Clinical Outcomes in type 2 Diabetes with in-stent Restenosis and Late/very Late Stent Thrombosis Undergoing Primary Percutaneous Coronary Intervention: A Single-center Observational Cohort Study


 BACKGROUND: This study aimed to compare differences in the risk factors and clinical outcomes of type 2 diabetes mellitus (DM) and non-DM patients with de novo lesions (DNL), definite in-stent restenosis (ISR), and late or very late stent thrombosis (LST/VLST). METHODS: A total of 4151 patients with acute coronary syndrome were screened angiographically to determine DNL, ISR, and LST/VLST. A total of 3976 patients were included in the analysis and divided into DM (n=1302) or non-DM (n=2674) group at admission. The primary endpoint was a composite of major adverse cardiovascular events (MACEs), defined as death, myocardial infarction, revascularization, and ischemic stroke within a 1-year follow-up period. RESULTS: In the group with total white blood cell count >10 10^9/L (p=0.004), neutral granular cell count >7 10^9/L (p=0.030), neutrophil-lymphocyte ratio >1.5 (p=0.041), DNL outperformed LST/VLST lesions in terms of revascularization at a median follow-up of 698 days. Among patients with DNL, the incidences of MACEs (log-rank p=0.0002), all-cause death (log-rank p=0.00032), cardiac-related death (log-rank p=0.021), and revascularization (log-rank p=0.029) were significantly lower in the non-DM group than in the DM group. However, no difference was observed in the event rates of endpoints among patients with ISR and LST/VLST between the non-DM and DM groups. Furthermore, among DM patients, there was a critically higher cumulative incidence of revascularization (log-rank p=0.0002) in the subgroup with ISR lesions and higher rate of ischemic stroke (log-rank p=0.033) in the subgroup with LST/VLST lesions. CONCLUSION: This study found that DM patients have a higher incidence of composite clinical outcomes than non-DM patients among patients with DNL. Compared with patients with DNL, patients with ISR lesions and LST/VLST lesions had more incidence of revascularization by long-term follow up. Thus, ISR and LST/VLST lesions are critical problems after coronary stenting, especially among DM patients.

death, recurrent MI, revascularization, and ischemic stroke. MACE was the composite of all-cause death, recurrent MI, and ischemic stroke.
The physicians in charge of the follow-up identi ed and extracted primary endpoints from hospital records, laboratory reports, and clinical notes in the event of death.

Statistical analyses
Time-to-event variables are presented as Kaplan-Meier (KM) curves by R language, and incidences of subgroups were compared using the log-rank test. Baseline patient characteristics were compared between patients with DNL, ISR, and LST/LVST among DM or non-DM patients.
Continuous variables are presented as means ± standard errors and categorical data as counts and percentages. Differences between continuous variables were compared using independent t-test, and those between categorical variables were compared using the χ2 test or Fisher test to assess the interaction between lesion types and baseline clinical, laboratory index, or angiographic characteristics. The Mantel-Cox method was used to calculate hazard ratios and 95% con dence intervals for comparisons of clinical outcomes including MACEs and all-cause death between groups, and the log-rank test was used to calculate corresponding p values. We conducted two-sided analysis of p values to allow conventional interpretation of results, and a p < 0.05 was considered statistically signi cant. Missing data were handled by single imputation. Most of the statistical analyses were conducted using the R language version I 386 3.6.2 (R Foundation for Statistical Computing, Vienna, Austria). Other analyses were performed using SPSS Statistics version 20.0 (SPSS, Inc., Chicago, IL). Table 1 presents the baseline demographic data, indicators of serum in ammation, lipids, angiographic features, and procedural characteristics of the entire study population. In total, 3976 patients were divided into the DM group (1178 DNL, 16

