Author + information
- Received May 22, 2014
- Revision received July 21, 2014
- Accepted August 28, 2014
- Published online January 1, 2015.
- Kefei Dou, MD∗,
- Dong Zhang, MD∗,
- Bo Xu, MBBS∗∗ (, )
- Yuejin Yang, MD∗∗ (, )
- Dong Yin, MD∗,
- Shubin Qiao, MD∗,
- Yongjian Wu, MD∗,
- Hongbing Yan, MD∗,
- Shijie You, MD∗,
- Yang Wang, MSc∗,
- Zhenqiang Wu, MSc∗,
- Runlin Gao, MD∗ and
- Ajay J. Kirtane, MD, SM†
- ∗State Key Laboratory of Cardiovascular Disease, Department of Cardiology, Cardiovascular Institute, Fuwai Hospital, and National Center for Cardiovascular Diseases, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, China
- †Columbia University Medical Center/New York Presbyterian Hospital, New York, New York
- ↵∗Reprint requests and correspondence:
Mr. Bo Xu or Dr. Yuejin Yang, Fuwai Hospital, National Center for Cardiovascular Diseases, A 167, Beilishi Road, Xicheng District, Beijing, 100037, China.
Objectives The purpose of this study was to establish a scoring system to evaluate the risk of side branch (SB) occlusion in patients undergoing coronary bifurcation intervention.
Background The risk of SB occlusion is the most important consideration affecting the selection of an optimal intervention strategy.
Methods A total of 1,545 consecutive patients undergoing percutaneous coronary intervention for bifurcation lesions (1,601 lesions treated with a single stent technique or main vessel [MV] stenting first strategy) were studied. A total of 1,200 lesions were used to construct the risk model and score system, and 401 lesions were used to validate the model. A multivariable risk score (RESOLVE [Risk prEdiction of Side branch OccLusion in coronary bifurcation interVEntion]) was constructed with incremental weights attributed to each component variable according to its estimated coefficients. SB occlusion after MV stenting was defined as any decrease in Thrombolysis in Myocardial Infarction flow grade or absence of flow in SB after MV stenting.
Results SB occlusion occurred in 118 (7.37%) of 1,601 bifurcation lesions. In multivariable analyses, 6 variables were independently associated with the risk of SB occlusion (model C-statistic = 0.80 [95% confidence interval: 0.75 to 0.85] with good calibration). For the 401 lesions included in the validation cohort, the RESOLVE score had a C-statistic of 0.77 (95% confidence interval: 0.69 to 0.86), with good calibration. SB occlusion rates in the validation cohort increased significantly across different risk groups, from 0.0% in the low-risk group, to 3.8% in the intermediate-risk group, and to 19.8% in the high-risk group (p < 0.001).
Conclusions The RESOLVE score, a novel angiographic risk stratification tool, can help identify patients at risk for SB occlusion during bifurcation intervention.
Percutaneous coronary intervention (PCI) of coronary bifurcation lesions is often considered a challenging task due to the risk of side branch (SB) compromise. The current favored approach to coronary bifurcations is a single-stent strategy with a provisional approach to the SB, in part due to shorter procedure times as well as its demonstrated noninferiority to a 2-stent strategy with respect to clinical outcomes (1–4). However, SB occlusion after main vessel (MV) stenting is an infrequent but serious procedural complication for the provisional approach. SB occlusion can result in vessel closure and ischemia, with clinically-significant myocardial infarction and even death depending upon the size of the SB (and the myocardial territory subtended by it) (5,6). Thus, particularly for large SB with significant disease, the optimal intervention strategy remains controversial.
Clinical studies of bifurcation PCI have primarily attempted to determine which strategy (e.g., provisional vs. 2-stent) is most optimal. However, it can be argued that no single approach should be the default strategy in all patients with bifurcation lesions (7). For example, the contribution of specific bifurcation anatomy in determining the optimal technique for bifurcation PCI has been recently examined (8). Ideally, the decision-making regarding the best strategy to treat bifurcation lesions should ultimately be based on the individual characteristics of the treated patient/lesion affecting the probability of SB occlusion.
