Complexity of Atherosclerotic Coronary Artery Disease and Long-Term Outcomes in Patients With Unprotected Left Main Disease Treated With Drug-Eluting Stents or Coronary Artery Bypass Grafting

Study group and design

The MAIN-COMPARE (Revascularization for Unprotected Left Main Coronary Artery Stenosis: Comparison of Percutaneous Coronary Angioplasty Versus Surgical Revascularization) study enrolled patients with unprotected LMCA stenosis who underwent CABG or PCI as the index procedure at 12 major cardiac centers in Korea between January 2000 and June 2006 (18). From January 2000 through May 2003, coronary stenting was performed exclusively with bare-metal stents, whereas from May 2003 through June 2006, DES were used exclusively. Because the SYNTAX score was developed in the DES era, as an integral part of the SYNTAX trial design, and methods of stent implantation have evolved from the bare-metal stent era to the DES era with an attempt to fully cover the diseased segments, we included only the post-DES cohort (wave 2 of the registry; DES vs. concurrent CABG between May 2003 and June 2006) among the overall cohort for these analyses. The details of the overall cohort and a separate cohort from bare-metal stents to DES have been described previously (12). Patients who had prior CABG, those who underwent concomitant valvular or aortic surgery, and those who had ST-segment elevation myocardial infarction (MI) or presented with cardiogenic shock were excluded. The local ethics committee at each hospital approved the use of clinical data for this study, and all patients provided written informed consent.

Patients underwent PCI, instead of CABG, because of either patient or physician preference or the high risk associated with CABG. Methods of stent implantation for patients with LMCA disease have been described previously (19,20). All procedures were performed using standard interventional techniques. Antiplatelet therapy and periprocedural anticoagulation followed standard regimens. After the procedure, patients were prescribed aspirin indefinitely and clopidogrel for at least 6 months, regardless of DES type. Surgical revascularization was performed using standard bypass techniques (21). Complete revascularization was performed when possible using arterial conduits or saphenous vein grafts. Clinical follow-up after PCI and after CABG was recommended at 1 month, 6 months, and 1 year and then annually thereafter. For this analysis, the follow-up period extended through October 30, 2009, to ensure that all patients had an opportunity for at least 3 years and approximately up to 7 years of follow-up information.

The primary safety outcomes were death and the composite of death, Q-wave MI, or stroke. The primary efficacy outcome was target vessel revascularization (TVR). Death was defined as death from any cause. Q-wave MI was defined as documentation of a new abnormal Q-wave after the index revascularization. Stroke, as indicated by neurological deficits, was confirmed by a neurologist on the basis of imaging studies. TVR was defined as any repeat revascularization in any left anterior descending coronary artery or left circumflex coronary artery, as well as in the target segment. All outcomes of interest were confirmed by source documentation collected at each hospital and were centrally adjudicated by an independent group of clinicians.

The SYNTAX score reflects a comprehensive anatomical assessment, with higher scores indicating more complex CAD. As previously defined (13,14), we categorized study patients according to the SYNTAX score; a low score was defined as ≤22, an intermediate score as 23 to 32, and a high score as ≥33. SYNTAX scores were calculated by the independent angiographers blinded to patient information and treatment types. The details of angiographic measurements and the SYNTAX scoring algorithm at the core laboratory have been described elsewhere (22).

Statistical analysis

For each category of the SYNTAX score, patient characteristics pertaining to the index revascularization, including demographic characteristics, the presence or absence of a variety of clinical and coexisting conditions, cardiac presentation, left ventricular function, and lesion characteristics were compared between DES and CABG with the use of the Student t tests or Wilcoxon nonparametric tests for continuous variables and the chi-square or Fisher exact test for categorical variables. Five-year cumulative incidence rates of clinical outcomes were estimated using the Kaplan-Meier method and compared using the log-rank test.

To reduce the impact of treatment selection bias and potential confounding in an observational study, we performed rigorous adjustment for differences in baseline characteristics of patients by use of weighted Cox proportional-hazards regression models using inverse-probability-of-treatment weighting (23). With that technique, weights for patients receiving CABG were the inverse of (1 − propensity score), and weights for patients receiving DES were the inverse of the propensity score. Propensity scores were estimated without regard to outcomes, using multiple logistic regression analysis. A full nonparsimonious model was developed that included all variables shown in Table 1. Model discrimination was assessed using c-statistics, and model calibration was assessed using Hosmer-Lemeshow statistics. For each SYNTAX category (low, intermediate, and high risk), a separate propensity score for DES versus CABG was derived. Interactions between the levels of SYNTAX score and treatment strategy were tested by the incorporation of formal interaction terms in the multivariate Cox model. Weighted Cox proportional-hazards regression models and adjusted survival curves were constructed using inverse probability weights (24). All statistical analyses were performed using SAS version 9.1 (SAS Institute, Inc., Cary, North Carolina), and a 2-sided p value of ≤0.05 was considered to indicate statistical significance.