Results
The ndings of revascularization were consistent across the strati ed analyses of subgroups, including variables representing serum in ammation, lipids, and thrombus levels ( Fig. 1). Especially, in the subgroup with total white blood cell (WBC) count > 10 10^9/L (p = 0.004,  trend toward superiority for the non-DM group, compared with the DM group, was apparent ( Fig. 2). At approximately 5 years, the KM curves began to continually diverge for the primary endpoint in favor of the non-DM group up to 8.2 years. However, no differences were found in the cumulative incidences of clinical outcomes composite of MACEs, all-cause death, cardiac-related death, recurrent MI, revascularization, and ischemic stroke between the DM and non-DM groups among patients with ISR lesions and LST/VLST lesions.  Table 3 and presented in Fig. 3. At a median follow-up of 698 days, the ISR group had inferior all-cause revascularization compared with the DNL and LST/VLST groups: 4 (25.00%) versus 175 (1.49%) and 21 (19.44), log rank = 0.0002.  [5,6]. In ammation and accumulation of reactive oxygen species and metabolic cytokines are primary mechanisms of vascular remodeling and progression of adverse myocardial diseases resulting from glycemic variability and hyperglycemia [7][8][9][10][11]. Furthermore, a previous study has reported that insulin resistance is higher in patients with cardiovascular disease [12]. Various biomarkers are proposed to play a role in the strati cation of ACS. Cyr61, which predicts primary endpoints in patients with ACS, is involved in cell adhesion, proliferation, migration, and in ammation [13].
Indeed, glycemic variability has previously been shown as an outcome predictor of patients with ACS undergoing PCI [14]. Furthermore, the risk for repeat revascularization has been related to DM severity, with insulin-dependent DM having the highest risk factor for repeat revascularization [15]. Elevated glucose level is markedly related to sympathetic stimulation, and catecholamine can stimulate glucose release and control hyperglycemia [16][17][18]. Mechanistically, an increase in the incidence of MACE, mortality, and stroke among DM patients might be a result of direct glucotoxic effects, which lead to the attenuation of endothelium-dependent vasodilatation and myocardial perfusion damage [19][20][21]. Furthermore, hyperglycemia can cause conformational changes in platelet glycoproteins and affect platelet function and intraplatelet signaling pathways; as a result, more solid coronary clots are formed [22][23].

DM and LST/VLST
In the implantation of rst-generation DES, the incidence of LST/VLST was correlated with incomplete stent apposition and delayed endothelial coverage, thereby leading to chronic in ammation [24]. However, second-generation DES, which is characterized with durable, biodegradable and biocompatible polymers, is not resistant to LST/VLST [25]. The mechanisms of thrombosed stent segments are brin deposition and chronic in ammation leading to strut malapposition, delayed healing, and heart remodeling, which are distinct from early ST [26][27][28][29][30][31][32]. Previous studies [33][34][35][36] have identi ed DM as an important clinically independent predictor of poor outcome in ST in the real world of mixed use of bare-metal stents (BMS) and DES. Longer lesion length, smaller vessel size, a higher rate of residual dissections, increasing thrombus burden, and bifurcation lesions might be the underlying reasons for a predisposition of DM patients to ST [37,38]. This study highlights that total WBC count (p = 0.021) and neutral granular cell count (p = 0.018) were independent risk factors of LST/VLST among DM patients. This is consistent with the severe in ammation status of DM patients with LST/VLST. In addition, this study found a signi cant increase in the incidence of ischemic stroke in patients with LST/VLST compared with those with DNL and ISR (log rank = 0.033). A previous study has reported that neovascularization, brin accumulation, and thrombus burden are accompanied by in ammation, which correlated with the early healing of thrombus [39]. Occasional accumulation of macrophages, giant cells, and lymphocytes is a main characteristic of the in ammatory response after percutaneous coronary stenting [40,41]. Presence of peristrut eosinophilic material in the plasma might be a marker of endothelial cell leakage. Therefore, it is necessary to compare the effects of hypercholesterolemia using a healthy model.

DM and in-stent thrombosis
DM patients have a two-to four-fold higher risk of developing ISR after PCI than non-DM patients and thus deserve additional attention.
Although new-generation DES have greatly decreased neointimal proliferation, ISR and late stent failure are common complications and crucial after coronary stenting. A recent study [42] con rmed that a higher hemoglobin A1c (HbA1c) variability in type 2 DM patients was more likely to cause higher incidences of neointimal hyperplasia and ISR and hypothesized that postprandial glucose variability might be more important than fasting glucose in the development of ISR. Compelling evidence from a notable study [43] has con rmed a signi cantly increased rate of ISR in DM patients undergoing PCI irrespective of speci c treatment modalities, such as BMS, DES, and balloon angioplasty. Another study reported that endothelial dysfunction and impaired bioavailability of endothelium-derived nitric oxide play a critical role in the pathogenesis of post-PCI restenosis [44].
The possible mechanisms of glycemic and HbA1c variabilities that affect the progression of ISR in DM patients remain unclear. Previous studies concluded that hyperglycemia [45,46], insulin resistance [47], and glycemic variability [48] result in adverse vascular and myocardial remodeling directly and indirectly by stimulating the production of in ammatory factors, metabolic cytokines, and reactive oxygen species.
This is consistent with our nding that the prognosis DNL outperformed ISR, especially in the subgroup with total WBC count > 10 10^9/L, neutral granular cell count > 7 10^9/L, and NLR > 1.5. Furthermore, accumulating evidence con rmed that delayed re-endothelialization [49] and endothelial dysfunction [50] play major roles in the development of ISR and are signi cant predictors of ISR after stent implantation.
Among patients with re-stenosis of the stent, insulin resistance was an established and acknowledged contributory element. The higher incidence of MACEs was correlated with endothelial dysfunction and dysregulated glucose homeostasis, which play a signi cant role in restenosis [51]. Therefore, delayed re-endothelialization and endothelial dysfunction are potential mechanisms in the progression of ISR under the setting of high glycemic variability [52]. Previous studies have reported [53] that endothelial vasomotor function in the systemic artery tree is signi cantly related to the pathobiological process of ISR by suppressing the proliferation of smooth muscle and inhibiting intimal hyperplasia. Endothelial vasomotor function has been shown to re ect nitric oxide-mediated dilation [54]. Furthermore, asymmetric dimethylarginine has been shown to be correlated with the pathogenesis of atherosclerosis and endothelial dysfunction [55]. A previous study [56] revealed that serum soluble triggering receptor expressed on myeloid cells-1 (sTREM-1) level, which is considerably affected by DM, is a predictive biomarker of ISR and an important mediator of migration, cellular in ammation, vascular smooth muscle cell proliferation, and sTREM-1 concentration. A high ISR rate may be related to dyslipidemia in DM, mainly due to increased remnant-like particle cholesterol, which is identi ed as lipoproteins rich in triglycerides, and in the fasting state, very low density lipoproteins are major components [57].

