Predicting cerebrospinal fluid α-synuclein seed amplification assay positivity in borderline hyposmia

Cerebrospinal fluid (CSF) α-synuclein seed amplification assays (SAAs) provide valuable biomarkers of underlying α-synuclein pathology. Although more severe olfactory impairment is associated with a higher likelihood of CSF α-syn SAA positivity in prodromal Parkinson's disease (PD), a simple tool to enrich candidates for CSF biomarker testing is lacking, particularly among individuals with borderline hyposmia. We aimed to develop and internally validate an exploratory clinical prediction model for CSF α-syn SAA positivity in this subgroup.

A total of 83 participants with age- and sex-adjusted University of Pennsylvania Smell Identification Test (UPSIT) percentiles >15 and ≤ 20 were included from the Parkinson's Progression Markers Initiative (PPMI) Phase 2 prodromal cohort. A prediction model was developed using least absolute shrinkage and selection operator (LASSO) logistic regression with 16 candidate predictors and 10-fold cross-validation. Internal validation was performed using 1000 bootstrap resamples repeating the complete model-development procedure.

Among 83 participants, 43 (51.8%) were SAA-positive and 40 (48.2%) were SAA-negative. The LASSO model retained age, sex, REM Sleep Behavior Disorder Screening Questionnaire (RBDSQ) score, and Symbol Digit Modalities Test (SDMT) score. The apparent area under the receiver operating characteristic curve (AUC) was 0.927, and the optimism-corrected AUC was 0.896, with an optimism-corrected Brier score of 0.149.

This exploratory four-variable LASSO model demonstrated promising discrimination for CSF α-syn SAA positivity among individuals with borderline hyposmia. External validation in larger, independent cohorts is required before clinical application.