First Among Equals
Professor Roger Greenhalgh, Professor Janet Powell, and Louise Brown address outside interpretations of the design and conclusions of the EVAR-2 trial.
Endovascular Today: The EVAR-2 trial concluded that patients who were unfit for open surgical repair and randomized to endovascular aneurysm repair had a considerable 30-day operative mortality and did not show improved survival over the group of patients treated with medical management alone. Have the recent questioning and subsequent review of that data demonstrated that the 30-day mortality of the EVAR group was overstated, or do you stand by your original conclusions?
EVAR-2 Trial Participants: We stand by our original conclusions. The EVAR-2 trial is not really seen in the US for exactly what it is. There has been no equivalent or similar trial performed in the US, and it is not designed like most of the so-called "trials" that have been performed in the US. The US is very orientated toward the need to provide PMA data for the FDA and in these studies, the start of follow-up is at the time of procedure. EVAR-1 and EVAR-2 are intention-to-treat trials and analyses that commence from the time of randomization and analyze according to randomized group. It is also possible to perform so-called "per protocol" analysis, but these can be controversial.
The question asked by EVAR-2 has never been asked before or since, and it has not been copied by anyone in the US or the rest of Europe. We would even go so far as to say that no one else has had the courage to do what we did in EVAR-2. We had a group of clinicians who, together with their anesthetists, decided based on various clinical criteria that the patients were unfit for open repair. These unfit patients were randomized either to surveillance and medical treatment or endovascular aneurysm repair with medical treatment. So EVAR-2 was designed to answer the question of whether endovascular aneurysm repair has anything to offer in patients whom the clinicians considered as being unfit for open repair.
Endovascular Today: Have advances with respect to device development and procedural techniques improved to ameliorate some of this risk?
EVAR Trial Participants: No.
Endovascular Today: Is it correct to say that half of the mortality reported in patients randomized to endovascular repair occurred in patients who were waiting to be treated?
EVAR-2 Trial Participants: No, this is not correct, there were 74 deaths in the endovascular repair group, and 14 of these occurred before any AAA repair was performed.
Endovascular Today: Why was there such an extensive delay in treating the patients in the endovascular aneurysm repair arm of the trial?
EVAR-2 Trial Participants: Clinicians were asked to perform AAA repair within 30 days of randomization, but clearly this was not enforceable. In the UK, there are waiting list issues, but these do not fully explain the median delay of 57 days between randomization and AAA repair. In some cases, these very unfit patients needed further optimization before the anesthetists would put them under a general anesthetic (something that was quite common during the early parts of this trial), and in other cases the patients were often frail and unwell and lapsed into periods where they were not ready even for an endovascular procedure. The short delay reflects the difficulties that clinicians experience when managing patients who are so unfit with other comorbidities. The EVAR-2 publication was a documentation of what actually happened during the period since 1999, and most of the problems with interpretation come from a lack of recognition that the starting point is the moment of randomization rather than the moment of intervention.
Endovascular Today: After further review of the data, do you still contend that endovascular repair in high-risk patients offers no improvement in survival rates over surveillance and/or medical management alone?
EVAR-2 Trial Participants: If you look at the definition for which EVAR-2 was performed, you will see it was performed in the situation where clinicians found the patient unfit to have open repair. There is nothing about "high risk" in that definition. We have used fitness scoring to indicate that the so-called "high-risk" patients as defined in the US actually reside within the population of the EVAR-1 trial. It was shown at the SVS meeting that the Kaplan-Meier survival curves for the US "high-risk" series were all similar to the population of EVAR-1, and the survival of EVAR-2 was inordinately worse than all of the other series (eg, the Cleveland Clinic single-center study and the Lifeline Registry). Therefore, the issue addressed by EVAR-2 is not endovascular repair in high-risk patients, but whether endovascular repair was beneficial in patients found by clinicians to be unfit for open repair.
Subsequently, we have shown that endovascular aneurysm repair is superior to open repair in the fittest patients. In the least fit patients (ie, the EVAR-2 cohort), there is no superiority demonstrated by treating these patients with endovascular repair. That is what is new about the data. The patients of EVAR-2 are so unfit (and much less fit than anyone reported in the US studies) that actually the clinicians got it right-they should not have had open repair and were correctly assigned into the EVAR-2 trial.
Endovascular Today: What should be done with the patients who are very unfit?
