RWD48 Introduction to Odyssey: Real-World Rare Disease Data Collection Program From Digitized Health Records of Patients With Glycogen Storage Disease Type III (GSDIII) in the United States

GSDIII is a rare inherited deficiency of glycogen-debranching enzyme that leads to liver, heart, and skeletal muscle damage. Real-world data (RWD) regarding GSDIII management and patient outcomes are limited. The GSD Odyssey study collects RWD on the impact of glycogen storage diseases to advance disease knowledge and identify areas to improve care. Non-interventional cohort study (target: 30+ patients) that collects prospective and retrospective data with the PicnicHealth digital record platform using a novel human-in-the-loop machine learning system to structure US medical records (clinical notes, medications, laboratory/imaging results, diagnostic reports). The study is HIPAA-compliant and IRB-approved; all patients or caregivers of patients <18 years provided written informed consent; and data are anonymized. Records are assessed for any mentions of specific complications or assistive devices and mentions of specific diets or nutrition management within the prior 3 years to indicate recent disease management. Preliminary results are reported. Eleven patients with GSDIII were enrolled from December 2020-May 2022. Median age was 46 (range, 7-57) years; 55% were female. Median years of retrospective data was 6.1 per patient, with 25 providers and 5 care sites. Common complications were fatty liver (91%), weakness (91%), hepatomegaly (82%), hypoglycemia (64%), and rhabdomyolysis (55%). Median age at first mention was 1 year for hepatomegaly and gastrostomy, 2 years for hypoglycemia, and 11 years for weakness. Most recent dietary management included cornstarch (55%), glycosade (18%), high-protein diet (73%), carbohydrate-restricted diet (45%), and low-sugar diet (9%). 18% used continuous glucose monitoring. Ever-use of assistive devices included wheelchair (64%), walker (45%), cane (36%), crutch (27%), and scooter (18%). Meaningful data can be extracted across the US healthcare system using this novel machine learning method, which is already providing insight to guide clinical development of a new therapy for GSDIII. More robust data are expected with additional enrollment.