Generalizability of Left Ventricular Assist Device Clinical Trial Findings: An Analysis of the INTERMACS Database

Abstract

Clinical trials assessing durable LVAD therapy have demonstrated survival benefits that have increased adoption. However, highly selected trial populations may not reflect the real-world setting. We assessed the generalizability of LVAD clinical trial findings by comparing mortality between LVAD recipients meeting current trial eligibility criteria to those ineligible. Patients undergoing LVAD implantation between 2012-2017 were identified in INTERMACS (n=15,077). MOMENTUM 3 (Multicenter Study of MagLev Technology in Patients Undergoing Mechanical Circulatory Therapy with HeartMate 3) trial inclusion/exclusion criteria were mapped to INTERMACS variables to categorize patients as trial eligible or ineligible and by number of exclusion criteria present: 0, 1, 2, and ≥3. Kaplan-Meier analyses were used to compare cumulative mortality by trial eligibility and number of exclusion criteria. Multivariable Cox models were used to report risk-adjusted mortality with hazard ratios (HR) and 95% confidence intervals (CI). A total of 6728 (44.6%) LVAD recipients were trial ineligible, of which 4331 (28.7%) had 1 exclusion, 1517 (10.1%) had 2, and 880 (5.8%) had ≥3. Mortality for trial ineligible recipients (1-year: 26±1%; 3-year: 44±2%) was higher than for trial eligible recipients (1-year: 16±1%; 3-year: 37±1%; p<0.001, Fig. A). Number of exclusion criteria per patient was directly associated with mortality (p<0.001, Fig. B). Trial ineligibility was independently associated with mortality (HR 1.29 [CI 1.21-1.37], p<0.001), while an increased number of exclusion criteria conferred increased mortality risk (1: HR 1.15 [CI 1.07-1.23]; 2: HR 1.49 [CI 1.35-1.64]; ≥3: HR 2.13 [CI 1.89-2.40]; all p<0.001). Nearly half of real-world LVAD recipients do not meet current clinical trial criteria, and these trial ineligible recipients demonstrated higher mortality. Providers should incorporate trial criteria when assessing LVAD candidate outcomes.