EVAR-2 Trial Participants: What we concluded from EVAR-2 was to concentrate on getting patients more fit before considering intervention for their aneurysms. In some patients you will be successful because they get further away from the time they had their coronary, or their chests improved, or their renal function improved. If you can alter the patient's level of fitness, you then get a revised population to intervene upon, and a different result. In fact, in EVAR-2, there is some evidence to suggest that the patients who crossed over and had an AAA repair were moderately fitter at baseline than the rest of the cohort, and this meant that it was possible to improve their fitness prior to surgery. The mortality in this group of previously unfit patients was equivalent to EVAR-1 levels.
That explanation fits together, and it also allows us to agree with the concept that endovascular repair may be useful in patients with high risk because the so-called high-risk patients in the US exist within the EVAR-1 population, where EVAR is recommended.
Endovascular Today: Are the patients in EVAR-2 much sicker than their counterparts in US trials?
EVAR-2 Trial Participants: In every population, both in the US and the UK, there are some patients who are so unfit that the surgeons do not consider them for intervention. There are some who are so bad that they might have a coronary within the next 3 weeks, or a myocardial infarction, or unstable angina. They may also have other terminal illness, dementia, or social care issues that mean they are in such a state that the surgeon would not perform open surgery upon them. We are sure that these patients exist in the US, but they are not included in the US trials because the entry point into these trials is AAA repair.
The patients were very unfit by definition. The same level of sickness exists in patients in the US, but no equivalently unfit series have been defined, attempted, or reported in the US. The series that have been reported in the US have been reported as "high risk," as was the case with the Cleveland Clinic's single-center experience. Our Kaplan-Meier survival curves over the years compared the EVAR-1 patients into three categories: Good fitness, medium fitness and poor fitness. The Cleveland Clinic and the Lifeline survival curves are very similar. The expectation of death in the EVAR-2 patients is far greater than any patient series discussed in the US.
Endovascular Today: After the publication of the EVAR-2 data, the SVS Outcomes Committee performed its own review of the patients from the US IDE trials who met the qualifying criteria in EVAR-2. What is your view of those efforts to identify patients similar to the EVAR-2 trial from the Lifeline Registry?
EVAR-2 Trial Participants: The US IDE high-risk series, which includes patients that are considered fit for open repair, is not even remotely equivalent to EVAR-2. It is no surprise that they would try to do that because there was automatic surprise at the findings of EVAR-2. It would be sensible, however, to note the reaction of Juan Parodi, MD, who introduced endovascular aneurysm repair. He was not at all surprised with the findings of EVAR-2. He said very early on that he performed endovascular repair in very sick patients because he thought that they were the patients who would benefit most, but he had come to realize that the importance of assessing and improving fitness before intervention was a major issue. We think the SVS Outcomes Committee's US IDE study, which is basically a trawl of the Lifeline registry, is an attempt to define high-risk patients and to have a justification for using endovascular repair for those patients. We do not disagree with that thesis, but we would personally not have trawled the Lifeline registry in that way looking for a mandate for surgery. We would have interpreted EVAR-1 and EVAR-2 differently in the first place and recognized that within EVAR-1, the high-risk patients exist.
Endovascular Today: What is your view regarding endovascular aneurysm repair based on the data they presented?
EVAR-2 Trial Participants: We have no comment to make about endovascular aneurysm repair based upon data that are not our own. Whatever others want to trawl up, they can. A national registry has a lot of credit to it, but the weakness of all registries is that not all of the possible results, particularly not all of the bad results, are submitted. Therefore, the registry is as good as the data received, but it might not be all of the data. In a randomized, controlled trial, there is a report on all of the data and a so-called CONSORT diagram is used to show what happened to every single patient. In a registry, there is a tendency for patients and serious events to not be reported.
Endovascular Today: What is your view on single-center series data that demonstrate markedly better survival than depicted in EVAR-2?
EVAR Trial Participants: A single-center study with a completely different cohort of patients is not to be compared with EVAR-2. EVAR-2 is unique, and there are no equivalent multicenter trials anywhere in the world. No one has had the courage to undertake anything like EVAR-2, in either multicenter or single-center analysis.
Endovascular Today: Is it appropriate for one country to base its reimbursement decision on the results of a study conducted in another country?
EVAR-2 Trial Participants: We think it all helps. We would be remiss to not take note of trials and data that occur in the US, and it would be unwise for the medical community in the US to do the same about UK trials and data.
Professor Roger Greenhalgh is Head of the Department of Vascular Surgery, Imperial College, London, England. He may be reached at firstname.lastname@example.org.