September 4th, 2012

ESC Trials: The Best And The Worst

Two trials presented at the ESC this year — WOEST and IABP-SHOCK II — are great examples of the way medicine is supposed to work. Another trial, FAME 2, is an example of so many of the things that can go wrong.


WOEST and IABP-SHOCK II are remarkably similar. Both trials tested conventional wisdom and found it lacking. WOEST examined the routine use of aspirin in “triple therapy”, which is when people already taking an anticoagulant undergo PCI and then receive an additional antiplatelet drug and aspirin. IABP-SHOCK II tested the routine use of circulatory support with intra-aortic balloon counterpulsation (IABP) for patients in cardiogenic shock following MI for whom early revascularization is planned.

Despite scant evidence, both of the ideas tested in these trials had received class 1 recommendations in the guidelines and were widely used in clinical practice. And both trials provided near definitive proof that the conventional wisdom was completely and utterly wrong.

This is the way science is supposed to work: an idea gets put to a fair test. Judging from the initial response to these trials, it seems likely that the cardiology community will rapidly accept the findings, and guidelines and clinical practice likely will change in short order to reflect the new evidence base.


FAME 2 also addressed, or claimed to address, an important question. Although the evidence base for PCI in stable angina had always been weak or nonexistent, its popularity had undergone exponential growth for many years, until COURAGE famously put the brakes on this growth. When fractional flow reserve (FFR) first came along, it was viewed with considerable suspicion in the interventional community, since in many respects it helped confirm the findings of COURAGE by appearing to demonstrate that a significant percentage of lesions intervened upon were not ischemic and therefore almost certainly didn’t benefit the patients who had undergone PCI.

Eventually, however, the interventional cardiology community found a path to renewal with FFR. Perhaps, it reasoned, instead of being used to illustrate the lack of utility of PCI, FFR could be used to guide PCI decisions, limiting interventions to ischemic lesions that would benefit from PCI.

This is where FAME 2 comes into the picture. In the trial, patients who had at least one functionally significant lesion, as defined by FFR, were randomized to FFR-guided PCI plus medical therapy or medical therapy alone. The trial was stopped early, after only about half of the intended number of patients were enrolled, because of a significant reduction in the primary endpoint (the composite of death, MI, or urgent revascularization) in the PCI group compared to the medical- therapy-alone group.

The FAME 2 investigators, along with many members of the interventional cardiology community, have presented the results of the trial as a definitive response to the questions about PCI raised by COURAGE. In an ESC press release, FAME 2 coordinator Bernard De Bruyne said:

“With this new knowledge, I believe that FFR should become the standard of care for treating most patients with stable coronary artery disease and significant coronary narrowings.”

The Society for Cardiovascular Angiography and Interventions (SCAI) was so excited about the results of FAME 2 that it rushed out an e-publication of a “President’s Page” perspective on FAME 2, written by SCAI president J. Jeffrey Marshall and interventional cardiologist Ajay Kirtane, arguing that “FAME 2 offers the best data currently available to guide” treatment. The perspective of most interventional cardiologists is probably best summarized by this headline published on TCTMD: “PCI Bests Medical Therapy in Stable Patients with Proven Ischemia.”

The sad thing about these simplistic responses to FAME 2 — and the reason why I use this trial as an example of a poor model for clinical trials — is that there is no acknowledgement of the extraordinary division of opinion about this trial and its meaning. The SCAI document discusses the FAME 2 publication but does not even reference, or respond to, the issues raised in an editorial accompanying the publication of FAME 2 in the New England Journal of Medicine. That editorial, by Bill Boden, the principal investigator of the COURAGE trial, delivered a trenchant attack on the view that FAME 2 represents anything like a definitive response to COURAGE.

I summarized Boden’s points in my previous news story about FAME 2:

  • There were few “hard” events in FAME 2 and urgent revascularization could be performed without objective evidence of ischemia or positive biomarkers.
  • Since the trial was unblinded, “investigators may have had a lower threshold for recommending revascularization” for patients in the medical group.
  • Patients in the FFR group did not have noninvasive testing demonstrating ischemia, so some may have had preserved myocardial perfusion.
  • Patients in FAME II were not at very high risk.
  • The short followup period (mean followup of 7 months) did not leave enough time for the risk of restenosis in the PCI group to fully emerge.

Boden is highly critical of the early termination of the study, writing that it leaves “more questions than answers… but the only enduring finding of the FAME 2 trial appears to be that  of a reduced short-term rate of unplanned revascularization with FFR-guided PCI, with little evidence of long-term, incremental benefit on prognostically important clinical outcomes.”