The risk of SB occlusion is the most important factor affecting the selection of an optimal intervention strategy in coronary bifurcation intervention. Previous studies have reported that the risk of SB occlusion could be affected by numerous factors like bifurcation lesion anatomy and the PCI procedure (6,9–11). However, the potential contributions of angiographic and PCI procedural indexes to SB occlusion have not been fully elucidated. Clinical prediction models involving clinical, angiographic, and PCI procedural indexes may, therefore, be helpful for medical decision making. Accordingly, this study was designed to establish a novel score system, RESOLVE (Risk prEdiction of Side branch OccLusion in coronary bifurcation interVEntion), to evaluate the risk of SB occlusion in coronary bifurcation intervention.
From January 2012 to July 2012, a cohort of 7,007 consecutive patients underwent PCI at Fuwai Hospital in Beijing, China. For the present study, specific inclusion criteria were: 1) patients with a coronary bifurcation lesion (defined as a coronary artery narrowing occurring adjacent to or involving the origin of a significant SB) undergoing PCI; and 2) patients with bifurcation lesion consisting of a significant SB confirmed by the treating physician and the core laboratory. Selection of SB was not based on the reference diameter of SB to avoid neglecting small SB that could have important clinical significance. The criteria for a significant SB were based upon those defined by the European Bifurcation Club (12); that is, a branch that the operator would not want to lose in the global context of a particular patient (symptoms, location of ischemia, viability, collateralizing vessel, left ventricular function and so on). The reference diameter of SB was principally ≥1.5 mm. Patients undergoing elective SB stenting before MV stenting were excluded. Among the 7,007 patients, 5,172 patients without bifurcation lesions and 290 patients with bifurcation lesions undergoing elective SB stenting were excluded. A total of 1,545 patients with 1,601 bifurcation lesions that met all the inclusion criteria and had no exclusion criteria were included in this study (Figure 1). PCI procedure and periprocedural medications were based on the operator’s discretion and current guidelines (Online Appendix).
The ethics committee of the Cardiovascular Institute and Fuwai Hospital approved this study.
Data collection and quantitative coronary angiography
Clinical data were obtained through a review of the medical records. All baseline and procedural cineangiograms were reviewed and analyzed by an independent core laboratory (for details, please see the Online Appendix). Blood sampling and cardiac enzyme measurement are also detailed in the Online Appendix.
Quantitative coronary angiography (QCA) was performed using standard quantitative analyses and definitions (13). Angiograms obtained at baseline and after pre-dilation were analyzed with a computer-based system dedicated to bifurcation analysis (Qangio XA, version 7.3, Medis, Leiden, the Netherlands). We obtained quantitative angiographic measurements of the 4 segments of the bifurcation lesion: the proximal MV segment, the distal MV segment, the SB segment, and the bifurcation core segment (Figure 2). Bifurcation core was defined as the central part of the bifurcation, which begins where the common vessel starts to split into 2 branches and ends at the carinal point (14), the area of which is calculated by the Qangio XA software. We also obtained the bifurcation angle (BA) (the angle between the distal MV and the SB) from the analysis system.
In addition to the intrinsic data outputted by the QCA system, another variable was derived based on the raw QCA data: the diameter ratio between MV/SB ([reference diameter of proximal MV + reference diameter of distal MV] ⁄2 [reference diameter of SB]), which is a parameter reflecting the relative plaque shift burden from MV to SB.
Statistical methods and score determination
Statistical methods are detailed in the Online Appendix. The database was divided into 2 subsets by procedural chronological order: a developmental dataset of 1,200 lesions that served to construct the risk model, and a validation subset of 401 lesions to test and validate the model. The multivariable model was built by stepwise variable selection with the same entry and exit criteria as in the univariable analyses. The score was then derived by attributing integer numbers to the variables retained in the multivariable model. Additionally, the scoring system was then used to define 3 risk groups (low, intermediate, and high risk). All p values were 2-tailed, and a p value <0.05 was considered statistically significant. All analyses were performed with the SAS version 9.3 system (SAS Institute, Cary, North Carolina).
