We screened all consecutive patients who underwent PCI at Korea University Anam Hospital between January 2009 and July 2015. Among them, 353 patients had ISR lesions (> 50% restenosis in the stent or within 5 mm of the stent edges) and underwent repeat PCI for ISR lesions. Thirty-six patients had missing data on serum UA level and were excluded. Finally, 317 patients (328 lesions) were analyzed in this study. Clinical events were monitored until June 2016 through medical record reviews and telephone calls. The present study was approved by the hospital’s institutional review board (IRB no. AN16238-002) and performed in accordance with the Declaration of Helsinki. The need for written informed consent was waived owing to the retrospective nature of the study.

A serum UA level of > 6.8 mg/dL was defined as hyperuricemia for both sexes [24]. This cutoff is the limit of urate solubility in the serum, and supersaturation of urate in extracellular fluid has been known to predispose a person to various pathologic conditions, including gout and cardiovascular diseases. Thus, the higher UA group included patients with hyperuricemia (serum UA level ≥6.8 mg/dL) or patients treated with anti-hyperuricemic agents such as allopurinol and febuxostat.

The primary end point, major adverse event (MAE), was defined as a composite of all-cause death; non-fatal myocardial infarction; any revascularization, including target-vessel revascularization (TVR) and non-TVR; and coronary artery bypass graft surgery. Myocardial infarction was defined as present when patients had elevated cardiac enzymes with compatible symptoms or electrocardiographic findings. Stent thrombosis was defined as definite stent thrombosis based on Academic Research Consortium Criteria [25].

Interventional procedures were performed according the standard clinical guidelines. Interventional strategies, including drug-coated balloon (DCB) angioplasty, DES implantation, and use of adjunctive devices and pharmacotherapy, were decided according to the operators’ discretion. Balloon pre-dilatation was performed for all ISR lesions. The first-generation DES included CYPHER® (Cordis, Johnson & Johnson, Miami Lake, FL, USA) and TAXUS™ (Boston Scientific Corp., Marlborough, MA, USA). The second-generation DES included XIENCE™ series (Abbott Vascular Devices, Temecula, CA, USA) and Endeavor® series (Medtronic Cardiovascular, Santa Rosa, CA, USA). The third-generation DES included BioMatrix (Biosensors, Singapore, Singapore) and Nobori (Terumo Corporation, Tokyo, Japan). The DCB (SeQent® Please balloon catheter; B.Braun, Melsungen, Germany) became available and was used from July 2010.

Laboratory profiles, including lipid panel, creatinine, glucose, hsCRP, and UA levels, were obtained within 4 weeks before the index PCI or at the index admission date. Serum UA level was measured by using an enzymatic method with an automatic biochemistry analyzer (Beckman Coulter AU 5800; Beckman Coulter Inc., Brea, CA, USA). Creatinine clearance was calculated using the Cockcroft and Gault formula [26].

Three radiologic technologists blinded to the patients’ treatment performed analyses with a quantitative coronary angiographic system (CASS system; Pie Medical Instruments, Maastricht, the Netherlands). By using the guiding catheter for magnification-calibration, the diameter of the reference vessel, minimal luminal diameter, and percent diameter stenosis were measured from diastolic frames in a single, matched view showing the smallest minimal luminal diameter. ISR lesions were classified according to the Mehran classification [27]. Multifocal, diffuse, proliferative, and occlusive ISR lesions were classified as non-focal-type restenosis lesions. Acute gain was calculated as the increase in minimal lumen diameter of the treated lesion immediately after the index procedure compared with that before the procedure. Late lumen loss was defined as a decrease in minimal lumen diameter of the treated lesion at the follow-up coronary angiography compared with that immediately after the index procedure. All quantitative angiographic measurements were obtained before and after PCI, and at the follow-up coronary angiography.

Categorical variables are reported as count (percentage), and continuous variables are reported as mean ± standard deviation. To compare the baseline clinical characteristics, angiographic features, procedural details, and the cumulative incidence of clinical events between the higher UA and normal UA groups, the chi-square test for categorical variables and Student’s t test (or Wilcox test) for continuous variables were performed. Kaplan–Meier survival curves with a log-rank test were generated to compare the long-term incidence of MAE between the two groups. In order to identify the risk predictors of MAE, the multivariate Cox proportional hazard model was used to evaluate the possible contributing factors. The following variables were included in the Cox regression model: age, sex, body mass index, current smoking, hypertension, diabetes mellitus, acute myocardial infarction at the index PCI, low-density lipoprotein (LDL)-cholesterol level, triglyceride level, UA level, creatinine clearance, left ventricular ejection fraction (LVEF), previous first-generation DES implantation, multivessel involvement, chronic total occlusion lesion, ISR type (III, IV), and PCI type (DES or DCB). Hazard ratios with 95% confidence intervals and p-values were reported. All tests were two-tailed, and p-values < 0.05 were considered statistically significant. All statistical analyses were performed using SPSS software (v20; IBM SPSS Corp., Armonk, NY, USA).

