Cardiac troponins and growth-differentiation factor-15 (GDF-15) for prediction of angiographic coronary artery disease in chest pain patients without myocardial injury

Abstract

Abstract Funding Acknowledgements Type of funding sources: Private grant(s) and/or Sponsorship. Main funding source(s): Grieg Foundation and Western Norway Regional Health Authority. Background There is a growing evidence that patients with angiographic signs of coronary artery disease should receive preventive therapy to reduce the risk of future adverse cardiac events. Purpose Evaluate whether cardiac troponin T and I (cTnT and cTnI) concentrations within the range from the limit of detection of the assay to the 99th percentile or GDF-15 may help identify patients with angiographic evidence of coronary artery disease (CAD). Methods A subgroup of 550 patients from the WESTCOR study with both cTnT and cTnI concentrations ≤99th percentile was evaluated with coronary computed tomography angiography (CCTA). CAD was defined as ≥10% narrowing of lumen of any coronary artery on CT angiogram or too high calcium score for anatomy to be evaluated. Overall diagnostic accuracy of cTnT, cTnI and GDF-15 and their combination were evaluated calculating the area under the receiver-operating characteristics curve (ROC-AUC). cTn analysis were performed using high-sensitivity assays from Roche Diagnostics (cTnT, limit of detection of 5 ng/L and gender-neutral 99th percentile of 14 ng/L) and Abbott Diagnostics (cTnI, limit of detection 1.6 ng/L and 99th percentile 16 ng/L for women and 34 ng/L for men). GDF-15 analyses were performed using the GDF-15 assay on Cobas e801 (Roche Diagnostics). Results A total of 53.5 percent (294 of 550 patients) had angiographic coronary disease. The risk of having CAD was significantly higher in patients with cTnT between 5 ng/ml and the gender-neutral 99th percentile compared to very low levels of cTnT <5 ng/ml, 65.7% (95% CI: 60-71%) vs 38.1% (95% CI: 32-44%, p-value for difference <0.001). Similar numbers for cTnI between 2 ng/ml and the gender-specific 99th percentile compared to very low levels <2 ng/ml was 62.9% (95% CI: 57-69%) vs 43.2% (95% CI: 37-49%, p-value for difference <0.001). The odds ratio for CAD if cTnT was ≥5 ng/ml or cTnI ≥2 compared to very low levels was 3.1 (2.1-4.4) for cTnT and 2.2 (1.6-3.1) for cTnI. For GDF-15, the risk of having CAD was lower in patients with low values <1200 ng/L (49.7%, 95% CI: 45-54%) compared to patients with higher concentrations >1200 ng/L (75.9%, 95% CI: 66-86%, p-value for difference <0.001). The accuracy of cTnT was significantly higher than for cTnI for identifying CAD (AUC 0.67, 95% CI 0.63-0.71 vs. 0.61, 95% CI 0.58-0.66, p-value for difference <0.01), and borderline higher than for GDF-15 (AUC 0.62, p-value for difference 0.067). Adding GDF-15 to cTn did not increase diagnostic performance of cTnT (AUC 0.67 vs. 0.68, p-value for difference 0.56) or cTnI (AUC 0.61 vs. 0.63, p-value for difference 0.23). Conclusion Overall diagnostic performance of cTnT is superior to cTnI in identifying patients without angiographic evidence of CAD. Adding GDF-15 does not increase overall diagnostic performance. Legend: Figure 1. Receiver-operating characteristics curve of cTnT, cTnI and GDF-15 for the identification of CAD (≥10% lumen reduction).