Background: Most studies supporting high-sensitivity cardiac troponin (hs-cTn) assays have been performed outside the United States (US), and primarily in patients with chest discomfort. These populations differ from US clinical practice, where cTn testing is performed in more heterogeneous populations. We compared the diagnostic performance of hs-cTnI between all-comers and those specifically with chest discomfort; incidence of type 1 (T1MI) and 2 myocardial infarction (T2MI) was also analyzed
Methods: This is a post-hoc analysis of UTROPIA (NCT02060760), which was a prospective study of 1,631 consecutive adults presenting to one US emergency department (ED) in whom at least 2 serial cTnI were obtained at clinicians’ discretion. Investigational hs-cTnI (Abbott) 99th percentile upper-reference limits (URL) were 16 ng/L for females and 34 ng/L for males. All cases were adjudicated by two physicians, with a third in cases of disagreement. Diagnostic performance of serial 0/3 hour hs-cTnI values ≤ URL for acute myocardial infarction (MI) was determined for both all-comers and those with any complaint of chest discomfort. Negative predictive value (NPV), positive predictive value (PPV), sensitivity and specificity are presented with associated 95%CIs. Comparisons of proportions were done via Chi-square.
Results: >Acute MI occurred in 10.4% (n = 170) of all-comers (n = 1,631) compared to 9.3% (n = 78) in the chest discomfort subset (n = 835) (p = 0.40). T1MI was more frequent in those with chest discomfort (62% of MIs) than in all-comers (40% of MIs) (p = 0.002).
Using 0/3h hs-cTnI ≤ URL, among all-comers, sensitivity was 95.4% (95%CI 91.5-99.3), specificity 85.7% (83.5-87.9), NPV 99.4% (98.9-99.9) and PPV 43.2% (36.9-49.4). Among patients with chest discomfort, sensitivity was 94.2% (87.9-100), specificity 91.9% (89.5-94.2), NPV 99.4% (98.7-100) and PPV 53.9% (43.6-64.1).
Conclusions: A more restrictive study design focused on patients with chest discomfort influenced the T1MI/T2MI ratio and impacted the diagnostic specificity and PPV for diagnosis of acute MI by hs-cTnI. For hs-cTnI ≤ URL at 0/3 hours, in both groups, sensitivity was approximately 95% and NPV 99%. We suggest that US-based hs-cTn studies focus on ED all-comers, rather than more restrictive subgroups, to mirror clinical decision making and avoid overestimation of specificity and PPV.