This retrospective cohort study enrolled 119 consecutive patients who underwent elective or urgent PCI with sirolimus-eluting stent (SES) implantation between November 2019 and November 2021 in three north-eastern Chinese provinces—Heilongjiang, Jilin, and Liaoning. All patients had a single, de novo coronary lesion; in five cases, overlapping stents were placed within the same target segment. Exclusion criteria were left-main coronary artery disease, symptomatic heart failure (New York Heart Association Class III–IV), renal dysfunction (serum creatinine >1.8 mg/dl), and significant vessel tortuosity or calcification, defined as a calcium arc ≥180° or an intravascular ultrasound calcium score ≥3. Eligible patients were stratified into three age groups on the basis of previously reported age-related differences in neointimal healing and strut-coverage patterns (9): ≤55 years (n = 46), 56–65 years (n = 39), and > 65 years (n = 34). The patient screening and exclusion process is summarized in Figure 1. The study was approved by the Ethics Committee of Harbin Medical University, which granted a waiver of informed consent because only anonymized retrospective data were analyzed.

To ensure adequate statistical power for detecting intergroup differences in MACEs, the minimum required sample size was calculated using the following formula for comparing proportions: n=Zα/22⋅p(1−p)δ2, where Z_α/2 = 1.96 (two-tailed α = 0.05), p = 0.20 (anticipated MACE incidence), and δ = 0.10 (margin of error). The calculated minimum sample size was 62. After accounting for potential data loss and subgroup analyses, a conservative threshold of 102 participants was set and ultimately exceeded by the final cohort of 119 patients.

Coronary vessel analysis was performed by two independent analysts using QAngio XA (version 7.3). Target segments encompassed the stented regions and the 5-mm proximal and distal margins. Key parameters, including reference vessel diameter, minimum lumen diameter, percent diameter stenosis, and late lumen loss, were measured according to the standard protocol (10). Inter-reader discrepancies greater than 5% were resolved by consensus.

OCT imaging was performed using the LightLab system (version 5.2) with a Dragonfly catheter. The system operated at a center wavelength of 1,310 nm, with an axial resolution of 15 μm and a lateral resolution of 50 μm. Pullback images were acquired at 1-mm intervals along each stented segment, excluding side branches ≥2 mm in diameter, at a pullback speed of 20 mm/s and a frame rate of 100 fps. Saline was used as the flush medium at an injection rate of 4 ml/s. Balloon occlusion was not applied during image acquisition. Image analysis was conducted independently by two reviewers following the ESC 2024 Consensus Guidelines (10). The following definitions and quantitative parameters were applied:

Follow-up was carried out at 6 and 12 months through outpatient visits and review of electronic medical records. The primary endpoint was the incidence of MACEs, defined as a composite of cardiac death, target-vessel myocardial infarction, or ischemia-driven revascularization. Secondary endpoints included all-cause hospital readmissions and bleeding events classified as Bleeding Academic Research Consortium type ≥2 (14). All clinical events were independently adjudicated by two cardiologists who were blinded to the OCT findings. Complete follow-up data were available for all patients, with no losses reported. The antiplatelet agent prescribed (clopidogrel or ticagrelor) was recorded at baseline and included as a covariate in all multivariable regression models assessing predictors of strut coverage and MACEs.

Cold exposure was assessed with the CEDQ, a self-developed instrument that captures outdoor exposure, indoor heating adequacy, and protective behaviors during periods of cold weather. Each item was rated on a 4-point Likert scale (0–3), producing a cumulative score from 0 to 36.

For analytical purposes, a composite cold exposure metric that integrates three components was developed: (1) daily outdoor exposure duration (with individual questionnaire items capped at 8 h to reflect realistic limits for rural and older populations; entries exceeding this threshold were verified by follow-up telephone calls), (2) periods of inadequate indoor heating (<16 °C), and (3) time spent without adequate protective measures during cold weather. High cold exposure was defined as ≥12 h/day of total cold stress accumulated across all three domains, rather than outdoor time alone. This metric was designed to capture cumulative environmental cold burden rather than any single exposure component.

