Prognostic impact of percutaneous coronary intervention in stable coronary disease

This editorial refers to ‘Clinical outcomes in patients with stable coronary artery disease with vs. without a history of myocardial revascularization’, by Y. Elbez et al ., on page 23. Evidence from recent trials, embedded into the latest clinical guidelines, 1 has led to fierce debates about the role of revascularization on prognosis in coronary artery disease. The COURAGE trial 2 seems to have made conservative cardiologists more conservative, whereas those who believe that revascularization has an impact on prognosis probably have not changed their practice. We seem to have split into two camps. On the one hand, there are those who believe that optimal medical therapy is the key to modifying prognosis via its impact on plaque stabilization, and that ischaemia per se is not so important. On the other hand, there are those who cite the studies showing the adverse effects of ischaemia and the better prognosis of those who have been through revascularization procedures. 3 The latter camp does not believe we can yet throw out the historical evidence base that has guided decision-making in patients with coronary artery disease over several decades. They are supported by the current guidelines, which still advocate risk stratification and evaluation of ischaemia. Ischaemia appears to be associated with a worse prognosis, even in the era of optimal medical therapy. 4 , 5 Leaving aside the debate on the relative merits of coronary artery bypass surgery (CABG) and percutaneous coronary intervention (PCI), it is remarkable that some decades into the history of revascularization, we still cannot clearly answer the question whether abolishing or reducing ischaemia is a good thing. Only randomized trials will give us clear answers of the true impact of a treatment and yet it is difficult to do some of the trials we might want to do or ought to do. The effect of inclusion and exclusion criteria in most trials means that an answer is provided for the randomized cohort, but then we have concerns about the generalizability of these results because so many patients get excluded. The consort diagrams of trials must be studied to try and analyse this, but even these may tell only part of the story as not every patient considered or approached for a trial gets recorded in the study databases. Clinicians have their own biases and may choose to either enter or not enter a patient into a study, because they have their own views of whether there is ‘equipoise’ or true uncertainty and whether they are prepared to randomize. It is always very useful to be able to look at trial registry data alongside the results from the randomized cohort to help us get a perspective. But because registry data cannot remove these intrinsic biases, it is sometimes as hard to interpret their results as the results from the more restrictive randomized cohorts. This is also true for ‘real-world’ registries hoping to accumulate data on an ‘all-comer’ population outside of clinical trials. Elbez et al.6 present a fascinating analysis from the Reduction in Atherothrombosis for Continued Health (REACH) registry. This is a large multinational registry that collected data between December 2003 and June 2004 from patients who presented in the outpatient setting. By definition, these patients were stable, although we do not know exactly why they had been referred to the outpatient clinics. Formal consent was required for participation, which would have excluded some patients. Even so, the investigators collected data on over 68 000 patients. In this study, only those patients with a previous history of coronary artery disease were analysed, but this represented over 30 000 patients. The results of the raw data are perhaps of most interest. They demonstrate the current natural history of those with and those without a previous history of coronary revascularization, and find that those with a history of revascularization (especially PCI) fare better than those with no history of revascularization, especially if the revascularization was within the last year. These data are useful as they allow us to provide broad-brush reassurance to those who have undergone revascularization, but perhaps the poorer prognosis of those who have not undergone revascularization is a cause for reflection. There were similar findings after propensity score analysis, but this does not mean that we can conclude that PCI offers a prognostic benefit. As the authors discuss, selection bias is an issue in this sort of study. Patients who have not previously undergone revascularization represent a mixed bag. Some will not yet have been investigated appropriately. Some will have undergone angiography and been found to have minor disease. Some will have extensive disease which is not amenable to either PCI or CABG. Some will have a previous myocardial infarction and now present with heart failure or atypical chest pain, and so on. The REACH registry does not allow us to dissect out these different cohorts, but it does raise the issue of how we should manage the unrevascularized patient. Should we still be measuring for ischaemia and looking at the coronary anatomy to help us guide treatment or does early angiography risk decisions made by the oculo-stenotic reflex? Although viewed disparagingly by some, could the oculo-stenotic reflex actually be a good thing in some cases if it leads to a long-term reduction in the ischaemic burden? Or should we offer optimal medical therapy first and then only offer angiography to those with continuing symptom, regardless of extent of