The present case-control study enrolled genetically unrelated patients treated at the Division of Cardiology, Bethune International Peace Hospital (Shijiazhuang, China), between August and December 2021. All patients suspected of CAD underwent coronary angiography (CAG). CAG images were independently evaluated by two experienced interventional cardiologists who were blinded to the clinical characteristics and laboratory data of the patients. Disagreements were resolved by consensus or by a third senior interventional cardiologist. CAD severity was assessed according to American Heart Association guidelines (26), with CAD defined as ≥50% luminal stenosis in at least one major epicardial coronary artery or its major branches. The number of diseased vessels was determined by examining the left anterior descending artery, left circumflex artery and right coronary artery (range, 0–3 vessels).

Two control groups were established: Group 1 comprised patients with coronary luminal stenosis ≤40% on angiography, normal left ventricular ejection fraction (LVEF) and normal regional wall motion, while group 2 consisted of healthy individuals randomly selected from those undergoing routine health examinations at the Bethune International Peace Hospital. Clinical and demographic data, including baseline characteristics and laboratory parameters, were extracted from electronic medical records. Serum lipid profiles, including total cholesterol, triglycerides, high-density lipoprotein-cholesterol (HDL-C), and low-density lipoprotein-cholesterol (LDL-C), were measured using standard enzymatic colorimetric methods on a Beckman Coulter AU5800 automated clinical chemistry analyzer (Beckman Coulter, Inc., Brea, CA, USA).

Retrospectively-collected residual clinical blood specimens obtained within 24 h of admission were utilized; serum samples were allocated for HP electrophoresis-based phenotyping and quantitative determination, and anticoagulated venous blood samples were reserved for HP PCR-based genotyping analysis.

Serum HP concentrations were quantified using a BN™ II nephelometer system (Siemens Healthineers) employing immunoturbidimetric methodology. Detection was performed using N Antiserum to Human Haptoglobin (Cat. No. OSAS09; Siemens Healthineers) with N Protein Standard SL (Cat. No. OQIM13) for calibration. Quality control was performed using N/T Protein Control SL (Cat. No. OQIN13) and N/T Protein Control L (Cat. No. OQIO13; Siemens Healthineers). The lower detection limit of the assay was 0.08 g/l. Routine instrument maintenance, calibration and quality control procedures were performed according to manufacturer specifications and ISO9001 quality management standards (27).

HP genotypes were determined using two complementary methods: Reverse transcription-quantitative PCR (RT-qPCR) and native polyacrylamide gel electrophoresis, ensuring result validation and accuracy. HP genotyping was performed using a TaqMan-based RT-qPCR assay adapted from previously validated methods (28,29). Genomic DNA was extracted from EDTA-anticoagulated peripheral blood samples using a GeneRotex 96 automated nucleic acid extraction system with the Magnetic Bead-based Whole Blood Genomic DNA Extraction kit (cat. no. T148H; Xi'an Tianlong Science and Technology Co., Ltd.,) according to the manufacturer's instructions.

This assay determined HP genotypes by evaluating the relative copy number of the HP2 allele-specific duplication junction region. The HP-5′ promoter region, which is present in all individuals regardless of genotype (28,29), was used as an internal genomic control for normalization. Primers and TaqMan probes were synthesized by Sangon Biotech Co., Ltd., with the following sequences: HP5, forward 5′-CACATTTACTGATTTCAGGCTGGA-3′, reverse 5′-CCTTTTCACAGTAATTTTCTCCACCT-3′, TaqMan probe 5′-FAM-AGCTTTTAAGCAATAGGGAGATGGCCACA-BHQ1-3′; HP2, forward 5′-GGAGCTGCTCTGCACATCAA-3′, reverse 5′-CCCTTTCAATGAATTTCAGGGA-3′, TaqMan probe 5′-VIC-ACCCCGAATAGAAGCTCGCGAACTGTA-BHQ1-3′; and HPdel, forward 5′-TCTTTATGGCACTGGGGAACA-3′, reverse 5′-AGCAAGACACTCGTGAGTGGAA-3′, TaqMan probe 5′-ROX-TGTGCAAGAGCCTTTCCAATTTTGATCA-BHQ2-3′. Each 20 µl PCR reaction contained 2 µl genomic DNA template, 10 µl Premix Ex Taq™ (Perfect Real Time; cat. No. RR039A; Takara Bio Inc.,), 1 µl each of HP2 forward and reverse primers (300 nM), 1 µl each of HPdel forward and reverse primers (300 nM), 0.5 µl each of HP5 forward and reverse primers (150 nM), 1 µl each of HP2-TaqMan and HPdel-TaqMan probes (83 nM), 0.5 µl of the HP5-TaqMan probe (42 nM) and nuclease-free water to reach the final volume. An AGS4800 Real-Time PCR System (AGS Technologies Co., Ltd) was used for amplification. The PCR cycling conditions were as follows: Initial denaturation at 95°C for 10 min, followed by 40 cycles of denaturation at 95°C for 10 sec and annealing/extension at 60°C for 1 min.

