Patients and study design

Inclusion criteria required that participants were aged between 18 and 90 years, capable and willing to provide written informed consent, and hospitalized with a confirmed diagnosis of CCS.

Key exclusion criteria comprised: 1) pre‑existing active inflammatory state (high‑sensitivity C‑reactive protein ≥10 mg/l); 2) uncontrolled infectious diseases; 3) known autoimmune or immune‑related diseases (eg, systemic lupus erythematosus, asthma, inflammatory bowel disease, gout, malignant tumor, etc.); 4) in‑stent restenosis; 5) poor‑quality OCT images; and 6) unsuitability for this trial as deemed by the investigators (eg, severe renal / hepatic dysfunction, hemodynamic instability, inability to provide informed consent).

A total of 102 consecutive patients with CCS who underwent OCT at the Cardiology Department of Shanghai East Hospital between May and October 2025 were included in this cross‑sectional study. A total of 22 patients were excluded for the following reasons: refusal to participate (n = 12), in‑stent restenosis (n = 5), poor‑quality OCT images (n = 4), and suspected systemic inflammatory disease (n = 1). Ultimately, a cohort of 80 patients underwent comprehensive evaluation (Figure 1). Informed consent was acquired from all participants, and the study received approval from the Ethics Committee of Shanghai East Hospital (2025505).

Atomic absorption spectroscopy

The patients scheduled for enrollment following coronary angiography (CAG) evaluation had approximately a 5‑ml of coronary artery blood sample drawn from the site of the target plaque with a 5F Tig catheter (Terumo Corporation, Tokyo, Japan). The samples were collected into copper‑free centrifuge tubes, and centrifuged at 3000 rpm for 10 minutes, after which the upper serum was carefully collected and preserved in liquid nitrogen for analysis. Serum copper levels were quantified using a TAS‑990 atomic absorption spectrometer (Beijing General Purpose, Beijing, China).¹²

Enzyme‑linked immunosorbent assay

Inflammatory markers, including interleukin 6 (IL‑6), IL‑10, and matrix metalloproteinase‑9 (MMP‑9), were evaluated using commercially available biotin / avidin‑based enzyme‑linked immunosorbent assay kits (Jianglaibio, Shanghai, China).

Data collection

Data for this study were systematically collected from the hospital electronic health record system. The extracted variables encompassed baseline demographics, medical history, relevant laboratory parameters, and angiographic findings. Furthermore, the complexity of coronary artery lesions was assessed for every patient using the SYNTAX score I, which was calculated via the official online calculator available at https://syntaxscore.org.

Optical coherence tomography imaging

The OCT imaging was performed by withdrawing the imaging catheter through the target vessel segment. The ILUMIEN OPTIS OCT imaging system (C408661, Abbott Medical, Santa Clara, California, United States) and a Dragonfly OCT catheter (C408645, Abbott Medical) were utilized. Analyses of the OCT images were conducted by 2 experienced OCT technicians blind to the patient clinical status. When there was a significant difference in the findings of the 2 technicians, a senior physician performed a comprehensive evaluation. Based on the OCT image characteristics, the plaque types were divided into 4 categories: 1) fibrous plaques, characterized by intimal thickening with a homogeneous bright signal; 2) calcified plaques, presenting as intimal thickening with low signal intensity and sharply delineated borders; 3) lipid plaques, identified by intimal thickening with hypointensity and poorly defined boundaries; and 4) mixed plaques, defined by the presence of 2 or more of the aforementioned components. In addition to qualitative plaque characterization, the following key quantitative metrics were systematically collected: the minimal lumen area and the plaque burden. The minimal lumen area was measured at the narrowest point of the lumen, while the plaque burden was calculated to quantify the proportion of the vascular cross‑sectional area occupied by the plaque relative to the total area defined by the external elastic membrane. To mitigate the risks associated with prolonged procedure time and minimize the potential for contrast‑induced renal toxicity, it is a conventional clinical practice to perform OCT solely on the vessel with the most severe stenosis for prepercutaneous coronary intervention lesion assessment and procedural guidance. Accordingly, in patients with multivessel disease, the examination was confined to the vessel exhibiting the most severe stenosis.

