Determinants of the effect of extracorporeal carbon dioxide removal in the SUPERNOVA trial: implications for trial design

Purpose To describe the variability and determinants of the effect of extracorporeal CO 2 removal (ECCO 2 R) on tidal volume ( V t ), driving pressure (Δ P ), and mechanical power (Power RS ) and to determine whether highly responsive patients can be identified for the purpose of predictive enrichment in ECCO 2 R trial design. Methods Using data from the SUPERNOVA trial (95 patients with early moderate acute respiratory distress syndrome), the independent effects of alveolar dead space fraction (ADF), respiratory system compliance (Crs), hypoxemia (PaO 2 /FiO 2 ), and device performance (higher vs lower CO 2 extraction) on the magnitude of reduction in V t , Δ P , and Power RS permitted by ECCO 2 R were assessed by linear regression. Predicted and observed changes in Δ P were compared by Bland–Altman analysis. Hypothetical trials of ECCO 2 R, incorporating predictive enrichment and different target CO 2 removal rates, were simulated in the SUPERNOVA study population. Results Changes in V t permitted by ECCO 2 R were independently associated with ADF and device performance but not PaO 2 /FiO 2 . Changes in Δ P and Power RS were independently associated with ADF, Crs, and device performance but not PaO 2 /FiO 2 . The change in Δ P predicted from ADF and Crs was moderately correlated with observed change in Δ P ( R 2 0.32, p < 0.001); limits of agreement between observed and predicted changes in Δ P were ± 3.9 cmH 2 O. In simulated trials, restricting enrollment to patients with a larger predicted decrease in Δ P enhanced the average reduction in Δ P , increased predicted mortality benefit, and reduced sample size and screening size requirements. The increase in statistical power obtained by restricting enrollment based on predicted Δ P response varied according to device performance as specified by the target CO 2 removal rate. Conclusions The lung-protective benefits of ECCO 2 R increase with higher alveolar dead space fraction, lower respiratory system compliance, and higher device performance. ADF and Crs, rather than severity of hypoxemia, should be the primary factors determining whether to enroll patients in clinical trials of ECCO 2 R.