Patient, lesion, and procedural characteristics
SB occlusion occurred in 118 (7.37%) of 1,601 bifurcation lesions treated with a single stent technique or MV stenting first strategy. Of SB occlusions, total occlusion occurred in 83 (70.3%) of lesions and a decrease in TIMI (Thrombolysis In Myocardial Infarction) flow occurred in 35 (29.7%) lesions. Blood flow in SB was restored spontaneously in 12 (10.2%) lesions and by SB intervention in 8 (6.8%) lesions of 118 occluded SB. In total, 98 (83.0%) lesions occluded permanently. SB occlusion was significantly associated with periprocedural myocardial infarction (6.1% vs. 1.3%, p < 0.001) (Online Table 1).
The overall patient characteristics are shown in Table 1. Except for the rate of myocardial infarction at 1 month and emergency PCI, all baseline characteristics were not significantly different between the 2 groups. Lesion and procedural characteristics are presented in Table 2. Among lesion characteristics, location of bifurcation, Medina classification, and distribution of plaque differed significantly between the 2 study groups. Among procedural characteristics, all data except the rate of SB pre-dilation were significantly different between the 2 groups. QCA data are presented in Table 3. There were significant differences between the 2 groups in reference diameter of all 4 parts of the bifurcation. However, with respect to length, only the lesion length of bifurcation core was significantly different between the 2 groups. The diameter ratio between MV/SB in the SB occlusion group (1.6 ± 0.3) was significantly higher than in the no SB occlusion group (1.4 ± 0.3) (p < 0.001).
Additional analyses of patient and lesion characteristics in the low BA group (BA <50°) and high BA group (BA ≥50°) are presented in Online Tables 2 to 5.
The resolve risk model and resolve risk score
After univariable and multivariable selection, 5 primary angiographic variables (plaque distribution, MV TIMI flow grade before stenting, pre-procedural diameter stenosis of bifurcation core, bifurcation angle, and diameter stenosis of SB before MV stenting) and 1 derived variable (diameter ratio between MV/SB) remained independently associated with the risk of SB occlusion. There was little correlation apparent between these variables; the variance inflation factor showed an absence of multicollinearity among variables in the model (shown in Online Table 6). Within the derivation sample, the C-statistic for this model was 0.80 (95% confidence interval [CI]: 0.75 to 0.85) (Figure 3A), and excellent calibration was observed (Hosmer-Lemeshow [HL] p = 1.00).
The scores attributed to each variable according to the estimated coefficients from the derivation dataset are shown in Table 4. The risks of SB occlusion associated with each point are presented in Online Table 7. The C-statistic for the risk score was only slightly worse than that of the original model: 0.76 (95% CI: 0.71 to 0.82; HL p = 0.12) (Figure 3A).
RESOLVE scores range from 0 to 43. The interquartile range and the frequency distribution of each variable across quartiles of RESOLVE score are displayed in Online Table 8. As shown in the table, event rates in the derivation cohort across quartiles of RESOLVE score were: 1.0% in quartile I (RESOLVE score: 0 to 2); 3.9% in quartile II (RESOLVE score: 3 to 6); 3.6% in quartile III (RESOLVE score: 7 to 9); and 17.5% in quartile IV (RESOLVE score: ≥10) (p < 0.001). The odds of SB occlusion were 0.049 (95% CI: 0.007 to 0.356) for quartile I versus quartile IV, 0.193 (95% CI: 0.110 to 0.338) for quartile II versus quartile IV, and 0.175 (95% CI: 0.092 to 0.333) for quartile III versus quartile IV. There were no differences in SB occlusion rate between quartiles II and III (p = 0.80). Thus, quartile I was determined to be the low-risk group (quartile I, score: 0 to 2); quartiles II and III were combined into an intermediate-risk group (quartiles II and III, score: 3-9); and quartile IV was the high-risk group (quartile IV, score: 10 to 43) (Table 5).
