In patients with complex coronary artery disease—namely, unprotected left main coronary artery (ULMCA) or de novo 3-vessel disease—undergoing coronary artery bypass graft (CABG) surgery or percutaneous coronary intervention (PCI), the long-term prognostic implications of complete (CR) versus incomplete (ICR) revascularization is unsettled ((1),(2),(3),(4),(5),(6),(7),(8),(9),(10),(11),12). In addition, recent studies have suggested that angiographic CR after CABG revascularization might not be associated with a significant improvement in long-term clinical outcomes ((6),7).

The impact of successful or unsuccessful treatment of total occlusion (TO) and its association with completeness of revascularization in PCI- or CABG-treated patients with ULMCA and/or multivessel disease also remains undefined. Although multiple registries have associated successful complete TO PCI with improved long-term mortality, these studies were conducted in patients with less complex coronary disease ((13),(14),(15),(16),17). Only 1 retrospective registry study has examined CTO PCI in patients with predominantly multivessel disease and associated successful CTO PCI in the context of CR with a survival advantage, compared with patients with ICR (18).

The purpose of this study was to report the long-term (4-year) clinical impact of complete and incomplete angiographic revascularization in the All-Comers SYNTAX (Synergy Between Percutaneous Coronary Intervention with Taxus and Cardiac Surgery) Trial. The All-Comers SYNTAX Trial provided an opportunity for the assessment of patients where selection bias in enrolling patients was minimal and therefore was potentially more representative of contemporary clinical practice. Secondly, the presence of TOs to influence the ability of PCI or CABG to achieve CR was examined, on the basis of the hypothesis that failure to treat TOs would have a significantly detrimental impact on long-term clinical outcomes in PCI- and CABG-treated patients.

The All-Comers SYNTAX Trial is a randomized, prospective, multicenter trial (n = 3,075) investigating patients with ULMCA disease (isolated or associated with 1-, 2-, or 3-vessel disease) or de novo 3-vessel coronary artery disease and has previously been described ((19),(20),21). In brief, exclusion criteria were appropriately limited and consisted of patients with prior coronary revascularization, the requirement of concomitant cardiac surgery, or ongoing acute myocardial infarction. During the local Heart Team meeting, the interventional cardiologist and cardiac surgeon specified the number of coronary lesions requiring treatment and their angiographic location and characteristics with the SYNTAX Score as a tool to aid in this process ((22),(23),24). All patients considered as potentially achieving “equivalent anatomic” revascularization with percutaneous or surgical revascularization by the Heart Team were randomized on a 1:1 basis (n = 1,800) to either PCI with TAXUS Express paclitaxel-eluting stents (Boston Scientific Corporation, Natick, Massachusetts) or CABG. Patients unsuitable for randomization were nested into PCI (CABG ineligible patients, n = 198) and CABG registries (PCI ineligible patients, n = 1,077) (Figure 17_gr1). All randomized patients underwent planned follow-up. Within the nested registries, all PCI patients and 649 randomly allocated CABG patients underwent follow-up on the basis of the original study protocol ((20),21).

An independent Clinical Events Committee, including cardiologists, cardiac surgeons, and a neurologist reviewed all the primary clinical endpoints ((20),21). Baseline and peri- and post-procedural data were prospectively collected by the individual participating centers. Pre-procedural left ventricular ejection fraction (LVEF) was defined as the percentage LVEF taken by transthoracic echocardiography or diagnostic coronary angiogram and was categorized as good (≥50%), moderate (30% to 49%), or poor (<30%). Four-year clinical outcomes were compared in groups with and without angiographic CR in the PCI and CABG arms, with analyses repeated in the respective TO/non-TO groups.

Clinical outcomes included all-cause death, cardiac death, major adverse cardiac and cerebrovascular events (MACCE) (a composite of all-cause death, myocardial infarction, stroke, and all-cause revascularization) and its components, stent thrombosis, and graft occlusion. Direct comparisons between the All-Comers PCI and CABG arms were not undertaken due to the previously reported outcomes of the SYNTAX trial ((19),(20),(25),26) and the presence of greater clinical comorbidity and more complex anatomical coronary disease in the nested PCI and CABG registries, respectively, which were the predominant reasons for entry ((27),28).

