Pre-operative mortality risk assessment in patients with continuous-flow left ventricular assist devices: Application of the HeartMate II risk score

Background Survival with left ventricular assist device (LVAD) therapy is dependent on appropriate patient selection. The HeartMate II risk score (HMRS) was recently derived and validated to predict 90-day mortality in clinical trial patients with continuous-flow LVADs. The aim of this study was to test HMRS validity in predicting survival at our institution. Methods We performed a retrospective analysis of patients implanted with HeartMate II (HMII; Thoratec, Pleasanton, CA) LVADs from March 31, 2004 to September 20, 2012 at the Columbia University Medical Center (CUMC). Patients were stratified according to HMRS profiles (HMRS Low < 1.58, 1.58 ≤HMRS Medium >2.48, HMRS High > 2.48) calculated using age, albumin, creatinine, international normalized ratio (INR) and center volume. Outcome was defined as survival at 90 days after device implantation. Results HeartMate II LVADs were implanted in 205 patients. Pre-operative data from 201 patients were categorized into HMRS Low (n = 101; 1.04 [0.64 to 1.31]), HMRS Medium (n = 73; 1.98 [1.78 to 2.25]) and HMRS High (n = 27, 3.07 [2.70 to 3.43]) (p < 0.0001). Kaplan-Meier survival estimates at 90 days (HMRS Low 91.0 ± 2.9%, HMRS Medium 91.7 ± 3.2%, HMRS High 88.7 ± 6.1%) and at 1 year (HMRS Low 85.5 ± 3.8%, HMRS Medium 79.3 ± 5.5%, HMRS High 82.4 ± 8.4%) after LVAD implantation were not statistically different (p = 0.43). Prediction of 90-day mortality by receiver operating characteristic was poor (AUC = 0.56). Conclusion HMRS stratification poorly discriminates 90-day mortality after HMII LVAD implantation at our institution. Its generalizability as a universal prognostic score may be limited.