SAT-777 Evaluation of FRAX in a U.S. Cohort of Primary Hyperparathyroidism

Abstract Disclosure: V.R. Sant: None. Y. ElNakieb: None. J.F. Rousseau: None. C.D. Rubin: None. N.M. Maalouf: None. Introduction: Fracture risk prediction in primary hyperparathyroidism (PHPT) remains challenging, particularly in identifying patients who may experience fracture risk reduction following parathyroidectomy (PTX). Despite multiple iterations of PHPT management guidelines acknowledging the uncertain but potential utility of the FRAX fracture prediction tool in this population, its performance in PHPT remains poorly characterized. We evaluated FRAX in a large U.S. PHPT cohort using an electronic health record dataset encompassing over 100 million patients. Methods: We analyzed a previously described PHPT cohort derived from the TriNetX database, restricting inclusion to those with at least 10 years of follow-up and complete clinical data within the range of FRAX input parameters. The primary outcome was incident major osteoporotic fracture (MOF) within 10 years, identified by a validated claims-based algorithm from ICD-9/10 codes, and stratified by treatment strategy (PTX vs non-surgical). Requisite FRAX variables (age, sex, race, height, weight, prior fracture, smoking status, alcohol use, glucocorticoid use, rheumatoid arthritis, and secondary osteoporosis) at time of cohort inclusion were extracted and used to generate FRAX 10-year MOF predictions without bone mineral density (BMD). Observed and FRAX-predicted MOF events were compared using receiver operating characteristic (ROC) analysis, and intervention thresholds were assessed. Results: Among 4,288 patients with PHPT, 27.6% underwent PTX. At diagnosis, the median age was 61 years (IQR 54-68), with 78% female, 75% White, and 19% Black individuals, and a median BMI of 28.8 kg/m2 (IQR 25.0-33.4). Observed 10-year MOF incidence was 4.5%. Median 10-year MOF FRAX prediction was 5.4% (IQR 3.1-8.5%, range 0.5-36%). The overall FRAX area under the ROC curve (AUC) was 0.709. By treatment strategy, AUC was 0.675 (PTX) and 0.720 (non-surgical). Stratified by sex, AUC was 0.714 (female) and 0.670 (male). By race/ethnicity, AUC was 0.695 (White), 0.685 (Black), 0.701 (Hispanic), and 0.773 (Asian). Common US FRAX intervention thresholds were used to stratify the cohort (<10% low-risk, ≥20% high-risk). Among low-risk, MOF was observed in 3.2% (non-surgical) vs 2.9% (PTX) (absolute risk reduction [ARR] 0.3%); high-risk: 24.1% vs 15.4% (ARR 8.7%). When evaluating the ARR in MOF with PTX across the span of FRAX intervention thresholds, MOF ARR ≥5% with PTX was observed consistently at all FRAX-MOF thresholds >16%. Conclusions: This represents the first evaluation of FRAX performance in a U.S.-based PHPT cohort. Although FRAX does not explicitly account for PHPT or its treatment strategy, it exhibits acceptable predictive performance in this population. Additionally, FRAX may help stratify fracture risk and inform surgical decision-making even in the absence of BMD. Presentation: Saturday, July 12, 2025