Abstract 17110: Racial Differences in the Performance of the Transthyretin Amyloid Cardiomyopathy Risk Score

Introduction: Early diagnosis of transthyretin cardiac amyloidosis is crucial, particularly with the availability of disease-modifying therapies. Tc-99m pyrophosphate scintigraphy (PYP scan) is widely used for diagnosis. A previously validated score derived from a nearly all Caucasian cohort has shown promise in predicting PYP scan positivity in patients with HFpEF; but further assessment in a racially diverse cohort is needed.

Aims: To evaluate the performance of the ATTR-CM score in predicting PYP scan positivity overall, and across racial groups.

Methods: We included all patients referred for PYP with SPECT/CT at the Cleveland Clinic between 1/2012 and 1/2020 who had undergone echocardiography within one year of PYP scan. The ATTR-CM score was calculated, consisting of the following variables: Age (if 60-69, +2; if 70-79, +3; if ≥80, +4), Sex (if Male, +2), Hypertension diagnosis (if present, -1), Ejection Fraction (if <60%, +1), Posterior Wall Thickness (if ≥12 mm, +1) and Relative Wall Thickness (if >0.57, +2). A score ≥6 was considered high-risk.

Results: The cohort consisted of 540 patients (64% white; 33% black). The overall prevalence of ATTR-CM was 28.5%, with no significant difference between racial groups (27.5% vs 29.9%; p = 0.52). A higher proportion of black patients had an EF < 40% (23.3% vs 35.4%, p = 0.01). The score showed good discrimination with an AUC of 0.816, with similar performance between racial groups (0.824 vs. 0.824; p < 0.001; Figure 2). For a score ≥6, sensitivity was 77.3% (82.7% white; 67.3% black), specificity was 70.2% (64.3% white; 80.6% black), PPV was 50.9%, and NPV was 88.6%. Similar findings were observed when considering only patients with EF ≥ 40%.

Conclusions: The overall performance of the ATTR-CM risk score by AUC was consistently good across races. However, it was less sensitive for detecting disease in black patients. Efforts to scale ATTR-CM diagnosis tools should be mindful of racial differences in risk prediction models.