In mitral valve surgery, cardioplegia is commonly dosed by body weight or body surface area (BSA), which may not reflect true cardiac mass. We evaluated whether indexing the initial dose to left ventricular (LV) mass improves dosing fidelity.

We retrospectively studied 323 patients undergoing mitral surgery with St. Thomas solution (2019–2022). The initial dose of cardioplegia was based on a flow of 150 mL · min^–1 · m^–2 of body surface area. To reduce heterogeneity related to differences in ventricular remodeling and cardioplegia delivery strategies, the study was restricted to patients undergoing mitral valve surgery. Echocardiography-derived LV mass yielded two indices: C1 = initial perfusion dose (mL)/LV mass (g) and C2 = initial perfusion dose (mL)/LV mass index (LVMI, calculated as LV mass divided by body surface area [BSA]). Patients were divided into 2 groups, Group A (C1 ≤ 6.6 mL·g^-1) and Group B (C1 > 6.6 mL·g^–1). Primary endpoint was CK-MB mass (a quantitative measure of the creatine kinase-MB) at 0, 7, 24, and 48 h. High-sensitivity cardiac troponin T (hs-cTnT), ICU and hospital stay were also evaluated. Restricted cubic splines and segmented regression assessed nonlinearity and thresholds; propensity score matching (PSM) generated 67 pairs. Model performance was evaluated using the Akaike and Bayesian information criteria (AIC/BIC).

In unmatched and matched analyses, C1 showed a nonlinear association with CK-MB mass, with a threshold near C1 = 12.2 mL·g^–1, above which CK-MB increased more steeply. C2 displayed higher thresholds (17.5). After PSM, Group B (higher C1) had higher CK-MB mass at 0, 24, and 48 h; hs-cTnT at 24 h was slightly higher in Group A. ICU and hospital stay were similar between groups. Model fit favored C1 over C2 at 24 and 48 h (lower AIC/BIC).

LV-mass–indexed dosing captures a nonlinear relevant dose–response and suggests that C1 values above approximately 12.2 mL·g^–1 may represent a potential upper range that warrants caution.