Finding the Value of High-sensitivity Troponin Assays
Abstract & Commentary
By Michael H. Crawford, MD
Sources: deFilippi CR, et al. Association of serial measures of cardiac troponin T using a sensitive assay with incident heart failure and cardiovascular mortality in older adults. JAMA. 2010;304:2494-2502; de Lemos JA, et al. Association of troponin T detected with a highly sensitive assay and cardiac structure and mortality risk in the general population. JAMA. 2010;304:2503-2512.
New high-sensitivity cardiac troponin assays detect blood levels in many patients with chronic heart failure or ischemic heart disease who do not meet clinical criteria for myocardial infarction. Elevated levels in such patients have been shown to predict cardiovascular events. Thus, deFilippi and colleagues hypothesized that measurable high-sensitivity cardiac troponin T levels (cTnT) in elderly ambulatory patients without a history of heart failure would predict new-onset heart failure or cardiovascular death independent of traditional clinical risk factors, and that changes in cTnT over time also would be predictive. The 5,888 potential subjects age 65 or older came from the U.S. Multicenter Cardiovascular Health Study, and were recruited in the early 1990s. The 4,221 subjects without a heart-failure history and with enough serum saved to measure cTnT at entry and 2-3 years later, were followed for a median of 12 years. cTnT ≥ 3 pg/mL was detectable in 66% of the subjects. New-onset heart failure occurred in 30% and cardiovascular death in 26%. Higher cTnT levels were associated with many traditional risk factors such as history of vascular disease and ejection fraction, but not to race or weight. cTnT was weakly correlated with BNP and CRP. After adjustments for clinical factors, the hazard ratio (aHR) of cTnT for new heart failure was 2.5 (95% CI 2-3) among those with the highest cTnT levels (> 12 pg/mL) and 2.9 for cardiovascular death (2.4-3.6) compared to those with the lowest levels of cTnT. Increases in cTnT of > 50% on the second measurement also increased heart-failure risk (aHR 1.6, 1.3-2.0) and cardiovascular death (aHR 1.65, 1.35-2.03). A decrease of < 50% was associated with a lower risk of heart failure (0.73, 0.54-0.97) and cardiovascular death (0.71, 0.52-0.97). The addition of cTnT to clinical risk-factor models significantly, but modestly, improved risk prediction; differences in the area under the curve were 0.015 for heart failure and 0.013 for death. The authors concluded that in older subjects without known heart failure, initially elevated cTnT levels and increases in cTnT levels predicted the occurrence of heart failure and cardiovascular death.
de Lemos and colleagues addressed similar issues in a younger population of 3,546 randomly selected subjects aged 30-65 years enrolled in the population-based Dallas Heart Study between 2000 and 2002 and followed for 5-7 years. The endpoints of their study were cardiac MRI assessment of cardiac structure and function and mortality, as it related to five levels of high-sensitivity cTnT. The prevalence of detectable high-sensitivity cTnT was 25% in this population sample compared to 0.7% using the standard assay. If patients with known or suspected vascular disease (i.e., diabetics) were excluded, the prevalence was 16%. Detectable cTnT increased with age, being 58% in those between ages 60 and 65 years. Also, men and black subjects had significantly higher levels. MRI detected left ventricular hypertrophy, systolic dysfunction, and also chronic kidney disease increased with higher cTnT. In addition, all-cause mortality was higher as cTnT increased. After adjustment for traditional risk factors, CRP and BNP, cTnT was independently associated with mortality (aHR 2.8, 1.4-5.2). Adding cTnT to an adjusted mortality model modestly increased risk prediction. The authors concluded that in a middle-aged population, cTnT predicted structural heart disease and mortality.
These two studies demonstrate that this new high-sensitivity assay is going to detect cTnT in up to two-thirds of patients without known vascular disease, a history of heart failure, or diabetes, as compared to the standard assay, which will be positive in 1%-8% of such subjects. The positivity rate will increase with age, male sex, African descent, and the presence of cardiovascular disease. Detectable cTnT will be associated with death, structural heart disease, and the development of heart failure in a gradient fashion. In older subjects, a marked change in baseline levels will signal a change in the risk of cardiovascular events and death. In those with known stable ischemic heart disease, the detection of cTnT will approach 100%. As these high-sensitivity assays are embraced by clinical laboratories, what will be the implications for clinicians?
Since troponin measurements are now the lynch-pin of the diagnosis of acute myocardial infarction, the rate of false-positive diagnoses is bound to increase, leading to a potentially dangerous, but certainly inappropriate application of further testing and treatment. On the other hand, the infrequent negative test should precipitate immediate discharge from the emergency department.
In acute coronary syndromes without ECG ST segment elevation, a positive troponin puts the patient in a higher-risk category for severe coronary artery disease and subsequent ischemic events. Thus, a positive troponin usually leads to an invasive approach. Consequently, the number of patients with acute coronary syndromes treated invasively should increase with these assays. However, since most of these patients are treated invasively anyway, it may not make much difference in their management.
Patients with atypical chest pain and other non-ischemic causes of chest pain often get troponin measured, and positive results usually trigger at least a non-invasive cardiac evaluation. The number of such patients subjected to further evaluation likely will increase, which will increase healthcare costs for unclear gains.
The issue raised these by these two papers is whether a high-sensitivity cTnT measurement should be used as a preventative-medicine screening test. In order to answer this question, a prospective study of the clinical value of cTnT for preventing cardiovascular events should be necessary. Theoretically, you could use the information to plan the aggressiveness of your preventative measures. I am sure it would influence physician behavior, but it is unclear whether it would increase asymptomatic individuals' compliance with prevention measures. Studies have shown that imaging studies, when shown to patients, do not influence their behavior much, but acute illnesses requiring hospitalization do. It is unlikely that an obscure lab test would. Also, cardiac event rates are declining, and it may be difficult to prove that one lab test could influence them further. In addition, when added to clinical risk-factor models, especially those involving BNP, the incremental gain from cTnT was modest. BNP also is a good predictor of mortality, and some have suggested both be done for screening.
Finally, the mechanism of elevated troponin levels is not clear from these studies. In ischemic syndromes, the origin of troponin seems obvious, but it is unlikely that 16%-66% of the general population has myocardial ischemia. Perhaps if we understand the mechanism of an elevated value, we would have a better idea of the value of the test in the general population.