Heart Failure Related Hospitalization and Mortality after Aortic Valve Replacement

Abstract

Patients undergoing surgical aortic valve replacement (AVR) are at risk for developing subsequent heart failure (HF) related events that can be predicted using a multivariate model. Based on the model age, preoperative NYHA class, left ventricular grade, atrial fibrillation, coronary artery disease, smoking, and redo AVR status predicted HF after surgery. Increased transprosthesis gradients were associated with HF after AVR while larger prosthesis size and effective orifice area were independently associated with freedom from HF. Select patient and prosthetic valve characteristics may be used to predict the incidence of HF related hospitalization and mortality after AVR. A semiparametric Cox proportional hazards model was developed based upon 1563 patients who underwent AVR at the University of Ottawa Heart Institute between 1976 and 2001. To validate the model a Kaplan-Meier analysis for the composite endpoint of HF related hospitalization or HF related death beyond 30 days post AVR was performed on an independent cohort of 1014 patients who underwent AVR in 2007-2009 as part of the IDE Trifecta trial. The HF endpoint occurred in a total of 41 subjects during 5 years of follow up. These events were driven by 34 HF hospitalization (83%) and 7 HF deaths (17%). Repeat HF hospitalizations occurred in 7 patients (20.6%). At baseline patients with HF events were older (79±8 vs 72±9 years; p<0.0001), had more females (63.4% vs 34.7%; p=0.0002), had a higher proportion of patients with left ventricular ejection fraction <50% (30% vs 12.8%; p=0.0018), and a higher proportion of patients with NYHA Class III or IV (78.1% vs 48.1%, p=0.0012). At 1-year post AVR patients with HF events had a higher trans-valvular gradient (10±4 vs 8±4 mmHg; p=0.004). The table compares the HF-free survival per Kaplan-Meier analysis of the Trifecta™ valve recipient cohort vs that predicted by the model over 5 years post-AVR. The HF-free survival values predicted by the model were well within the confidence intervals of the Kaplan-Meier analysis, confirming the accuracy of the predictive model over the 5-year time interval.Table 1Time post AVR30 days1 year2 years3 years4 years5 yearsCHF-free survival per Kaplan-Meier analysis of Trifecta cohort, % (95% Confidence Interval)100% (N/A)98.2 (97.2 - 98.9)97.0 (95.7 - 97.9)96.0 (94.3 - 97.1)94.7 (92.6 - 96.1)94.4 (92.3 - 95.9)CHF-free survival per Ruel predictive model customized to reflect Trifecta cohort100%98.7%97.4%96.0%94.4%92.7% Open table in a new tab A model for predicting HF events after AVR was validated. While patient characteristics cannot be improved at the time of surgery, a lower valve gradient and/or a larger EOA may be achieved with better valve selection, and thereby result in significantly reduced HF events. Moreover, referral of patients to AVR before advanced symptoms and LV dysfunction develop can result in lower HF recurrence.