Identification of multiple system atrophy mimicking Parkinson's disease or progressive supranuclear palsy

Abstract

We studied a subset of patients with autopsy-confirmed multiple system atrophy who presented a clinical picture that closely resembled either Parkinson's disease or progressive supranuclear palsy. These mimics are not captured by the current diagnostic criteria for multiple system atrophy. Among 218 autopsy-proven multiple system atrophy cases reviewed, 177 (81.2%) were clinically diagnosed and pathologically confirmed as multiple system atrophy (i.e. typical cases), while the remaining 41 (18.8%) had received an alternative clinical diagnosis, including Parkinson's disease (i.e. Parkinson's disease mimics; n = 16) and progressive supranuclear palsy (i.e. progressive supranuclear palsy mimics; n = 17). We also reviewed the clinical records of another 105 patients with pathologically confirmed Parkinson's disease or progressive supranuclear palsy, who had received a correct final clinical diagnosis (i.e. Parkinson's disease, n = 35; progressive supranuclear palsy-Richardson syndrome, n = 35; and progressive supranuclear palsy-parkinsonism, n = 35). We investigated 12 red flag features that would support a diagnosis of multiple system atrophy according to the current diagnostic criteria. Compared with typical multiple system atrophy, Parkinson's disease mimics more frequently had a good levodopa response and visual hallucinations. Vertical gaze palsy and apraxia of eyelid opening were more commonly observed in progressive supranuclear palsy mimics. Multiple logistic regression analysis revealed an increased likelihood of having multiple system atrophy [Parkinson's disease mimic versus typical Parkinson's disease, odds ratio (OR): 8.1; progressive supranuclear palsy mimic versus typical progressive supranuclear palsy, OR: 2.3] if a patient developed any one of seven selected red flag features in the first 10 years of disease. Severe autonomic dysfunction (orthostatic hypotension and/or urinary incontinence with the need for a urinary catheter) was more frequent in clinically atypical multiple system atrophy than other parkinsonian disorders (Parkinson's disease mimic versus typical Parkinson's disease, OR: 4.1; progressive supranuclear palsy mimic versus typical progressive supranuclear palsy, OR: 8.8). The atypical multiple system atrophy cases more frequently had autonomic dysfunction within 3 years of symptom onset than the pathologically confirmed patients with Parkinson's disease or progressive supranuclear palsy (Parkinson's disease mimic versus typical Parkinson's disease, OR: 4.7; progressive supranuclear palsy mimic versus typical progressive supranuclear palsy, OR: 2.7). Using all included clinical features and 21 early clinical features within 3 years of symptom onset, we developed decision tree algorithms with combinations of clinical pointers to differentiate clinically atypical cases of multiple system atrophy from Parkinson's disease or progressive supranuclear palsy.

Keywords: Parkinson’s disease; autonomic dysfunction; multiple system atrophy; progressive supranuclear palsy; red flag.

Figure 1

Study design of the present study.

Figure 2

Decision tree analysis: Parkinson’s disease mimic versus typical Parkinson’s disease. Using all clinical features examined (75 items) outlined in Tables 2 and 3, Supplementary Table 3, and Supplementary Figs 1 and 2, decision tree analysis was performed to distinguish Parkinson’s disease mimics from typical Parkinson’s disease. The decision tree algorithm systemically selects the most determinant feature in the diagnostic process for the present cohort. It suggests which clinical feature should be heeded at the time of examination and/or in the future. Percentages indicate the probability of the underlying pathology [atypical MSA (Parkinson’s disease mimics) or typical Parkinson’s disease].

Figure 3

Decision tree analysis: PSP mimic versus typical PSP (PSP-P). Using all clinical features examined (62 items) outlined in Tables 4 and 5, and Supplementary Table 4, decision tree analysis was performed to distinguish PSP mimics from typical PSP (PSP-P). The decision tree algorithm automatically chooses the most determinant feature in the diagnostic process for the present cohort. It suggests which clinical feature should be heeded at the time of examination and/or in the future. Percentages indicate the probability of the underlying pathology [atypical MSA (PSP mimics) or typical PSP].