, insula, amygdala and right pallidum (family-wise error (FWE)-corrected p <  0.05). While lower putaminal uptake on the contralateral side to clinically more affected side was associated with higher MDS-UPDRS III score (p = 0.022), we found a trend association between higher geriatric depression scale and lower pallidum uptake (p = 0.09). Higher SCOPA-AUT gastrointestinal subscore was associated with lower uptake in mean putamen and caudate nucleus (p = 0.01 to 0.03), whereas urological subscore was inversely correlated with mean caudate nucleus, putamen, and pallidum uptake (p = 0.002 to 0.03). REM sleep behaviour disorder screening questionnaire was associated with lower 123I-FP-CIT binding in caudate nucleus, putamen and pallidum (all p <  0.05). No significant association was found for Montreal Cognitive Assessment (all p > 0.45) or excessive daytime sleepiness (all p > 0.29). In addition to the well-established striatal deficit, this study provides evidence of a major extrastriatal 123I-FP-CIT impairment, and therefore of an altered serotonergic transmission in early PD.


Participants
As of 4th February 2019, 423 PD and 196 Controls (CTL) were enrolled in the PPMI study. For the present work, we included all subjects who had both 123 I-FP-CIT SPECT imaging and high-resolution 3 T T1-weighted magnetization-prepared rapid gradient echo (MPRAGE) MRI performed within 3 months. In addition, inclusion was limited to individuals aged over 40 years to ensure that only sporadic rather than genetic cases were taken into consideration. We also collected clinical data from the PPMI database, i.e. total scores of the MDS-UPDRS III scale (off-state, ≥6 h after dopaminergic treatment discontinuation) at baseline (within 3 months of scans) and at 1-year follow-up, as well as Hoehn and Yahr stage [17]. Non-motor impairment at baseline was assessed using measures of cognition (MoCA) [13], depression (GDS) [12], dysautonomia with SCOPA-AUT scale [14], including gastrointestinal (SCOPA-GI, 7 items) and urinary (SCOPA-URO, 6 items) subscores, as well as excessive daytime sleepiness (ESS) [15] and REM-sleep behaviour disorder (RBDSQ) [16].

MRI imaging
All patients were scanned on a 3 T Siemens Magnetom TRIO Tim MRI scanner (Siemens Medical Solutions, Erlangen, Germany) using a 32-channel head coil. MPRAGE structural T1-weighted images were acquired with the following parameters: TR = 2300 ms, TE = 3 ms, flip angle = 9°, 256x256x240 mm matrix and voxel size = 1 mm3 (isotropic).

ROI-based analyses
Uptake extraction of the significant striatal and extrastriatal ROIs found in the voxel-wise analysis was performed with PETPVE12 [18] for correlation with clinical scales. After SPECT/MRI coregistration and MRI segmentation described above, we proceeded to ROI uptake extraction using geometric transfer matrix (GTM), including partial volume effect (PVE) correction. SPECT images were smoothed using an 8 mm full width at halfmaximum (FWHM) Gaussian kernel. The Desikan-Killiany was used for ROI uptake extraction [19] (Fig. 1). Semiquantitative values were obtained using the occipital lobe as the reference (REF) for DAT-rich ROIs (caudate nucleus, putamen and pallidum) or cerebellar grey matter for SERT-rich regions (amygdala and insula) [20] using the specific/non-specific binding ratio (SBR) as SBR = (Uptake ROI -Uptake REF )/Uptake REF .

Statistical analysis
Statistical analyses were performed with Stata Version 14.2 software (College Station, TX). Continuous variables were assessed for normality with plotted histograms and Shapiro-Wilk test. We used t-test or Mann-Whitney U test (MWU) for continuous variables with a normal and non-normal distribution, respectively. χ 2 test was performed for categorical variables.
Correlation analyses were performed for PD using univariate linear regressions between clinical motor/nonmotor scales and ROIs with significantly reduced uptake in PD. Between-ROI correlations were also performed using the same covariates. When assessing a correlation with a motor scale (MDS-UPDRS III), we used age, sex, antidepressant medication and centre as covariates and striatal ROIs uptake on the contralateral side to clinically more affected side according to MDS-UPDRS III lateralising items. These included appendicular rigidity, all items of bradykinesia (finger tapping, hand movements, pronation-supination movements of hand, toe tapping, leg agility), postural and kinetic tremor of hands. Axial items included speech, facial expression, rigidity of the neck, arising from chair, gait, freezing of gait, postural stability and posture.
For non-motor scales (GDS, MoCA, SCOPA subscores, ESS and RBDSQ), MDS-UPDRS III score was added as an additional covariate to adjust for motor severity and the mean (average of left and right) ROI uptake was used as the independent variable.

