Stents or Surgery

In Meredith Willson's “The Music Man,” sleepy River City, Iowa, comes under musical assault from a smooth and self-assured confidence man, Professor Harold Hill. He declares that restless youth (who are not so restless) are in moral peril but may be saved through an effortless acquisition of musical skill requiring only the purchase of uniforms and instruments, and the use of the “think” method. Of course, the think method is a sham promulgated to convince these offbeat Iowans to enrich the good professor and gain little in return. Fortunately, the sham and its purveyor are very entertaining.

You may ask how this pertains to revascularization therapy. The common ground lies in the glee with which our surgical colleagues point to the application and failures of balloon angioplasty, which offers a comparison in which cardiologists are Professor Hill and percutaneous transluminal coronary angioplasty (PTCA) is the “think” method. In this scenario, the cardiologist identifies every coronary lesion as a source of peril. Salvation is offered, with less musical distraction but equal reward, in the form of morbidity-free angioplasty, and patients are diverted from a more effective means of revascularization.

There is a kernel of truth in this argument. Not every lesion requires revascularization. Angioplasty is not without risk and frequently fails within the first 6 months after the procedure. In populations whose survival is unlikely to be affected by any revascularization procedure, PTCA compares favorably to coronary bypass surgery, albeit with a substantially greater need for more procedures. Worse still, in the high-risk diabetic subpopulation with multivessel coronary stenosis, a 1st choice of PTCA carries a heavy toll in both the need for repeat procedures and, more importantly, survival. However, I hope to convince you that with the recent advances in stent-assisted angioplasty (recently acquiring the abbreviation PCI, for percutaneous coronary intervention) proper application now offers a very good alternative to surgery. But, as Professor Hill was admonished as he disembarked in River City to begin his confidence game, “you've gotta know the territory.”

The introduction of coronary artery bypass grafting (CABG) was an enormous advance in the treatment of coronary artery disease. Successful bypass reduced angina and improved exercise capacity. When studied in randomized trials of early surgical referral (Fig. 1),^1–4 CABG proved effective in reducing mortality. However, 3 central principles of that survival impact quickly became apparent: 1) the amount of potential survival benefit was roughly proportional to the patient's risk of death from the next coronary event, whether estimated using clinical, angiographic, or physiological variables; 2) revascularization was not a treatment for atherosclerosis, but a means of reducing the impact of later disease progression; and 3) benefit lasted only as long as the grafts remained patent.

Fig. 1 Survival curves from the VA Cooperative trial⁶ compare early referral for CABG with an initial strategy of medical therapy. Superimposed on the survival curves are the long-term patency of SVGs and IMA-to-LAD bypass grafts. Extending from the surgical survival curve is an estimate of what might have been, had the usefulness of the internal mammary artery been recognized before the design of the VA Cooperative trial.

CABG = coronary artery bypass grafting; IMA = internal mammary artery graft; LAD = left anterior descending coronary artery; MED = medical therapy alone; SVG = saphenous vein graft; VA = Veterans Administration

When we examine the outcome of the VA Cooperative trial,^4–6 these principles can be seen clearly (Fig. 1). Early on, procedural risk obscured the impact of successful revascularization; but by 5 years, improved survival in surgically treated patients was apparent. This apparent difference was statistically verified only in patients at high risk (3-vessel disease with low left ventricular ejection fraction [LVEF]), and it disappeared after 11 years.⁵ In fact, follow-up out to 22 years showed that low-risk patients were harmed by an early referral for surgical therapy.⁴ If we compare the survival curves from the VA study to a superimposed graph of saphenous vein graft (SVG) patency, it is apparent that the decline in benefit from early surgical referral follows the natural history of an SVG. If, instead, we compare the survival curves to a graph of internal mammary artery patency when the internal mammary artery (IMA) is used to bypass the left anterior descending coronary artery (LAD), we might imagine (in fact, we all believe it to be the case) that the outcome of this and other trials might have been very different (Fig. 1). In a small subgroup of patients from the Coronary Artery Surgery Study who received an IMA, this suspicion was borne out.⁷ The lessons of the randomized trials and the outcome differences when using the IMA as a bypass conduit are that the untreated natural history of a patient's disease determines the acceptable procedural risk and the possibility of obtaining a survival benefit from revascularization, and that benefit is afforded by an open vessel. The contention of advocates of PCI borrows a phrase from golf's 19th hole, “it's not how, it's how many”—and our way is safer.