ischaemia? Or should we always look for demonstrable ischaemia (whether invasively or non-invasively)? If there is ischaemia, should the extent of ischaemia play a role in the decision-making process for both angiography and revascularization? We seem to have gone from a time when we thought we knew what to do to a time when we have become uncertain. There is a lot of evidence to suggest that reduction of ischaemia seems to be of benefit, but critics argue that most evidence was collated before the era of optimal medical therapy. An additional argument is that CABG offers prognostic benefit because it bypasses vulnerable plaque, not because it reduces ischaemia, and that PCI to one or two lesions may reduce ischaemia but has little impact on plaque vulnerability. So, surgical revascularization has been shown to improve medium-term survival in some subsets of patients (in an era of minimal medical therapy 3 ), whereas more recent studies like Bypass Angioplasty Revascularization Investigation 2 Diabetes (BARI-2D) show some clinical benefits but no major impact on survival in the medium term, even in diabetics. 7 , 8 On the other hand, registry data suggest that revascularization with PCI improves prognosis in patients with chronic total occlusions, 9 , 10 although we need randomized trials to be certain. We have to remind ourselves though that patients randomized into most current trials fall into the lower end of the risk spectrum and so trying to demonstrate an impact on survival in these patients is harder. We can conclude that reducing ischaemia is not mandated, but we cannot conclude that reducing ischaemia never has a prognostic impact. Indeed, various meta-analyses suggest that PCI does have a prognostic impact in stable coronary artery disease, especially with the use of new generation drug-eluting stents. 11 , 12 We are going to have to await the results of the International Study of Comparative Health Effectiveness with Medical and Invasive Approaches (ISCHEMIA) trial to see if we can get a better handle on this, remembering that the most ischaemic patients will not be randomized. The aim in ISCHEMIA is to randomize only those patients with >10% of myocardium at risk. Other studies, such as Revascularisation for Ischaemic Ventricular Dysfunction (REVIVED-BCIS2), will also help clarify whether revascularization by PCI offers prognostic benefit in those with dysfunctional but viable ischaemic myocardium. But in both studies, the symptom level has to be such that the patients have to be content to stay on medical therapy if randomized to the conservative arm, and crossover should be kept as low as possible. We have to be careful to differentiate between truly stable coronary disease (with or without demonstrable ischaemia) and those with a syndrome which has not led to hospitalization or an adverse clinical event, but which suggests a transition from a stable situation to an unstable one. We were taught as medical students that new onset and rapidly progressive symptoms cannot be viewed as stable angina and the severity and type of symptoms were also important, e.g. nocturnal, decubitus, or post-prandial angina or angina associated with syncope. The studies that demonstrated an adverse prognosis in these scenarios led to the introduction of rapid access chest pain clinics, and no one has yet had the ‘courage’ to randomize this cohort of patients to a conservative vs. invasive strategy, even in the era of optimal medical therapy. Recent onset angina should not be considered to have the same prognosis as patients with stable coronary artery disease with longer histories, and patients with stable angina have varying levels of risk for death and future myocardial infarction depending on a multiplicity of clinical variables. 13 It is likely that revascularization has an important prognostic impact in these patients, but we may not be able to design a randomized trial to prove it. Comparison of outcomes with historical controls may be the best level of evidence we can achieve, although that itself may be difficult if new evidence regarding medical therapy emerges over time. This also raises the debate about what we mean by optimal medical therapy, and whether different patients get different levels of protection from each of the drug classes in current use: anti-platelet drugs, beta-blockers, lipid-lowering medication, ACE inhibitors, and angiotensin receptor blockers. There is uncertainty about whether all patients should be offered all classes of drugs, but until we have a clearer idea about which patients get no benefit, then we continue with a blanket policy according to the current guidelines. What is agreed for now, then, is that all patients, whether revascularized or not, should be considered for optimal medical therapy. Unless the evidence base changes, patients with unstable symptoms and the acute syndromes should be offered an invasive strategy unless there is a reason not to. For patients with stable coronary disease, guidelines recommend risk assessment and currently this includes assessment of the extent of ischaemia. Although different healthcare systems might exhibit a different threshold for the use of angiography, the use of revascularization should be guided by the current appropriate use criteria. 14 We will have to await further trial evidence to see if these will change. We do not have to be polarized and as we gain additional knowledge, we should be able to reach a growing consensus on the role of revascularization in individual patients. Observations from REACH and other registries add to our understanding, but continue to raise as many questions as answers.