Coronary artery disease patients, who receive PCI in our institute, were enrolled in a stent registry at the time of the index procedure (i.e. Everolimus, Zotarolimus, or Biolimus eluting stent registry). According to the protocol of the individual registries, a follow up CAG was recommended (which is not mandatory) at 9 to 13 months post-PCI. Of those that agreed to and received the first follow up angiogram, a second routine follow up angiogram was recommended at 24 months. During the study period (July 2008 to March 2013), 3044 patients were enrolled in various stent registries, and 1486 patients (48.8%) received 9 to 13 month follow up CAG. Among these patients, 320 patients (21.9%) had a second routine follow up CAG at 24 months post-PCI. Patients with lesions of a lesion length of ≥2 mm and luminal narrowing ≥30% at baseline angiography were included in the analysis. As a sensitivity analysis to check the possibility of selection bias, the baseline demographics were compared with that of the entire 3044 patients that received PCI during the study period (Additional file 1: Table S2).

The study was approved by the ethics committee and institutional review board and was conducted according to the principles of the Declaration of Helsinki. All patients provided written, informed consent for participation in the registry.

Coronary angiograms were recorded at baseline and at two serial follow-up periods. Analysis was done at the angiographic core laboratory by 3 specialized quantitative coronary angiography technicians at the Seoul National University Hospital Cardiovascular Clinical Research Center Angiographic Core Laboratory. Standard qualitative and quantitative analyses and definitions were used for angiographic analysis [11]. The ACC/AHA lesion classification system, comprising 11 angiographic variables with all lesions categorized into types (A, B1, B2 and C) were used to characterize lesions [12]. Measured variables included the reference vessel diameter, the minimal luminal diameter (MLD), and the diameter stenosis (DS). Delta DS was defined as the last DS minus the initial DS (DS at 1st follow-up when an adverse event occurred at the 1st follow-up, DS at 2nd follow-up when an adverse event occurred at the 2nd follow-up or in event-free cases.). Regarding reliability analysis, the intra-observer intraclass correlation (ICC) was 0.941 (95% CI 0.929–0.951) and the inter-observer ICC was 0.986 (95% CI 0.973–0.992).

The study endpoints were any event of nonculprit-ischemia driven revascularization (IDR) and the rate of DS progression during the follow-up period. Revascularization was defined as ischemia-driven if there was stenosis of at least 50% of the diameter with evidence of ischemia, as documented by a positive functional study, ischemic changes on an electrocardiogram, or ischemic symptoms. In the absence of documented ischemia, DS of at least 70% was required.

The definition of ‘rapid progressors’ was derived using the following method. A previous study reported that the rate of non-culprit lesion related clinical event was approximately 10% during 3 years of follow-up [2]. Therefore, we analyzed the cutoff point for DS progression in the upper 10% of the population using a histogram analysis. From this analysis ‘rapid progressors’ were defined as patients with at least one lesion that had DS progression more than 1% per month (or 12% per year). This is the rate of DS progression that would turn an average lesion in our study, into a significant one during the 24 month follow-up period.

Continuous variables were presented as mean ± standard deviation and were compared using Student’s t-test or the Mann-Whitney U test. Categorical variables were presented as proportions, and chi-square test or Fisher’s exact test was applied to compare differences between groups, as appropriate. A comparison of baseline and follow-up values of the QCA results were analyzed by paired t-tests. To determine the independent predictors of nonculprit-IDR, a Cox proportional hazard model was used. Factors included into the multivariate model were lesion type, diameter of reference vessel, lesion location of the coronary artery as lesion-related factors, and age, gender, body mass index, diabetes, hypertension, chronic kidney disease, dyslipidemia, previous MI, smoking, and clinical diagnosis of ST segment elevation myocardial infarction as clinical factors. A multiple logistic generalized estimating equations (GEE) modeling using the autoregressive structure, was performed to analyze the longitudinal changes on lesion characteristics and clinical parameters as independent variables versus the progression of nonculprit lesions as the dependent variable. Variable included in the GEE model were identical to that included in the Cox proportional hazard model. Two-sided P values less than 0.05 were considered statistically significant for all tests. All statistical analyses were performed using SPSS version 20.0 (IBM Co., Chicago, IL, USA).

Among the 640 baseline nonculprit lesions, 220 (34.4%) were located in the left anterior descending coronary artery, 181 (28.3%) in the left circumflex coronary artery, 237 (37.0%) in the right coronary artery, and 2 (0.3%) in the left main coronary artery. Also, 164 (25.6%) were proximal lesions, 152 (23.8%) were mid lesions, 171 (26.7%) were distal lesions and 153 (23.9%) were branches of epicardial coronary arteries. Baseline QCA revealed a lesion length of 11.4 ± 7.7 mm with a DS of 45.1 ± 10.5, 29.8% had a DS > 50, and 2.5% had a DS > 70%. There was no difference in lesion character in DM and non-DM patients (Additional file 1: Table S4).

