Diagnostic performance of enzymes in the discrimination of myocardial infarction

We apply rigorous statistical methods to assess the diagnostic value of creatine kinase, creatine kinase-MB, aspartate aminotransferase, lactate dehydrogenase, and alpha-hydroxybutyrate dehydrogenase in realistic circumstances. The trade-off between false positives and false negatives obtained with various decision rules is examined with the receiver-operator characteristic function and with discriminant analysis. During the three days after an infarction, lactate and alpha-hydroxybutyrate dehydrogenase can provide diagnostic thresholds of constant sensitivity and specificity. By contrast, sensitivity progressively decays for creatine kinase, creatine kinase-MB, and aspartate aminotransferase. The diagnostic uncertainty introduced by the infarction's varying age at hospitalization is evaluated by subjecting the mixed patient population to discriminant analysis. For some enzymes, repeating the same assay on the second day contributes to sensitivity, while adding a different assay on the first day enhances specificity. The effects of lower or higher infarction prevalences on sensitivity and specificity are most favorable when creatine kinase-MB is combined with lactate or alpha-hydroxybutyrate dehydrogenase. Adding a third assay is ineffective. The further differentiation according to infarction stage produces only 81% correct classifications, even when five different assays are used on two consecutive days. As a general strategy, we recommend the assay of two enzymes on two consecutive days.