Since PD diagnosis is still mainly set on clinical assessments based upon the presence of characteristic features, an additional technical tool to support the clinically suspected diagnosis would be beneficial. Unfortunately, broadly available non-invasive neuroimaging tools are limited mainly to transcranial sonography which crucially depends on the rater’s expertise, whereas MRI is primarily performed to exclude other underlying pathologies.
For these reasons, we investigated MRI signal intensities of SN and GPI in a clinical routine setting using a clinical 1.5T MRI scanner with a standard protocol for acquisition and a simple ROI-based post-processing. We observed T2-hyperintense signal changes of both the SN and the GPI areas to be useful to support the diagnosis of PD, whereas controls did not show these features. Additionally, signal intensities of SN correlated with the GPI area of the same side. In addition to these observations in PD and consistent with previous MRI-based studies in atypical Parkinsonian syndromes, signal changes in GPI were confirmed which can also be potentially used to differentiate PSP patients from controls
. In order to serve as an additional technical tool in the diagnosis of PSP, these data have to be reproduced and validated using a dedicated study design with investigation of larger numbers of PSP patients.
The thresholds (using the “or” combination) allowed for a high sensitivity (86%) and NPV (90%) for the sample of all investigated pathologies. (Although MSA and CBS showed no significant differences in the comparison with the control group, most probably due to the low sample size, the complete patient group was compared with controls in order to encompass all investigated differential diagnoses.) Specificity and PPV were similar for SN and GPI investigations, while sensitivity and NPV were lower for SN than for GPI. The combined analysis showed similar statistical values for sensitivity and NPV as GPI alone. Taken together, we suggest the use of a combination of GPI and SN in order to consider supplementary information of SN rather than investigating GPI alone.
Our results are in general agreement with previous studies focusing on conventional MRI-based techniques in the diagnosis and differential diagnosis of PD, which revealed a broad spectrum of signal alterations in T1-, T2- or T2*-weighted imaging within the SN
[8, 25]. In contrast to most previous studies, however, we did not investigate iron or hemosiderin depletion within the SN, visualised by an increase of the T2 and T2* decay resulting in selective shortening of these relaxation times, but rather T2-hyperintensities as a correlate of selective and focal gliosis illustrating the upmost relevant neurodegenerative component of PD. In this context, our results partly reflect the well-known neuroanatomical projectional system of dopaminergic neurons of the SN to the GPI without the necessity of advanced imaging techniques such as tractography
. Beyond that, the presented altered signal intensities within the ROIs in the deep brain structures were derived on an individual basis without the need for any post-processing procedures.
What may be the pathophysiological correlate of the signal alterations observed in the SN and GPI of our PD patients? According to Braak et al., pathophysiological processes of PD can be divided into six stages each of which is marked by the continuous development of distinctive inclusion Lewy-bodies and evolves sequentially with the beginning at definite predisposed vulnerable sites, advancing from there in a predictable form through the grey matter of the brain
. Remarkably, the course of the disease can clinically be divided into a premotor (stages 1–3) and a symptomatic phase (stages 4–6), while the SN and the basal ganglia are pathophysiologically not impaired in the premotor stages. In short, in Braak stage 3, the continuous, self-propagating process reaches the mid- and forebrain including the pars compacta of the SN, followed by the anteromedial temporal mesocortex, the basal ganglia, the limbic system as well as the prefrontal cortex in stage 4 while in the last two stages, pathoanatomical involvement includes the association areas of the neocortex
[1, 13, 28–31]. On the basis of this knowledge, hyperintensities of SN and GPI, as observed in our MRI data, may be a morphological sign of neurodegenerative processes in the SN with consecutive degeneration of projection fibers to the corresponding GPI. This assumption may also be strengthened by recent reports on stage-dependent SN signal reduction as a putative marker of neuromelanin loss in PD patients in high-resolution T1-weighted imaging with magnetization transfer effect at 3 Tesla
. As an alternative theory, one could postulate local magnetic field disturbances as a correlate of iron accumulation in the SN, but we do not favour this hypothesis since T2*-weighted images did not show any substantial signal alterations of the SN in our sample that should be the case in iron deposition (data not shown). However, this lack of evidence for iron deposit might not only be a bias effect in our patient sample, but could also be related to the applied T2*-sequence parameters and also the field strength of our scanner. Here, recent studies report a traceable increase in susceptibility of the pars compacta in PD patients compared with controls by use of susceptibility mapping at ultrahigh field (7 Tesla) and advanced post processing methods
There are some additional limitations to the interpretation of these results. The major limitation is the application of the MRI data analysis to a sample of clinically diagnosed patients in a retrospective design. However, due to the explorative approach of this pilot study, the design with inclusion of clinically diagnosed patients in different (including advanced) clinical stages was chosen. Due to its design, the study cannot overcome the challenge of missed PD diagnosis since it aimed to provide imaging markers for identification of PD patients using clinical assessment as the gold standard rather than pathology. In the future, in order to test the value for early PD diagnosis or differential diagnostics, a prospective design in early affected patients or cases of clinical doubt or a retrospective design comprising patients with neuropathologically proven PD diagnosis has to be used. In addition, it has to be considered that a possible shortcoming may be the use of the 1.5T MRI scanner which implies a limited local resolution in comparison to higher field strengths although a 12-channel head coil was applied. With respect to data analysis, computerised analysis is necessary instead of pure visual inspection - however, the ROI-based processing can be performed within a few minutes on a standard workstation and does not require any advanced qualification beyond neuroanatomical identification of the basal ganglia structures. In addition, the thresholds used for the calculation of sensitivity and specificity were defined ex post.
In summary, although the demonstrated results are based on clinically diagnosed patients and therefore might include a selection bias, our MRI-based findings seem to have the potential to serve as an additional non-invasive neuroimaging-based technical tool within the diagnostic work-up for individual diagnosis of PD. Future studies should set focus on the potential of this ROI-based MRI analysis in early PD patients and might include further technical assessments of the patients such as other imaging modalities including dopamine transporter (DaT) scan. Furthermore, future studies might address the item of the comparison of different computerised MRI-based techniques in the imaging assessment of Parkinsonian syndromes including susceptibility mapping and the analysis of mean diffusivity values in DWI data.