Astonishingly, the ESC press release didn’t even mention that the ESC’s own discussant of the trial, Frans Van de Werf, concluded that FAME 2 did not provide the “final answer to the question how to treat stable CHD patients.” Here on CardioExchange, Rick Lange and David Hillis provided another deeply skeptical perspective on FAME 2, and their view received endorsements from Sanjay Kaul, David Cohen, and Harlan Krumholz.

PCI supporters are acting as if the new evidence provided by FAME 2 forges a new consensus in support of FFR-guided PCI, but only by ignoring a chorus of dissent.

It is perhaps worth noting here that WOEST and IABP-SHOCK II were investigator-driven trials in which industry played no significant role. Of course, this idyllic situation may have been possible only because no significant commercial interests were at stake in the trial. With FAME 2, by contrast, the commercial stakes — for industry, for hospitals, and for interventional cardiologists — could not be higher. Perhaps these influences have helped induce a self-interested reality distortion field.

9 Responses to “ESC Trials: The Best And The Worst”

  1. Right on in every way.

  2. Ajay J Kirtane, MD, SM says:

    The SCAI commentary that was co-authored by Dr. Marshall and me was based upon data in the the actual NEJM manuscript, NOT the editorial commentary. In fact, anticipating a critical editorial response such as the one by Dr. Boden (and some of the comments above), we thought that it was reasonable to discuss the trial from the contextual perspective of the prior studies in the literature.

    Re the points above:
    1. FAME 2 was NOT POWERED to look at “hard events”. FAME 2 does not provide a definitive answer regarding death/MI, but neither did COURAGE! This is precisely why the ISCHEMIA trial is being conducted. With regards to the much maligned endpoint of unplanned hospitalization leading to urgent revascularization: Even if one were to exclude 50% of unplanned hospitalizations leading to urgent revascularization from the medical therapy arm because one believed the investigators were biased, this would leave 7 events in the PCI arm and 24 events in the medical therapy arm, generating a p-value of 0.002 in favor of PCI for that endpoint! And why is that endpoint so insignificant to patients? Would YOU want to come back to the hospital with a suspected ACS and have another procedure if it could have been safely prevented?

    2. See #1 and my blog comments. Dr. Boden and others cite COURAGE and state that we should leave angiographically significant and ischemia producing stenoses alone and treat them medically. But when these stenoses get treated in another trial, then systematic bias is raised. Don’t you remember the crossover rates in COURAGE? I personally in fact agree that this is the biggest potential weakness of the trial, but it certainly does not earn this trial the moniker of “worst trial of ESC”. And again, even if we account for a 50% bias as in #1 above, the trial still shows a benefit of PCI in reducing unplanned hospitalizations leading to urgent revascularization, no harm with a PCI-first approach, and clear and early relief of symptoms. What’s so bad about that for patients?

    3. FFR is very well correlated with non-invasive testing demonstrating ischemia in numerous studies (in fact publised in the NEJM in 1996); it is safe to say that the ischemia in this trial was real. Admittedly, this trial did not include ischemia without stenosis, but that is a different entity that wasn’t studied in COURAGE either.

    4/5. The true clinical risk profile of these patients is yet to be determined by the number of events at longer term follow-up. While I agree that restenosis won’t be measured at 7 months, given the 97% DES use in FAME 2 (virtually all with 2nd generation DES), we can anticipate clinical restenosis will be <5% in this population, which would minimally alter the outcomes observed (unlike in COURAGE, where the use of POBA and BMS would clearly affect the durability of effect as well as outcomes). Additionally, look at the KM curves shown in the FAME 2 manuscript – they continue separating rather than converging.

    I'm very glad there is now debate regarding this trial and I also feel that it's nice to see people discussing the strengths and weakness of the clinical evidence base regarding revascularization in CAD so openly. But please read our editorial carefully. It does not advocate FFR or PCI unequivocally. It simply states that there are data out there that PCI – if applied to the right patients – can help those patients in ways that are important to THEM. In fact, I'd pick a different sentence of our editorial to quote: "Notwithstanding the early cessation of the trial, the findings from FAME 2 represent a further critical piece of evidence in favor of the ischemia hypothesis in SIHD; namely, that the safe and effective revascularization of ischemia-producing lesions can lead to improved patient outcomes and quality of life metrics above and beyond symptom relief."

    Please also see my comments to the blog post by Drs. Lange and Hillis.

    • Dr. Kirtane– I’ll leave the in-depth analysis of the data to yourself and all the other great contributors and commenters on this site. It’s a fascinating discussion, and I don’t pretend to know how this issue will eventually be resolved. It’s entirely possible that FFR-guided PCI will play an enormous role in coming years, but it’s equally plausible that medical therapy will prove to be the best overall starting strategy for these patients. Thus the equipoise behind the ISCHEMIA trial.