HRPs on OCT were identified by the presence of at least 1 of the following criteria: 1) macrophage infiltration, defined as the presence of signal‑rich superficial clusters or punctate bright spots with dorsal shadowing^16,17; 2) thin‑cap fibroatheroma (TCFA), identified as a signal‑poor lipid core encompassing at least 90 ° of the vessel circumference and covered by a fibrous cap with a thickness of up to 65 µm¹⁸; 3) fibrous cap rupture, presenting as a clear discontinuity of the fibrous cap that establishes communication between the lumen and the underlying necrotic core; 4) severe stenosis, defined as a reduction in the lumen area of at least 90%³; 5) spotty calcification, appearing as a punctate calcific deposit with a calcification arc below 90 ° and a length below 10 mm; 6) tortuous microchannels (<⁠300 µm in diameter) traversing the plaque^19-22 (Figure 2).

Statistical analysis

Continuous variables were reported as mean (SD) or median (interquartile range [IQR]) and compared using the t test or the Mann–Whitney test. Categorical variables were reported as percentage and compared using the χ² test or the Fisher exact test. Spearman rank correlation coefficient was used to evaluate bivariate correlations between variables. Univariable and multivariable logistic regression analyses were conducted to assess the independent association of clinical characteristics and laboratory parameters (listed in Table 1) with HRP occurrence. Variables with a P value below 0.1 in the univariable analysis were included in the multivariable analysis, and the multivariable regression model was built using stepwise variable selection (entry and removal thresholds, 0.05 and 0.1, respectively). A sensitivity analysis was conducted using an alternative definition of HRP, which required meeting at least 2 of the following 4 criteria: 1) lipid arc of at least 90 °; 2) minimum fibrous cap thickness below 65 µm; 3) presence of plaque rupture or thrombus; and 4) macrophage infiltration. Receiver operating characteristic (ROC) curves were used to assess the prognostic value of biomarkers for HRP progression. Areas under the curve (AUCs) were calculated, optimal cutoff values were determined based on the maximum Youden index, and the DeLong test was used to compare AUCs among the parameters. All analyses were performed using R software (version 4.4.1, R Foundation for Statistical Computing, Vienna, Austria). A 2‑tailed P value below 0.05 was considered significant.

Study cohort

The study cohort comprised 80 patients, at a mean (SD) age of 66.7 (9.2) years, of whom 72.5% were men. HRP was detected in 39 patients (48.8%; Figure 1). Among these, 28 patients presented with a single high‑risk feature, specifically: macrophage infiltration (n = 7), TCFA (n = 5), plaque cap tear (n = 4), severe stenosis (n = 2), spotty calcification (n = 3), and microchannels (n = 7). The remaining 11 patients exhibited multiple concurrent high‑risk features, including combinations, such as macrophage infiltration with TCFA (n = 4), macrophage infiltration with severe stenosis (n = 1), macrophage infiltration combined with microchannels (n = 1), TCFA with plaque cap tear (n = 2), TCFA with microchannel (n = 1), severe stenosis with microchannel (n = 1), and spotty calcification with microchannel (n = 1).

As detailed in Table 1, significant elevations in median (IQR) levels of low‑density lipoprotein cholesterol (LDL‑C; 2.3 [1.6–3.1] vs 1.8 [1.2–2.3] mmol/l; P = 0.005), copper (23 [17.3–34] vs 13.9 [9–19.7] µmol/l; P <⁠0.001), and IL‑6 (14.2 [8.9–20.1] vs 5.6 [1.9–13.5] pg/ml; P <⁠0.001) were observed in the HRP vs non‑HRP group. However, no significant intergroup differences were found in demographic characteristics, CAG results, or other laboratory parameters. OCT parameters are summarized in Table 2, indicating lower rates of fibrous plaque (25.6% vs 39%) and calcified plaque (10.3% vs 26.8%) but a higher rate of lipid plaque (38.5% vs 7.3%; P = 0.005 for plaque type distribution) in the HRP group. There were no significant differences in the distribution of the imaged coronary artery, plaque burden, or minimal lumen area between the HRP and stable plaque groups.

Associations of copper levels with demographic characteristics and inflammatory markers

Serum copper levels showed no significant differences across subgroups stratified by sex, age (<⁠65 vs ≥65 y), as well as hypertension, or diabetes status (Supplementary material, Figure S1). In contrast, serum copper levels demonstrated significant correlations with inflammatory markers: a positive correlation with IL‑6 (R = 0.519; P <⁠0.001) and MMP‑9 (R = 0.235; P = 0.04), and a negative correlation with IL‑10 (R = –0.387; P <⁠0.001; Figure 3).

