Abstract 3370: Which Troponometric Best Predicts Outcome Following Coronary Artery Surgery?
Aims Various timed or cumulative troponin measurements have been used in surgical myocardial protection studies as surrogates of patient outcome. Our aim was to define the troponometric best able to predict in-hospital and late mortality.
Methods We followed all cause mortality (censored at 09/08) in 440 patients (1/00 – 09/04) undergoing isolated on-pump coronary artery surgery with standardised anesthesia, perfusion, cardioplegia and post-operative care. Each subject had troponin I (cTnI) estimation at baseline, 6, 12, 24, 48 and 72 hours post-operatively, Individual time point cTnI (T6, T12, T24, T48, T72), peak cTnI (Cmax), increase in cTnI between 6 to 12 and 6 to 24 hours (TΔ6–12 and TΔ6–24), cumulative area under the curve (CAUC) cTnI (24, 48 and 72 hours) and occurrence of cTnI>13ng.ml−1 (T>13) at any time point were analysed as predictors of survival. Univariate logistic regression and multivariate Cox models (incorporating EuroSCORE) were used to model the probability of in-hospital or late death. The Akaike Information Criteria (AIC) was used to determine goodness of fit.
Results There were 56/440 deaths (100% follow-up) after a median (interquartile range) of 6.3(4.9–7.4) years. Univariate logistic regression analysis (odds ratio [95% CI]) found that T12 (1.03 [1.01–1.06], p=0.018), T24 (1.04 [1.02–1.07], p=0.002), T48 (1.08 [1.03–1.15], p=0.004), T72 (1.13 [1.02–1.24], p=0.020), T>13 (2.06 [1.07–3.96], p=0.031), TΔ6–12 (1.04 [1.01–1.07, p=0.003), TΔ6–24 (1.03 [1.01–1.05], p=0.018]), logeCAUC24 (1.65 [1.17–2.32], p=0.004), logeCAUC48 (1.77 [1.29–2.43], p<0.001) and logeCAUC72 (1.69 [1.25–2.30], p<0.001) predicted mortality but of these, TΔ6–24 and logeCAUC72 appeared superior based upon the lowest AIC. On Cox multivariate analysis, when EuroScore was added to the model (hazard ratio [95%CI]) Cmax(1.02 [1.01–1.03], p<0.001) and logeCAUC72 (1.57 [1.20–2.05], p=0.001) and EuroScore(1.46 [1.31–1.63]; p<0.001 were independent predictors of mortality.
Conclusions In human myocardial protection studies, serial troponin data should be collected up to 72 hours post-operatively to allow calculation of Cmax and logeCAUC72 as outcome surrogates. As these troponometrics best predict mid-term mortality.