What is the accuracy of the clinical diagnosis of multiple system atrophy? A clinicopathologic study

Background: The presentation of symptoms for multiple system atrophy (MSA) varies. Because there are no specific markers for its clinical diagnosis, the diagnosis rests on the results of the neuropathologic examination. Despite several clinicopathologic studies, the diagnostic accuracy for MSA is unknown. Objectives: To determine the accuracy for the clinical diagnosis of MSA and to identify, as early as possible, those features that would best predict MSA. Design: One hundred five autopsy- confirmed cases of MSA and related disorders (MSA [n=16], non-MSA [n=89]) were presented as clinical vignettes to 6 neurologists (raters) who were unaware of the study design. Raters identified the main clinical features and provided a diagnosis based on descriptions of the patients' first and last clinic visits. Methods: Interrater reliability was evaluated with the use of κ statistics. Raters' diagnoses and those of the primary neurologists (who followed up the patients) were compared with the autopsy-confirmed diagnoses to estimate the sensitivity and positive predictive values at the patients' first and last visits. Logistic regression analysis was used to determine the best predictors to diagnose MSA. Results: For the first visit (median, 42 months after the onset of symptoms), the raters' sensitivity (median, 56%; range, 50%-69%) and positive predictive values (median. 76%; range, 61%-91%) for the clinical diagnosis of MSA were not optimal. For the last visit (74 months after the onset of symptoms), the raters' sensitivity (median, 69%; range, 56%-94%) and positive predictive values (median, 80%; range, 77%-92%) improved. Primary neurologists correctly identified 25% and 50% of the patients with MSA at the first and last visits, respectively. False-negative and -positive misdiagnoses frequently occurred in patients with Parkinson disease and progressive supranuclear palsy. Early severe autonomic failure, absence of cognitive impairment' early cerebellar symptoms, and early gait disturbances were identified as the best predictive features to diagnose MSA. Conclusions: The low sensitivity for the clinical diagnosis of MSA, particularly among neurologists who followed up these patients in the tertiary centers, suggests that this disorder is underdiagnosed. The misdiagnosis of MSA is usually due to its confusion with Parkinson disease or progressive supranuclear palsy, thus compromising the research on all 3 disorders.