Patients demographics
According to our inclusion criteria (subjects aged ≥40 years, available SPECT and MPRAGE MRI performed within 3 months interval), 158 early PD and 63 CTL from 12 different medical centres were eligible for the present work. During the image preprocessing, 4 PD and 1 CTL subjects were excluded because of technical issues related to SPECT/MRI coregistration. Eventually, 154 PD and 62 CTL were included for group comparisons. Demographics and inclusion flowchart are available in Table 1

Voxel-based 123 I-FP-CIT SPECT analyses
Compared to the CTL group, PD subjects showed decreased uptake in a first cluster (total 1109 voxels (vx), peak level T-score 16.3) including right insula, putamen, amygdala and pallidum. The second cluster (total 1195 vx, peak level T-score 14.4) included left insula, putamen and caudate nucleus and amygdala. A third smaller cluster (140 vx, peak level T-score 7.4) included right caudate nucleus (all FWE-corrected p < 0.05) (Fig. 3).

Correlation analysis of clinical scales and ROIs 123 I-FP-CIT uptake
ROI-based uptake values of striatal (caudate nucleus and putamen) and extrastriatal (pallidum, insula and amygdala) regions of interest are available in Table 2. Strong correlations were observed between caudate nucleus, putamen and pallidum 123 I-FP-CIT uptake (all p < 0.001), while insula binding correlated with caudate nucleus and putamen binding (both p < 0.001) but not with pallidum (p = 0.83). Uptake in the amydala correlated with putamen (p = 0.004) and insula (p < 0.001), but not with caudate nucleus (p = 0.14) or pallidum (p = 0.20). Similar findings were observed when studying the subgroup of PD patients with GDS ≥ 5. However, when assessing only PD patients with mild cognitive impairment (MOCA ≤25, n = 30), we did not observe any significant correlation between striatal Higher MDS-UPDRS III was associated with lower 123 I-FP-CIT uptake in the putamen contralateral to clinically more affected side (p = 0.022, coefficient = − 9.2, 95% confidence interval (CI) -17 to − 1.4). No significant correlation was found between ROIs uptake and MoCA (all p > 0.45). We found a trend association between higher GDS and lower mean 123 I-FP-CIT pallidal uptake (p = 0.08, coefficient − 0.65, 95% CI − 1.39 to 0.09). Higher SCOPA-GI correlated with lower mean putamen (p = 0.01, coefficient − 2.2, 95% CI − 3.9 to − 0.5) and  No significant correlation was observed between ESS and ROIs 123 I-FP-CIT uptake (all p > 0.29). However, we observed thatamong PD participantshigher RBDSQ was associated with lower 123 I-FP-CIT uptake in caudate nucleus, putamen and pallidum (all p < 0.05). All correlations are available in Table 3.
Regarding antidepressant medication, we observed that PD patients on SSRI/SNRI treatment had non-significantly lower 123 I-FP-CIT uptake values in caudate nucleus, putamen, pallidum and insula compared to patients without antidepressants (4-8% reduction, all p > 0.11, MWU). Conversely, 123 I-FP-CIT binding in the amygdala was slightly higher in patients on SSRI medication (9% increase, p = 0.10, MWU). SSRI/SNRI treatment was added as a dichotomised variable in the regression analyses and did not show any significant effect on striatal or extrastriatal uptake (all p > 0.10).