Gary, Indiana

When questioned about his musical credentials, the erstwhile Professor Hill claimed matriculation from “the Gary Conservatory, class of '05.” In '05, the conservatory didn't exist, but most of the players were either ignorant of or oblivious to that fact. When engineering advances made PTCA an easier procedure, one that could be used in attempts to treat multivessel disease, the hope was to apply a safer, easier procedure to the task of repairing native vessels, thus obviating the need for CABG with its temporally limited SVGs. However, with a procedural mortality rate of 1% and a 6-month restenosis rate after PTCA that is conservatively estimated at 40%, we encountered a problem with one of the 3 principles of revascularization: benefit lasts only as long as the vessel is open. In short, most patients will need multiple procedures to maintain patency, thus being exposed to an aggregate procedural risk of death that is essentially the equivalent of a single bypass procedure, or their protection from angina and death will be inferior. Like the class of '05, the grounds upon which PTCA was held to be superior to surgery didn't exist. This was put to the test in the BARI^8,9 (Bypass Angioplasty Revascularization Investigation) trial, in which patients with multivessel disease amenable to either procedure were randomized to CABG or PTCA.

There are a couple of caveats about BARI. First, almost all of the patients in the trial had normal left ventricular systolic function. Therefore, with respect to survival, the gain from a revascularization procedure may have been difficult to detect had either procedure been compared with medical therapy. Second, the interventionalist's eyes may have been bigger than his balloons. In patients who had an average of 3.5 treatable lesions, only 1.9 were successfully treated in the PTCA arm of the trial. Meanwhile, 3.1 lesions were treated in the surgical arm. Fortunately, in nondiabetic patients, no apparent harm came of 1st referral for PTCA. In diabetics, the outcome was quite different. Patients randomized to CABG who got a left internal mammary artery (LIMA) graft to the LAD fared far better than their counterparts who were treated by means of PTCA.⁹

In a very interesting post-hoc analysis of both the trial and the accompanying registry, Detre and colleagues¹⁰ examined the impact of revascularization upon the outcome of the next myocardial infarction (MI). In diabetic patients who did not receive CABG either primarily or later when problems resurfaced, there was a 10-fold increase in mortality after any subsequent MI. Performance of coronary bypass surgery at any time after entry into the study allowed a patient to be considered among the CABG group. After 5 years, 64% of diabetic patients initially treated with PTCA underwent CABG. Therefore, it is possible that Darwinian bias, or surgery for the fittest, resulted in overestimation of the relative impact of CABG. However, it is very unlikely that bias is responsible for all of the rather dramatic difference between the groups. The observed difference between PTCA and CABG in the frequency of potentially significant coronary stenoses, implying less effective revascularization with PTCA, is a more likely explanation.¹¹

A registry of eligible patients who underwent the procedure of their own or their physicians' choice accompanied the BARI randomized trial. The overall survival of patients in the registry was slightly but not significantly better than among the randomized patients. Unlike patients in the randomized trial, registry diabetic patients enjoyed the same survival probabil-ity whether treated with surgery or PTCA (Table I).¹² However, patients referred for surgery had more LAD lesions, more type-C lesions, and just plain more lesions. More patients referred for surgery had 3-vessel disease and diffuse disease. In addition, some patients were treated medically.¹² The natural conclusion drawn from these observations is that the cardiologists cherry-picked their patients, saving the easy lesions for themselves. They did, but they did so to the patients' benefit. They chose patients whom they could treat well and did so, with all the weaknesses of PTCA, without adverse performance compared with surgery. The real conclusion is that PTCA can be used safely and well in comparison with bypass surgery. But because of its weaknesses, as well as the risk of failure of emergency surgical referral and the risk of restenosis within 6 months, choosing the right patient and lesion is of supreme importance.

TABLE I. Survival of Patients in the BARI Trial12

Marian

But … then along came stents. The good professor's undoing as a confidence man begins with his attraction to a spinster, Marian the librarian. His pursuit of her is not only the source of a whimsical, musical alliteration but also the beginning of attempts to truly fulfill his promises. Unfortunately for him, the wherewithal is lacking.

While not proved in a randomized trial, it is pretty clear that stents make a PCI procedure safer by affording the interventionalist the capacity to get out of trouble and avoid emergency surgical referral and its attendant risks.^13,14 Stents are also highly touted for their ability to reduce the risk of restenosis after angioplasty but, in fact, their effect is rather small. There are 2 reasons. Stents actually worsen intimal hyperplasia, which is the major determinant of restenosis. Using the loss in lumen diameter at 6 months after stent implantation as a measure of intimal hyperplasia, if even the newest stents are compared to PTCA, it is apparent that all stents fall in the same range, at a loss of about 0.8 mm —and that's worse than PTCA, at a loss of 0.32 mm (Fig. 2^{12–27, 15–30}). Second, with a lower likelihood of early procedural failure, the interventionalist can be a bit more daring and attempt to tackle longer, more complex lesions, thus nullifying any potential gain in terms of restenosis.