HP genotypes were determined using the comparative Cq method (2^−ΔΔCq method) (30). For each sample, the following calculations were performed: ΔCq=Cq(HP5′)-Cq(HP2); ΔΔCq=ΔCq(sample)-ΔCq(reference); HP2/HP5′ ratio=2^−ΔΔCq; where the reference was genomic DNA from a confirmed HP2/HP2 homozygotic individual. Genotypes were assigned based on the HP2/HP5′ ratio: i) HP1/HP1 was assigned when no notable HP2 signal was detected; ii) HP2/HP1 was defined as a ratio ranging from 0.34–0.50; and iii) HP2/HP2 individuals displayed ratios ranging from 0.79–0.98. The presence of HPdel was identified by detection of the HPdel-specific signal, and samples showing HPdel signals were further categorized as HP1/HPdel, HP2/HPdel or HPdel/HPdel based on the combination of HP2 and HPdel signal patterns.

HP phenotypes were confirmed by native polyacrylamide gel electrophoresis of HP-hemoglobin complexes (10,31,32). Briefly, 10 µl serum samples were mixed with 2 µl 10% (v/v) hemoglobin solution and incubated at room temperature for 10 min to allow formation of HP-hemoglobin complexes. Samples were then mixed with an equal volume of non-denaturing loading buffer, consisting of 125 mM Tris-HCl, 20% glycerol and 0.001% bromophenol blue at pH 6.8.

Electrophoresis was performed using a discontinuous polyacrylamide gel system, which consisted of a 4% stacking gel (125 mM Tris-HCl; pH 6.8) and a 4.7% resolving gel (360 mM Tris-HCl; pH 8.8). Initial electrophoresis was conducted at 120 V until the bromophenol blue front approached the resolving gel, followed by 150 V until completion. Gels were stained using a chromogenic substrate solution containing 5 ml 0.2% (w/v) 3,3′,5,5′-tetramethylbenzidine in methanol, 0.5 ml dimethyl sulfoxide, 10 ml 5% (v/v) glacial acetic acid, 1 ml 1% (w/v) potassium ferricyanide and 150 µl 30% (w/w) hydrogen peroxide. Staining was performed at room temperature for 5–15 min.

Statistical analyses were conducted using Python (version 3.9; Python Software Foundation) with pandas (33), NumPy (34), scipy.stats (35) and scikit-learn libraries (36). Continuous variables were presented as mean ± standard deviation and categorical variables as frequencies (percentages). Between-group comparisons were conducted using χ² test for categorical variables and one-way ANOVA) or Kruskal-Wallis tests for continuous variables, depending on data distribution assessed by the Shapiro-Wilk test. Following significant ANOVA results, post hoc pairwise comparisons were performed using Bonferroni correction. For Kruskal-Wallis tests, Mann-Whitney U tests with Bonferroni correction were applied for post hoc pairwise comparisons. For χ² tests involving multiple groups, post-hoc pairwise comparisons were performed using individual χ² tests with Bonferroni correction for multiple comparisons (adjusted α=0.05/number of comparisons).

Correlation analyses employed Pearson's or Spearman's coefficients as appropriate. Multivariate logistic regression identified independent CAD risk factors, including variables with P<0.2 from the univariate analysis. Model performance was evaluated using receiver operating characteristic (ROC) curve analysis and area under the curve (AUC) calculations.

Genotype-stratified analyses explored HP genotype-specific associations by performing the aforementioned correlation analysis methods within each genotype subgroup (n≥10). Between-genotype HP concentration comparisons were conducted using Kruskal-Wallis tests with η² effect sizes.

Missing data were handled by complete case analysis, for example the quantity of LVEF missing data was 18.3%. Statistical significance was set at α=0.05.

The present study enrolled 230 patients with CAD, 83 participants in the control 1 group and 192 participants in the control 2 group. Key baseline characteristics are summarized in Table I. The CAD group was predominantly male (74.8%), significantly more compared with the control groups 1 (56.6%; P=0.003) and 2 (53.1%; P<0.001). Smoking prevalence was significantly higher in the CAD group (30.0%) compared with the control groups 1 (16.9%; P=0.029) and 2 (15.1%; P<0.001). Hypertension was significantly more frequent in the CAD group (62.6%) compared with in the control 2 group (26.0%; P<0.001). Similarly, diabetes mellitus was significantly more prevalent in the CAD group (21.3%) compared with in control group 2 (8.3%; P<0.001). Genetic history was rare but present only in the CAD and control 1 groups.

CAD group demonstrated significantly lower high-density lipoprotein-cholesterol (HDL-C) levels (1.04±0.24 mg/dl) compared with control groups 1 (1.12±0.16 mg/dl; P=0.004) and 2 (1.33±0.27 mg/dl; P<0.001) and significantly higher triglyceride levels (1.62±0.88 mg/dl) than control group 2 (1.12±0.32 mg/dl; P<0.001). LVEF was significantly lower in the CAD group (0.619±0.077) than in control group 1 (0.647±0.044; P=0.003). No significant differences were observed in median age, drinking prevalence, total cholesterol or low-density lipoprotein-cholesterol between the CAD group and either control group.