Factors associated with high‑risk plaque occurrence

In Table 3 we show the results of the univariable and multivariable logistic analyses of predictors for factors associated with HRP occurrence. In the univariable analysis, the variables strongly associated with HRP occurrence were: copper levels (odds ratio [OR], 1.14; 95% CI, 1.07–1.22; P <⁠0.001), LDL‑C (OR, 2.21; 95% CI, 1.27–3.84; P = 0.005), and IL‑6 (OR, 1.13; 95% CI, 1.06–1.21; P <⁠0.001). In the multivariable logistic regression model, copper (OR, 1.1; 95% CI;, 1.03–1.17; P = 0.005), LDL‑C (OR, 2.54; 95% CI, 1.25–5.16; P = 0.01), and IL‑6 (OR, 1.1; 95% CI, 1.02–1.18; P = 0.01) yielded similar results. The Hosmer–Lemeshow goodness‑of‑fit test indicated adequate model calibration (P = 0.8). Moreover, in the sensitivity analysis, where 25 participants met the alternative HRP definition, the association between HRP occurrence and these variables remained essentially unchanged.

Diagnostic performance of individual and combined biomarkers

A combined biomarker panel incorporating serum copper, LDL‑C, and IL‑6 was developed using multivariable logistic regression to enhance the diagnostic accuracy of HRP. The predicted probability of HRP (P) was calculated as: logit(P) = 0.094 × Copper + 0.93 × LDL‑C + 0.093 × IL‑6 − 4.999, where logit(P) is the log‑odds of HRP presence derived from the biomarker levels. The corresponding probability is derived as P = 1 / (1 + e^−logit(P)).

ROC curve analysis revealed superior diagnostic performance of the combined panel’s predictive probability, as compared with individual biomarkers (Figure 4, Table 4). The combined panel achieved an AUC of 0.864 (95% CI, 0.783–0.945), markedly outperforming single biomarkers, that is, copper (AUC, 0.802; 95% CI, 0.706–0.897; P = 0.04), LDL‑C (AUC, 0.68; 95% CI, 0.563–0.797; P = 0.002), and IL‑6 (AUC, 0.729; 95% CI, 0.617–0.84; P = 0.004) as determined by the DeLong test. Based on the maximum Youden index (J = Sensitivity + Specificity − 1), the optimal probability cutoff value of 0.4 for the combined panel was determined. Using this cutoff value, the panel achieved sensitivity of 84.6%, specificity of 73.2%, positive predictive value of 75%, and negative predictive value of 83.3%.

Prevalence of high‑risk plaque in patients with different elevated biomarkers

Optimal cutoff values for serum copper, LDL‑C, and IL‑6 were established as 15.6 μmol/l, 2.3 mmol/l, and 10.4 pg/ml, respectively (Table 4). Biomarker levels exceeding these thresholds were classified as elevated. The patients were stratified into 4 groups based on the number of elevated biomarkers (0, 1, 2, or 3). The prevalence of HRP increased markedly with the number of elevated biomarkers, occurring in 7.1% of the patients with no elevated markers, 21.1% of those with 1 elevated marker, 70.4% of individuals with 2 elevated markers, and 75% of those with 3 elevated markers (Figure 5).

Limitations

Several limitations of our study must be acknowledged. First, the relatively small sample size restricts the generalizability of our findings. Larger and more diverse cohorts will be needed to validate these results and establish their broader applicability. Second, the cross‑sectional nature of the study limits our ability to draw conclusions about causal relationships. For instance, atherosclerosis is a chronic inflammatory disorder, and definitive evidence clarifying the causal relationship between the levels of serum copper and inflammatory cytokines, particularly IL‑6, would substantially strengthen the pathophysiological and clinical relevance of serum copper level in the context of atherosclerosis. Third, our study focused solely on CCS patients who underwent CAG, and we did not include patients with unstable angina, a subtype of ACS, which may have been underrepresented due to the absence of ST‑segment elevation on electrocardiography. This could have led to potential misclassification or omission of patients who might exhibit similar plaque vulnerability features. Lastly, the study cohort was predominantly male (72.5%), which limits generalizability to female populations. Women often exhibit different coronary plaque phenotypes, such as plaque erosion rather than rupture, potentially affecting the applicability of our findings.

Despite these limitations, our analysis provides novel insights into noninvasive biomarkers for assessing vulnerable plaque burden and may help refine cardiovascular risk stratification without contrast administration.

Conclusions

In conclusion, elevated serum copper is independently associated with OCT‑defined HRP in CCS patients. A combined model incorporating copper, LDL‑C, and IL‑6 significantly improves noninvasive HRP identification, suggesting a potential multimarker strategy for CCS risk stratification.