Fig. 2 The angiographic loss in luminal diameter after 6 or 9 months is a measure of the severity of intimal hyperplasia after stent placement. The late loss reported for various forms of bare-metal stent is shown with their average value (solid line) ± 2 standard deviations.^15–30

How then do stents compare to CABG? The same as PTCA, of course (Fig. 3).³¹ Several trials have been performed. Their data were recently compiled in a meta-analysis²⁸ that reported a virtually identical 3-year survival expectation after either procedure, with the exception, at least in the population studied in one of the trials, of diabetic patients, who did not do quite as well when treated with PCI.³² With the protection of a stent, the interventionalist now will not give up on tougher lesions and leave them untreated, as was done in BARI. But, by extending our reach, we further expose the Achilles' heel of PCI: restenosis, the need for repeat procedures, and less effective revascularization.

Fig. 3 Survival curves from a meta-analysis compare the outcomes after an initial revascularization strategy of PCI-stent vs coronary artery bypass grafting (CABG).³¹

HR = hazard ratio; PCI-percutaneous coronary intervention

76 Trombones

The thrilling musical conclusion of “The Music Man” follows the near exposure of Professor Hill's duplicity and an essentially magical transformation of the town's youth into a talented marching band. So too—even as the disappointments of stent-assisted PCI were looming—interventional cardiologists were given the drug-eluting stent (Fig. 4).16,19,23–28,33–36

Fig. 4 The late loss in luminal diameter after trials of various forms of drug-eluting stent, compared with bare-metal stent.

Using a polymer-based drug delivery system that enables high local drug concentration but avoids systemic toxicity, sirolimus- and paclitaxel-eluting stents have been used to reduce the growth of intimal hyperplasia. The impact is really rather striking. If we use late lumen loss after angioplasty as our measure of intimal hyperplasia, drug-eluting stents reduce the primary determinant of restenosis by 50% to 100% (Fig. 5).24–28,37,38 In large-scale clinical trials, this corresponds to an angiographic restenosis rate of 4% to 6% and a very low likelihood of a repeat procedure on the same lesion.^22,23 We can assume that a lasting, successful resolution of occlusive lesions will provide lasting relief from angina and protection from death in the event of a future MI. Building upon the results of prior trials comparing PCI to CABG, we can surmise that the comparison between drug-eluting stents and surgery will be much more favorable for stents, if the proper lesions are chosen.

Fig. 5 Populations and lesion characteristics differ between trials of drug-eluting stents. Therefore, the percent reduction (Control Late Loss – Experimental Late Loss/Control Late Loss) in late loss from several reported trials of drug-eluting stents is shown.24–28,37,38

Let's try to get a peek at the future, drawing from a recent article by Sawhney's group³⁹ (Table II^39–42). Three studies^40–42 have compared stent-PCI to the real gold standard procedure, LIMA-to-LAD bypass. By enrolling patients with disease limited to the LAD, we can get a look at the performance of the methods. Each of the studies has used slightly different endpoints to perform their comparisons. Diegeler and associates⁴⁰ used cardiac death, MI, and target-lesion revascularization. Drenth and coworkers⁴¹ used all-cause death, MI, and target-vessel revascularization, but added stroke. Goy and colleagues⁴² used cardiac death, MI, and any repeat revascularization. Surely there is no surprise that the stented population did not do quite as well as the LIMA population. Stented patients were roughly 2 to 3 times more likely to have a bad outcome—a result driven mostly by repeat revascularization. The drug-eluting stent trials have used some different definitions to look at outcomes after implantation of a bare-metal stent or a drug-eluting stent. One of the terms is target-vessel failure, which is defined as death, MI, or target-vessel revascularization. Because this definition is pretty comparable to the outcomes used for the stent versus LIMA trials, we can estimate what may come of a comparison between drug-eluting stents and LIMA bypass. In the SIRIUS trial of the sirolimus-eluting CYPHER™ stent,³⁹ a small subgroup of patients with isolated proximal LAD stenosis was treated. Their target-vessel failure rate of 10% compares very favorably to the reported event rates for LIMA bypass.³⁹