The previously described protocol demonstrated robust psychometric performance, with a content validity index of 0.92 and a test–retest intraclass correlation coefficient (ICC) of 0.91 (P < 0.001), as well as strong convergent validity with meteorological data (r = 0.82, P < 0.01).

In a subset of participants (n = 34), wearable temperature sensors (iButton DS1922l) recorded skin-proximal temperatures at 10-min intervals over 7 days. Participants spanning a range of age groups, occupations, and housing conditions showed a high level of agreement between CEDQ entries and sensor readings (r = 0.89, P < 0.001), with a bias of 0.7 h/day (limits of agreement: −2.1 to +3.5), supporting the criterion validity of the CEDQ.

Multivariable models assessing the impact of cold exposure on vascular healing were adjusted for two additional covariates: self-reported indoor temperature and a deprivation index derived from postcode-based estimates of housing insulation and household income. A threshold of ≤–20  °C was applied, consistent with the 2022 China Climate Bulletin (15).

Data normality was assessed with the Shapiro–Wilk test. Continuous variables are reported as mean ± standard deviation or median (IQR), as appropriate. For comparisons among three groups, normally distributed variables were analyzed using one-way ANOVA with Bonferroni correction, whereas non-normally distributed data were analyzed with the Kruskal–Wallis test followed by Dunn's post hoc analysis. Categorical variables are expressed as counts (%) and compared using the Chi-square test or Fisher's exact test. For two-group comparisons, the Student's t-test or the Mann–Whitney U-test was applied, depending on data distribution.

Multivariable logistic regression was performed to identify independent predictors of delayed vascular healing and 1-year MACE, with adjustment for age, sex, comorbidities, and stent characteristics. Variables with P < 0.1 in the univariate analysis were entered into the final multivariable model.

Longitudinal changes in NIH were assessed with linear mixed-effects models that included fixed effects for age group, time point, and their interaction, along with random subject intercepts. Inter-observer agreement for NIH measurements and strut coverage was excellent (ICC = 0.92; Cohen's κ = 0.85, P < 0.001).

Time-to-event outcomes were analyzed with Cox proportional-hazards models, adjusted for stent length, duration of diabetes, and serum 25-hydroxyvitamin D [25(OH)D] levels.

To enhance individualized risk stratification after SES implantation—particularly in older patients exposed to environmental cold—we developed a composite prediction tool, the 4D Risk Score.

The model integrates four validated predictors of MACEs: (1) uncovered-strut rate ≥10%, (2) neointimal growth rate ≥2.0 μm/month; (3) neointimal heterogeneity (IQR) ≥100 μm, and (4) prolonged cold exposure ≥72 h at ≤–20  °C (assessed with the CEDQ). Each parameter contributes one point to the total score (range, 0–4). The model was internally validated by receiver-operating-characteristic (ROC) analysis and decision-curve analysis. Risk categories were defined as low (score 0–1), moderate (2–3), and high (4) to guide post-PCI clinical management.

All statistical analyses were performed with SPSS (version 28.0) and R (version 4.2.1); statistical significance was set at P < 0.05.

This study was approved by the Ethics Committee of Harbin Medical University. For the cohort of patients who underwent OCT examinations, anonymized retrospective data were used, and therefore, a waiver of informed consent was granted. In contrast, for patients who participated in the cold exposure survey, informed consent was obtained prior to participation. All data were handled in accordance with ethical guidelines, and patient confidentiality was strictly maintained throughout the study.

