Piergiorgio Tozzi, Roger Hullin, Fabrizio Gronchi, Matthias Kirsch
Institution : Service de chirurgie cardiaque, service de cardiologie, service d’anesthésie, CHUV, Lausanne, Suisse
Objectives : In the era of continuous flow left ventricular assist devices (LVAD), the decision of whether a patient will tolerate isolated LVAD support or will need biventricular support (BIVAD) can be challenging. Incorrect decision making with delayed right ventricular (RV) assist device implantation results in increased morbidity and mortality. Several risk stratification tools have been developed to predict the need of right assist device (RVAD), but none seems to outperform the others. We reviewed our experience to validate the most recent strategies to identify patients who are at risk of post LVAD right ventricular failure (RVF).
Methods : We collected variables for 34 consecutive patients who underwent VAD implant from 2010 through June 2015 (LVAD=32, BIVAD=2). We defined pre-operative RVF based upon echocardiographic parameters. Post-operative RV failure was defined according to INTERMACS Protocol 3 definition (CVP >18mmHg, CI <2l/min/m2, need for nitric oxide or inotrope therapy >7 days after LVAD implant). For each patient, we calculated the CRITT score and applied the Pittsburgh Decision Tree (PDT). We assessed the performance of the score by entering it in a logistic regression model.
Results : The mean age was 52±12; 70.5% were men. The indication was bridge to transplant in 100%. Etiology was ischemic in 42%; 14.7% required additional mechanical circulatory support after LVAD implant (BVAD=2, ECMO=3). All patients received continuous flow LVADs (Heartmate II=13, HeartWare=21). By 30 days, 22 patients (64.7%) developed RVF defined as the need for RVAD (n=2); pulmonary vasodilator use >48h (n=18); inotropes for > 14 days post LVAD (n=22). According to CRITT score the RVF expected rate was 29% (10 patients had 4 or 5 corresponding to 80% risk of RVF). According to PDT, 8 patients needed BVAD. CRITT score achieved best performances (CI 95%; sensitivity = 90%; specificity 8.3%; accuracy 61.8%).
Conclusion : These data indicate that even clinical RVF risk prediction models developed in the era of continuous flow pumps have rather limited clinical applicability. We definitely need new clinical test based on quantitative pre-operative imaging data, able to quantify the functional reserve of RV to drive our clinical decision about using an isolated LVAD versus BiVAD, especially in the destination therapy perspective.