Data-Driven Subtypes of Multiple System Atrophy and Their Implications for Prognosis.

Background: While multiple system atrophy (MSA) presents with high heterogeneous motor and nonmotor symptoms, the associations between clinical phenotypes and prognosis are unclear. Objective: We aimed to evaluate clinical phenotypes of MSA using data-driven approach and measure the impact of phenotypes on survival and bedbound status. Methods: 193 MSA patients were recruited from Xuanwu Hospital Capital Medical University, whose history, motor and non-motor symptoms were examined using cluster analysis. Ninety-five participants were followed-up via telephone after a mean of 31.87 months. We employed Kaplan- Meier analysis to examine survival and performed Cox and logistic regression analyses to identify factors associated with survival and bedbound status. Results: We identified four clinical profiles of MSA: cerebellar symptom-dominant, sleep and mood disorder-dominant, rigid akinetic-dominant, and malignant diffuse. The overall median survival was 7.75 years (95% CI 7.19-8.31). After adjusting for years from symptom onset to diagnosis, age and sex, patients in the malignant diffuse and rigid akinetic-dominant clusters had greater risk of death than sleep and mood disorder-dominant cluster. Furthermore, patients in the malignant diffuse and rigid akinetic-dominant clusters had higher risk of being bedbound than cerebellar symptom-dominant cluster. Conclusions: The malignant diffuse and sleep and mood disorder-dominant were identified besides the two classical subtypes, parkinsonism, and cerebellar symptom-variant. Patients with rigid-akinetic motor profiles have a worse prognosis than cerebellar symptom-dominant profiles in general. Diffuse symptoms, especially postural instability, and cognitive alterations at diagnosis, indicate rapid functional loss and disease progression. The different profiles and prognoses might indicate varied underlying pathological mechanisms.