Population

Between May 2003 and June 2006, 1,474 patients with unprotected LMCA disease were treated with DES implantation (n = 784) or CABG (n = 690). During the study enrollment period, the SYNTAX score algorithm was not available for the physicians. A retrospective retrieval of a baseline angiogram for a detailed measurement of the SYNTAX score was available in 1,146 patients (78%) of the overall cohort (645 patients [82%] in the DES group and 501 patients [73%] in the CABG group). The mean SYNTAX score was significantly lower in the DES group than in the CABG group (24.8 ± 10.9 vs. 38.7 ± 13.3, p < 0.001).

Comparisons of the baseline characteristics between 2 treatment groups in each of the SYNTAX categories are shown in Table 1. In each category of the SYNTAX score, there were some imbalances of baseline clinical and angiographic features between the 2 treatment groups with statistical significance. Overall, with increasing SYNTAX score, patients had higher risk clinical and angiographic profiles.

Procedural characteristics of the study patients were as follows: 1) among DES patients, 78% received sirolimus-eluting stents and 22% received paclitaxel-eluting stents; 2) the mean number of stents implanted in a patient's LMCA lesions was 1.2 ± 0.5, the mean total length of the stents was 32.8 ± 22.0 mm, and the mean stent diameter was 3.4 ± 0.2 mm; 3) among CABG patients, 46% underwent off-pump surgery; and 4) 96% underwent revascularization of the left anterior descending coronary artery with an internal thoracic artery. Complete revascularization was achieved in 64% of DES patients and 80% of CABG patients.

Outcomes

The median follow-up duration was 55.1 months (interquartile range: 45.8 to 65.5 months) in the overall patient group. Complete follow-up for major clinical events was obtained in 98.3% of the overall cohort (98.4% for the DES group and 98.0% for the CABG group; p = 0.57).

During 5 years of follow-up, the observed long-term rates of death and the composite of death, Q-wave MI, or stroke were significantly lower in the DES group than in the CABG group in patients with low SYNTAX scores (Table 2,Fig. 1). These safety outcomes were similar between the 2 treatment groups among patients with intermediate scores, but there was evidence of a nonsignificant increase in the rate of safety outcomes associated with DES compared with CABG in patients with high scores. The advantage of CABG was quite consistent for TVR.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Unadjusted Kaplan-Meier Survival Curves for Long-Term (5-Year) Outcomes According to SYNTAX Score Category

(A) Outcomes (death; a composite outcome of death, myocardial infarction [MI], or stroke; and target-vessel revascularization) in patients with low SYNTAX (Synergy Between PCI With Taxus and Cardiac Surgery) scores (≤22). (B) Outcomes in patients with intermediate SYNTAX scores (23 to 32). (C) Outcomes in patients with high SYNTAX scores (≥33). CABG = coronary artery bypass grafting; DES = drug-eluting stent(s).

After adjustment of baseline covariates using inverse-probability-of-treatment weighting in each of the SYNTAX score categories, the adjusted hazard ratios for death and serious composite outcomes favored DES treatment in patients with low SYNTAX scores; in contrast, the adjusted hazard ratios for these safety outcomes favored CABG in patients with high scores (Table 2, Fig. 2). When we assessed the effect on outcomes of treatment strategy and its interaction with SYNTAX score category, treatment effects for death and composite outcomes, but not TVR, were modified by the level of SYNTAX score (interaction p = 0.047 for death, interaction p = 0.08 for composite outcome, and interaction p = 0.45 for TVR).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Adjusted Survival Curves for Long-Term (5-Year) Outcomes According to SYNTAX Score Category

(A) Outcomes (death; a composite outcome of death, MI, or stroke; and target-vessel revascularization) in patients with low SYNTAX scores (≤22). (B) Outcomes in patients with intermediate SYNTAX scores (23 to 32). (C) Outcomes in patients with high SYNTAX scores (≥33). Abbreviations as in Figure 1.

In the DES group, the rates of death and the composite of death, Q-wave MI, or stroke were significantly increased in patients with high SYNTAX scores compared with those with low or intermediate scores. In the CABG group, however, there was no uniform increase of safety outcomes according to SYNTAX score.

In the overall population, the unadjusted risks for death and the composite of death, Q-wave MI, or stroke were significantly lower in the DES group than in the CABG group, whereas the rate of TVR was significantly higher in the DES group (Table 3). After adjustment of the SYNTAX score, however, the 5-year risks for death and the composite of death, Q-wave MI, or stroke were similar in the 2 groups, but the adjusted risk for TVR was consistently higher with DES. Similar results were observed even after adjustments of the SYNTAX score and all potentially explanatory factors for clinical outcomes.

Study limitations

First, the inherent limitations of a nonrandomized, registry study should be acknowledged. The choice of revascularization was at the discretion of the treating physician and/or patient. Despite rigorous adjustment for selection bias and confounding with the inverse-probability-of-treatment weighting method, hidden biases or residual confounding may have affected the results. Therefore, our findings should be confirmed or refuted through large, randomized clinical trials with long-term follow-up. Second, our analysis was underpowered to detect significant differences in mortality and serious composite outcomes. Nonsignificant trends for these outcomes might have been significant with a larger cohort of patients. Third, in the present analysis, SYNTAX scoring was performed retrospectively and only in 78% of cases, introducing at least the potential of an unmeasured bias. Finally, because this study evaluated the first generation of DES, the direct application of these findings to second-generation or next-generation DES may be limited.