Cardiometabolic biomarker profiles associated with estimated 10-year ASCVD risk by the China-PAR model in an older adult health examination cohort: a cross-sectional study

Figures

• 

  Figure 1. Participant flow diagram. Among 566 individuals initially assessed, 55 were excluded because of incomplete clinical or laboratory data. Missingness involved hematologic/platelet measurements in 49 records, platelet indices only in five records, and WHR in one record. The final analytic cohort comprised 511 participants with complete data for the reported analyses. WHR, waist-to-hip ratio.

• 

  Figure 2. Scatter plots of laboratory indicators vs. China–PAR 10-year ASCVD risk scores (top 16 variables). This figure displays the 16 indicators most strongly correlated with the calculated China-PAR risk score. Each scatter plot illustrates the relationship between a single variable and the 10-year ASCVD risk score. Blue dots represent individual data points, the yellow trend line indicates the linear fit, and the light-yellow shaded area represents the 95% confidence interval. rho denotes the Spearman rank correlation coefficient, the P-value indicates the two-sided significance level of the correlation test, and N is the number of valid samples. WHR, waist-to-hip ratio; FPG, fasting plasma glucose; CysC, cystatin C; Waist, waist circumference; hs_cTnI, high-sensitivity cardiac troponin I; AC, atherogenic coefficient; GGT, gamma-glutamyl transferase; Creatinine, serum creatinine; HCY, homocysteine; NEUT_abs, neutrophil absolute count; UA, uric acid; LYM%, lymphocyte percentage; NT_proBNP, N-terminal pro-B-type natriuretic peptide; NEUT%, neutrophil percentage; PDW, platelet distribution width; MONO_abs, absolute monocyte count.

• 

  Figure 3. Results of feature selection by LASSO regression. (A) Ten-fold cross-validation curve: The x-axis represents the penalty parameter on a logarithmic scale [log(λ)], and the y-axis represents the mean cross-validated AUC with its standard deviation. The dashed vertical line marks lambda.min, at which the cross-validated AUC was 0.824, and the dotted vertical line marks lambda.1se, at which the AUC was 0.794. These AUC values quantify discrimination of the China-PAR-defined high-risk category and do not represent prediction of observed ASCVD events. (B) LASSO coefficient path plot: This plot shows the trajectories of coefficients for the top 25 variables as log(λ) changes. The x-axis is log(λ), and the y-axis is the coefficient. The dashed vertical line indicates lambda.min and the dotted line indicates lambda.1se. At lambda.min, four nonzero features were retained: CysC, FPG, AC, and NT-proBNP; at lambda.1se, CysC and FPG remained nonzero.

• 

  Figure 4. Distribution of calculated 10-year ASCVD risk scores and China–PAR, defined higH–risk proportions by quartile stratification of the four LASSO–selected biomarkers. (A) Cystatin C (CysC) stratification; (B) Fasting plasma glucose (FPG) stratification; (C) Atherogenic coefficient (AC) stratification; (D) N-terminal pro-B-type natriuretic peptide (NT-proBNP) stratification. The left bar charts show the mean ± SD of the calculated 10-year ASCVD risk score within each quartile group. P for trend indicates the P-value for the linear trend test. The right stacked bar charts show the proportional distribution of China-PAR risk categories (low, intermediate, and high) across quartiles, and P for trend (high-risk proportion) indicates the P-value for the monotonic trend in the proportion classified as high risk. Sex-stratified trend tests are descriptive and do not represent formal biomarker-by-sex interaction tests.