Discussion
In the present work, we showed that in addition to a well-known decrease in 123 I-FP-CIT uptake involving the striatum, early PD subjects exhibit a significant uptake reduction in extrastriatal regions including the pallidum, amygdala and insula. As extrastriatal 123 I-FP-CIT binding mainly derives from SERT [2], we took advantage of the same scanning session of a large PD cohort to confirm early binding changes in regions outside the striatum compared to similarly-aged Controls. 123 I-FP-CIT impairment in pallidum was recently observed by Lee et al. [21]. 11 C-PE2I and 11 C-DASB PET studies showed that while the lateral part of the pallidum (globus pallidum externus) has a similar proportion of serotonergic and dopaminergic terminals, the medial part (globus pallidus internus) is mainly a serotonergic nucleus [20]. In addition, our PD subjects exhibited an impaired extrastriatal uptake in the insula and amygdala. These findings confirm recent data by Pilotto et al. showing decreased 123 I-FP-CIT uptake in insula, thalamus and cingulate in both PD and dementia with Lewy bodies [9]. Moreover, they confirm previous neuropathological evidence and in vivo 11 C-DASB-PET and 123 I-FP-CIT SPECT studies showing an altered serotonergic uptake for PD in the insula [6,8], providing additional value for 123 I-FP-CIT SPECT, a major diagnostic imaging tool in daily clinical practice. Findings of a reduced 123 I-FP-CIT uptake in the amygdala are also in line with post-mortem studies using chromatography and enzyme-linked immunosorbent assay showing  that PD subjects had reduced dopamine and noradrenaline levels in the amygdala [22]. Our correlation analysis confirmed that motor impairment as measured by MDS-UPDRS III total score was associated with decreased putaminal 123 I-FP-CIT uptake [23]. It might seem surprising that MDS-UPDRS III was not more strongly correlated with striatal ROIs 123 I-FP-CIT uptake. However, a recent psychometric assessment of MDS-UPDRS part II and III scales in PD subjects enrolled into the PPMI cohort showed an important floor effect, which can explain the moderate association for our cohort [24].
Clinical association with an extensive serotonergic deficit may include apathy, emotional disturbances, depression, Table 3 Correlations between clinical motor and non-motor scales and regional 123I-FP-CIT uptake in PD cognitive deficits, and dysautonomic manifestations in PD, all of which being potentially present early in the disease course or even in its prodromal phase [25]. Previous studies did not show any significant correlation between striatal 123 I-FP-CIT binding and non-motor symptoms such as fatigue, depression and excessive daytime sleepiness in PD [26]. Using a ROI-based approach to assess the contribution of extrastriatal ROIs, Qamhawi et al. observed a reduced 123 I-FP-CIT binding in the raphe nucleus in early PD patients from the PPMI initiative compared to Controls (although only 12.5% of PD had values below 1.5 standard deviation). They also found that raphe nucleus binding was associated with rest tremor amplitude, constancy and severity, but not with nonmotor symptoms such as depression, REM-sleep behaviour disorder and excessive daytime sleepiness [27].
While a trend association was found between GDS and pallidal 123 I-FP-CIT uptake in the present study, we did not find a significant correlation between SERT-rich regional uptake and cognitive impairment as measured with MOCA. One reason might be that other monoaminergic pathways are involved in the pathophysiology of cognitive impairment [23] and that larger samples would be necessary to tackle such complex processes. In addition, our PD cohort mainly consisted of early subjects with relatively preserved cognition (mean MoCA 27.4), so admittedly, a significant correlation between SERT binding and depression/cognitive scales could have been observed in cohorts with a broader range of impairment [28].
Interestingly, correlations between striatal and extrastriatal regions binding among PD patients showed that caudate nucleus and putamen 123 I-FP-CIT uptake impairment was associated with concomitant reductions in insula and amygdala binding in the full cohort analyses and also in the PD subgroup with GDS ≥ 5. However, when assessing these potential associations in the subgroup with mild cognitive impairment (MOCA ≤25, n = 30), we did not observe such correlations, except for a trend association between putamen and amygdala uptake (p = 0.09). The smaller subgroup could explain the lack of significant association. In addition, changes in striatal and extrastriatal binding can be of a different magnitude in patients with cognitive impairment, although we did not find any significant relationship between extrastriatal binding and cognitive impairment.