The calculation of the SYNTAX Score was undertaken by the study sites and an independent core laboratory (Cardialysis BV, Rotterdam, the Netherlands) blinded to the treatment assignment ((22),(23),24). The age and angiographic characteristics of TOs were specified as part of the SYNTAX Score calculation. The definition of the TOs required that there was absolutely no flow through the lesion (Thrombolysis In Myocardial Infarction flow grade 0). Antegrade flow beyond the TO maintained by bridging collaterals and/or ipsi-collaterals was permitted. Core laboratory-calculated SYNTAX Scores are reported in this study. Procedural success rates of recanalization/bypass of TOs were available in the electronic case records of the randomized SYNTAX population and are reported.

Completeness of revascularization was prospectively determined after the revascularization procedure by the operator, on the basis of the intended “equivalent anatomic” revascularization agreed during the local Heart Team conference before revascularization. The degree of revascularization was angiographic, on the basis of segment numbering of vessels with a diameter ≥1.5 mm used in the calculation of the SYNTAX Score ((22),(23),24). The exact definition of CR defined in the original trial protocol (NCT00114972) is detailed in the following text. The interventional cardiologist and surgeon during the local Heart Team conference will specify the number of lesions and locations requiring treatment to achieve “equivalent anatomic” revascularization (decision). Complete revascularization is defined as the treatment of any lesion with more than 50% diameter stenosis in vessels ≥1.5 mm as estimated on the diagnostic angiogram during the local Heart Team conference. Outcomes will be documented by the operator based on whether the intended equivalent anatomic revascularization was achieved.

The means ± SD for continuous variables were compared with the Student t test. Binary variables are reported as counts and/or percentages and compared with the chi-square test. Four-year clinical outcomes were compared in groups with and without angiographic CR and expressed as Kaplan-Meier estimates, with curves examined visually. Comparisons were made with Cox proportional hazard ratios (HRs) and the log-rank test. In addition, TO and non-TO group-specific HRs for all recorded 4-year clinical outcomes were obtained. The significance of the interaction effect between CR versus ICR in the PCI and CABG arms and the respective TO and non-TO groups were tested with the chi-square test. This was undertaken to assess whether the HR of all 4-year clinical outcomes for CR versus ICR was different across the TO and non-TO groups.

Binary logistic regression analyses were undertaken to identify univariate predictors of ICR with baseline characteristics (20) and components of the SYNTAX Score, respectively. The variables examined are detailed in (Table 1). The SYNTAX Score was excluded from these analyses to allow for the appropriate identification of its components (e.g., TO, calcification) associated with ICR. Univariate predictors of ICR were entered into a multivariable model. The enter method was implemented to determine independent predictors, with a variable entry/stay criteria of 0.05/0.1. Similar analyses for ICR were repeated in the TO treated groups. A 2-sided p value < 0.05 was considered significant for all tests. Analyses were conducted with SPSS (version 19.0, SPSS, Inc., Chicago Illinois) and STATA (version 11.0, StataCorp, College Station, Texas).

Baseline characteristics of the randomized and nested registry arms (All-Comers population) of the SYNTAX Trial have previously been described ((20),28). Within the All-Comers population (n = 2,636), angiographic CR was achieved in 52.8% of the PCI arm (n = 1,095) and 66.9% of the CABG arm (n = 1,541) (Figure 17_gr1).

Baseline and procedural characteristics of CR and ICR in the PCI (n = 1,095) and CABG (n = 1,541) arms are shown (Table 1). Within the PCI and CABG arms, more anatomically complex disease (higher SYNTAX Score)—including the presence of TOs, greater clinical comorbidity (higher EuroSCORE [(29)] and Parsonnet Score [30]), and a reduced prevalence of ULMCA disease—were evident in patients with ICR compared with CR.

Within the PCI and CABG arms, ICR (compared with CR) was associated with significantly higher frequencies of 4-year mortality, all-cause revascularization, stent thrombosis (PCI arm), and MACCE (Figure 17_gr2).

