This was a retrospective observational cohort study carried out at Shengjing hospital of China medical university. The consecutive unstable angina pectoris patients with coronary second generation, sirolimus eluting stent implantations and planned follow-up angiographies in our hospital from March 2015 to September 2015 were studied. All patient’s data were de-identified prior to data collection and the need for ethics approval was waived by our hospital. We excluded the patients with failed stenting; acute myocardial infarction; stable angina pectoris; major adverse cardiac events (death, myocardial infarction, stroke, coronary artery bypass graft, repeated PCI) after PCI; and patients with other conditions that can cause increased RDW (for example: anemia, heart failure, current acute infection, malignancy, chronic inflammatory disease and autoimmune disease) before the procedure and during the follow-up. The initial assessments of the patients included a medical history, physical examination, blood biochemistry tests for a complete blood cell count; total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol and triglycerides; blood urea nitrogen; serum creatinine; nitrogen terminal-pro brain natriuretic peptide. The blood samples were drawn from participating patients after an overnight fast of more than 12 h in the morning of the initial PCI and were tested at the center for clinical chemistry of our hospital. The levels of RDW were measured with the use of the BECKMAN Ac. T 5 diff as a continuous variable. The normal range of RDW is 10.0–15.0 in our hospital.

Drug eluting stents were implanted in all patients according to the current guidelines. All patients received antiplatelet therapy before and after the procedures: 100 mg aspirin was continued indefinitely and 75 mg clopidogrel was continued for at least 1 year after stent implantation. The follow-up angiographies were routinely performed 9–12 months after the initial PCIs. The evaluations of restenosis were carried out using the conventional QCA technique. ISR at follow-up was defined as luminal narrowing of more than 50% occurring in the segment within the stent or within a 5 mm segment proximal or distal to the stent.

IBM SPSS Statistics 23.0 was used to analyze the data. Categorical variables were reported as frequencies (percentage) and were compared using chi-square analysis. Continuous variables were reported as means±standard deviations and were compared by Student’s t test. A one-way analysis of variance was used to determine differences among different categories of RDW. Differences at a two tailed P value less than 0.05 were considered significant. Binary logistic regression analysis was carried out to assess the risk factors of ISR. Odds ratio (OR) and 95% confidence interval (CI) were calculated. Among the variables tested, only those variables with statistical significance set at P less than 0.05 at univariate analysis together with age and gender were included in a multivariate logistic regression model to determine the independent predictors of ISR. We also performed a receiver operating characteristics (ROC) curve analysis to determine the optimal cutoff, diagnostic cutoff, screening cutoff value of RDW for prediction of in-stent restenosis. The optimal cut-off value is the point on the curve that is closest to the top of the left hand y-axis. This is the point at which the true positive rate is optimized and the false positive rate is minimized. The screening cut-off value is chosen to maximize the sensitivity of the test while maintaining an appropriate speciality. The diagnostic cut-off value is chosen to maximize the specificity of the test while maintaining an appropriate sensitivity. The determinations of diagnostic cutoff and screening cutoff value are often an expert decision and subjective to some degree.

A total of 233 patients with unstable angina pectoris and drug eluting stent implantations were screened. Eleven patients lost follow-up; 2 patients died; 6 patients suffered heart failure during follow-up. So 214 patients were eligible for this study. After an average follow-up of 11.1 ± 5.8 months, 55 patients suffered in-stent restenosis. The baseline characteristics of these patients were summarized in Table 1. There were no differences in the follow-up period between the ISR group and non-ISR group. Diabetes were more common in patients with ISR. Patients with ISR were more likely to be older; had a higher baseline and followup RDW; and lower total bilirubin, direct and indirect bilirubin; had longer stents of smaller diameters implanted.

We compared the RDWs prior to the PCI (RDW_baseline) with the RDWs at follow-up (RDW_follow-up) and interestingly we found that the RDWs increased significantly during follow-up in both the non-ISR group (13.0 ± 0.8 vs. 13.2 ± 0.8; P = 0.003) and the ISR group (13.7 ± 0.8 vs. 14.0 ± 1.4 p = 0.022). The increases of RDW during follow-up were not significantly different between in non-ISR and ISR groups (P = 0.472). (Table 2).

We divided the participants into three groups according to the tertile of RDW values before the initial PCI (low RDW group: ≤ 12.5%; intermediate RDW group: > 12.5% and ≤ 13.5%; high RDW group: > 13.5%) to explore the relationship between the baseline RDWs and the ISR at follow-up. The clinical and angiographic characteristics were compared among different RDW groups (Table 3). We found that the patients with higher baseline RDWs were more likely to develop ISR at follow-up (low RDW:12.3%; intermediate RDW:19.7%; high RDW:47.7%; P < 0.001). There were no statistically significant differences in clinical and laboratory characteristics among different RDW categories except age, stent diameter and hemoglobin.

Binary logistic regression analysis was used to assess independent predictors of ISR at follow-up. All variables that were clinically important and potentially correlated with ISR were included in the univariate logistic model (Table 4). We found that diabetes, indirect bilirubin, baseline RDW, hemoglobin were the significant univariable predictors of the ISR. Then we applied multivariate logistic regression analysis that included all the significant univariable predictors found in Table 4 to determine the independent predictors of ISR. The odds ratio and P value were adjusted for age and gender (Table 5). In the multivariate analysis model, only the baseline RDW (OR, 5.179; 95% CI, 2.568 to 10.446, P<0.001); indirect bilirubin (OR, 0.413; 95% CI, 0.305 to 0.559, P<0.001) and diabetes (OR, 4.077; 95% CI, 1.654 to 10.054, P = 0.002) were significant independent predictors of ISR at follow-up.

ROC curve analysis of the diagnostic accuracy of the RDW for in-stent restenosis showed an area under the curve (AUC) of 0.743 (95% CI, 0.664 to 0.822; P<0.001) (Fig. 1), and the optimal cutoff value of RDW for discrimination of in-stent restenosis was 13.37 (sensitivity:65.5%, specificity:73.6%). The diagnostic cutoff value of RDW was 13.89 with a specificity 91.8% and sensitivity 40%. A RDW 12.99 was used for the screening cutoff value with a sensitivity 83.6% and specificity 49.1%.