The primary goal of the present study was to describe the late clinical outcomes of patients undergoing contemporary PCI based on a large, prospective registry. We demonstrated that contemporary PCI in a cohort of unselected patients referred to a single centre was associated with excellent long-term outcomes. Survival at seven years was high, and long-term MACE rates were low and consistent with outcomes from RCTs, reflecting the safety of PCI in this high-risk population.
In contrast to randomized trials, the present study described very late clinical outcomes and their predictors in a large cohort of ‘all-comers’. At seven years, all-cause mortality was 10.6% and repeat revascularization (PCI or CABG surgery) was 17.2%. Our results are comparable with registry data of approximately 1500 unselected patients receiving sirolimus-eluting stents (9
) and a pooled analysis from a four-year follow-up of patients enrolled in the SIRolImUS-coated Bx Velocity balloon-expandable stent in the treatment of patients with de novo coronary artery lesions (SIRIUS) trials (21
). Total cumulative CVEs including admission for acute MI, heart failure or stroke, in addition to repeat revascularization, were high (31.4%) but reflected the relative complexity of the cohort.
The proportion of diabetic patients, patients older than 75 years of age, and those with renal impairment, multivessel disease and LMS involvement, as well as the large burden of high-risk procedural characteristics (multivessel PCI, residual stenosis and incomplete revascularization) found in the study population highlight the complex nature of disease presenting in everyday practice. Additionally, more than one-half of the procedures (53%) were performed urgently, with 1.3% of patients presenting in cardiogenic shock. The present patient population is not generally well represented in RCTs, although evidence for the role of PCI, particularly with DES, in specific high-risk subsets can be inferred from various subgroup analyses (22
) and smaller targeted trials (23
Our data suggest that late unadjusted cumulative survival was significantly worse in women and diabetic individuals, and was directly proportional to the degree of LV dysfunction. This observation is consistent with one-year data reported previously (9
) from DES registries. Consistent with other studies (9
), the presence of diabetes was associated with higher rates of revascularization procedures. When late CVEs were assessed, the presence of cardiogenic shock and severe LV dysfunction were again most significantly predictive of adverse events, with clinical markers of extensive disease such as diabetes, peripheral vascular disease and LMS involvement also independently associated with a higher event rate. This may suggest that late outcome in these patients is related as much to the state of underlying atherosclerosis as to the procedural success of a particular intervention.
In the present analysis, we found that a suboptimal procedural result with incomplete revascularization or residual stenosis of greater than 20% was associated with significantly worse long-term survival. Although pre-existing clinical factors (cardiogenic shock, severe LV dysfunction, peripheral vascular disease, diabetes, renal dysfunction, increasing age and requirement for emergent PCI) independently predicted late mortality, such procedural characteristics were also independently predictive of long-term outcomes. Some of these procedural characteristics (multivessel PCI, undelivered stents and residual stenosis) may reflect the burden of coronary disease in these patients, rather than merely problems with technical aspects of the procedure. In addition, the strongest independent predictors of the need for late revascularization were procedural rather than clinical in nature. Residual stenosis of greater than 20% in the treated artery was most significantly associated with repeat PCI or CABG surgery, although multivessel PCI and a history of PCI (due to atherosclerotic burden and/or restenosis) were also strongly predictive of the need for further procedures.
Coronary stenting occurred in 96% of all procedures, and DES were deployed in 23% of cases. When comparing patients from mid-2003 onward – the time at which DES became available at our institution – unadjusted cumulative survival at four years was 91±1% in DES-treated patients versus 89±1% in patients who received BMS. The results of our study suggest that the use of DES was associated with significantly improved late outcomes, demonstrating increased survival, decreased CVEs and reduced repeat revascularization events. While the superiority of DES in reducing target vessel revascularization is well established, the observation that DES are associated with improved survival has not been substantiated in randomized trials that eliminate selection bias. However, recently published registries have reported similar data in Ontario (12
) and Massachusetts (13
) – populations with very different health care systems and practices. Tu et al (12
) demonstrated an absolute all-cause mortality reduction at three years between patients treated with BMS (7.8%) and those who received DES (5.5%), while Mauri et al (13
) reported a mortality benefit at two years. Evidently, any absolute mortality benefit attributable to DES would need to be confirmed by a randomized study with long-term follow-up. It is hypothesized that this mortality benefit may, in part, be explained by a reduction in the need for repeat procedures (with a small subsequent reduction in procedure-related mortality), and a possible benefit associated with prolonged dual antiplatelet therapy in patients receiving DES (13
Another finding of significant interest was that the use of GPIs during PCI was associated with reduced late cardiac events. GPIs have been demonstrated to improve early outcomes in both elective (25
) and high-risk (26
) PCI patients, and there is evidence for their efficacy up to three years (28
). This therapy is used more often at our institution in higher risk patients, and in those undergoing urgent or emergent procedures. Therefore, although a selection bias exists in the use of GPIs, it might be expected to be associated with worse clinical outcomes. The mechanism by which transient and early use of a GPI leads to reduced long-term MACE is unclear, but may relate to long-term vessel healing and stability as well as reducing the early post-PCI events associated with reduced long-term mortality (27
The present study has several important limitations. It describes a single-centre experience and lacks the obvious advantages of a multicentre, multinational randomized study. It is unlikely, however, that a randomized study will be conducted in the context in which the present study was performed, with virtually no exclusion criteria. Despite reporting outcomes from a large prospective registry, our study also suffers from the limitations of retrospective, observational data. Specifically, we cannot account for all possible confounding factors and, because follow-up data were collected from linkage to a provincial administrative database, we cannot report on certain types of data – such as causes of death, types of clinical presentation for subsequent revascularization (ACS, non-ST elevation myocardial infarction, elective, etc) – or specify target-vessel revascularization. In addition, the cohort did not include a high number of primary PCI procedures because this procedure was not introduced as the standard of care in our centre until 2008. All procedures, including stent and pharmacotherapy choice, were at the discretion of the operator and, therefore, open to bias. As a result, the present Toronto-based registry may not be widely applicable to other health care systems or regions where treatment decisions differ.