Article Text
Abstract
Background Novel biomarkers are constantly emerging in heart failure (HF), assisting in risk stratification and decisions involving therapies of limited resource. Cardiac troponins are elevated in a substantial proportion of patients with decompensated HF and have recently been recognised as predictors of poor prognosis. We aimed to determine whether small rises in troponin I, within the range normally reported as negative, would provide further prognostic information, and if so, how does this compare with the established biomarker B-type natriuretic peptide (BNP).
Methods Consecutive patients admitted to two large city hospitals between 1 December 2006 and 4 November 2008 with decompensated HF were invited to participate in an ongoing observational study. A confirmed diagnosis required typical clinical findings and BNP greater than 100 pg/ml. Patients with an acute coronary syndrome complicated by pulmonary oedema were excluded. The troponin I assay used (Abbot Architect) reports negative as less than 0.04 ng/ml: the level that achieves 10% coefficient of variance. However, analytical sensitivity for this assay is 0.015. Patients were classified according to the actual troponin I value: less than 0.02 (group1), 0.02–0.039 (group2) and 0.04 or greater (group3). Survival for each group was determined using the Kaplan–Meier method. A Cox proportional hazard model was employed to estimate hazard ratios (HR) for troponin group univariately, then within a multivariable model alongside age, sex and BNP.
Results 835 patients with clinical and BNP criteria for HF were enrolled. 765 patients had troponin I measured. 342 patients had a troponin I initially reported as negative (<0.04 ng/ml) and actual values were available for 273 patients. Those without actual values were excluded from this analysis. Median troponin I was 0.05 ng/ml (interquartile range 0.02–0.18). Troponin I was less than 0.02 ng/ml in 166 patients (group 1) and 0.02–0.039 ng/ml in 107 patients (group 2). Group 3 comprised 429 patients with troponin I of 0.04 ng/ml or greater. Follow-up was available for a median of 367 days (range 9–725) and 171 deaths occurred. The troponin I group predicted outcome in survival analysis: group 1, 89%; group 2, 65.2% and group 3, 58.3% (log rank p<0.001), see fig. Univariate analysis demonstrated that group 3 was at highest risk of death during follow-up, with group 2 at intermediately increased risk compared with group 1, see table 1. Following multivariable adjustment for age, sex and BNP, the troponin group remained a significant predictor of death, see table 2.
Conclusion Using the exact troponin I identified patients at incrementally increased risk of mortality following admission with HF, independent of age, sex and BNP. Patients conventionally seen collectively as “troponin I negative” therefore comprise a heterogeneous population, and include a proportion with a clinically significant and identifiable increased risk of death. More sensitive troponin assays may improve prognostication following admission with HF.