Differences in measurement of high-sensitivity troponin in an on-demand and batch-wise setting

Background: Most studies assessing the diagnostic value of high-sensitivity troponin in the diagnosis of myocardial infarction used batch-wise analyses of frozen samples for high-sensitivity troponin measurements. Whether the accuracy of these batch-wise high-sensitivity troponin measurements described in diagnostic studies is comparable to clinical routine is unknown. Methods: We enrolled 937 patients presenting with suspected myocardial infarction in this prospective cohort study. Measurements of high-sensitivity troponin I (Abbott Architect) and high-sensitivity troponin T (Roche) were performed in two settings: (a) on-demand in clinical routine using fresh blood samples; and (b) in batches using frozen blood samples from the same individuals at three timepoints (0 hours, 1 hour and 3 hours after presentation). Results: Median troponin levels were not different between on-demand and batch-wise measurements. Troponin levels in the range of 0 to 40 ng/L showed a very high correlation between the on-demand and batch setting (Pearson correlation coefficient (r) was 0.92-0.95 for high-sensitivity troponin I and 0.96 for high-sensitivity troponin T). However, at very low troponin levels (0 to 10 ng/L) correlation between the two settings was moderate (rfor high-sensitivity troponin I 0.59-0.66 and 0.65-0.69 for high-sensitivity troponin T). Application of guideline-recommended rapid diagnostic algorithms showed similar diagnostic performance with both methods. Conclusions: Overall on-demand and batch-wise measurements of high-sensitivity troponin provided similar results, but their correlation was moderate, when focusing on very low troponin levels. The application of rapid diagnostic algorithms was safe in both settings.