TABLE II. Trials of Stent Implantation for Isolated Proximal LAD Stenosis³⁹

Why should we pursue stents as the 1st option for revascularization? Coronary bypass surgery has been historically plagued by the invasive nature of the sternotomy and its attendant risk of infection, the need to manipulate the aorta (producing risk of stroke), the finite lifespan of an SVG, and the many problems associated with cardiopulmonary bypass. The wound after a stent procedure is minimal, the stroke risk small, and the promised success rate equal to or better than that of an SVG. Some may argue that surgical techniques are improving to minimize risks, and they are. Recently performed surgical trials suggest that new techniques, including limited sternotomy and warm-heart surgery, are a significant advance. Difficulties with the creation of the graft anastomosis and with graft failure must still be conquered, however. The oft-heralded weakness of the SVG has been partially overcome by vigorous risk-factor management and by the increasingly popular use of all-arterial-conduit bypass procedures. However, the free radial grafts have patency rates that are eerily similar to those of SVGs, and even the mammary, when not directed to the high flow region of the LAD, is less than perfect.^43–45 At the same time that surgical technique is evolving, PCI technique is evolving as well. Techniques of groin management are improving to reduce the risk of access complications after stent placement. Restenosis, although not completely conquered, will be drastically reduced by drug-eluting stents, and the risk of stent-procedural MI is falling with improved antithrombotic therapy and the introduction of distal protection devices—a risk that has not been confronted surgically. The main point of difference between the procedures, and perhaps the most important difference, is the nature of the procedure. After PCI, recovery time is a matter of days. And if the lasting success that is promised by the early results of drug-eluting stents can be delivered, there is no reason for patients to undergo the more intrusive surgical procedure.

However, in keeping with good debate technique, I will point out some of the weaknesses of PCI when drug-eluting stents are used. First and foremost, certain lesion characteristics carry a high risk of procedural complication and eventual failure. Most prominent among these is a bifurcation lesion with large daughter vessels. A bifurcation lesion increases the likelihood of procedural failure. If both daughter vessels receive a bare-metal stent, the risk of any adverse event, either immediately or within 6 months, is greater than 50%.⁴⁶ Drug-eluting stents reduce the risk of restenosis after stent placement in this situation, but not to the extraordinary extent reported for simpler lesions. Rather than a rate of 4% or 5%, the risk is 26% and the incidence of out-of-laboratory thrombosis of the lesion is increased by about 3-fold.⁴⁷

In truth, the major areas of weakness for bare-metal stents—diabetes, long lesions, small vessels, and final result—plague drug-eluting stents as well, although the impact is somewhat diminished. In a nondiabetic patient with a lesion 10- to 15-mm long in a vessel of 3-mm or greater diameter, the likelihood of target-vessel failure is very low, less than 5%. This failure rate compares very favorably to the expected failure rate of a bypass conduit and argues that treatment of multivessel disease can be performed with the expectation of excellent outcome in terms of both survival and repeat revascularization. But failure rates are higher in diabetic patients (12%), in long lesions (12%), and in small vessels (18%).^48–50 Therefore, if we approach a patient who has several characteristics suggestive of an early or late suboptimal result, we clearly cannot expect the excellent outcomes reported for lower-risk patients or lesions, and comparisons with CABG are likely to echo the outcomes of BARI and ARTS.^8,32

An unrelated and fortunately uncommon point of concern is the need for coronary revascularization before a patient undergoes a noncardiac surgical procedure that requires general anesthesia. The only patient truly in need of revascularization in this setting is one who has the anatomic and physiologic characteristics that indicate a survival benefit from CABG. Meanwhile, the application of stent-assisted angioplasty in this setting meets with difficulty when the surgical procedure demands the withdrawal of antiplatelet therapy. When patients who have received coronary stents undergo a surgical procedure, the risk for stent thrombosis, myocardial infarction, and death is substantially increased.^51,52 A minimum waiting time of 6 weeks has been recommended after placement of bare-metal stents before surgery.⁵² The safe waiting time after placement of a drug-eluting stent is unknown.

Conclusion

As the techniques for surgical and percutaneous revascularization have been refined and their long-term outcomes better understood, surgeons and interventional cardiologists have begun to treat different patient populations, and appropriately so. In a low- or intermediate-risk patient pool that requires revascularization primarily for the treatment of angina, a 1st choice of PCI—assuming the reasonable evaluation of lesions and the likelihood of success—results in a survival expectation that does not differ from that for a surgically treated patient. The introduction of drug-eluting stents will certainly reduce the problematic need for repeat procedures. Meanwhile, in a patient with a potential survival benefit from revascularization and a greater risk of treatment failure even when treated with drug-eluting stents, surgery is, and will remain the preferred treatment. Specifically, patients with long lesions, diffuse disease, bifurcation lesions, heavy calcification, insulin-treated diabetes mellitus, or another surgical lesion remain within the realm of surgery. However, patients with focal lesions, no bifurcations, and large vessels can and should be safely treated with PCI (Table III). Moreover, the population at high risk for surgical complications, a population that includes comorbidities such as severe chronic obstructive pulmonary disease or patients who have previously undergone sternotomy, may be more safely treated with PCI and with reasonable efficacy.

TABLE III. Some of the Factors Now Affecting Choice between Surgery and Stents

Clearly, technological advances, most notably the drug-eluting stents, have improved both the safety and the efficacy of PCI. But in regard to the question “Will stents replace surgery?” I would suggest a change. We should instead ask, “Are stents now a good alternative to surgery in the right circumstances?” The answer, at long last, is “Yes.”