In this cohort, 71% of patients received second-generation DES. The baseline demographic, clinical, and environmental characteristics stratified by age group are summarized in Table 1. Elderly patients (> 65 years) experienced significantly longer daily cold exposure (11.5 ± 4.1 h/day; P < 0.001) and reported lower indoor heating temperatures (20.3 ± 1.8  °C; P < 0.001) than their younger counterparts. They also had higher fasting glucose concentrations (6.3 ± 1.4 mmol/L; P = 0.002), increased low-density lipoprotein cholesterol levels (2.9 ± 0.8 mmol/L; P = 0.019), and lower serum 25(OH)D concentrations (24.6 ± 6.2 ng/ml; P < 0.001). Collectively, these findings indicate that elderly patients face both metabolic dysregulation and environmental cold stress, which may synergistically impair vascular healing after DES implantation.

To evaluate age-related effects on vascular healing, serial OCT assessments were performed at 6 and 12 months. At the 6-month follow-up, younger patients (≤55 years) exhibited greater neointimal formation (median NIH thickness: 90 μm, IQR 60–130) and higher strut coverage (72.1%) than elderly patients (median NIH thickness: 60 μm, IQR 50–90; strut coverage: 55.0%; both P < 0.001). The elderly cohort also showed the highest proportion of uncovered struts (11.7%), suggesting impaired early endothelialization.

By 12 months, elderly patients displayed a distinct change in healing trajectory, characterized by accelerated neointimal growth (3.33 μm/month vs. 0.83 μm/month; P < 0.001) and greater variability in NIH thickness compared with younger patients, indicating a biphasic and disorganized repair pattern.

Age-stratified healing patterns are summarized in Table 2. Patients older than 65 years exhibited significantly delayed endothelialization at 6 months and accelerated NIH progression compared to younger groups (both P < 0.001). Median NIH thickness increased over time in all groups, with the elderly displaying a late-phase catch-up (Figure 2A). The elderly cohort also showed a significantly higher monthly growth rate (Figure 2B). In addition, Figure 2C demonstrates wider IQRs in the elderly, indicating greater neointimal heterogeneity during the late healing phase. Collectively, these findings suggest that advanced age is associated with biphasic, spatially heterogeneous vascular repair after stent implantation. Figure 3 illustrates the hallmark biphasic healing pattern in an elderly patient exposed to high levels of cold, showing simultaneous under- and over-healing within a single vessel segment. Figure 4 demonstrates the temporal evolution of three elderly patients with similar cold exposure, revealing pronounced heterogeneity in healing trajectories despite comparable baseline characteristics. The divergent patterns, including delayed coverage (panels A–D), persistent non-healing (panels B–E), and excessive proliferation (panels C–F), were differentially associated with clinical outcomes, with persistent non-healing exhibiting the highest rate of MACEs (45.5% vs. 18.2% for delayed healing, P = 0.023). These OCT findings provide a foundation for understanding the quantitative impact of cold exposure on clinical outcomes.

Building on the heterogeneous healing patterns observed in elderly patients, we systematically evaluated the dose-response relationship between cold exposure and clinical outcomes. Among the 119 participants, cumulative cold exposure during a 7-day period ranged from 28.0 to 64.5 h, with a median of 49.5 h. Individuals exposed to ≥56 h of ambient cold had a significantly higher incidence of MACEs than those with lower exposure (38.1% vs. 15.5%, P = 0.02). A positive association was observed between CEDQ score and MACE risk. In unadjusted logistic-regression analysis, each 1-point increase in CEDQ score was associated with a 24% rise in the odds of MACEs [odds ratio (OR): 1.24, 95% confidence interval (CI): 1.11–1.38, P < 0.001]. This relationship remained significant after adjustment for age, sex, comorbidities, indoor temperature, and the residential deprivation index (adjusted OR: 1.21, 95% CI: 1.08–1.35, P < 0.001).

Restricted cubic-spline modeling at an ambient temperature of 20 °C demonstrated a near-linear dose–response relationship, whereby each additional hour of exposure to ambient temperatures of 20 °C was associated with a 12% increase in MACE risk (P = 0.012) (Figure 5A). A forest plot of the multivariable-adjusted hazard ratios showed that high daily exposure to cold (≥ 12 h/day, corresponding to the ≥56 h/week threshold) was linked to a markedly increased risk of MACEs (hazard ratio 3.42, 95% CI: 1.68–6.97, P = 0.001) (Figure 5B). These findings further support a dose-dependent relationship between cold exposure and the occurrence of MACEs.

