PT-019 - ESTABLISHING A PARAMOUNT QUANTITATIVE FRAMEWORK LINKING DISEASE PROGRESSION AND CLINICAL OUTCOMES FOR PARKINSONS DISEASE

C. Gupta¹, C. Kim¹, V. Siripuram¹, T. Yoneyama², S. Schmidt¹, S. Guzy¹, V. Vozmediano¹; ¹University of Florida, ²Takeda Pharmaceuticals.

Background: Parkinson's disease (PD) is a neuro-degenerative disease characterized by gradual motor and non-motor impairments. We established a quantitative framework that integrates imaging biomarkers (dopamine striatal binding ratio (SBR)) and clinical MDS UPDRS scores to better understand heterogeneity in the onset and progression of the disease and ultimately predict long-term outcomes based on early biomarkers.

Methods: Data was obtained from Parkinson’s Progression Markers Initiative, for SBR and MDS-UPDRS scores over a period of 8 years, for 419 diagnosed PD patients and 196 healthy control (HC) participants. A piecewise disease progression function with two exponential functions was introduced to describe the SBR decline in left/right putamen and left/right caudate prior to and after onset of PD. HC data informed the impact of aging, and PD data was used to calibrate the post onset function. These patient-level disease progression profiles were used as input in an item response theory (IRT) model instead of a hidden disability variable to account for interindividual differences in disease progression. Changes in non-motor and motor items over time were linked to disease progression in caudate and putamen, respectively.

Results: Our analysis showed that pathophysiological changes already start within 4-15 years prior to clinical diagnosis. The lag time between onset of PD and diagnosis is 8-15 years in putamen and 4-6 years in caudate. The mean annual decline rate for SBR was estimated to be 0.4-0.8% for putamen and caudate pre onset, and in the range of 7.5-11% post PD onset. The fully qualified IRT models were able to successfully simulate the clinical scores of 398 of the total 416 patients (96%) driven by the patient specific disease progression, despite varying age of diagnosis for the PD patients. Further, VPC plots and simulations indicated that out of the two putamen’s, the less damaged one would be the better driver of clinical motor symptoms. Both the caudates were shown to be good drivers of cognitive symptoms, as expected, since they are impaired in the later state of the disease.

Conclusion: The two-pronged modeling effort establishes a strong link between SBR score and clinical symptom, opening doors for predictive modeling after diagnosis via establishing the disease profile from early DaTScan and predicting clinical symptoms.