Abstract 12703: Transthyretin Amyloid Cardiomyopathy Among Patients Hospitalized for Heart Failure and Performance of an Adapted Wild-Type ATTR-CM Machine Learning Model: Findings From GWTG-HF

Abstract

Introduction: An 11-factor random forest model has been previously developed among ambulatory patients for identifying potential wild-type amyloidogenic TTR cardiomyopathy (wtATTR-CM), but model performance in a large sample of patients hospitalized for HF has not been evaluated. Methods: This study included Medicare beneficiaries hospitalized for HF in the Get With The Guidelines-HF® Registry from 2008-2019. Patients with and without a diagnosis of wtATTR-CM were compared. ICD9/10 codes were used as proxies to identify wtATTR-CM patients on inpatient/outpatient claims within ±6 months of hospitalization. Within a cohort matched 1:1 by age and sex, univariable logistic regression was used to evaluate relationships between each of the 11 factors of the established ATTR-CM model and wtATTR-CM. Discrimination and calibration of the 11-factor model was assessed. Results: Among 205,545 patients hospitalized for HF across 608 US hospitals, 627 patients (0.31%) had a diagnosis code for wtATTR-CM. wtATTR-CM patients were more likely to be male (69% vs 47%), Black race (23% vs 9%), and had lower systolic blood pressure (median 124 vs 139 mmHg). Univariable analysis within the 1:1 matched cohort of each of the 11-factors in the wtATTR-CM model found pericardial effusion, carpal tunnel syndrome, lumbar spinal stenosis, and elevated serum enzymes (e.g., troponin elevation) to be strongly associated with wtATTR-CM ( Figure, Panel A ). The 11-factor model showed modest discrimination with c-statistic of 0.65 ( Figure, Panel B ), with good calibration within the matched cohort ( Figure, Panel C ). Conclusions: Among patients hospitalized for HF in US practice, the number of wtATTR-CM as defined by diagnosis codes was low. Patients with a diagnosis of wtATTR-CM have a distinct clinical profile, and most factors within the prior 11-factor model were associated with greater odds of wtATTR-CM diagnosis. In this population, the wtATTR-CM model demonstrated modest discrimination.