Table 3 summarizes the full statistical results from the two models. Model A treated the CEDQ score as a continuous variable, whereas Model B categorized cold exposure with a threshold of 56 h week. Participants in the highest exposure group (≥56 h week) had significantly greater odds of MACEs than those in the lowest exposure group (<42 h week; adjusted OR: 2.87, 95% CI: 1.32–6.25, P = 0.008). Both models showed significant associations between cold exposure and MACE risk, supporting the predictive validity of both continuous and categorical cold-exposure metrics.

Spatial heterogeneity of NIH, defined as an IQR >120 μm, emerged as an independent predictor of MACEs (16). ROC analysis demonstrated moderate discriminative performance, with an area under the curve (AUC) of 0.72. Kaplan–Meier survival analysis confirmed that patients with pronounced spatial heterogeneity had significantly lower MACE-free survival (P < 0.001) (Figure 6).

In the composite 4D Risk Score (Table 4), all four parameters (i.e., uncovered strut rate, neointimal growth rate, spatial heterogeneity, and cold exposure) were independently associated with MACEs, yielding hazard ratios of 2.91, 3.12, 3.42, and 3.25, respectively. The model displayed excellent performance, with an internally validated AUC of 0.83.

A stepwise algorithm illustrating how the 4D Risk Score can inform management after DES implantation is provided in Figure 7. In ROC analysis of the test dataset, the AUC was 0.63 (Figure 8A), indicating only moderate discrimination. This value is lower than initially reported, likely reflecting optimism inherent to internal validation. Nevertheless, decision-curve analysis (Figure 8B) demonstrated a consistent net clinical benefit across a wide range of threshold probabilities, outperforming both “treat-all” and “treat-none” strategies and underscoring the clinical utility of the score in optimizing therapeutic decisions. The proposed 4D Risk Score helps in personalized risk assessment and management strategies after PCI. The detailed criteria for scoring and management recommendations are outlined in Table 5.

Collectively, these findings establish the 4D Risk Score as a practical and clinically meaningful tool for individualized risk stratification after PCI, particularly in older patients exposed to environmental cold stress.

This study identifies a biphasic vascular-healing trajectory in elderly patients after SES implantation in cold environments. This pattern, conceptually illustrated in Figure 9, consists of two distinct phases: In the early phase (0–6 months), healing was markedly delayed, as reflected by higher proportions of uncovered struts and thinner neointimal coverage than in younger adults. During the late phase (6–12 months), neointimal growth accelerated and spatial heterogeneity increased, indicating a compensatory yet dysregulated repair response. Importantly, a dose–response relationship was evident: prolonged cold exposure was associated with a higher uncovered-strut rate and greater neointimal proliferation. These findings suggest that cold exposure functions as a cumulative vascular stressor that amplifies age-related impairments in endothelial recovery.

The delay in early healing may be driven by cold- and age-induced disruptions of vascular homeostasis. Acute cold exposure precipitates sympathetic over-activation and vasoconstriction by elevating circulating norepinephrine and stimulating α-adrenergic receptors, thereby impairing endothelial reactivity. Chronic cold stress, meanwhile, promotes the release of pro-inflammatory cytokines (e.g., interleukin-6 and tumor necrosis factor-α), leading to delayed endothelial regeneration and destabilized neointimal scaffolding. In addition, mitochondrial dysfunction in endothelial progenitor cells (EPCs) restricts β1-integrin-mediated migration (17, 18), further contributing to delayed endothelialization.