Multivariate analyses of ICR in the PCI and CABG arms are shown (Table 2). The strongest independent predictor of ICR in the PCI arm was the presence of a TO (HR: 2.70, 95% confidence interval [CI]: 1.98 to 3.67, p < 0.001). The strongest independent predictor of ICR in the CABG arm was left arterial dominance (HR: 2.19, 95% CI: 1.05 to 4.57, p = 0.038). Left arterial dominance, any right coronary artery lesion, number of lesions, and clinical variables (EuroSCORE/Parsonnet Score) were common independent predictors of ICR in the PCI and CABG arms.

Baseline and procedural characteristics of patients with and without TOs in the PCI (n = 1,095) and CABG (n = 1,541) arms are shown (Table 3). Eight hundred and forty patients (PCI: 26.3%, CABG: 36.4%, p < 0.001) were identified to have 1,007 TOs. All patients with TOs (PCI and CABG arms) were significantly more likely to have had a previous myocardial infarction, a lower LVEF, diffuse or small vessel disease, more coronary lesions, lesion length >20 mm, and a reduced prevalence of ULMCA disease. Patients with TOs were less likely to achieve CR in the PCI and CABG arms (CR PCI: non-TO 59.8%, TO 34.3%, p < 0.001; CR CABG: non-TO 69.8%, TO 64.8%, p = 0.048).

(Table 4) shows the baseline angiographic characteristics of the TOs in the entire SYNTAX population and PCI and CABG arms. The age of the TOs was documented as >3 months or unknown in 95.6% of the study population. Approximately two-thirds of TOs (68.1%) were located in the proximal-mid coronary vessels on coronary angiography (Figure 17_gr3).

(Figure 17_gr4) demonstrates the procedural success rates of TOs (n = 543) on an intention to treat basis by the study sites in the randomized SYNTAX population. More TOs were successfully treated with CABG compared with PCI (68.1% vs. 49.4%, p < 0.001).

(Figure 17_gr5) demonstrates comparisons of HRs for CR versus ICR in the PCI and CABG arms for all 4-year clinical endpoints in the respective TO and non-TO groups. The findings associating ICR (compared with CR) with higher frequencies of 4-year adverse clinical outcomes, as seen in the CABG and PCI arms of the SYNTAX population, remained consistent in the respective TO groups (nonsignificant p values for interaction), with a few notable exceptions.

Within the PCI arm (p value for interaction <0.001), CR in patients with TOs was associated with substantially fewer episodes of 4-year stent thrombosis compared with ICR (CR: 0%, n = 0; ICR: 4.6%, n = 7; no HR available as 0 events in patients with CR) (Figure 17_gr5). Within the CABG arm (p value for interaction = 0.0008), CR in patients with TOs was related to a significantly lower risk of 4-year graft occlusion, compared with ICR (CR 2.7% vs. ICR 7.6%; HR: 0.34, 95% CI: 0.14 to 0.84, p = 0.019); conversely, in patients with CR without TOs, the risk of 4-year graft occlusion was substantially higher (CR 5.2% vs. ICR 0.86%; HR: 6.22, 95% CI: 1.49 to 26.07, p = 0.012).

Despite limitations of power in the TO groups, significant reductions in 4-year MACCE were evident in patients with TOs who achieved CR (vs. ICR) treated with either PCI or CABG (Figure 17_gr6).

(Table 5) lists independent predictors of ICR in the TO groups of PCI- and CABG-treated patients. Within the cohort of TO treated PCI patients, heavy calcification and long lesions were additional independent predictors of ICR compared with the main study findings (Table 2).

In this post hoc study of the All-Comers SYNTAX Trial, there were several notable findings. First, within the PCI and CABG arms, ICR (compared with CR) seemed to be a surrogate marker of a greater burden of anatomical coronary complexity and clinical comorbidity and was associated with a significant increase in 4-year mortality, all-cause revascularization, MACCE, and stent thrombosis (PCI arm). Second, the presence of a TO was less likely to result in CR in the PCI and CABG arms, with this effect being substantially more pronounced in the PCI arm, where the presence of a TO was shown to be the strongest independent predictor of ICR. Third, ICR (compared with CR) was associated with poorer clinical outcomes (4-year mortality and MACCE) across the TO and non-TO groups.

