Troponin in Sepsis

To the Editor:

Frencken and colleagues measured high-sensitivity cardiac troponin I (hs-cTnI) levels in patients with community-acquired pneumonia and sepsis, and reported elevations above the upper limit of normal in 85% of their cohort (1). Their interpretation of this result was that myocardial injury due to oxygen supply–demand mismatch was responsible for the elevated hs-cTnI. The authors’ findings are interesting, and the associations between elevated hs-cTnI and abnormalities in laboratory tests related to inflammation and coagulation deserve exploration. Nevertheless, we are troubled by certain aspects of the report.

First, the upper limit of normal for elevated hs-cTnI is based on levels in a reference population of healthy volunteers without apparent disease. To apply that cutoff to patients with severe acute disease may not be appropriate (2). Indeed, recent data suggest that the cutoff for abnormal hs-cTnI in acutely ill hospitalized patients may be over four times higher (3). The results of the current study serve mainly to confirm prior studies showing that elevated troponin is a common finding in patients with sepsis (4).

Second, the claim that elevated hs-cTnI represents myocardial ischemia appears to be largely unsupported. Hs-cTnI is a specific marker for myocardial ischemia only in the appropriate clinical scenario. Outside of a scenario that enriches the pretest probability of ischemic cardiac disease (e.g., angina in a patient at risk), the significance of elevated hs-cTnI is uncertain. Indeed, the authors suggest this by reporting that only 30% of the cohort had troponin levels sent for clinical indications, with only 16 of 29 patients having 12-lead electrocardiography that showed signs of ischemia. Elevated hs-cTnI in the absence of other signs of an acute coronary syndrome is nonspecific and has been documented in many diseases, and even in endurance athletes after strenuous exercise (5). The authors posit “myocardial oxygen supply–demand mismatch,” but they offer only indirect evidence for this. They base this postulate on a logistic regression model that associated risk factors for coronary atherosclerosis with elevated troponin levels, but offer no direct evidence of myocardial ischemia as a cause of elevated hs-cTnI. For example, they did not report whether hs-cTnI levels were higher in the 16 patients who had electrocardiographic findings of ischemia than in the 13 patients without such signs. The logical extension of the authors’ conclusions would be that endurance athletes with elevated hs-cTnI levels also have myocardial oxygen supply–demand mismatch, which is preposterous. A more likely explanation for the reported observation is that hs-cTnI levels are elevated nonspecifically by a variety of stressors, including serious illness, where elevated hs-cTnI is a marker of disease severity.

Third, the mechanism of hs-cTnI elevation and its causal significance is open to speculation and further exploration. To conclude that hs-cTnI release was caused by myocardial injury due to impaired oxygen delivery is a false syllogism that equates a positive blood test with the presence of a disease (6). This is a form of the base rate fallacy: when a large, undifferentiated population is tested without establishing the true prevalence of the disease, we expect false positives. If a test with less than 100% specificity is used as the sole criterion for diagnosing a disease, the prevalence of the disease will increase in proportion to the prevalence of testing. To suggest that we use hs-cTnI as a screening test for sepsis-induced organ injury and hope for a way to accelerate its clearance is likely to lead to overdiagnosis and therapeutic misadventure.

Frencken and colleagues add interesting observations to the substantial evidence base on troponin elevations in the critically ill. However, mechanistic explanations and clinical applications will require much additional work.