Study Finds All-Comers Design Not So All Inclusive

Download this article's Factoid (PDF & PPT for Gold Subscribers)

Even an all-comers trial design does not result in the inclusion of all consecutive patients, according to a single-center experience published online May 12, 2011, ahead of print in the European Heart Journal. Not only that, but the patients who do enroll differ significantly from nonparticipants in terms of both baseline characteristics and subsequent mortality.

Eric Boersma, PhD, and colleagues from Erasmus Medical Center (Rotterdam, The Netherlands), retrospectively examined their center’s recent experience in 2 large all-comers trials comparing DES: LEADERS and RESOLUTE-III. The LEADERS trial enrolled 1,707 patients in 10 European hospitals between November 2006 and May 2007, while RESOLUTE-III enrolled 2,292 patients at 19 European hospitals between May and October 2008.

Both studies were prospective, randomized, controlled, noninferiority trials whose only requirement for inclusion was routine PCI for any indication. The few exclusion criteria were participation in another clinical trial; pregnancy; known intolerance for aspirin, clopidogrel, ticlopidine, heparin, or stent components; surgery planned within 6 months; large diameter of the study vessel; or lack of informed consent.

Nonparticipation Tied to Different Baseline Factors, Outcomes

During the enrollment period of the 2 trials, Erasmus Medical Center treated 1,242 consecutive PCI patients, of whom 579 (48%) actually participated in the studies. The main reasons for nonparticipation were:

• Inability to provide informed consent (33.5%), likely due to poor physical condition
• Refusal to participate (19%) for reasons including home situation, severe comorbidity, or study requirements
• Patient met 1 of the other exclusion criteria (26.9%)

Baseline characteristics differed between participating and nonparticipating patients. Those who enrolled were more likely to have stable angina (42.5% vs. 34.4%), hypertension (52.8% vs. 49.1%), and hypercholesterolemia (56.3% vs. 49.1%) but less likely to present with AMI (31.4% vs. 42.4%) or heart failure (2.1% vs. 4.4%).

Both 30-day and 1-year mortality were lower among all-comers trial participants compared with those who did not enroll. However, the discrepancy in 1-year mortality disappeared when the endpoint was evaluated only in those who survived the first 48-hours (table 1).

Table 1. Mortality Risk

“These results reveal that the all-comers design does not fully represent daily ‘real world’ clinical practice,” the investigators conclude. “Furthermore, our data indicate that an all-comers design is suitable for studying long-term treatment effects, but probably less so for studying acute treatment effects in the broad spectrum of patients undergoing PCI.”

Admittedly, the current findings raise the question of “whether the nomenclature ‘all-comers’ is a misnomer,” the researchers say, but they stress that an all-comers design still includes more patients than is typical with classical randomized controlled trials, which have been reported to exclude approximately 84% to 98% of eligible patients.

In an e-mail communication with TCTMD, study coauthor Sanneke P.M. de Boer, MD, maintained that the all-comers design, despite its limitations, is still the best way to improve generalizability of randomized controlled trial results.

“It leads to the inclusion of less selected patients when compared with classical randomized controlled trials [and] results in a more representative sample of the target population,” he said, adding that the use of proxy consent might help enroll more patients whose acute conditions would otherwise discourage participation. Of note, the medical ethics committee of Erasmus Medical Center prohibited proxy consent.

Dr. de Boer asserted that concerns related to an all-comers trial are not inherent to the design but instead hinge on how it is implemented in ‘real-world’ clinical practice. Even when working with broad study criteria, the paper explains, “[p]hysicians tend to avoid including patients with a perceived high risk of adverse effects; high-risk patients themselves are less likely to volunteer to participate in a clinical study; and finally regulatory authorities may restrict the inclusion of higher risk subjects.”

Good But Room for Improvement

In an e-mail communication with TCTMD, Gregg W. Stone, MD, of Columbia University Medical Center (New York, NY), said he agrees that “the ‘all-comers design’ is a step forward, . . .  but it is by no means perfect, still excluding many high-risk patients.” Therefore, he added, although longer-term outcomes may be comparable between participants and nonparticipants, short-term results from these trials may not be generalizable to high-risk patients.

“However, there is no a priori reason to believe that the results of these trials are not applicable to higher-risk patients who don’t participate—it is just impossible to prove that they are,” Dr. Stone emphasized.

More participation can be encouraged “by allowing health care proxies to be involved in the [informed consent] process, and by fully staffing research departments on a 24/7 basis,” he suggested. “It is also likely that the rate of enrollment varies from center to center depending on institutional commitment.”

Results Speak to Larger Issues in Trial Design

Nor is every all-comers trial created equal, David E. Kandzari, MD, of Piedmont Heart Institute (Atlanta, GA), told TCTMD in a telephone interview.

For instance, RESOLUTE-III, he reported, features detailed independent adjudication of endpoints, complete monitoring of all patients, and rigid Academic Research Consortium definitions. The study is also multicenter, company sponsored, and involves a high level of oversight by regulatory bodies. On the other hand, the SORT-OUT research group, which also performs all-comers trials, works “in a system of 5 hospitals in Denmark. They don’t measure periprocedural cardiac enzymes, because they don’t believe in them. They don’t contact the patient individually for follow-up [but] just query a database from a national health care system.”

Despite their limitations and variability, Dr. Kandzari said, “all-comers trials are clearly a better step forward compared with the more historical vanilla populations studied in randomized controlled trials. And they’re still probably our best opportunity for direct head-to-head comparisons.” Registries, meanwhile, are better suited for evaluating how a single stent performs in real-world practice, he noted.

All-comers trials also help address another challenge in designing clinical trials that evaluate DES: getting appropriate statistical power when event rates are now typically so low. “The bottom line is, to do [a randomized noninferiority trial with target lesion failure] event rates of 3% or less in a so-called simple patient population, you now need on the order of 8,000 to 9,000 patients,” he explained. “We’re now almost mandated to do these all-comers trials” because they capture a higher event rate.

Dr. Kandzari saw no easy way to encourage better enrollment when using an all-comers approach. “[Investigators] are doing the best that they can while still maintaining the integrity of the study and the design, and the intent to make the most accurate comparison between 2 therapies,” he said, adding that while studies may differ in terms of endpoints and follow-up, the upfront process of approaching and gaining consent from patients in all-comers trials “is probably as good as it gets.”

 

---

Source:
de Boer SPM, Lenzen MJ, Oemrawsingh RM, et al. Evaluating the ‘all-comers’ design: A comparison of participants in two ‘all-comers’ PCI trials with non-participants. Eur Heart J. 2011;Epub ahead of print.

 

 

Related Stories:

• Paper Highlights Pitfalls in Interpreting Randomized Clinical Trials
• Real-World Routine Primary PCI Equals Clinical Trial Experience