Abstract Details

Title PREDIGT Score: Diagnosing Parkinson's Disease Without a Movement Disorder Examination

Topic Movement Disorders

Presentation(s) S17 - Parkinson's Disease Biomarkers and Pathophysiology (1:36 PM-1:48 PM)

Poster/Presentation
Number 004

Objective To validate the PREDIGT score and test the model’s discriminative accuracy.

Background We recently designed an incidence prediction model founded on the concept that PD pathogenesis is multifactorial. We postulated 5 factors to determine cumulative incidence rates: DNA variants (D); Environmental factors (E); Gene–environment interactions, such as inflammation (I); Gender (G); and Time (T). The proposed formula for calculating the PD incidence rate (PR; %) in an individual is PR=(E+D+I)*G*T.

Design/Methods We began to validate this model using baseline data of 651 PD patients and 355 age- and sex-matched healthy controls (HC) from two case-control cohorts: (1) the multicentre Parkinson’s Progression Marker Initiative (PPMI) study; and (2) the single-centre De Novo Parkinson (DeNoPa) study. Risk and protective factors were selected and assigned concrete values. The PREDIGT formula was applied to each subject based on the aggregate score in each category (E,D,I,G,T). Receiver operating characteristic (ROC) curves and area under the curve (AUC) analyses were used to measure the model’s performance. Plots of score density were also used to illustrate the distinction between PD and HC. A complementary model with logistic regression coefficients was also included to refine the initial validation model.

Results The original formula and PREDIGT coefficients were able to discriminate between PD and controls: We first included variables that were shared between the cohorts and calculated AUC values of 0.86, 0.85, 0.85 for PPMI, DeNoPa, and for both cohorts combined, respectively. The validation model was then refined by adding variables that are only available in one (but not the other) cohort, which generated AUC values of 0.89 in PPMI and 0.86 for DeNoPa. In a logistic-regression-based analysis to weigh variables’ relative impact on each other, the AUC outcomes were increased at an AUC 0.92 for PPMI and 0.92 for DeNoPa.

Conclusions Our results suggest promising, early validation of the original PREDIGT model.

Authors/Disclosures
Juan Li, PhD (Ottawa Hospital Research Institute) PRESENTER	No disclosure on file
Tiago Mestre, MD, MSC (University of Ottawa)	Dr. Mestre has received personal compensation in the range of $500-$4,999 for serving as a Consultant for CHDI. Dr. Mestre has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for PTC. Dr. Mestre has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Abbvie. Dr. Mestre has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Ipsen. Dr. Mestre has received personal compensation in the range of $500-$4,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Abbvie. Dr. Mestre has received personal compensation in the range of $500-$4,999 for serving on a Speakers Bureau for Abbvie. The institution of Dr. Mestre has received research support from CIHR. The institution of Dr. Mestre has received research support from Ontario Research Fund. The institution of Dr. Mestre has received research support from MJFF. The institution of Dr. Mestre has received research support from Parkinson Canada. The institution of Dr. Mestre has received research support from University of Ottawa/PRC.
	No disclosure on file
Claudia Trenkwalder, MD (Center For Parkinsonism and Movement Disorders)	No disclosure on file
Brit Mollenhauer	No disclosure on file
Tim Ramsay	No disclosure on file
	No disclosure on file
Michael G. Schlossmacher, MD (Ottawa Hospital)	No disclosure on file

��ɫ��

��ɫ��