In the present study, we found a high accuracy of 6-[18F]FDOPA and only a moderate accuracy of N1 evaluation in discriminating PS from Non-PS.
6-[18F]FDOPA is a well-established PET tracer that illustrates terminal dopa decarboxylase activity, providing a valuable tool for the identification of early PD [3]. The high sensitivity, specificity and accuracy in our study go in line with other studies, who reported sensitivities and specificities of 90–100% and 91–100%, respectively [19, 20]. However, presynaptic dopaminergic function is known to be impaired not only in IPS, but also in APS, thus resulting in a low specificity for discriminating both of them [21, 22]. This was underlined by the results of our semi-quantitative metabolism analysis, which revealed no substantial differences in striatal 6-[18F]FDOPA uptake between both entities, but a profound caudal-rostral gradient in the putamen on both hemispheres in IPS and APS. To address this issue, further studies regarding specific 6-[18F]FDOPA uptake patterns in the striatum are still required.
Nigrosome 1 is the largest cluster of dopaminergic cells in the substantia nigra pars compacta and can be visualized by high-resolution SWI MRI. In IPS patients, the otherwise physiological swallow tail appearance of the N1 is lost, which indicates nigral degeneration [23]. Schwarz et al. reported almost perfect sensitivity, specificity and accuracy of 95% [7] for N1 assessment to discriminate IPS patients from healthy controls. A recent meta-analysis confirmed the excellent diagnostic accuracy of N1 assessment [24]. However, in our study, N1 evaluation had a moderate sensitivity, specificity and accuracy in discriminating PS from parkinsonism without dopaminergic deficit and was inferior to 6-[18F]FDOPA. Those discrepancies are due to completely different patient cohorts: our study cohort comprised only of patients with movement disorders of initially unknown origin that were recruited from our highly specialized ambulatory movement disorders clinic of the department of neurology in a university setting. No healthy controls were examined, which somewhat impairs comparability to earlier studies. In the general population, the IPS to APS ratio and the PS to Non-PS ratio are very dissimilar to our study cohort. A case–control study from Perez Akly et al. compared N1 evaluation results of 16 patients with IPS and 16 patients with essential tremor [25]. The sensitivity and specificity was 94% and 75–88%, respectively, which is higher than reported in our data (83% and 66%, respectively). In our study cohort, 2 of 6 patients with ET showed a loss of N1 signal. A possible explanation to this phenomenon might be the relatively long disease duration of 7 and 10 years, respectively, that could have led to nigral cell damage without significant involvement of dopaminergic cells, thus producing false-positive results.
Intermodality agreement of 6-[18F]FDOPA and N1 evaluation was very good in patients with clinical IPS, but declined in patients with APS and non-parkinsonian movement disorders. The main reasons for this may be the different principle of depicting the nigrostriatal system, alongside with our highly preselected cohort of patients. While there are several 6-[18F]FDOPA studies that describe qualitative and quantitative ways to discriminate between IPS and APS, such as unilateral uptake loss in the posterior putamen [26] or the presence of a rostro-caudal uptake gradient [27], the N1 assessment on 3 T MRI in daily routine happens solely on a visual basis. To enhance the potential of N1 imaging, quantification methods, like quantitative susceptibility mapping combined with histogram analysis, may be promising in the future [28].
In almost every case of disagreement between both imaging modalities, 6-[18F]FDOPA would have suggested the correct diagnosis, according to the clinical gold standard. The pool of patients with a mismatch result was very heterogenous and did not differ significantly from the matching group of 35 patients in terms of clinical diagnosis, age, duration of disease, etc.
Our study patients presented to the PET/MR center with a wide range of disease durations. In order to address this issue, we compared imaging results of subjects that were scanned within short (≤ 2 years) or long (≥ 2 years) intervals after symptom onset. In both intervals, overall sensitivity, specificity and accuracy of 6-[18F]FDOPA PET appeared to be higher than that of N1 imaging. Performances of both modalities declined slightly with longer duration of disease, but remained to be fair in terms of intermodality agreement. The number of false-positive and false-negative cases was too small to postulate any significant differences between short and long scanning intervals. However, for any scanning time point after symptom onset, IPS identification accuracy was excellent in both 6-[18F]FDOPA PET and N1 imaging. This finding goes in line with an earlier study of Stezin et al., who reported high sensitivity, specificity and accuracy of N1 imaging in PD patients, compared to healthy controls, with 5.2 ± 3.9 years mean duration of disease [29].
Our patient cohort included several disease entities, which can alternatively be divided into synucleopathies, like IPS and MSA, or tauopathies, such as PSP, CBD and AD, respectively. It is known that presynaptic dopaminergic imaging of synucleopathies and tauopathies may have different regional patterns of metabolism. Our study results may indicate a potential advantage of N1 and 6-[18F]FDOPA PET accuracy in synucleopathies, but, due to the small sample size, this statement remains speculative. Furthermore, there are several studies that do not show any differences between detection rate of synucleopathies or tauopathies in 3 T SWI MR imaging [30,31,32,33].
This study has several limitations. First, the retrospective study design, which bears a possible bias in our highly preselected patient cohort. Second, a relatively small sample size per particular diagnosis, because of missing clinical information for many patients. Third, the definition of clinical diagnosis as the gold standard. The sensitivity and specificity of the clinical diagnosis are 88% and 68%, respectively, as confirmed by post-mortem findings in patients with IPS [34]. Although the clinical follow-up was at least six months in all of our patients, the results of 6-[18F]FDOPA PET and N1 evaluation could have influenced the final diagnosis, because our neurological specialists were not blinded to the scan results. To address this bias, we performed a secondary analysis, where four patients were excluded because their former clinical diagnosis of IPS had to be corrected to Non-PS after normal PET results, according to MDS diagnostic criteria. Overall statistical performances of both imaging modalities did not change significantly to our primary analysis. This may indicate that the superior performance of PET imaging is not a consequence of their inclusion in the MDS diagnostic criteria.
Our divergent imaging results of patients with APS could be explainable by neuropathological considerations: dopaminergic neurons are thought to degenerate by a "dying back" mechanism, i.e., the extent of terminal loss in the striatum is generally larger than the extent of neuron loss in the substantia nigra [35]. This could account for higher sensitivity and specificity of 6-[18F]FDOPA PET than that of N1 imaging. Yet, the relationship between imaging findings and neuropathology is not static. It is unclear what the 6-[18F]FDOPA PET abnormalities in APS mean on a patho-anatomical level, and whether N1 abnormalities really reflect neurodegeneration or other processes like glial reactions.
However, as presynaptic dopaminergic imaging is an established and reliable tool for assessing parkinsonism, its outcomes might be generally better accepted by clinicians, thus resulting in another possible bias towards it. To finally overcome these uncertainties, histopathological examination might be the only way, which is not feasible in clinical studies.
In summary, long-term prospective studies with a higher number of patients and healthy controls are required to confirm the findings of the present study and to assess disease progression and severity.