The robustness of our AUC72 estimations were assessed using a series of sensitivity
analyses. Firstly we conducted trapezoidal integration for area under the curve measurement as
above, and also considered additional covariates for the imputation model as follows: age,
gender, TIMI flow pre procedure, LAD culprit artery, symptom to intervention time and
procedural success. In the second sensitivity analysis, a repeated measures analysis was used to
estimate the overall profile of cTnI/CK release over the 72 hour window. All available biomarker
data were analyzed using linear mixed-effects regression with patient as a random effect together
with treatment group, time of assay, and an interaction term between treatment group and time of
assay included as fixed effects. For this analysis, the non-significant interaction term between
treatment group and time of assay was removed from the model. In the final sensitivity analysis,
trapezoidal integration was used for the estimation of area under the curve. Patients with one or
more missing biomarker assays were replaced by linear interpolation and extrapolation.