Many studies have reported that hyperuricemia is associated with cardiovascular disease. A recent meta-analysis including 29 prospective cohort studies also showed that hyperuricemia is an independent risk factor for cardiovascular morbidity and mortality [28]. Biologically, UA exerts pro-oxidant or nitric-oxide-reducing effects depending on its concentration and chemical microenvironment [29]. When the urate concentration exceeds 6 mg/dL, the risk of urate crystal formation and precipitation increases. Therefore, hyperuricemia is generally defined as a serum UA level of > 6.8 mg/dL [30]. The present study adopted this cutoff value. However, the optimal threshold for serum UA level remains debatable. Some studies used different cutoff values based on sex, considering the significant difference in reference ranges of serum UA levels between men and women. Recently, the clinically detrimental effect of serum UA seems to be evident even below its saturation limit, likely independent of urate crystal formation in cardiovascular diseases. Receiver-operating characteristic curve analysis of serum UA level for MAE in the present study showed an area under the curve of 0.544 (95% confidence interval 0.474–0.615, data not shown). In addition, when we further analyzed the clinical outcomes between two groups determined using the median UA level (5.3 mg/dL), the results also showed similar clinical outcomes between patients with lower UA level (≤5.3 mg/dL) and patients with higher UA level (> 5.3 mg/dL) (Additional File 1: Table S4 and Additional File 1: Figure S1). These data suggested that the association between serum UA level and poor clinical outcomes was very weak, and the optimal cutoff value of hyperuricemia might be obscure in those high-risk patients who underwent repeat PCI for ISR lesions.

In the present study, patients with hyperuricemia were predominantly male and somewhat obese. Additionally, they had lower creatinine clearance and showed a trend of higher serum triglyceride levels. Interestingly, patients with hyperuricemia had a higher frequency of non-focal-type restenosis lesions than normouricemic patients. Previously, elevated serum hsCRP level was reported as a risk predictor of non-focal-type ISR after DES implantation, suggesting that inflammatory activity might contribute to aggressive restenosis [31]. In addition, old age, hypertension, diabetes mellitus, and paclitaxel-eluting stent implantation were also reported to be associated with the non-focal type of ISR [32‐34]. Thus, considering that hyperuricemia is associated with elevated hsCRP level and other inflammatory markers, it could also be another possible biomarker for non-focal-type ISR. In addition, the present study showed a significant difference in the interval between previous PCI and index PCI between the low UA group and the high UA group. The high UA group took a longer time to develop ISR than the lower UA group. A previous study using an intravascular imaging modality demonstrated that neointimal hyperplasia is associated with earlier ISR, whereas neoatherosclerosis is associated with later ISR [35]. It also suggested the potential role of a high serum UA level in the development of neoatherosclerosis and ISR.

The present study did not show an association between hyperuricemia and clinical outcomes after repeat PCI for ISR lesions. Previous stent type, stent number, bifurcation lesion, ISR type, and repeat first-generation DES implantation were suggested as risk predictors of poor prognosis [36‐39]. Conventional demographic risk factors, such as diabetes mellitus, failed to reach clinical significance after repeat PCI for ISR [40]. These findings suggested that the pathologic mechanisms of recurrent ISR are rather different from those of de novo coronary atherosclerosis, and implied that lesional, technical, and mechanical factors might play important roles in recurrent ISR development after repeat PCI for ISR. A recent study even suggested DCB angioplasty as a predictor of target lesion failure in the second-generation DES era [41]. When we analyzed the impact of hyperuricemia in patients treated with DES or in patients treated with DCB separately, there were no significant differences in clinical outcomes between the low UA group and the high UA group in both the DES- and DCB-treated patients (Additional File 1: Table S5).

In addition, it was previously demonstrated that serum LDL-cholesterol level was significantly associated with the development of neoatherosclerosis, which has been studied as an important pathologic process related to poor clinical outcome after PCI in the DES era [42]. There was also a case of recurrent neoatherosclerosis after repeat PCI for ISR [43]. These data suggested that the residual risk of altered lipid metabolism should be considered after repeat PCI for ISR lesions. The present study indicated LDL-cholesterol level and LVEF as important risk predictors of MAEs (Table 4). However, the Cox proportional hazard model for TVR failed to suggest any independent risk factor from the 17 potential risk factors including age, sex, body mass index, current smoking, hypertension, diabetes mellitus, presentation of acute myocardial infarction, LDL-C, triglyceride, UA, creatinine clearance, LVEF, prior first-generation DES use, multivessel involvement, chronic total occlusion, ISR type, and PCI strategy (data not shown). The Cox proportional hazard model for non-TVR proposed LDL-C and LVEF as the independent risk factors for non-TVR in patients after repeat PCI for ISR (Additional File 1: Table S6). These results suggested that LDL-C and LVEF contribute to MAE development mainly driven by non-TVR rather than TVR. Management of lipid profile and heart failure could be emphasized as a fundamental strategy to prevent adverse clinical outcomes in patients after repeat PCI for ISR, although their association with TVR is obscure. However, the present study showed that TVR rather than non-TVR formed a majority of MAEs (70.7%, 70 of 99). Thus, although the present study failed to suggest the important risk predictor for repeat target vessel failure, further studies should resolve this issue.

The present study has several limitations. First, this is a single-center, retrospective study. The study population was enrolled for a long duration and the baseline characteristics were heterogeneous. Moreover, the PCI strategy was dependent on the discretion of the operators, and a selection bias should be considered in the interpretation of our results. Second, the sample size was too small to discriminate the clinical impact of hyperuricemia, although the patients were followed-up for a long duration. Third, the present study did not analyze intravascular imaging data (intravascular ultrasound or optical coherence tomography) because of their limited usage (36.9%). Considering that mechanical and technical factors may contribute to ISR, detailed lesional information could provide an insight into the clinical relevance of hyperuricemia. Therefore, our findings should be extended and validated further by other studies.