Statistical Analysis

Data normality was assessed with the Shapiro–Wilk test. Comparisons between CAD and non-CAD groups or low p75^NTR and high p75^NTR groups were done using the t-test or the Mann–Whitney U test according to the distribution, or with the Chi-square test for dichotomous variables. Univariate linear models were done using IBM SPSS Statistics 25. Multivariate linear models including 23 clinical, biochemical and demographic variables, the adaptive least absolute shrinkage and selection operator (ALASSO) (Zou, 2006; Friedman et al., 2010), and random forests (Liaw and Wiener, 2002) were implemented using R version 4.1.0 (R Development Core Team, 2021). For ALASSO compatibility purposes, skewed continuous covariates were log-transformed then standardized. The 95% post-selection confidence intervals were constructed for the estimated covariate coefficients selected by ALASSO (Tibshirani et al., 2019). For random forests, the number of candidate variables, the maximum number of nodes per tree and the number of trees were tuned by selecting the lowest mean-squared error using 10-fold cross validation. Adjusted p-values < 0.05 were considered statistically significant. Graphical representations were plotted using GraphPad Prism Software version 8 for Windows (San Diego, CA, United States).