Our early PD group did not show any major signs of excessive daytime sleepiness, with ESS scores similar to Controls (p = 0.68). At variance with Yousaf et al. [29], no significant correlation between ESS and striatal 123 I-FP-CIT uptake was observed in PD (all p > 0.29). Differences in the inclusion criteria could explain the apparently opposite findings. In fact, Yousaf et al. [29] performed between-group comparisons of PD with or without excessive daytime sleepiness, whereas we assessed correlations with striatal/extrastriatal 123 I-FP-CIT using a cohort with relatively low ESS score.
In contrast, we observed that PD group had a significantly increased RBDSQ score (p < 0.001) relative to Controls, with 51/154 PD (33.1%) having RBDSQ ≥5. Moreover, higher RBDSQ was associated with lower 123 I-FP-CIT binding in caudate nucleus, putamen and pallidum (all p < 0.05). These findings are in line with Pagano et al., who showed that RBD was associated with faster striatal dopaminergic decline [30].
Finally, a significant association was found between SCOPA-GI subscore and caudate/putamen uptake. This confirms previous observations from Hinkle et al. [31], who described an association between SCOPA-GI score (especially constipation items) and both regions. Additionally, we observed that the SCOPA-URO subscore was also negatively correlated with striatal uptake. This is in line with previous findings from Kim et al. [32] who recently found a correlation between putamen uptake and SCOPA-URO subscore also including the PPMI cohort. For both studies from Hinkle and Kim, SPECT/ MRI coregistration was not performed. While our findings are of interest, they should be interpreted with an important caveat in mind. In fact, while bowel motility in late-life has been associated with postmortem neuron density in the substantia nigra [33], a causal relationship between presynaptic striatal dopamine uptake and gastrointestinal impairment in early PD would be hasty. Indeed, one hypothesis would be that dysautonomia would appear concurrently to striatal dopaminergic degeneration and not necessarily be driven by it.
Studies on rats and humans have shown conflicting results regarding the effect of antidepressant medication (SSRI/SNRI) on 123 I-FP-CIT uptake values. In rats, subchronic administration of fluvoxamine was not associated with significant changes in striatal 123 I-FP-CIT uptake [34]. In humans, higher binding in striatal regions and lower binding in extrastriatal regions was observed after acute treatment of paroxetine [35], while chronic treatment was associated with lower striatal binding [36], which is in line with our findings.
The present study has several major strengths: first, it is based on a large cohort of well-characterized PD and CTL subjects who underwent extensive clinical motor and non-motor evaluation. In addition, we included subjects whose SPECT was acquired within 3 months of a high-resolution anatomical MRI in order to proceed to MRI/SPECT coregistration and to provide PVE-corrected results. Although our included subjects represent about half of the total PD subjects in the PPMI cohort, given the stringent inclusion criteria we applied for analyses purposes, we believe this provides a major insight into the pathophysiology of monoaminergic degeneration in PD.
The present work also presents some limitations. As this is the case in similar clinical studies, diagnoses are not based on neuropathology, so we cannot exclude diagnostic misattribution, especially since some PD cases were enrolled at a very early stage (10 subjects with < 6-month disease duration). In addition, SPECT acquisition was performed 3-4 h after 123 I-FP CIT injection, which is the ideal timeframe for DAT evaluation, whereas the recommended time window for extrastriatal SERT is 2-3 h [2]. Nonetheless, due to a slow 123 I-FP-CIT washout, we expect SERT binding to be relatively stable at 3-4 h [37].

Conclusion
We presented evidence of a widespread 123 I-FP-CIT extrastriatal impairment in early PD, spanning the amygdala and insular region, which is in keeping with neuropathological and 11 C-DASB PET imaging. As 123 I-FP-CIT SPECT imaging is used daily to confirm nigrostriatal degeneration, these results bring novel evidence for a potential role in helping to discriminate PD from nondegenerative conditions by assessing extrastriatal uptake. Further studies are warranted to assess whether atypical parkinsonian syndromes harbour similar extrastriatal 123 I-FP-CIT impairment, therefore confirming complex monoamine transmission abnormalities in degenerative parkinsonisms.