Learning from the Learning Curve in Heartmate II Implantation: Low Volumes Do Not Equate Bad Outcomes

Purpose Use of continuous-flow left ventricular assist device (CF-LVAD) for end-stage heart failure is increasing, both as bridge to transplantation (BTT) and destination therapy (DT). Optimal volume thresholds are debated, but most implantations are still done in low volume centers. This study aims to evaluate the procedural learning curve of HeartMate II (HM2) implantations by comparing outcomes between two time periods in a low volume center. Methods All 51 consecutive patients undergoing HM2 implantation at the Montreal Heart Institute between January 2009 and December 2017 were evaluated. Patients were divided in an early era group (from 2009 to 2014) (n=25) and late era group (from 2015 to 2017) (n=26). The primary outcome was all-cause mortality and the secondary outcome was a composite of mortality, documented neurological event (i.e. stroke or transitory ischemic attack), reoperation for bleeding, need for temporary right ventricular assist device, and pump thrombosis. A cumulative sum control (CUSUM) analysis was used to establish a threshold of implantations that optimizes outcomes. Results The composite endpoint was significantly lower in the late era (early: n=19, 76% vs. late: n=11, 42%; p=0.01). Significant changes in patient selection lead to reduced comorbidities in the late era, notably insulin-treated diabetes (n=4, 16% vs. n=0, 0%; p=0.01) and previous stroke (n=11, 44% vs. n=3, 12%; p=0.01). Likewise, postoperative medical treatments using epinephrine, norepinephrine, nitric oxide, and sildenafil use was increased (p Conclusion Patients undergoing HM2 implantation in a low volume center have improving outcomes with number of cases and optimized results after a threshold of 23 cases. This is possibly caused by improvement in patient selection and improvement in surgical technique.