Aging further exacerbates this dysfunction through telomere attrition, which activates the ataxia-telangiectasia mutated/nuclear serine/threonine protein kinase Chk2/p53 pathway in vascular smooth muscle cells (VSMCs), leading to G1 cell cycle arrest and a 42% reduction in proliferative capacity compared to younger individuals (19). As a result, neointimal coverage becomes thinner and displays greater spatial heterogeneity (median thickness: 60 μm vs. 90 μm; a 75% increase in IQR).

During the chronic phase (6–12 months) of healing, telomere-dysfunction-driven stress responses sustain inflammation and epigenetic drift. Up-regulation of the long non-coding RNA TERRA triggers Toll-like receptor-9/nuclear factor κB signaling, enhancing monocyte recruitment and extracellular-matrix remodeling. Concomitantly, hypermethylation of the Krüppel-like factor-4 promoter (β > 0.8) induces phenotypic switching in VSMCs, evidenced by an 80% reduction in the contractile marker α-smooth-muscle actin and a 320% increase in the osteogenic marker osteopontin (20, 21). This phenotypic plasticity drives NIH and metabolic reprogramming via glycolytic enzymes (e.g., 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase-3), increasing cellular-proliferation heterogeneity (IQR 1.7 → 4.3) (22). However, these pathways remain hypothetical because molecular assays (e.g., methylation profiling, RNA sequencing) were not performed. Multivariable analysis in this study identified aging as an independent predictor of delayed vascular healing, highlighting the need for age-adapted OCT thresholds.

Our results align with those of Tomaniak et al. (23), who reported delayed neointimal coverage in elderly patients at 9 months. Unlike single-time-point studies, we propose a temporal healing trajectory comprising an initial phase of EPC-related dysfunction followed by inflammation-driven HIN. This biphasic pattern, supported by serial OCT data and spatial-heterogeneity metrics, reflects the interplay between aging and cold-induced vascular stress.

Biological aging and chronic cold exposure exert a synergistic “dual-hit” effect on vascular healing, delaying re-endothelialization and increasing the risk of MACEs. Cold stress promotes oxidative damage and accelerates telomere attrition—two hallmarks of vascular senescence—thereby impairing smooth-muscle repair and endothelial regeneration. In north-eastern China, extreme winter conditions intensify vascular vulnerability through two primary mechanisms:

This study has several limitations. First, its retrospective, non-randomized design restricts causal inference, particularly regarding the effect of cold exposure on vascular healing. Second, the choice of antiplatelet therapy (ticagrelor vs. clopidogrel) was not standardized and may have influenced endothelial recovery through distinct platelet–vascular interactions (30). Third, although OCT provides high-resolution structural information, it does not capture vascular function (e.g., shear stress, nitric-oxide bioavailability, or endothelial nitric-oxide synthase activity) (31); multimodal imaging (such as fractional flow reserve or near-infrared spectroscopy) should therefore be considered. Fourth, the 12-month follow-up may have missed very-late events, including stent thrombosis or chronic restenosis. Longer-term studies are needed to assess sustained vascular responses to aging and cold exposure. Fifth, although most patients received second-generation DESs (71%), the findings may not be generalizable to third-generation stents or bioresorbable scaffolds, which could differ in healing kinetics under environmental stress (32). Sixth, the elderly subgroup was relatively small (n = 34), limiting the power to detect sex-based differences or gene–environment interactions (e.g., TRPM8 polymorphisms or oestrogen-related endothelial effects) (33). In addition, the absence of physiological measurements (e.g., flow-mediated dilation, circulating endothelial microparticles) limits mechanistic interpretation beyond OCT-derived metrics. Future studies should incorporate such biomarkers to clarify healing heterogeneity. Lastly, while cold-induced vasodilation is known to provide insights into vascular health, this study did not employ non-invasive methods such as diffuse speckle pulsatile flowmetry to assess it. Given its potential as a non-invasive technique to study endothelial function under cold stress, incorporating such methods in future studies could provide valuable additional data and help clarify the mechanisms of vascular healing in response to cold exposure.