Author + information
- Cian McCarthy,
- Roland Van Kimmenade,
- Hanna Gaggin,
- Mandy Simon,
- Nasrien Ibrahim,
- Parul Gandhi,
- Noreen Kelly,
- Arianna Belcher,
- Shweta Motiwala,
- Rhonda Rhyne,
- Craig Magaret and
- James Januzzi
Background: Clinical prediction of major adverse cardiac events (MACE) in patients undergoing cardiac evaluation or entering clinical trials remains challenging due to low sensitivity and specificity. We assessed whether a multiple biomarker approach improves prediction of incident MACE (composite of cardiovascular death, myocardial infarction and stroke) in patients referred for diagnostic coronary angiography.
Methods: In a 649 participant training cohort undergoing coronary angiography, predictors of MACE within one year after the procedure were identified using least-angle regression; over fifty clinical variables and 109 biomarkers (obtained at the time of coronary angiography) were analyzed. Predictive models were generated using LASSO with logistic regression. A score derived from the final model was developed and evaluated with a 278 patient validation set.
Results: The scoring system developed from the training cohort consisted of four biomarkers (N-terminal pro B-type natriuretic peptide [NT-proBNP], kidney injury molecule 1, osteopontin and tissue inhibitor of metalloproteinase-1). In the validation cohort, each marker improved model discrimination or calibration for MACE; the final score had an area under receiver operating characteristic curve (AUC) of 0.79 (p < 0.001), higher than AUC for clinical variables (0.75). In net reclassification improvement (NRI) analyses, addition of other markers to NT-proBNP resulted in significant improvement (NRI=0.45; p =0.008); similar improvement was seen in comparison to clinical variables alone (NRI= 0.54, p = 0.002). In the validation cohort, bimodal distribution of events was noted; at optimal score cut-off, we found 64% sensitivity, 76% specificity, 28% positive predictive value, and 93% negative predictive value for MACE. In Kaplan Meier survival analyses, time to first MACE was substantially shorter in those with an elevated score (p <0.001); though derived for one year prediction, such risk extended to at least to 4 years.
Conclusions: In a general population undergoing diagnostic coronary angiography we describe a multiple biomarker score with high accuracy for predicting incident MACE within one year of follow up.
Poster Hall, Hall C
Friday, March 17, 2017, 10:00 a.m.-10:45 a.m.
Session Title: Traditional and Novel Factors Used to Assess the Risk of, and Used for the Treatment of, Coronary Artery Disease
Abstract Category: 2. Acute and Stable Ischemic Heart Disease: Clinical
Presentation Number: 1126-339
- 2017 American College of Cardiology Foundation