The total follow-up duration was 640.3 patient-years. During the median follow-up period of 737 days, 20 lesions in 20 patients (6.3%) required nonculprit-IDR. Clinical and lesion characteristics between those with and without IDR are shown in Table 1. Patients who received nonculprit-IDR were more likely to have diabetes mellitus or chronic renal failure, and discharge medication pattern were similar between the two groups. Lesions that received IDR had a smaller MLD, larger DS, and were more likely to be B2/C lesions.

The independent baseline patient-level and lesion-level correlates of nonculprit-IDR were diabetes (hazard ratio [HR] 3.698, 95% confidence interval [CI] 1.377–9.933, p = 0.009) and lesion type B2/C (HR 3.510, 95% CI 1.376–8.955, p = 0.009), respectively. The rate of nonculprit-IDR for lesions increased along with the number of risk factors, showing a rate of 1.8, 1.8 and 11.9%, for lesions that included 0, 1, or both risk factors, respectively. Kaplan Meier curve of nonculprit-IDR events, according to the number of risk factors is shown in Fig. 2. The combination of these two risk factors increased the risk of IDR by at least 6-fold compared with any single risk factor.

During the total follow-up period, the mean MLD decreased (1.58 ± 0.49 mm vs. 1.50 ± 0.48 mm vs. 1.46 ± 0.48 mm, p < 0.001, at baseline, 1st follow-up and 2nd follow-up, respectively) whereas the mean angiographic DS increased (45.1 ± 10.6 vs. 47.1 ± 11.2 vs. 48.2 ± 11.2, p < 0.001, at baseline, 1st follow-up and 2nd follow-up, respectively), suggesting continuous progression of angiographic stenosis (mean delta DS: 3.73 ± 7.09%) A dot plot of the baseline DS and delta DS of each lesion is shown in Fig. 3. Also, the velocity of DS progression was 2.46 ± 6.70%/year until the 1st follow-up, 1.49 ± 4.17%/year from the 1st follow-up to the 2nd follow-up period, and 2.19 ± 5.47%/year during the total follow-up period. A histogram of the velocity of all lesions is in Additional file 1: Figure S1. Lesions with nonculprit-IDR events had a larger delta DS (22.1 ± 15.4% vs. 3.1 ± 5.8%, p < 0.001) and faster rate of DS progression (21.12 ± 17.17%/year vs. 1.58 ± 3.14%/year, p < 0.001) compared to those without events.

To investigate the predictors of DS progression, we used a GEE model for repeated measures. After correcting for the size and location of the coronary artery, male sex, low BMI (< 25 kg/m²), diabetes mellitus, and lesion type B2/C were significant risk factors for a higher DS. Among these factors, diabetes mellitus and lesion type B2/C lesions showed interaction with time during the follow-up period. In other words, patients with diabetes mellitus and type B2/C lesions showed faster progression of angiographic stenosis compared with those without these factors (Additional file 1: Table S5). The mean DS and mean delta DS according to the presence of 0, 1 or 2 risk factors are shown in Table 2. Compared with the angiographic stenosis progression of lesions with 0 or 1 risk factor, those with 2 risk factors had a significantly faster progression (delta DS of lesions with both risk factors: 2.3-fold larger compared to those with no risk factors and 1.9-fold larger compared to those with 1 risk factor). This was also consistent with comparison of the velocity of DS progression, where lesions with both risk factors showed a 2.8-fold larger velocity of DS progression.

Among the 640 lesions, 38 lesions (5.9%) in 33 patients were defined as rapid progressors, defined as a progression of DS more than 1% year month or more than 12% per year. The GEE model for repeated measures revealed lesion type B2/C as a lesion-related risk factor for rapid progressors with a OR 2.139 (95% CI 1.066–4.294, p = 0.032) and diabetes mellitus as a patient-related risk factor (OR 2.782, 95% CI 1.349–5.737, p = 0.006). Lesions with both risk factors of diabetes and lesion type B2/C were at the highest risk of rapid progression (13/273 [4.8%] vs. 12/283 [4.0%] vs. 13/84 [15.5%], p < 0.001 for lesions with 0, 1, and 2 risk factors, respectively), with an odds ratio of 3.250 (95% CI, 1.451–7.282), compared to type A/B1 lesions in non-diabetic patients (Fig. 4).

The pathophysiology of coronary plaque formation has been studied for decades. Inflammation was identified as the primary driving force for activation and proliferation of smooth muscle cells, a processes mediated by growth factors. [14] Lesion progression can be induced by lipid accumulation, endothelial damage, or plaque erosion which involves many factors such as clinical risk factors of the individual, mechanical forces, etc. [15] Progression to a vulnerable plaque (known as thin cap fibroatheroma) [16], or injury such as deep plaque fissures or ulcerations, can cause acute coronary syndrome. [17]

Imaging techniques such as IVUS and OCT have been used to evaluate lesions and provide valuable information on plaque and lumen character. For example, the PROSPECT (Providing Regional Observations to Study Predictors of Events in the Coronary Tree) trial demonstrated that vulnerable plaques which are most likely to cause sudden unexpected adverse cardiac events can be pre-identified through IVUS imaging techniques. [2] In this study, the 3-year cumulative rate of major adverse cardiovascular events was 20.4%, among which, 11.6% were related to non-culprit lesions. Predictors for non-culprit lesions related events included insulin dependent diabetes and previous PCI as patient related factors, and a plaque burden of 70% or greater or a minimal luminal area of 4.0 mm² or less or thin-cap fibroatheromas lesions as lesion related factors. In addition, the PROSPECT II trial is ongoing trial which will assess the ability of intracoronary near infrared spectroscopy (NIRS) to identify vulnerable plaques which subsequently lead to coronary events.

