Percent Increase in Left Ventricular Stroke Work Index and Right Ventricular Stroke Work Index after Milrinone Infusion Predicts Escalation of Therapy and Mortality

Purpose We hypothesized that changes in biventricular function would approximate myocardial reserve and predict outcomes. Methods We reviewed 178 patients undergoing right heart catheterization and milrinone loading at our center between January 2013 and January 2019. Percent increase in left ventricular stroke work index (piLVSWI) was calculated as 100 * (left ventricular stroke work index after infusion - left ventricular stroke work index before infusion)/left ventricular stroke work index before infusion, and percent increase in right ventricular stroke work index (piRVSWI) was calculated as 100 * (right ventricular stroke work index after infusion - right ventricular stroke work index before infusion)/right ventricular stroke work index before infusion. Cutpoints were determined by ROC analysis, and patients were stratified as follows: (1) High piLVSWI, High piRVSWI; (2) High piLVSWI, Low piRVSWI; (3) Low piLVSWI, High piRVSWI; and (4) Low piLVSWI, Low piRVSWI. Cox regression analyses were performed to determine association with escalation of therapy, defined as left ventricular assist device implantation, heart transplantation, or death. Results Cutpoints of 18.9% and 17.5% were determined for piLVSWI and piRVSWI, respectively. Low piLVSWI, High piRVSWI was associated with escalation of therapy or death at 30-days (HR 3.78, 95% CI 1.41 - 10.17, p = 0.0084). Multivariate analysis identified Low piLVSWI, High piRVSWI as a predictor of escalation of therapy or death at one-year (HR 1.96, 95% CI 1.02 - 3.78, p = 0.043). Grouping patients according to whether they possessed Low piLVSWI, High piRVSWI revealed a greater hazard for the endpoint at 30-days and one-year (HR 2.98, 95% CI 1.41 - 6.28, p = 0.004; HR 1.79, 95% CI 1.0002 - 3.2, p Conclusion Change in biventricular function after milrinone infusion, particularly Low piLVSWI, High piRVSWI, may prognosticate worse clinical outcomes at 30-days and one-year.