In this single-center retrospective study, we consecutively screened 497 patients with ISR confirmed by coronary angiography (CAG) at the Affiliated Hospital of Zunyi Medical University between January 2018 and October 2022. ISR was defined as a percent diameter stenosis exceeding 50% within the stent implantation segment [10, 11, 12]. The inclusion criteria were first ISR on CAG follow-up and availability of OCT images before re-PCI. The exclusion criteria were combined multiple ISRs, ISR occurring 6 months after stenting, and poor-quality OCT images. Ultimately, 227 patients were included for analysis (Fig. 1), and details are displayed in the Supplementary Materials. The Supplementary Materials also show definitions of early in-stent restenosis (E-ISR), late in-stent restenosis (L-ISR), and very-late in-stent restenosis (VL-ISR) and reasons for conducting follow-up CAG. The NA and NV characteristics of ISR at different periods and the relationship between NA and NV in ISR lesions were compared. Additionally, to evaluate lipid control on OCT features of ISR lesions at follow-up, patients with ISR were further divided according to LDL-C levels into 1.8 mmol/L and $\ge$1.8 mmol/L groups [5].

Fig. 1.

Study flow diagram. ISR, in-stent restenosis; CAG, coronary angiography; DES, drug-eluting stents; E-ISR, early in-stent restenosis; LDL-C, low-density lipoprotein cholesterol; L-ISR, late in-stent restenosis; NV, neovascularization; OCT, optical coherence tomography; PCI, percutaneous coronary intervention; VL-ISR, very late in-stent restenosis.

Quantitative coronary angiographic (QCA) and OCT analyses are presented in the Supplementary Materials. NA was defined as neointimal formation in the presence of lipids or calcifications in $\ge$3 consecutive frames on OCT images [5, 16]. Lipidic neointima was defined as an intimal area with a diffused border and signal-poor region with marked attenuation [17]. Calcified neointima was defined as a well-defined area with well-defined contours and poor signal intensity [18]. Plaque type was considered mixed if there were both lipidic and calcific NA within the stenosis segment [16]. NV was defined as small (µm) vesicular or tubular structures identified on at least three consecutive cross-sectional OCT images [8, 9, 17, 19]. NV was considered endointimal if located in the superficial 50% of the neointimal thickness or peri-stent if located in the deep 50% of the neointimal thickness; in addition, some of the stenosis segments may have both endointimal and peri-stent NV. Other definitions of neointimal morphologies are displayed in the Supplementary Materials. Representative OCT images of ISR are displayed in Fig. 2.

Fig. 2.

Representative optical coherence tomography images of restenosis. (A) Homogeneous neointima. (B) Heterogeneous neointima. (C) Layered neointima. (D) Calcified neoatherosclerosis. (E) Both lipidic (asterisks) and calcified neoatherosclerosis (arrowhead). (F) Neoatherosclerosis (asterisks) with neovascularization (arrowheads). (G) Intraintima neovascularization (white arrowhead), peri-stent neovascularization (yellow arrowhead), and PLIA (asterisk). (H) Lipidic neoatherosclerosis (asterisks) with intimal disruption (arrowhead). (I) Macrophage infiltration (arrowhead). PLIA, peri-stent low intensity area.

Details of data analyses are presented in the Supplementary Materials.

A total of 227 patients with 227 lesions were included, with 55, 78, and 94 cases in the E-ISR, L-ISR, and VL-ISR groups, respectively. A comparative analysis of the basic data and QCA analysis of the three groups showed no significant differences among the groups, except for the clinical manifestations during OCT follow-up and the time of stent implantation. The number of patients presenting with acute coronary syndrome (ACS) at the early, late, and very late stages of CAG follow-up was 10 (15.4%), 17 (21.8%), and 42 (44.7%), respectively (p 0.001) (Table 1). There were significant differences in median stent implantation time for patients with E-ISR, L-ISR, and VL-ISR (11, 35, and 84 months, respectively; p 0.001). The QCA data are summarized in Supplementary Table 1.

Before comparing the relationship between NA and NV, we first compared the characteristics of NA and non-NA patients, and found that the NA group had higher LDL-C and creatinine levels and longer stent implantation time compared with those in the non-NA group (p 0.05). No significant differences were observed in diabetes mellitus, stent type, lesion characteristics of CAG, and OCT quantitative analysis of the MLA site between the two groups (p $>$ 0.05) (Supplementary Table 3).

Next, we compared the relationship between NA and NV and found no significant differences in general clinical data, CAG, and quantitative OCT findings of the MLA site between NV and non-NV groups, except for stent implantation time, MLA, and stent area; the median stent implantation time was longer in the NV group than in the non-NV group (60.0 months vs. 36.0 months, p = 0.008); the MLA and stent area of the NV group were larger than those of the non-NV group (p 0.05). Qualitative OCT analysis showed that the prevalence of NA was higher in ISR patients with NV lesions than in those without (68.8% vs. 41.8%, p 0.001) (Fig. 4). Moreover, patients with ISR combined with NV had a higher incidence of heterogeneous intima, macrophage infiltration, TCFA, intimal rupture, peri-stent low intensity area (PLIA), and thrombosis (p 0.05) (Table 3, Fig. 4).

