Objective: To assess the prognostic relevance and predictive ability of a PD subtype classification system generated in LABS-PD cohort.

Background: PD subtype classification systems have a potential value for the study of epidemiology, pathogenesis and therapeutic development in PD. Prognostic relevance and ability to predict future membership to a PD subtype are key aspects for the validation of a PD subtype classification system. These attributes have been rarely explored in previously reported PD subtypes.

Methods: To identify PD subtypes, we conducted cluster analyses in the LABS-PD cohort, a longitudinal cohort of 461 PD patients enrolled in the PRECEPT randomized controlled trial within two years of diagnosis. Prognostic relevance for each subtype was assessed over two years of follow-up after definition of PD subtypes after at least 5 years from diagnosis. We used a linear multiple regression model with the change in total score of the Schwab & England ADL (SE-ADL) and PDQ-39 as prognostic outcomes. We used a logistic multiple regression model with stepwise variable selection to determine the set of variables measured at the time of enrollment in the PRECEPT study that most accurately predicted membership to a PD subtype defined five years later.

Results: We identified a three-cluster solution. Clusters were named as follows: “Tremor predominant/benign” (n = 253), “Predominance of motor complications” (n = 183), “Rapid progression with axial problems and depression” (n = 25). For prognostic relevance, the PD subtype variable explained 3.1 (PDQ-39) and 6.9 (SE-ADL) % of the model variance, with greater discrimination among PD subtypes for SE-ADL. The combination of age at diagnosis, UPDRS Motor Score, Beck Depression Inventory score and “classical motor subtypes (TD/PIGD/indeterminate)” at baseline predicted with good accuracy the membership to PD subtypes defined in LABS-PD five years later: 1) “Predominance of motor complications” vs. “Tremor predominant/benign” (AUC=0.72), and 2) “Rapid progression with axial problems and depression” vs. “Tremor predominant/benign” (AUC=0.70).

Conclusions: PD-subtypes generated in LABS-PD have prognostic value. Baseline clinical and demographic characteristics predict PD-subtype membership at 5 years with good accuracy. External validation in similar cohorts required.

« Back to 2017 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/prognostic-relevance-and-predictive-ability-of-data-driven-pd-subtypes-in-the-longitudinal-and-biomarker-studies-in-parkinsons-disease-labs-pd/