Computable Phenotype for Polycystic Kidney Disease (PKD)

The overall objective of this proposal is to accelerate clinical trials for autosomal recessive polycystic kidney disease (ARPKD) and other hepatorenal fibrocystic diseases (HRFD) by creating PKDnet, a robust data resource for rapid cohort identification using PEDSnet. 

ARPKD causes a substantial burden of morbidity and mortality in children due to progressive chronic kidney disease, congenital hepatic fibrosis, and portal hypertension. Although several targeted disease-modifying therapies for ARPKD have shown promise in animal studies, and one drug is in a Phase 1 clinical trial, there are several key barriers to conducting clinical trials: 1) limited data on the natural history of kidney and liver disease progression; 2) lack of sensitive biomarkers of progression to serve as surrogate endpoints for clinical trials; and 3) difficulty in identifying and recruiting sufficient numbers of patients who may be eligible for clinical trials. Existing natural history data for ARPKD have been collected either in cohort studies requiring active enrollment of patients (e.g. NIH cohort, CKiD study), or by manual entry of clinical data into research registries by investigators (e.g. North American ARPKD8 and HRFD Core Center databases in the U.S., ARegPKD in Europe).

However, the number of patients enrolled in U.S. databases (e.g. ~122 patients in the HRFD Database, PI: Guay-Woodford) represents only a small fraction of the estimated 1,500 individuals (age 0-29) with ARPKD living in the U.S. In addition, clinical characteristics of patients participating in these studies may not be representative of the ARPKD population as a whole.Improved methods are therefore needed to identify and characterize children with ARPKD, and to facilitate their entry into high priority observational studies and clinical trials. By developing PKDnet, we aim to create a robust data resource for rapid cohort identification of children with ARPKDin PEDSnet, allowing the collection of rich natural history data from the electronic health record (EHR),and the recruitment of children into studies to develop novel biomarkers of disease progression and into clinical trials of promising investigational therapies.

Our specific aims are: 1) To develop a computable phenotype for ARPKD based on combinations of diagnosis codes and other EHR data, using an iterative process of probabilistic algorithm development and refinement based on manual chart review at CHOP (n≈50); and 2) To validate the ARPKD computable phenotype across all PEDSnet sites, using manual chart review to evaluate its sensitivity and specificity(n≈150). A preliminary PEDSnet data inquiry in 2016 identified 467 patients with a diagnosis code for ARPKD, and an additional 2650 patients with “polycystic kidney, unspecified type.” We therefore hypothesize that PKDnet may identify and characterize up to 400 patients with ARPKD, representing a sizable proportion of affected children in the U.S. The expertise of the PCEN Learning Health Systems Core will be integral to the success of this project. The study timeline will be 9-12 months for Aim 1 and 12-15 months for Aim 2 (two years total).