Clinical impact of the 2010–2012 low-end shift of high-sensitivity cardiac troponin T

The clinical implications of the 2010–2012 low-end shift of high-sensitivity cardiac troponin T (hs-cTnT) regarding possible misdiagnosis of acute myocardial infarction are largely unknown.

We aimed to quantify the impact of the 2010–2012 low-end shift and adjustment issue in 857 patients presenting to the emergency department with suspected acute myocardial infarction by comparing measurements performed with affected 2010–2012 lots with recalculated 2010–2012 values using a linear regression formula (provided by the manufacturer) and the corrected assay (re-measured in 2013). The final diagnosis was adjudicated by two independent cardiologists using all information including coronary angiography, echocardiography and serial hs-cTnT levels (with the corrected 2013 assay).

Acute myocardial infarction was the adjudicated diagnosis in 195 patients (22.7%). Median hs-TnT values were 8.5 ng/l for affected lots, 11.1 ng/l with recalculated and 10 ng/l with the corrected assay (P<0.001 for all comparisons). Spearman correlation coefficient was 0.937 (<0.001) for correct and affected respective correct and recalculated values. The Cusum test indicated significant deviation from linearity (P<0.01) for both correlations. Deviations nearly exclusively affected hs-cTnT levels below the 99th percentile (14 ng/L). Among the 195 patients with an adjudicated diagnosis of acute myocardial infarction, no patient was misclassified using affected lots if using conventional serial sampling. In contrast, misdiagnosis of acute myocardial infarction was significantly increased by affected lots if applying the novel ESC 0 h/1 h algorithm for the early rule-out of acute myocardial infarction (negative predictive value with affected lots 97.7% versus 99.7% with corrected lots).

The 2010–2012 hs-cTnT low-end shift affected nearly exclusively levels below the 99th percentile cut-off. While it did not affect the diagnosis of acute myocardial infarction when using conventional serial sampling as done in 2010–2012, it would impact on new early rule-out strategies using very low levels of hs-cTnT such as the ESC 0 h/1 h algorithm.

NCT0047058, NCT00470587