Aortic Valve Replacement for Severe Asymptomatic Aortic Stenosis-A Systemic Review and Bayesian Meta-analysis.

BACKGROUND

Previous meta-analyses have concluded that early aortic valve replacement (AVR) is associated with reduced cardiovascular events compared to clinical surveillance (CS) for severe asymptomatic aortic stenosis (AS). However, individual patient data (IPD) was not used and the possibility of biases not considered.

METHODS

A systematic review of randomized clinical trials (RCTs) whose primary outcome was mortality and unplanned cardiac hospitalization. IPD data was reconstructed, permitting calculation of time varying risks. Bayesian analyses, with vague priors, were performed for each trial and combined in a meta-analysis.

RESULTS

Four RCTs were identified. The pooled average long-term primary outcome was hazard ratio (HR) 0.52 (95% credible interval (CrI) 0.25 - 0.93) and the predicted HR for the next study was HR 0.52 (95% CrI 0.13 - 1.87) with increased uncertainty from between study heterogeneity. Only the EARLY_TAVR trial showed improved outcomes in year one (HR 0.33, 95% CrI 0.24 - 0.46), driven by a 142% increase in the crossover rate compared to previous trials. A one-year landmark analysis, that eliminates possible early unblinding bias, showed no long term AVR benefit for any individual study or for the pooled result (RR 0.59, 95% CI 0.20 - 1.40).

CONCLUSIONS

One-year landmark analysis showed no definitive longer-term benefit for early AVR compared to CS in asymptomatic AS. The EARLY_TAVR benefit is possibly the result of performance bias. Further research is required before early AVR for asymptomatic severe AS can become the new standard of care.