In this study, we quantified the expression of 2 previously characterized disease-related metabolic patterns, the MSARP and PDRP, in a sizeable cohort of patients with MSA, and compared the resulting network values to those from a group of patients with PD with a similar degree of motor disability. As expected for an MSA biomarker, MSARP expression was elevated in patients diagnosed clinically with this disorder, while PDRP expression was normal in the same subjects. Analogously, the patients with PD were characterized by elevated PDRP expression and normal MSARP scores. The specificity of the group differences seen for the 2 networks, and their distinctive relationships with clinical disability ratings, underscores the “disease-relatedness” of the MSARP and PDRP as functional imaging biomarkers.
Natural history studies in MSA have found that the time from symptom onset to death varies widely between patients with MSA with survival ranging from 3 to 15 years.18
By contrast, functional disability is closely associated with the severity of motor symptoms,16
and standardized rating scales are currently being used to measure disease progression in MSA experimental trials.19–21
Given the variability inherent in such clinical descriptors, alternative quantifiers of disease severity/progression may enhance the accuracy and speed of experimental trials of new MSA therapies. In this study, cross-sectional analysis of the MSA cohort indicated that the expression of the corresponding disease-related pattern increased in patients with longer symptom duration and worse motor disability. Similarly, the 2 patients with MSA who underwent longitudinal imaging studies exhibited increasing MSARP expression over time, while PDRP expression fluctuated within the normal range. These data suggest the MSARP is likely to be a disease-specific and clinically sensitive biomarker of MSA progression. Nonetheless, a systematic longitudinal study of early-stage MSA is required prior to adopting this network as an outcome measure in clinical trials.
The subjects with MSA included in this study represent patients who initially presented to a movement disorders specialist with an uncertain parkinsonian diagnosis, including patients with early symptoms. Indeed, 30% of the patients with MSA had symptoms for 2 years or less at the time of imaging. Certainly, objective disease biomarkers are most valuable in patients who are early in the disease course, as this is the target population for inclusion into clinical trials. However, diagnostic accuracy can be challenging in such subjects. Therefore, we took several measures to verify the diagnosis of MSA in this cohort. Clinicopathologic studies suggest the sensitivity for initial clinical diagnosis in MSA is comparatively low (22%–56%).22,23
Nonetheless, after follow-up by a movement disorders specialist, the sensitivity for an MSA diagnosis is as high as 88%–100%.23,24
For this reason, we included only patients who both met diagnostic criteria for MSA23
on the last evaluation after PET and in whom the clinical diagnosis was confirmed by a movement disorders specialist after at least 6 months, with an average of 2.1 ± 0.5 years of follow-up after PET. In addition, in 9% of the patients with MSA included in our cohort, the diagnosis was confirmed by postmortem examination, providing further diagnostic verification.
As previously reported, we found that MSA was associated with significant localized reductions in basal ganglia and cerebellar metabolic activity, which were not present in patients with PD.13,25
However, in contrast to the network-based MSARP measurements, these regional changes exhibited only a weak correlation with clinical motor ratings. While patients with MSA can have clinical parkinsonism without ataxia, and vice versa, data from a recent clinicopathologic study from 100 patients with MSA suggest that basal ganglia and cerebellar regional pathology do not exist in isolation.26
Furthermore, although the putamen and cerebellum are key elements of the MSARP topography, this metabolic network includes significant contributions from other brain areas.1
Thus, this spatial covariance pattern is likely to be more representative of the widespread neurodegenerative changes that underlie this disorder. We note that for our primary analysis, patients with MSA-P and MSA-C were analyzed together as one pathologic disease group. However, when these clinically defined subgroups were considered separately, only the patients with MSA-P exhibited significant reductions in putamen metabolism—and neither subgroup displayed significant metabolic reductions in the cerebellum. Indeed, despite the modest degree of regional change seen in these patients, both MSA subgroups exhibited substantial MSARP elevations compared to the PD and normal control groups. In aggregate, these observations are consistent with the notion that network biomarkers based on whole brain imaging data provide more robust descriptors of neurodegenerative processes than measurements derived from one or more isolated brain regions.4
To evaluate the specificity of the relationships between these metabolic networks and disease severity, we correlated MSARP and PDRP subject scores with corresponding motor disability ratings obtained at the time of imaging. Indeed, the correlations between network expression and motor ratings were observed to be fundamentally different in the MSA and PD patient groups. In the MSA group, we noted a correlation with higher MSARP expression in patients with more severe motor dysfunction. An analogous correlation was evident in the PD group, with greater expression of the relevant disease-related pattern (i.e., PDRP) in the more advanced patients. In keeping with the disease specificity of the MSARP, the expression of this pattern in patients with PD failed to correlate with individual motor ratings.
Interestingly, a significant clinical–network correlation was also observed in the MSA group, in which higher (more severe) motor ratings were associated with lower PDRP values. Of note, PDRP expression was normal in the MSA group and individual differences in the corresponding network values were relatively small. This is consistent with our observations in the 2 MSA patients with longitudinal imaging, in whom PDRP scores fluctuated within the normal range over the 4–7 years of follow-up. By contrast, in the MSA group, concurrently measured MSARP scores were abnormally elevated and increased with worsening motor symptoms. In this vein, the magnitude of the slope of the line correlating clinical ratings with MSARP scores was twice that for the corresponding correlation with declining PDRP values.
Finally, we note that there was no interaction effect between the 2 sets of network values in their prediction of the clinical severity ratings. Thus, the distinct clinical–network correlations observed for the 2 patterns in the MSA cohort are unlikely to represent the divergent effects of disease on basal ganglia glucose metabolism.13,15
This point is supported by an exploratory analysis in which we prospectively measured MSARP and PDRP expression in the individual scan data following the removal of the putamen using an anatomically standardized volume of interest mask.27
We found that the resulting pattern scores in the MSA group continued to correlate with the motor ratings (MSARP: r
= 0.42, p
= 0.02; PDRP: r
= −0.51, p
= 0.004). Similarly, a priori removal of the putamen from the PD scans did not substantially alter the clinical–network correlations observed in this group (PDRP: r
= 0.60, p
= 0.005; MSARP: r
= 0.10, p
= 0.69). An interesting possibility is that the correlation observed between decreased PDRP expression and increased motor ratings in patients with MSA is attributable to a predominance of nigral over striatal cell loss early in the disease process. While speculative, such an effect is consistent with the transient presence of a dopaminergic treatment response in some patients with early-stage MSA.18
The findings we describe support the proposition that network analysis of rest-state metabolic imaging data can provide robust, useful biomarkers for the diagnosis and assessment of patients with neurodegenerative disorders.3,28
Nonetheless, longitudinal imaging studies in patients with early MSA will be necessary to confirm the relationship between pattern expression and disease progression, and to quantify the annual rate of network change in these subjects.