Insight from optical coherence tomography (OCT)
OCT has shown micro-and high-resolution intra-vascular imaging of culprit vessels to meet better understanding of mechanisms of underlying pathophysiological process of stent thrombosis. 1) Patients with acute/subacute ST suffer from high incidence of underexpansion of stent which was identi ed as key morphological features of ST by OCT. Furthermore, it has been shown that ow disturbance and non-streamlined ow along malapposed stent struts are in keeping with the known association between acute thrombogenicity of stents and acute procedural results [58]. 2) Malapposed struts and uncovered struts were commonly found in the patients with ST. 3) In patients with LST and VLST, neoatherosclerosis [59] and uncovered struts were relatively frequent ndings.
Furthermore, a heterogeneous pro le has been observed [60] that malapposed, uncovered, underexpansion and severe restenosis are predominant features among patients with LST/VLST within the rst year and in-stent neoatherosclerosis phenomenon, residual edge dissection and plaque rupture within the proximal or distal edge segment are more commen among patients with LST/VLST beyond 1 year.
In addition, severe ISR has underlying association with VLST/LST and the linkage could be interpreted by a procoagulant state resulted from deceleration of ow within the restenotic stented segment. Representative images of dominant ndings for very late ST in patients with DM are shown in Fig. 4.

Limitation
This study has some limitations. First, we retrospectively collected clinical data on de nite ST of patients who underwent primary PCI, as self-reported by site investigators in this study. Furthermore, the trial was conducted in a single-center in China. Therefore, we cannot exclude geographical variations in PCI practice outside China or in higher-volume centers. Third, we did not enroll patients with ST in terms of probability, possibility, or secondary to chance, which may lead to underestimation of de nite ST incidence. However, ST according to the de nitions of ARC (26) was a pre-speci ed endpoint. All ST events were adjudicated independently by a blinded clinical events committee according to established criteria, and the incidence of de nite ST continued to diverge between the two investigated devices up to 5 years, which would render chance unlikely. Thus, it is necessary to closely evaluate demographic covariates and longitudinal management of therapeutic options.

Conclusion
In conclusion, the study found that DM patients have a higher incidence of composite clinical outcomes than non-DM patients with DNL.

Data Availability
The datasets used and/or analyzed during this study are available from the corresponding author on reasonable request.

Funding
This work was supported by the Chinese Academy of Medical SciencesInnovation Fund for Medical Sciences (2016-I2M-1-009) and National Natural Science Funds (number: 81970308).. Figure 1 Strati ed Analysis of the Revascularization at median Follow-up of 698 days in Patients with DNL or LST/VLST lesion Values are n (%). The primary endpoint is revascularization. *Interaction is for risk ratio -2 to 1 year and risk ratio 1 to 4 years for LST/VLST and DNL. DNL, denovo lesion; LST, late stent thrombosis; VLST, very late stent thrombosis; CI, con dence interval; WBC, white blood cell; NEUT, neutral granular cell counts; NRL, neutrophil lymphocyte ratio; Ldl, low densith lipoprotein; eGFR, estimated glomerular ltration rate; PCI, percutaneous coronary intervention Kaplan-Meier curve analysis for MACE/all caused death/cardiac caused death/revascularization at follow up between DM group and non DM group. DNL, denovo lesion; ISR, in stent restenosis; LST, late stent thrombosis; VLST, very late stent thrombosis groups=0,non-diabetes mellitus group; groups=1, diabetes mellitus group Time-to-Event Curves for the Primary Endpoint of revascularization and ischemic stroke at follow up between DNL group, ISR group and LST/VLST group among patients with DM. DNL, denovo lesion; ISR, in stent restenosis; LST, late stent thrombosis; VLST, very late stent thrombosis groups=1, denovo lesion group; groups=2, in stent restenosis group; groups=3, late stent thrombosis or very late stent thrombosis group