Fig. 4.

Comparison of OCT characteristics of ISR lesions in the NV and non-NV groups. NA, neoatherosclerosis; NV, neovascularization; PLIA, peri-low intensity area; TCFA, thin-cap fibroatheroma; OCT, optical coherence tomography. ***p 0.001; **p 0.01; *p 0.05; ns, not significant.

Comparison of the general data and CAG findings between the two groups revealed that the prevalence of diabetes mellitus (DM) was higher in the LDL-C 1.8 mmol/L group. The OCT data of the two groups showed that the incidences of lipidic NA, heterogeneous intima, TCFA, intimal rupture, and thrombus were higher in the LDL-C $\ge$1.8 mmol/L group (p 0.05). No significant differences were observed between the two groups in the prevalence of calcified NA, plaque erosion, macrophage infiltration, NV, cholesterol crystal, and PLIA (p $>$ 0.05) (Table 4, Fig. 5).

Fig. 5.

Comparison of plaque characteristics between LDL-C 1.8 mmol/L and LDL-C $\mathrm{\ge }$1.8 mmol/L groups. LDL-C, low-density lipoprotein cholesterol; NA, neoatherosclerosis; NV, neovascularization; PLIA, peri-low intensity area; TCFA, thin-cap fibroatheroma. **p 0.01; *p 0.05; ns, not significant.

A previous study found that the incidences of NA after the first- and second-generation of DES implants were 45.5% and 10.8%, respectively; moreover, the incidence of NA gradually increased with the extension of follow-up time [5]. Nakamura et al. [4] reported the NA incidence of 47.0% among 64 bare-metal stent (BMS) ISR and 241 DES ISR lesions. Chen et al. [16] reported an NA incidence as high as 75% after $>$7 years of stent implantation. Therefore, emerging evidence suggests that the incidence of NA increases with stent implantation time. In this study, the overall prevalence of NA was 52.9%, and it exhibited a time-dependent pattern: E-ISR, 40.0%; L-ISR, 51.3%; VL-ISR, 61.7%. The incidence of NA and its time dependence are similar to those previously reported.

Although evidence suggests that the incidence of NA is time-dependent, systematic studies on the manifestations of NA at different stages are lacking. Previous studies by Yonetsu et al. [20] reported that the incidence of lipid-rich neointima (lipid NA) in BMS/DES was time-dependent, with 8%/37%, 28%/63%, and 77%/75% in the early (9 months), middle (9–48 months), and delayed ($>$48 months) stages, respectively. Jinnouchi et al. [12] also reported that after second-generation DES implantation, the frequency of lipid-laden neointima was significantly higher in the L-ISR (beyond one year) than in the E-ISR group (within one year). In this study, the time dependence of NA was also observed, specifically manifested as lipid NA. Moreover, Garcia-Guimaraes et al. [21] demonstrated that calcified NA was predominantly observed in very late ISR cases ($>$3 years). In contrast, our study data suggested a trend of increasing calcified NA from early ISR to very late ISR; however, this observation did not reach statistical significance. Another study found that the presence of ISR with calcified nodule formation (i.e., calcified NA) was associated with female sex, combined hemodialysis, calcified lesions, and stent malapposition, but not with stent implantation time [22]. Our results are inconsistent with those of Garcia-Guimaraes et al. [21], which may be related to the use of a retrospective design with a small sample size in the present study or the phenomenon that calcified NA is not associated with stent implantation time but is related to whether the in situ lesions are calcified. Therefore, our data preliminarily revealed that the progression of lipid NA is primarily associated with L-ISR and VL-ISR but may not be associated with calcified NA, which needs to be confirmed in large-sample randomized prospective trials.

Previous studies showed that first-generation DES implants significantly reduced the prevalence of BMS-associated ISR but also increased the incidence of stent thrombosis [23]. Second-generation DES is associated with fewer stent thrombotic events, but NA formation remained unavoidable [24, 25]. Lee et al. [5] also showed that second-generation DES did not prevent NA better than the first-generation DES. Furthermore, the development of NA after DES implantation has been described as a late catch-up phenomenon, as it has been observed that neointimal growth is highly inhibited in the first year after DES implantation; however, subsequently, it shows sustained progression, accompanied by rapid lipid-laden macrophage deposition, thus becoming the final common pathway for stent failure in the late stages [6, 26]. Therefore, both pathological studies and endovascular imaging findings have shown that NA formation was an important cause of late failure of BMS, first- and second-generation DES [6, 27]. In an OCT analysis of 2139 patients with ACS, Amabile et al. [28] showed that NA is common in patients with very late stent thrombosis. Habara et al. [10] showed that VL-ISR ($>$5 years) had a significantly higher incidence of heterogeneous intima, NV, intimal rupture, and red thrombus than E-ISR (1 year). Furthermore, in their first-generation DES ISR study, Habara et al. [11] showed a gradual increase in TCFA, intimal rupture, and intimal NV from early to late and very late post-operative periods. Jinnouchi et al. [12] reported that after second-generation DES implantation, the frequency of neointima with lipid-laden tissue, macrophage infiltration, NV, and TCFA were significantly higher in the L-ISR than in the E-ISR group. This study also showed that heterogeneous intima, TCFA, intimal rupture, plaque erosion, macrophage infiltration, and thrombus were more common in the VL-ISR group than in the E-ISR group. Similarly, the results of the MLA site analysis were similar to those of the entire stent analysis.