The low rate of CR in the PCI arm (34.3%) compared with the CABG arm (64.8%) was predominantly related to the lack of procedural success of TO PCI, as evident by a 49.4% recanalization rate in the randomized SYNTAX population (compared with procedural success rate of 68.1% in the CABG arm) on an intention-to-treat basis (Figure 17_gr4). Furthermore, the clinical impact of failure to recanalize TOs by PCI would potentially have been greater, because two-thirds of TOs were located in the proximal-mid coronary major epicardial vessels (Figure 17_gr3). Comparatively, other anatomical factors associated with ICR in the PCI and CABG arms might be of lesser importance, because the amount of jeopardized myocardium would have been lower compared with a TO—for example, a diseased side branch of a coronary bifurcation. Previous studies have shown that myocardial territory with >10% reversible ischemia derives the greatest clinical benefit from revascularization ((31),(32),33).

In the CABG arm, CR was achieved less frequently in patients with TOs compared with patients without TOs (non-TO: 69.8%, TO: 64.8%, p = 0.048). Notably, a similar phenomenon was reported in the ARTS (Arterial Revascularization Therapies Study) (34). A potential explanation for this finding might be related to the presence of a TO being associated with more diffuse or small vessel disease (non-TO: 20.9%, TO: 33.0%, p < 0.001) and consequently hindering the anastomosis of a bypass graft distally (Figure 17_gr4).

One of the reasons the long-term prognostic implications of completeness of revascularization has historically remained controversial has been a lack of consistency in the definition of ICR ((1),(4),(5),(6),(10),(),). For example, variations in the size of coronary vessels included and diameter stenosis on the basis of coronary angiography have made results difficult to compare. Furthermore, inconsistencies in defining the clinical endpoints, quality of clinical endpoint monitoring and adjudication, and selection biases in observational registries have compounded this issue ((1),5).

The strength of the present study was that the angiographic-based definition of CR/ICR was prospectively and rigorously systematically applied across all PCI- and CABG-treated patients in an All-Comers setting. Although more anatomically complex patients were enrolled into the nested CABG registry and more with clinical comorbidity were enrolled in the nested PCI registry ((27),28), this is representative of contemporary clinical practice. Within the present study, the decision to use a 1.5-mm vessel as the minimum size of vessel that could be revascularized was devised during the design of the SYNTAX Trial to reflect the ability of the cardiac surgeon to anastomose bypass grafts to vessels of this size. To allow appropriate randomization of the patient, the interventional cardiologist was required to match this practice.

In patients treated with PCI or CABG, ICR seemed to be a surrogate marker of a greater burden of anatomical coronary complexity and clinical comorbidity and was associated with a poorer long-term prognosis in PCI- and CABG-treated patients. These findings are not unexpected, given the historical association of more extensive baseline preoperative coronary disease with reduced long-term survival in the CASS (Coronary Artery Surgery Study) (35) and Rotterdam (36) registries. Furthermore, within the BARI (Bypass Angioplasty Revascularization Investigation) trial (37)—consisting of patients treated with PCI or CABG who underwent entry and 5-year coronary angiographic follow-up—native coronary disease progression (and not the extent of initial revascularization) was shown to be the predominant determinant of recurrence of angina and jeopardized myocardium at 5 years. Notably in this study, two-thirds of the increase in myocardial jeopardy was in previously untreated coronary vessels.

Moreover, the findings of the present study directly contradict that reported from the 3-year randomized SYNTAX trial (7), where no differences in clinical outcomes between CR and ICR were seen in the CABG population. Similar findings were shown in the 4-year randomized data (unpublished data). The disparity in results seems to reflect the fact that the nested CABG registry contains more anatomically complex patients compared with the randomized CABG cohort (SYNTAX Score: nested CABG registry 37.8 ± 13.3, randomized CABG 29.1 ± 11.4, p < 0.002) ((27),28). The ICR in the All-Comers CABG population therefore potentially resulted in more jeopardized myocardium, secondary to more anatomically complex coronary artery disease and greater clinical comorbidity (higher EuroSCORE and Parsonnet Scores) and their association with a likely increase in coronary atherosclerotic burden (4) and adverse long-term clinical outcomes as previously discussed. Within the PCI arm, potentially more jeopardized myocardium remained in patients with ICR compared with CABG, secondary to the failure to recanalize TOs or other (perhaps lesser important) anatomical factors, such as the presence of bifurcations, in addition to the similar factors associated with ICR in CABG patients.