However, even these sophisticated imaging tools have shortcomings. For example invasive imaging tools cannot evaluate distal lesions, increases the time and expense of PCI and may cause coronary dissection or plaque erosion through physical contact. Also, evaluating the entire coronary vasculature with these tools is a significant clinical burden to both the clinician and patient. [18] Therefore, we still need more feasible angiographic standards for lesion evaluation.

In our study, we analyzed non-culprit lesions with serial coronary angiography and QCA, which permitted more detailed analysis. Compared to the PROSPECT trial, we could analyze the angiographic progression in every non-culprit lesion, including those which were not related to clinical events. We calculated the average angiographic progression rate in non-culprit lesions, the predictors of rapid progression, and also the predictors of clinical events. Regarding the fact that IVUS cannot be performed in every patient that we meet in the clinic, our results may give clinicians a more practical guideline for future events related with non-culprit lesions.

Our population, in general, was receiving optimal medical therapy, with dual antiplatelet agents and lipid lowering agents. Previous studies have shown that statins could induce atheroma regression in coronary artery disease patients. [19, 20] Regarding lipid lowering agents, 93.8% were on statin, and the LDL level was very well-controlled, being 97 ± 35 mg/dL at the initial state and 67 ± 22 mg/dL and 65 ± 20 mg/dL at the 1st and 2nd follow-up respectively, suggesting that the study population, on average, were receiving appropriate statin therapy.

During the 2-year follow-up period, nonculprit-related events occurred in 6.3% of the patients, which were all revascularization events with no events of MI or nonculprit-related cardiac death. This was a similar rate to a previous study, where revascularization occurred in 10.5% during a 3-year follow-up. [2] Among lesion factors showing difference between those who and who did not experience nonculprit-IDR (i.e. smaller initial MLD, larger DS and had a higher proportion of lesion type B2/C), we were able to identify one lesion characteristic that was a significant predictor of subsequent events: lesion type B2/C. Diabetes was the only significant clinical factor to predict nonculprit IDR.

Nonculprit lesions showed progression in DS at a velocity of approximately 2.2%/year. Factors associated with a faster progression over the follow-up period, were identical to those risk factors of nonculprit-IDR; diabetes and lesion type B2/C. It is well-known that local factors, such as tortuosity and irregular contours of the lesion (incorporated in the lesion type definition [12]) and systemic factors, such as diabetes both are related to lesion progression. [21, 22] From our result, lesions with risk factors had a 1.5-fold faster velocity of DS progression, compared to those without risk factors.

Among the total lesions, 5.9% were rapid progressors, with a progression of DS more than 1% year month or more than 12% per year. From the histogram of velocity of DS progression, there was a distinct group of lesions with faster progression. Predictors of rapid progression were identical to that of nonculprit-IDR (i.e. diabetes and lesion type B2/C); however there was a distinct pattern in the effect size of each risk factor. For predictors of nonculprit-IDR, lesion type B2/C had a larger effect size compared to diabetes, whereas predictors of rapid progressors, the effect size of diabetes was larger than that of lesion type B2/C. This can be partially explained by the character of the outcomes; nonculprit-IDR and rapid progressors. As nonculprit-IDR is outcome that embodies the discretion of the operator, high risk morphology of the coronary lesion, such as lesion type B2/C, could have influenced the procedure. On the other hand, rapid progressors was an angiographic finding which is free from any interference by the eye of the operator. Therefore, a more ‘ugly’ morphology of the nonculprit lesion, could have had more influence on nonculprit-IDR, compared to rapid progressors.

Although we performed a serial angiographic analysis of nonculprit lesion, our study has several important limitations. One inherent limitation is due to angiography itself. Even to the most experienced eye, angiography is a lumenogram yielding little insight into plaque composition or lesion pathology. It is well-known that plaques that show different composition have different outcomes, and that plaque composition does not always correlate with DS. Information regarding plaque character is something that is unobtainable from angiography. Also, other than morphology, fluid mechanics such as coronary wall shear stress is known to be associated with plaque progression, which was not analyzed. Second, due to the retrospective nature of our study, there could have been a selection bias in patient selection. We compared the baseline clinical characteristics with the total parent population whom received PCI during the study period, and we found minimal difference between the two populations. However, we cannot complete deny possibilities of other selection bias within our study population.