This study was designed to validate [
18F]FE-PE2I PET as a biomarker for dopaminergic neurodegeneration in non-advanced PD, through a cross-sectional study on its correlations with clinical symptom severity scores. To increase the statistical power compared with an earlier study [
21], we enlarged the sample size by including an additional cohort of PD patients with similar clinical characteristics. Patients in this cohort were evaluated with MDS-UPDRS-III in practically defined ‘OFF’ state on the day of the PET measurement so that correlations between motor scores and DAT availability were performed while patients were examined under the same conditions (OFF). Furthermore, correlations were performed using the MDS-UPDRS-III score including and excluding the tremor score, with the knowledge that bradykinesia and rigidity are symptoms more strongly associated with striatal DAT availability than tremor. Additionally, correlations with clinical measures were evaluated in functional subdivisions of the striatum.
Pattern in dopaminergic neurodegeneration
[
18F]FE-PE2I BP
ND data in the whole group of 41 PD patients were compared to those obtained in a group of 37 control subjects with similar age and sex distribution. To our knowledge, this is the largest reported cohort of PD patients examined with [
18F]FE-PE2I PET. We confirmed our earlier findings of a PD-related pattern of DAT deficit and extended the quantification also to the sensorimotor striatum. Overall, the findings of this study corroborate the knowledge that dopaminergic neurodegeneration in PD follows a retrograde pattern, affecting neuronal terminals first (striatum) before the cell bodies (substantia nigra) [
21]. Within the striatum, the posterior putamen (
sensorimotor striatum) is affected first, followed by the anterior putamen, caudate nucleus, and ventral striatum.
BPND in the substantia nigra
In the substantia nigra, there was no clear asymmetry of BP
ND between the less and more affected hemispheres. DAT density in the substantia nigra is much lower than the DAT in the striatal regions, and we have also previously shown that in the substantia nigra BP
ND estimates have lower reliability than in the striatal regions, making accurate measurement of DAT in this area challenging. Neuronal loss in the substantia nigra is mainly observed in the pars compacta, with the ventrolateral tier being relatively spared [
26]. The substantia nigra ROI template that we used in this study covers the entire region and is based on the distribution of uptake of [
18F]FE-PE2I [
22]. Therefore, it is possible that we were not able to detect an asymmetry of BP
ND because DAT was measured in the whole region rather than in the part of the substantia nigra that is most severely affected. Partial volume effect on the small area might have influenced the results as well. In a previous study with [
11C]PE2I PET in mild-to-moderate PD patients [
20], BP
ND was measured in subdivisions of the substantia nigra. Although it was found that the DAT deficit was larger in the ventral tier than the dorsal tier (
p < 0.001) [
20], no significant difference in BP
ND was observed between the more vs. less affected substantia nigra, in agreement with the findings of this study.
Correlations with clinical measures
In the publication of the first cohort [
21], no statistically significant correlation was found between BP
ND and disease duration, H&Y, or UPDRS motor score. Our improved sample size and clinical data collection protocol enabled a more powered assessment of the correlations.
In agreement with other previous DAT imaging studies [
12,
27], we observed a significant negative correlation between symptom duration and DAT availability measured with [
18F]FE-PE2I in the putamen, and more specifically the sensorimotor striatum. This correlation was not significantly influenced by the inclusion of the patient with a statistical outlier symptom duration (14 years). The relation between symptom duration and DAT availability in the putamen, caudate nucleus, and sensorimotor striatum was better described by exponential fitting than linear fitting. This non-linear relationship between symptom duration and dopaminergic deficit has been previously described in a large cohort of PD patients examined with the vesicular monoamine transporter type 2 PET radioligand [
18F]FP-DTBZ [
28]. The findings are not surprising, since it is known that clinical signs of Parkinson’s disease do not appear until about 50–70% of the nerve terminals are already lost [
26,
29,
30,
31,
32]. This might result in larger inter-individual variability of the dopaminergic deficit at the onset of the disease, with a more uniform decline as the disease progresses. The exponential correlation suggests that early on in the disease course, there is some degree of variability of DAT availability that probably reflects the fact that patients develop symptoms at different levels of dopaminergic deficit, probably due to different motor or dopamine reserve.
A significant negative correlation was found between the H&Y stage and [18F]FE-PE2I BPND in all nigrostriatal regions (caudate nucleus, putamen, and substantia nigra). The H&Y stage is a measure of the patient’s global motor disability. These findings suggest that the H&Y stage is the clinical measure that is most directly related to the overall nigrostriatal dysfunction, compared to the other clinical variables used.
A significant negative correlation between the MDS-UPDRS-III score and DAT availability in the putamen was only found when the tremor score was subtracted from the total score. It has to be highlighted, though, that one subject presented BP
ND values that were relatively high compared to their MDS-UPDRS-III score (Fig.
3), was identified as an outlier by the Grubb’s test, and indeed influenced the statistical results significantly. Methodological reasons for this subject’s relatively high BP
ND values were not found, and, therefore, the subject was not excluded from the analysis, despite being an outlier from a statistical point of view. Exploratory re-analysis with the removal of this subject gave correlations of BP
ND with total MDS-UPDRS-III scores in the putamen and sensorimotor striatum both with linear fitting (r
s between -0.43 and -0.56, p values between 0.006 and 0.05, see Additional file
2: Fig. S1a), as well as with exponential fitting (p-values < 0.05, see Additional file
2: Fig. S1b).
Previous [
18F]FE-PE2I studies in PD patients have reported no significant correlations between striatal DAT and UPDRS-III [
21,
33]. Possible explanations for these differences might be related to whether UPDRS-III was measured ‘ON’ or ‘OFF’ medication, or whether the motor score included also the tremor score or only the bradykinesia and rigidity scores. Two PET studies performed with the close analogue [
11C]PE2I found that striatal DAT-binding significantly correlated with total UPDRS-III (OFF) and bradykinesia-rigidity scores [
3] or with bradykinesia, rigidity, and axial symptoms, but not with tremor score [
20]. Overall, the findings of this study support the latter and suggest that the tremor score should be subtracted from the total MDS-UPDRS-III score in studies that intend to examine the correlation between DAT and motor symptoms.
There was no correlation of DAT availability in the substantia nigra with any of the clinical measures in patients examined ‘OFF’ medication, thereby not confirming our initial hypothesis. A possible explanation is that in the early stages of PD, there is still a relative sparing of neurodegeneration in substantia nigra, and the relationship might be more clearly expected later on in the disease process. Furthermore, a floor effect due to the low density of DAT could also be an explanation for the lack of correlation findings. However, in the combined group (n = 41), the correlation between BPND in the substantia nigra and H&Y stage became statistically significant likely due to increased sample size.