For the 401 lesions included in the validation cohort, the RESOLVE model displayed good prognostic accuracy, with a C-statistic of 0.81 (95% CI: 0.73 to 0.89; HL p = 0.77) (Figure 3B). The RESOLVE score also displayed good prognostic accuracy, with a C-statistic of 0.77 (95% CI: 0.69 to 0.86; HL p = 0.58) (Figure 3B). As shown in Table 5, SB occlusion rates in the validation cohort increased significantly across different risk groups: 0.0% in the low-risk group, 3.8% in the intermediate-risk group, and 19.8% in the high-risk group (p < 0.001). We performed an additional analysis to validate the performance of RESOLVE score in the subset of patients in whom the jailed wire technique was utilized (shown in Online Figure 1). Although the C-statistic was slightly worse than that of the original RESOLVE score, it was still predictive in patients in whom the jailed wire technique was utilized.
The major findings of this study are: 1) a novel risk stratification score system (the RESOLVE risk score) consisting of 6 variables was developed; 2) the 6 variables independently predictive of adverse outcomes included 5 angiographic variables and 1 novel derived variable (diameter ratio of MV/SB); and 3) the RESOLVE score was able to accurately predict the risk of SB occlusion in patients undergoing coronary bifurcation intervention with good discrimination and calibration in both the derivation and validation datasets.
The RESOLVE risk score was created specifically for patients with coronary bifurcation disease undergoing PCI. Bifurcations frequently contain a nonuniform geometrical distribution of atherosclerotic plaque, and the presence of a large plaque burden at the core of the bifurcation has been demonstrated to be associated with SB ostial deterioration and occlusion during MV treatment, even if significant disease in the SB is absent at baseline (15). It is, therefore, not surprising that plaque located at the same side of the SB was an important correlate of SB occlusion. Among other covariates included in the model, arteries with lower TIMI flow grade are more likely to have thrombus or other complex lesion features, and therefore, are more likely to have SB occlusion. Of note, the diameter stenosis of the bifurcation core, but neither the proximal nor the distal MV stenosis, was independently predictive of SB occlusion. The bifurcation core is the area most adjacent to SB, and as such, a smaller diameter stenosis could be a correlate of plaque shift during MV PCI. We also found that the risk of SB occlusion increased as the ratio of the diameter of MV/SB increased, which is concordant with the results of previous studies (16). The mechanism of SB occlusion is a question that remains to be clarified, although both plaque shift and carina shift have been thought to be potential mechanisms (17). Plaque shift seemed to explain most of the predictors in the present study, which was in consistent with a previous study (6). Future large sample size studies using intraluminal imaging may help to further clarify the actual mechanism of SB occlusion. Although previous studies have reported that lesion length of SB was associated with SB occlusion, in the present study, based on 1,545 consecutive patients with 1,601 bifurcation lesions, lesion length of SB was not selected as an independent predictor of SB occlusion after univariable and multivariable analysis. In our study, we found diameter stenosis of bifurcation core and SB were independent predictors of SB occlusion. Neither the MV nor SB lesion length was significantly correlated with SB occlusion. The severity of lesion rather than the length of lesion seemed to be more predictive for SB occlusion in our study. However, it is possible that due to the relatively short length of SBs within the current study, we may have lacked the ability to detect lesion length as an independent predictor of occlusion. Additional prospective clinical studies with a broader range of SB lengths may further help to clarify the question.
The effect of BA on the rate of SB occlusion during PCI is controversial. Previous studies have reported that smaller angle in coronary bifurcations predicted higher SB compromise, restenosis, and major adverse cardiovascular event rates based on small sample size (16). In addition, Goto et al. (18) reported that BA was not associated with SB compromise after MV stenting. However, Dzavik et al. (19) found that BA >50° was an independent predictor of major adverse cardiovascular events after bifurcation crush stenting. In our study, a wide BA predicted SB occlusion after MV stenting. Part of the explanation was that bifurcations with smaller BA were easier for flow diversion into the SB, and a BA that is too large might increase the pressure drop and flow resistance (20), thus increasing the SB occlusion risk. Another potential explanation was that increasing BA decreased wall shear stress and increased oscillatory shear index significantly around the carina (21), which might induce plaque proliferation at the bifurcation region (22,23). Higher plaque volume in the bifurcation core may contribute to the higher SB occlusion risk. Our study found that lesions with high BA (BA ≥50°) had significantly higher pre-procedural diameter stenosis of bifurcation core (37.7 ± 29.2% vs. 29.9 ± 26.5%, p < 0.001) as compared with lesions with low BA (BA <50°) (Online Appendix). In addition, lesions with high BA (BA ≥50°) had a significantly lower incidence of jailed wire in SB (22.7% vs. 32.2%, p < 0.001) compared with lesions with low BA (BA <50°); the inability to protect the SB may also contribute to SB occlusion.