      But I’m not sure you’ve addressed my main point, which is that FAME 2 is being held up by SCAI, industry, and the interventional cardiology community as a clear “win” for FFR-guided PCI over medical therapy, almost completely ignoring the strong criticisms of the trial leveled by the NEJM editorialist, the ESC discussant, and others like Rick Lange and David Hillis right here on CardioExchange. Presenting the trial in this manner could be considered propaganda rather than a scientific discussion, in which case SCAI has the appearance of being more like a union than a medical society.

  3. The study could generate either of the following questions…..

    1. “Would YOU want to come back to the hospital with a suspected ACS and have another procedure if it could have been safely prevented?”


    2. “Would you want to have a procedure that you’re not likely to need?”

  4. Ajay J Kirtane, MD, SM says:

    I’m sure most people would answer YES to #1 and NO to #2. That is exactly my point regarding picking the right patients. The PCI’s that were performed in the trial helped people feel better AND avoided urgent hospitalizations without increasing complications. Isn’t that what PCI is supposed to do?

  5. Ajay J Kirtane, MD, SM says:

    Dr. Husten, since when does recognition of the editorial commentary, etc. of a published NEJM manuscript require equal or more attention than the actual trial itself? Speaking on behalf of myself (not any society)… the trial was published through a peer-reviewed process in the NEJM and was in fact positive for its primary endpoint. Isn’t that important to share with one’s colleagues (in a society or otherwise)? Does that mean that it (or any trial for that matter) is free of controversy and/or nuance? Of course not. You and I both realize that headlines only tell part of the story: recall the negative headlines that came out after COURAGE, or after the DES scare in 2006 (which actually prompted many patients and physicians to underutilize potentially life-saving procedures – e.g. my patient who refused PCI with an anterior MI because he heard that PCI was no good on the news).

    Please be assured (as I am sure you know), this trial will be debated, discussed, and dissected in the months to come because that is what we Cardiologists (independent of sub-specialization) do. All of these viewpoints will not only be heard, but will be actively solicited at debates, conferences, and in the literature because ultimately it isn’t about whether anyone “wins” or “loses” – it’s about trying to take care of patients in the most responsible and informed way.

  6. Robin Motz, M.D., Ph.D. says:

    I just would comment from my prior experience as a physicist that if you have to argue about the significance and interpretation of an experimental result, then it isn’t significant and it has no certain interpretation.

  7. William Edward Boden, MD says:

    So, I’m joining the fray a bit late here, but have some perspectives I’d like to share:

    1) I disagree that FAME-2 represented the “worst of the ESC” Hot-Line Sessions; I do believe it is an important trial and does move the ball upfield (it’s week #1 in the NFL) but not into the end zone as many hoped. FAME-2 does add to our knowledge base.

    2) FAME-2 raises a critical issue in the design of trials and the choice of a primary endpoint. There have been now 16 RCTs of PCI vs. medical therapy and all 16 show no difference for death or MI between randomized strategies. The temporal range of these RCTs span 20 years from ACME (212 pts with single-vessel LAD disease using PTCA to FAME-2 with current generation DES). It is also impt to emphasize how far medical therapy has evolved over these 2 decades (from ACME and the use of only ASA, nitrates, and BB to the present day of “optimal medical therapy” and the use of disease-modifying Rx.) What most docs and the public and the payers (though still largely mute on the sidelines) want to know is whether PCI is a durable procedure beyond angina relief, so D/MI is an important and worthy endpoint. In designing ISCHEMIA, we debated using an endpoint like in FAME-2, and were shot down by the scientific community and NHLBI. This should tell us all something.

    3) As an industry-sponsored trial by a manufacturer of the prdcut being tested to assess FFR, there has to be acknowledgement that St. Jude medical “had a dog in the fight”. Why else would the all-important “urgent revasc. endpoint” be defined in a way that did not mandate objective evidence of ischemia? In an unblinded trial where the identity of assigned strategy is known, there is just too much temptation to NOT give medical therapy a fair trial.

    4) Thus, for all the continued distortion of “high COURAGE cross-over” from OMT to PCI, we see an even more exaggerated cross-over, if you will, in FAME-2. This is why revasc. is a poor component of a primary endpoint. And, by the way, the X-O rate of 33% in COURAGE was over 2.5 to 7 years! what gets distorted is the fact that medial X-o was at 11 months–16.5% crossed over from OMT to PCI within ~ 1 year, and the remaining 16.5% between years 1 through 7. Viewed from that perspective, the X-O rate after 1 year was only 2.7%/year.

