Introduction
Parkinson’s disease (PD) is a neurological condition characterized by degeneration of dopaminergic nigrostriatal neurons, and by the presence of cytoplasmic protein deposits in the affected neuronal pathways [
1]. A role for inflammatory processes in PD is supported by evidence from studies of activated microglia [
2,
3] and inflammatory cytokines [
4] in post-mortem brain tissue from PD patients. However, the relationship between inflammation and the neurodegenerative and behavioural features of PD remains poorly understood.
The 18 kDa translocator protein (TSPO), is known to be expressed in microglia cells, and can be quantified using PET and TSPO radioligands, and is thus a putative biomarker of neuroinflammation in vivo [
5]. Among the TSPO PET radioligands developed, [
11C]PK11195 has been the most widely used in applied studies [
6]. However, [
11C]PK11195 shows low specific binding to TSPO, which limits its use in the assessment of brain microglia in vivo. For this reason, several new TSPO radioligands have been developed with improved signal to noise ratio, including [
11C]PBR28, [
11C]DPA713, [
18F]FEPPA, and [
11C]DAA1106 [
6].
Previous PET studies using [
11C]PK11195 as a marker of activated microglia have provided support for elevated brain TSPO in PD patients [
7‐
10]. However, these findings have not been consistently confirmed [
11]. Two studies combining PET measurement of TSPO and markers of dopaminergic neurodegeneration in the same patients have yielded contrasting findings regarding the relationship between these measures [
7,
8]. Moreover, findings in subsequent studies using second generation radioligands have been inconclusive. Whereas elevated TSPO binding was found in a study using the radioligand [
11C]DPA713 [
12], this finding was not supported by investigations using the radioligand [
18F]FEPPA for assessment of brain TSPO binding in PD patients [
13,
14]. The inconclusive findings warrant further study of TSPO in PD using second generation PET radioligands and assessment of the relationship between TSPO binding and dopaminergic pathology in PD.
In the present investigation, the second generation radioligand [11C]PBR28 was used for quantitative evaluation of brain TSPO in 16 control subjects and 16 PD patients. To analyse the relationship between brain TSPO binding and dopaminergic pathology, PET measurements of the dopamine transporter (DAT) were undertaken using the radioligand [18F]FE-PE2I.
Discussion
Neuroinflammatory processes, involving expression of activated microglia, are considered to be of key relevance for processes underlying brain pathology in PD; however, previous PET studies of the microglial TSPO protein in PD have yielded inconsistent results. In the present study 16 PD patients and 16 matched control subjects were examined using PET and the second generation TSPO radioligand [11C]PBR28 as a marker of microglial activation. The results confirmed the well-known effect of TSPO genotype on [11C]PBR28 binding, whereas no evidence was found for a difference in TSPO binding between PD patients and control subjects, or for a correlation between the binding to TSPO and DAT in PD patients.
The results corroborate previous findings with the second generation TSPO radioligand [
18F]FEPPA [
13,
14], but are inconsistent with results showing elevated TSPO binding using the radioligands [
11C]PK11195 [
7‐
10] or [
11C]DPA713 [
12] to study brain TSPO in PD. PBR28 has been found to display two distinct affinity sites, including a high-affinity binding site (
Ki 3.4 nM) and a low-affinity binding site (
Ki 188 nM [
37]), a feature also shared by [
18F]FEPPA [
38]. Such a bimodal binding pattern has not been observed with [
11C]PK11195 and is less evident for DPA713, which displays a difference of about fourfold in affinity for high-affinity than for low-affinity binding sites [
37]. Affinity of PBR28 for the high-affinity binding site has been found to be higher than that of PK11195 (
Ki 28 nM) and DPA713 (
Ki 15 nM), confirming a superior signal-to-noise ratio for detecting group differences in TSPO binding to the high-affinity binding site with [
11C]PBR28 than with the latter tracers. However, owing to low affinity for low-affinity binding sites, binding to these sites cannot be quantified using second generation radioligands, such as [
11C]PBR28 and [
18F]FEPPA. Thus, it cannot be excluded that findings of studies using [
11C]PK11195 as a radioligand reflect the PBR28 low-affinity binding site.
Given the progressive neurodegenerative nature of PD, inflammatory processes are expected to vary during the course of the disease. It may thus be suggested that the discrepant findings of the current and earlier investigations could reflect differences in disease severity between the groups of patients recruited for these studies. However, differences in disease severity are unlikely to explain the inconsistency in the findings of previous investigations, in which increased TSPO binding was found in patients examined during the early stage [
8,
9] or later stages [
7,
10,
12] of PD, and the findings of the more recent studies, in which increases in TSPO binding were not observed in PD patients with various disease durations and severities [
13,
14]. Nevertheless, it cannot be excluded that clinical or neurochemical heterogeneity related to the disease course between the groups of patients examined may account for the inconsistent findings of TSPO PET studies in PD.
Another possible explanation for the discrepant findings is the difference in methods for quantitative evaluation of brain TSPO employed in the studies. The present findings, similar to previously reported results [
13,
14], rely on the work-demanding use of the arterial plasma concentration as an input function, whereas the other studies are based on cluster analysis for definition of nondisplaceable radioligand binding. Because of the different methodology for PET data quantification and outcome measures (
VT or BP
ND) used for the assessment of TSPO binding, the findings cannot be directly compared.