Beyond the main study findings of increased stent thrombosis associated with ICR in the PCI arm (CR 3.7%, ICR 6.5%, p = 0.046), further analyses indicated that there was a substantial increase in stent thrombosis in patients with TOs who had ICR (CR: 0 episodes [0%], ICR: 7 episodes [4.6%]). Notably, patients with ICR had fewer stents (p = 0.009) and shorter total length of stents implanted (p = 0.002) (Table 1). Attempting to understand the mechanisms underlying this possible phenomenon are speculative. As to whether episodes of stent thrombosis are secondary to neoatherosclerosis (38) related to the more adverse clinical comorbidity in patients with ICR: stent under-deployment due to greater coronary calcification associated with ICR, failure to fully expand a TO after successful coronary wire passage (Figure 17_gr4), or the potential risk of delayed re-endothelialization—particularly if a subintimal technique of coronary wire passage is undertaken (39)—are all hypothetical mechanisms that require further study.

In the present study, the graft occlusion rate was substantially lower when CR was achieved in patients with TOs (and greater when CR was achieved in patients without TOs). These findings are consistent with previous studies, where early graft failure has been shown to be higher when anastomosed to functionally nonsignificant lesions or where competitive filling with the treated native vessel is seen ((40),(41),(42),43).

Recently the concept of “reasonable” or “appropriate” ICR has been highlighted ((1),(3),(4),5). This is either anatomically guided, on the basis of revascularizing most of the major epicardial vessels, or functionally/physiologically guided (on the basis of the burden of residual ischemia). The underlying principle is that, even if CR was technically possible and achievable (i.e., in all vessels >1.5 mm in diameter), this would not necessarily improve clinical outcomes—given the small territory a small side branch would, for example, supply. Future studies, possibly using a noninvasive functional approach ((44),45), might help best distinguish what acceptable level of revascularization is appropriate that would have a significant impact on long-term clinical outcomes. The newly devised Residual SYNTAX (46) and CABG SYNTAX (47) Scores each allows for more objective assessment of the degree of revascularization and residual anatomical disease complexity post-PCI or -CABG. Use of these methodologies might in the future aid in determining a “reasonable” threshold of revascularization.

The present study represents a post hoc analysis of the SYNTAX trial, and the results should be considered as hypothesis-generating. There was limited statistical power in the TO-treated patients and for the comparison of low-frequency events, such as stroke, graft occlusion, and stent thrombosis. The age of TOs could not be clearly defined, with most cases being unknown. Further procedural details of recanalization of TOs are not available beyond what is reported in the study. Core laboratory assessment of CR/ICR was not undertaken, because no post-operative (CABG) coronary angiograms were mandated in the SYNTAX trial. The possibility of other unmeasured adverse clinical characteristics to have confounded the data toward ICR cannot be excluded, although the intention of the study was to achieve CR in all study patients. The “All-Comers” concept of the SYNTAX trial, although more representative of contemporary clinical practice compared with the randomized approach, has been reported to potentially not result in the inclusion of consecutive patients, predominantly due to the inability to gain appropriate informed consent and refusal to participate ((48),49). Although the SYNTAX trial was based on contemporary revascularization practice at the time, improvements in technology in PCI and CABG might yield differences in clinical outcomes in future trials.

Within the PCI and CABG arms of the All-Comers SYNTAX Trial, angiographically determined ICR seemed to be a surrogate marker of a greater burden of anatomical coronary complexity and clinical comorbidity and had a detrimental impact on long-term clinical outcomes, including mortality. This effect remained consistent in patients with and without TOs.

The impact of ICR on clinical outcomes is dependent on the complexity of coronary disease and risk profile of the study population, on the basis of the observations associating ICR with: 1) no effect on long-term clinical outcomes (randomized CABG population); and 2) more adverse long-term clinical outcomes (All-Comers CABG population).