Objective: To develop and test an enhanced algorithm for the prediction of Parkinson’s disease (PD) risk.
Background: The PREDICT-PD algorithm estimates risk of PD using published risk factor odds ratios (ORs)1. We have shown that it can identify incident cases of PD and enrich a population with so-called ‘intermediate’ markers (finger tapping speed, hyposmia and REM sleep behaviour disorder (RBD)). Here we used likelihood ratios (LRs) and incorporated intermediate markers into the algorithm. Whereas other algorithms tend to dichotomise information on the intermediate markers, we estimated LRs for their specific test values.
Method: To develop test value specific LRs, smell and probable RBD data for PD patients were taken from the Tracking Parkinson’s study (n=1046); motor impairment data using the Bradykinesia-Akinesia Incoordination (BRAIN) test came from published sources which were supplemented (n=87), and controls were PREDICT-PD study participants (n=1314). Smell was assessed using the University of Pennsylvania Smell Identification Test (UPSIT) and probable RBD using the RBD Screening Questionnaire (RBDSQ). UPSIT and RBDSQ data were analysed using logistic regression to determine which items predicted PD. Gaussian distributions were fitted to BRAIN test scores. LRs were calculated from logistic regression models or score distributions. Test value specific LRs were applied to the PREDICT-PD cohort to calculate enhanced PREDICT-PD risks, which were compared with basic risks.
Results: 16 UPSIT odours were associated with PD (LR range 0.005-5511), 6 RBDSQ questions were associated with PD (LRs range 0.34-69) and BRAIN test LRs ranged 0.16-1311. For a 70% detection rate (DR), the false-positive rate (FPR) for the 16 odours was 2.4%; for a 50% DR, the BRAIN test FPR was 6.6% and RBDSQ FPR was 12.2%. Risk of PD increased more with increasing risk scores from the enhanced algorithm than from the basic algorithm (hazard ratios per standard deviation increase in log risk 2.73 [95% CI 1.68–4.45] v 1.47 [0.86-2.51] respectively). The enhanced algorithm also correlated more closely with DaT-SPECT dopaminergic denervation (R2=0.14, p=0.01 v R2=0.043, p=0.17).
Conclusion: Incorporating intermediate markers from the pre-diagnostic PD phase and considering test specific LRs enhanced the performance of the PREDICT-PD algorithm.
References: 1. Noyce AJ, Bestwick JP, Silveira-Moriyama L, et al. PREDICT-PD: identifying risk of Parkinson’s disease in the community: methods and baseline results. J Neurol Neurosurg Psychiatry. 2014;85(1):31-37. doi:10.1136/JNNP-2013-305420
To cite this abstract in AMA style:J. Bestwick, S. Auger, D. Grosset, S. Kanavou, G. Giovannoni, A. Lees, J. Cuzick, C. Simonet, R. Rees, D. Rack, M. Jitlal, A. Schrag, A. Noyce. Improving estimation of Parkinson’s disease risk – enhancing the PREDICT-PD algorithm [abstract]. Mov Disord. 2020; 35 (suppl 1). https://www.mdsabstracts.org/abstract/improving-estimation-of-parkinsons-disease-risk-enhancing-the-predict-pd-algorithm/. Accessed December 2, 2023.
« Back to MDS Virtual Congress 2020
MDS Abstracts - https://www.mdsabstracts.org/abstract/improving-estimation-of-parkinsons-disease-risk-enhancing-the-predict-pd-algorithm/