Using prodromal non-motor symptoms to predict Parkinson’s disease onset and motor phenotype

BACKGROUND: Non-motor symptoms are highly prevalent in prodromal Parkinson's disease (PD); however, their impact on PD trajectory remains largely unexplored. We aimed to assess whether prevalent prodromal non-motor symptoms could predict future motor phenotype and time-to-PD diagnosis. METHODS: We studied the prodromal cohort of the ongoing Parkinson's Progression Markers Initiative (n=958), which prospectively assesses individuals with prodromal PD features (genetic: n=361, hyposmia: n=298, rapid eye movement behaviour disorder: n=136, combination: n=163) with up to 10 years of follow-up. The presence of prevalent prodromal symptoms was defined by evidence-based cut-off scores. In unmedicated or OFF-state PD converters (total n=52), binary logistic regression models established whether these predicted non-tremor-dominant (n=35) and tremor-dominant (n=17) motor phenotypes at diagnosis. Cox proportional hazards models determined whether identified prodromal symptoms predicted a shorter time-to-phenoconversion across all PD converters (n=59) and non-converters (n=343). Both models adjusted for age and sex. RESULTS: Prodromal anxiety and hyposmia were each associated with an increased risk of subsequent non-tremor-dominant PD, compared with other motor phenotypes (adjusted OR=4.45, 95% CI 1.34 to 15.27 and adjusted OR=3.90, 95% CI 1.01 to 15.16, respectively). Concurrent prodromal anxiety and hyposmia predicted an increased risk of PD phenoconversion over time (HR=4.93, 95% CI 2.71 to 8.98). CONCLUSION: In this exploratory analysis, individuals with prodromal hyposmia and anxiety phenoconverted to PD sooner and more often had a non-tremor-dominant phenotype, potentially reflecting more widespread pathology or specific pathophysiology underlying these symptoms. This may improve phenotyping prodromal PD and stratifying poorer prognostic trajectories for earlier and more personalised management.