Objective: To establish a data harmonisation framework for classifying cognitive status in Parkinson’s disease (PD) across multiple study and data types to facilitate large-scale genetic research.

Background: Cognitive impairment is one of the most disabling non-motor complications of PD, and progression to PD dementia (PDD) occurs in a large proportion of patients. Trajectories of cognitive decline are highly heterogeneous across individuals, and understanding its genetic aetiology may provide new insights into biology and potential future treatments. However, harmonising clinical data for large genetic studies is a challenge due to heterogeneity of study designs and data collection.

Method: We used longitudinal clinical data from the Global Parkinson’s Genetics Program (GP2; release 11). We developed a standardised operating procedure to harmonise available measures and ensure robust classification of PDD. Using the cohorts with multiple data types available, we performed exploratory strategies for dementia classification and performed head-to-head comparisons between different classifications, considering clinician or investigator-defined dementia as the gold standard.

Results: We included 2874 PD individuals of European ancestry (15 cohorts). Mean (SD) age at baseline was 64.94 (10.20) and the mean baseline MoCA was 26.34 (3.15). Mean follow-up time was 39.1 months (36.88). Longitudinal investigator/clinician-defined dementia classification was available for 16.53% of participants, while 67.31% and 89.37% had longitudinal MoCA and MDS-UPDRS1.1 scores available, respectively. The rate of dementia was 8.8% based on clinical diagnosis, 6.97% based on MoCA <21, and 6.87% based on MDS-UPDRS1.1 >2. When considering cohorts with all data types available, MoCA sensitivity/specificity was 88.10%/98.12%, while MDS-UPDRS1.1 was 97.62%/97.65%, compared to the gold standard. Mean (SD) time from baseline to dementia diagnosis was 4.29 (2.20) in the clinically defined group, compared to 3.80 (2.41) and 3.70 (2.40) in the MoCA and MDS-UPDRS1.1 groups, respectively.

Conclusion: MDS-UPDRS1.1 had the highest sensitivity to identify dementia compared to the gold standard. This provides a standardised and scalable approach for defining PDD across cohorts, enabling the inclusion of heterogeneous clinical data in genome-wide investigation of genetic determinants of dementia progression in PD.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/strategies-for-harmonising-clinical-data-for-multi-centre-large-scale-genome-wide-association-studies/