Objective: To test and refine the use of probability-based algorithms to identify individuals at high-risk of developing Parkinson’s disease (PD).

Background: Neurodegeneration preceding a formal diagnosis of PD is associated with identifiable motor and non-motor features. Evidence-based algorithms have been developed to attempt identifying individuals in this pre-diagnostic phase according to exposure to common risk factors and results of simple screening tests. Two notable approaches to risk estimation are the PREDICT-PD algorithm (Noyce at al 2013) and the MDS research criteria (Berg et al 2015).

Methods: The PREDICT-PD pilot cohort comprised 1,323 healthy 60 to 80 year olds who completed annual online surveys and a keyboard-tapping task. Here, we considered those in the cohort who were diagnosed with PD during follow-up and tracked their risk estimation in the years preceding formal diagnosis, according to both the PREDICT-PD and MDS research criteria algorithms. A key difference between the MDS and PREDICT-PD algorithms, is that whereas PREDICT-PD previously only used demographic information, environmental exposures and simple symptomology to estimate risk with tapping speed, hyposmia and REM sleep behaviour disorder as intermediate outcome markers, MDS-criteria combines an extensive set of clinical and radiological tests. PREDICT-PD scores are also not updated in the known absence of a risk factor and are based on odds ratios, rather than likelihood ratios. We sought to refine the PREDICT-PD algorithm’s risk estimation by including our motor and non-motor intermediate markers (hyposmia and REM sleep behaviour disorder) in the risk score.

Results: 8 people in the PREDICT-PD cohort have been diagnosed with PD from a total 864 with follow-up at year 4. For these 8 with PD, mean probability risk score was higher over the 4 years according to the MDS criteria than with PREDICT-PD basic score (25.5% vs 8.4%), both of which were higher than the rest of the cohort (2.7% & 2.8% respectively, both p<0.0001). Incorporating intermediate markers into the PREDICT-PD algorithm produced markedly greater probability risk scores than without for PD patients (38.7% vs 8.4%, p<0.0001), with values remaining greater than healthy comparators’ (6.6%, p<0.0001).

Conclusions: Probability-based algorithms offer a means of identifying people at highest risk of developing PD prior to a formal diagnosis. Their accuracy can be significantly improved by incorporating simple, remotely administered screening tests.

References: Noyce AJ, Bestwick JP, Silveira-Moriyama L, et al. 2013. PREDICT-PD: Identifying risk of Parkinson’s disease in the community: methods and baseline results. J Neurol Neurosurg Psychiatry. Berg D, Postuma R, Adler C, et al. 2015. MDS research criteria for prodromal Parkinson’s disease. Movement Disorders.

« Back to 2018 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/risk-estimation-in-the-years-preceding-diagnosis-of-parkinsons-disease-in-the-predict-pd-cohort/