Cardiac Amyloidosis AI‐ECG Model is Useful in Predicting Outcomes in Multiple Myeloma Patients Without Amyloidosis Undergoing Autologous Stem Cell Transplant

AL amyloidosis often presents with nonspecific symptoms, leading to delayed diagnosis and increased early mortality, even in patients with known multiple myeloma or other monoclonal gammopathies. This study assessed the cardiac amyloid artificial intelligence electrocardiogram (CA‐AI‐ECG) tool’s utility in predicting treatment outcomes and detecting CA in a cohort of multiple myeloma patients undergoing autologous stem cell transplantation (ASCT). Among 2934 multiple myeloma patients studied, 81% received early ASCT (< 12 months from diagnosis) and therefore had an early ECG performed (early ASCT/ECG cohort), with 5% having a positive pre‐ASCT CA‐AI‐ECG score. In the early ASCT/ECG group, a positive CA‐AI‐ECG score did not correlate with decreased overall survival. However, in the delayed ASCT/ECG group, a positive CA‐AI‐ECG score was associated with significantly lower overall survival (HR: 1.73, 95% CI: 1.07–2.78), remaining an independent risk factor on multivariate analysis. The observed overall survival difference between cohorts may be attributable to the longer time interval from diagnosis to ECG in the delayed ASCT group, leading to prolonged light chain exposure. During follow‐up, 25 patients developed AL amyloidosis, of which, 12 patients had systemic involvement. Those with a positive pre‐ASCT CA‐AI‐ECG were five times as likely to progress to systemic amyloidosis compared to those with a negative score (LR 5.0 p = 0.025). These findings suggest that the CA‐AI‐ECG tool may enhance subclinical CA detection, especially as an early screening tool.