Compared with previous studies (6,9–11), the strength of our study is that we have established the RESOLVE risk score system as a risk stratification tool for SB occlusion in coronary bifurcation intervention. Another strength of our study was that the study was conducted in a consecutive cohort of bifurcation patients, which could reflect the real-world clinical practice. The RESOLVE risk score, derived from a large cohort of patients undergoing provisional bifurcation PCI, has excellent discriminative ability (C-statistic for this RESOLVE risk model was 0.80 [95% CI: 0.75 to 0.85], and the C-statistic for RESOLVE risk score was 0.76 [95% CI: 0.71 to 0.82]) that was slightly greater than that described in a previous study (6). The RESOLVE risk score system contains only 6 variables and is simple to calculate, but does include QCA data, which may add to its discriminatory capacity. Coronary angiography before stenting may represent the most useful angiographic time point for physicians to choose a bifurcation intervention strategy.
First, the RESOLVE risk score was developed in consecutive patients enrolled at a single center. Thus, although the discrimination of the RESOLVE risk score was confirmed in a separate cohort of patients, its predictive accuracy of SB occlusion in patients with coronary bifurcation disease undergoing PCI should be further validated from a different study dataset. Second, the extent of coronary artery disease was determined by QCA. Although QCA may be time-consuming and/or not immediately available, it provides a more objective determination of the extent and severity of coronary artery disease than visual assessment (24). Therefore, scores that use QCA may be more reproducible than those based only on visual estimation of angiographic variables.
The RESOLVE risk score is a novel and simple risk stratification tool specifically developed for patients with coronary bifurcation disease undergoing PCI. The RESOLVE risk score is accurately predictive in both discrimination and calibration, and further validation of its performance in other patient populations is warranted.
This research was supported by the Beijing Municipal Science & Technology Commission. (grant number Z141107002514096). Dr. Yang has received research grants from Abbott Vascular, Boston Scientific, Medtronic, MicroPort, Lepu Medical, Sino Medical, and Essen Technology. Dr. Kirtane has received research grants from Medtronic, Boston Scientific, Abbott Vascular, Vascular Dynamics, Abiomed, Eli Lilly, and St. Jude Medical. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. The first 2 authors contributed equally to this work.
- Abbreviations and Acronyms
- confidence interval
- Hosmer-Lemeshow test
- main vessel
- percutaneous coronary intervention
- quantitative coronary angiography
- side branch
- Thrombolysis In Myocardial Infarction
- Received May 22, 2014.
- Revision received July 21, 2014.
- Accepted August 28, 2014.
- American College of Cardiology Foundation
- Maeng M.,
- Holm N.R.,
- Erglis A.,
- et al.
- Ferenc M.,
- Gick M.,
- Kienzle R.P.,
- et al.
- Hildick-Smith D.,
- de Belder A.J.,
- Cooter N.,
- et al.
- Muramatsu T.,
- Onuma Y.,
- Garcia-Garcia H.M.,
- et al.
- Hahn J.Y.,
- Chun W.J.,
- Kim J.H.,
- et al.
- Latib A.,
- Colombo A.
- Girasis C.,
- Farooq V.,
- Diletti R.,
- et al.
- Lansky A.J.,
- Dangas G.,
- Mehran R.,
- et al.
- Koo B.K.,
- Waseda K.,
- Kang H.J.,
- et al.
- Kimura B.J.,
- Russo R.J.,
- Bhargava V.,
- et al.
- Rodriguez-Granillo G.A.,
- Garcia-Garcia H.M.,
- Wentzel J.,
- et al.
- Beauman G.J.,
- Vogel R.A.