Consequently, NA formation may lead to stent failure and can trigger MACEs [29, 30, 31]. In this study, the number of patients presenting with ACS at follow-up was significantly higher in the VL-ISR group than in the E-ISR group; this may be related to the significantly higher prevalence of NA in the VL-ISR than in the E-ISR group. The incidence of NA with intimal rupture and thrombus in the VL-ISR group (21 intimal ruptures and 30 thrombi among 58 NA lesions) was significantly higher than that in the E-ISR group (3 intimal ruptures and 5 thrombi among 22 NA lesions), suggesting that NA plays an important role in stent failure and MACE. Notably, both the incidence of NA and NV are time-dependent, and there may be a potential relationship between them, but no reports have been found on the relationship between NA and NV in ISR lesions. Based on an in situ lesion study, the incidence of NA is reportedly associated with the formation of NV [32]. Therefore, we hypothesized that there might be a correlation between the occurrence of NA and the presence of NV in ISR lesions. To investigate and verify this hypothesis, we further explored the relationship between NA and NV in ISR lesions.

NV has been regarded as an important pathway for the delivery of erythrocytes and inflammatory cells involved in lipid plaque formation and has been identified as a contributor of plaque vulnerability [33]. In this study, NV, especially intraintimal NV, increased gradually from E-ISR to VL-ISR. Another study found that NV dilatation in the plaques for native coronary arteries was closely related to plaque vulnerability; plaques with NV have thinner fibrous caps and a higher incidence of plaque rupture [8]. NV inhibition effectively prevents atherosclerosis in situ [34]. Lipid-laden intima is thought to be closely associated with intimal NV after BMS implantation [17]. Tian et al. [32] also reported that NA was more common in stents with NV within the intima. In addition, Gao et al. [35] reported no difference in the incidence of NA between diabetic and non-diabetic patients, but NV in NA lesions was more common in diabetic patients. Our results also showed that the prevalence of NA was similar in diabetic and non-diabetic patients, and no difference in the incidence of NA was observed in diabetic patients with and without NV lesions; this may be related to the inclusion of a special population (ISR patients) in this study, which is expected to be further confirmed by randomized studies. Significantly, we found that NA was more frequently observed in ISR patients with NV lesions. Moreover, patients with ISR plus NV had higher incidences of macrophage infiltration, TCFA, intimal rupture, and thrombosis. These findings suggest that NV formation and dilatation in ISR lesions may be associated with NA progression and plaque vulnerability.

LDL-C is well known to be involved in atherosclerosis development and progression, and high levels promote cardiovascular events. In situ studies have shown that active control of LDL-C levels can stabilize or reverse coronary plaque, thus reducing the risk of MACE [36, 37, 38]. A first- and second-generation DES study showed that LDL-C $>$70 mg/dL was related to NA. Higher LDL-C levels were more prevalent in patients with NA [5], and Nakano et al. [39] showed that increased LDL-C levels are associated with plaque rupture in patients with ACS; however, in real clinical practice, the effect of LDL-C level control on OCT characteristics of ISR lesions remains unclear. Therefore, we further compared clinical data and OCT characteristics of different LDL-C levels, and unexpectedly, compared with the LDL-C $\ge$70 mg/dL (1.8 mmol/L) group, the prevalence of DM was higher in the LDL-C 70 mg/dL group. This may be because ISR patients with DM pay more attention to lipid management. However, when comparing OCT characteristics, ISR patients with poorly controlled LDL-C $\ge$70 mg/dL had a higher incidence of lipidic NA, heterogeneous intima, TCFA, intimal rupture, and thrombus compared to patients with LDL-C 70 mg/dL. These findings suggest the importance of enhanced lipid management for both in situ and ISR lesions. However, large-sample randomized controlled trials are needed to determine target LDL-C levels for different populations in complex clinical settings.

The current research has some limitations. First, this was a single-center, retrospective study with a relatively limited sample size. Second, some individuals diagnosed with ISR were excluded from the study owing to the absence of OCT imaging, which may introduce a potential selection bias, and the data derived from this study may not represent the wider patient population. Consequently, the findings of this investigation are solely descriptive and intended to generate hypotheses, which require further validation through large-sample randomized prospective trials. Third, not reporting the clinical results of these patients is an important limitation of this study. However, we are currently collecting the clinical follow-up data of these patients, and the data are not yet complete. Therefore, they were not included in the analysis of this study. We will continue to collect and analyze the clinical outcomes of these patients and report the results in future studies. Last, there was a lack of histological validation of neointimal tissue characteristics. Although OCT is the preferred intravascular imaging method for diagnosing NA in vivo, it has its own limitations and may not accurately assess qualitative neointimal characteristics.