Our findings obtained using [
18F]FE-PE2I as a radioligand show a marked reduction in nigrostriatal binding to the DAT in the PD patients examined. However, despite the marked reduction in dopaminergic innervation, no evidence was found for a change in brain TSPO binding in the same patients. The findings of the two earlier PET studies combining imaging of TSPO using [
11C]PK11195 as a radioligand and dopaminergic markers in PD were inconclusive. An inverse correlation between midbrain TSPO binding and dopaminergic innervation in the putamen was found in a group of drug-naive patients with early PD [
8], but this finding was not confirmed in a group of more severely affected patients [
7]. Taken together, the findings suggest that increased TSPO binding may parallel dopaminergic pathology during the early stages of the disease, but does not seem to be a marker of ongoing dopaminergic degeneration in PD.
Plasma protein binding measurements indicated differences in the estimates of the free fraction of [
11C]PBR28 between PD patients and control subjects. Given that differences in the plasma free fraction would be expected to yield biased estimates of group differences in cerebral TSPO binding when using
VT as an outcome measure, correction of
VT for the plasma free fraction has been suggested as a more suitable approach to the quantitative analysis of [
11C]PBR28 data [
39]. Accuracy in the determination of protein binding is limited by the sensitivity of the detection methods employed. However, the reliability of the estimates of plasma protein binding has been reported to be poor for [
11C]PBR28 [
19]. A correction for the free fraction of radioligand in plasma was consequently not applied in the present study.
To overcome methodological limitations in the measurement of plasma radioactivity, the use of ratio-based reference tissue methods has been proposed as an alternative for quantitative analysis of [
11C]PBR28 PET data [
35]. In the present study the DVR relative to the cerebellum was used as an alternative outcome measure. Using this method, regional DVR values were found to be statistically significantly lower in PD patients than in controls. In a recent test–retest study, DVR showed poor reliability and validity as an outcome measure for the quantitative assessment of cerebral [
11C]PBR28 binding [
40]. Because regional
VT values are highly correlated, normalizing these values by
VT for another region results in low residual variability and consequently high sensitivity to noise [
40]. For this reason
VT was considered as the preferred outcome measure for quantification of [
11C]PBR28 PET data in the present study.
PD patients and control subjects were examined in two studies conducted approximately 3 years apart. Methodological factors that could have varied between the two series should be considered as possible limitations of the investigation. Molar radioactivity for [
11C]PBR28 was found to be higher in PD patients than in controls, resulting in a higher injected mass in control subjects. However, the low masses of PBR28 injected (0.2–1.5 μg, 7–53 pmol/kg) would be expected to have had only a minor impact on
11C-PBR28
VT values. In support of this notion, no evident correlation was found between
VT values and molar radioactivity or injected mass of PBR28. In addition, based on examples with other tracers with nanomolar affinities for the target protein, such low masses have been predicted to induce less than 1% target occupancy when administered to human subjects [
41,
42]. For these reasons, the impact of the mass of PBR28 is considered to have been negligible when comparing
VT values between PD patients and controls.
The use of different MR scanners with magnetic field strengths of 1.5 T and 3 T in patients and control subjects, respectively, is another possible limitation of the study. To minimize the impact of MR image quality on PET data quantification, ROIs were defined in a standardized space, thus limiting the use of information from individual MR images to the spatial normalization of PET images. While different MR acquisition procedures are unlikely to have confounded the results of the PET quantitative analysis, the group comparison of brain morphometry data (Table
S1) should be interpreted with caution, as differences in magnetic field strength may have introduced bias in the estimates of regional brain volumes obtained by the methodology employed [
43]. A possible influence of partial volume effects resulting from a difference between groups in grey matter volumes cannot be excluded, although the homogeneous binding of [
11C]PBR28 across regions suggests that the impact of partial volume effects was minimal.
With the limitations inherent to the quantification of TSPO PET data, in vitro studies of post-mortem tissue allowing direct measurement of protein levels in the brain are of critical importance for interpreting findings obtained in vivo. Although in vitro studies have provided evidence for microglial activation in PD [
2,
3], investigation of TSPO radioligand binding is to our knowledge limited to a study of tissue samples from three PD patients [
44] that showed a marked increase, albeit with high within-group variability. Thus, the field would benefit from large-scale post-mortem investigations further addressing the topic of TSPO binding in PD.
While TSPO imaging is widely applied as a noninvasive marker of neuroinflammation, TSPO is not an ideal target for assessment of inflammatory processes. This protein is known to be expressed by multiple cell types other than activated microglia [
6]. Moreover, endogenous ligands including cholesterol and porphyrins [
45,
46] display high affinity for this protein. Such factors could account for the interindividual variability and discrepant findings observed in TSPO PET studies. Future development of tracers specifically targeting microglia activation is required to elucidate the role of inflammatory processes in neurodegeneration.
In conclusion, the present findings obtained using [11C]PBR28 PET do not support the hypothesis of elevated cerebral TSPO binding, or a relationship between TSPO binding and nigrostriatal degeneration, in PD. The results are consistent with previous findings obtained using the second generation radioligand [18F]FEPPA.