Of 505 samples analyzed, 450 showed concordant results between the two genotyping methods, yielding an overall concordance rate of 89.11% (450/505; data not shown). The remaining 55 samples required complementary use of both methods for accurate genotyping. A representative electrophoretic typing profile of HP is presented in Fig. S1.

Serum HP levels demonstrated significant genotypic and group variations (Table II). Within each genotype, the CAD group consistently displayed the highest HP concentrations. Specifically, Hp2-2 carriers in the CAD group demonstrated significantly higher serum HP levels (1.50±0.84 g/l) compared with those in control groups 1 (1.11±0.54 g/l; P=0.004) and 2 (0.85±0.42 g/l; P<0.001). Similarly, Hp2-1 carriers in the CAD group (2.01±0.86 g/l) had significantly higher HP levels than their counterparts in control groups 1 (1.39±0.39 g/l; P<0.001) and 2 (1.19±0.47 g/l; both P<0.001). The Hp1-1 genotype also showed significantly higher levels in the CAD group than in control group 2 (2.03±0.79 vs. 1.31±0.61 g/l; P=0.001).

HP genotype distribution differed significantly among CAD, control 1 and control 2 groups (χ²=25.014; degrees of freedom, 8; P=0.002; data not shown). In CAD vs. control 2 group comparisons (Table III), Hp2-2 was significantly more prevalent in patients with CAD than in control group 2 [54.8 vs. 41.1%; χ²=7.254; P=0.007; odds ratio (OR), 1.73; 95% confidence interval (CI), 1.18–2.55]. Hp2-1 showed a non-significant trend toward lower frequency in the CAD group compared with control group 2 (30.0 vs. 39.1%; χ²=3.431; P=0.064; OR, 0.67; 95% CI, 0.45–1.00). No significant differences were observed for Hp1-1 (P=0.201) or deletion genotypes (both P>0.999). Detailed statistical comparisons are listed in Table III. This comparison was selected for detailed analysis because control group 2 provided greater statistical power (n=192 vs. n=83 for control group 1) and demonstrated the most substantial differences in genotype distribution compared with the CAD group.

Table IV presents univariate and multivariate logistic regression analyses to identify independent CAD risk factors. In the multivariate analysis, HP concentration emerged as the strongest independent predictor (OR, 4.556; P<0.001), followed by hypertension (OR, 3.772; P<0.001), triglycerides (OR, 2.682; P=0.002) and HDL-C as a protective factor (OR, 0.011; P<0.001). Among HP phenotypes, Hp2-2 showed an independent association with CAD risk (OR, 1.781; P=0.045) compared with a Hp1-1 reference, representing a 78% increased risk even after adjusting for HP concentration and traditional cardiovascular risk factors. Other HP phenotypes showed no significant associations.

Visualization by forest plot demonstrated the magnitude and direction of associations for the previously identified significant predictors of CAD, with HP concentration and hypertension showing the strongest positive associations with CAD risk, whereas HDL-C levels demonstrated the most pronounced protective effect (Fig. 1A). ROC curve analysis demonstrated the superior discriminative ability of the multivariate model (AUC=0.841) compared with HP concentration alone (AUC=0.686), representing a clinically meaningful improvement of 0.155 in AUC value (Fig. 1B). The multivariate model achieved good discriminative performance, indicating its potential clinical utility for the assessment of CAD risk.

Significant differences in serum HP levels were observed across genotypes (ANOVA F=12.28; P<0.001). Hp1-1 and Hp2-1 genotypes exhibited the highest HP concentrations (2.03±0.79 and 2.01±0.86 g/l, respectively), whereas Hp2-2 showed intermediate levels (1.50±0.84 g/l), and Hp2-del demonstrated the lowest concentrations (0.66±0.92 g/l) (Fig. 2A).

Genotype-stratified correlation analysis revealed distinct patterns of association between HP levels and cardiac function (Fig. 2B). In the Hp1-1 group (n=17), serum HP concentration showed a moderate negative correlation with LVEF (r=−0.607; P=0.010), explaining 36.8% of the variance in cardiac function (Fig. 2C). Similarly, the Hp2-2 group (n=99) demonstrated a notable negative correlation with cardiac function (r=−0.249; P=0.013) (Fig. 2D). By contrast, Hp2-1 and Hp2-del genotypes showed no significant associations with LVEF (P>0.05).

Regarding CAD severity, only the Hp1-1 group showed a trending positive correlation between HP levels and vessel count (r=0.410; P=0.065), whereas other genotypes demonstrated no significant associations (data not shown). These findings suggest that HP concentration serves as a clinically relevant biomarker primarily in specific genotype subgroups, supporting the implementation of genotype-stratified HP assessment in cardiovascular risk evaluation.