    5) What I really don’t get is why so many of the “best available medical therapy alone” pts needed revasc. within such a short follow-up duration of 7 months when 75% had single-vessel CAD, only 3% had FFR-proven 3-vessel CAD, 11% were asymptomatic at baseline (along with 16% who had silent ischemia), and 66% of the enrolled pts had CCS Class 1-2 sx. What this tells me is that this was a really low-risk group, YET they needed “urgent revasc” in an incredibly short time period. This is very puzzling, and suggests a superficial commitment to more intensive medical therapy to avoid or defer the need for subsequent PCI.

    6) I had thought that FAME-2 woudl have “made the case” for an FFR-guided PCI strategy for all visually stenotic lesions but it would seem hard to accept the case that this is justified, based on the trial outcomes.

    7) One last point I could not include in the editorial because of word count limitations. In the CHARISMA Trial, the 12,163 pts (77%) of stable CAD pts who had established disease (like in FAME-2) had a positive outcome when randomized to clopidogrel + ASA vs. ASA alone for the endpoint of death, MI, stroke, or the need for revasc (this was a CHARISMA secondary endpoint), and this was statistically significant. But, because the trial included 3,284 pts who were asymptomatic and did not have established CVD. When added together, the entire study population of ~15,000+ pts did not show benefit for clopidogrel + ASA for the primary endpoint of death, MI, or storke. So, in FAME-2, it is possible that the use of thienpopyridines in all PCI pts and the lack of use in the medically-treated pts could also be a factor in the trial results.

    In any event, great to see such good discussion and debate.

  8. I would like to suggest an alternative explanation for FAME II results that differs from Drs. Boden disagrees with the conclusion that FFR guided PCI in stable CAD benefits patients over OMT by reducing progression to ACS and cites all prior trials including COURAGE that failed to show a benefit of PCI over OMT. Here is my reasoning:

    1. At the outset I agree that the main evidence of progression to ACS was based on worsening symptom/angina severity and not Acute MI, and hence subject to bias in an unblinded trial and the study should have allowed to continue to complete enrollment. Indeed, I suggest a larger trial sufficiently powered to detect differences in Ac MI be performed to demonstrate this is needed because the there is a pathophysiologically plausible reason for benefit of FFR guided PCI in reducing ACS in stable CAD as I suggest below.

    2. The difference between FAME II and all prior trials of PCI Vs OMT in stable CAD lies in the ACCURACY of identifying/selecting OBSTRUCTIVE CAD CAUSING ISCHEMIA in the patients enrolled: FAME II accurately identified ischemia causing obstructive CAD lesions by a method validated against the gold standard of absolute coronary blood flow reserve in animals (and also humans using PET) whereas all prior trials used visually estimated angiographic severity that is clearly not accurate in the 50-80% severity range or, non-invasive tests to detect ischemia which are equally inaccurate. Thus the apparent paradox that despite clear demonstration that high grade/obstructive lesions are more likely to progress to occlusion (asymptomatic or with ACS) than non-obstructive lesions revascularization has not been shown to reduce risk of ACS & associated mortality in stable CAD (of course most patients presenting with ACS result from non-obstructive lesion rupture/erosion mediated progression to ACS simply because they are more numerous).

    3. It therefore seems to me quite plausible that what FAME II demonstrated is that the therapeutic index of the benefit of PCI in stable CAD is quite narrow: i.e., the benefit is strictly confined to only obstructive lesions (of major epicardial vessels as used in FAME II, (which ensures a clinically significant size of ischemic myocardium at risk), and that if you use a method such as visual estimate of angiographic severity not sufficiently accurate, the benefit ‘signal’ is not only lost in the ‘noise’, but as FAME I defer and FAME I demonstrated, counterbalanced by increased adverse PCI related events of non-obstructive lesions, and hence the negative prior trials of PCI Vs OMT in stable CAD. So to me the interpretation of FAME II is, in order to derive the dual benefit of PCI (relief of ischemia/angina, and, prevention of progression to ACS) in stable Obstructive CAD lesions, it has to be obstructive beyond doubt and FFR is needed in most visually apparently obstructive lesions to ascertain this, involve a major vessel.

    4. To therefore test the hypothesis that PCI of truly obstructive stable CAD lesions not only relieves ischemia better than OMT (has been clearly demonstrated especially for lesions especially subtending >10% LV), but also reduces ACS and mortality (resulting from a local effect at the level of the lesion, i.e. dual mechanism of benefit) what one needs is a larger FAME II trial sufficiently powered and appropriately designed (? blinded to the patient and managing cardiologist distinct from the study interventionalist) to detect this endpoint. The ISCHEMIA trial may achieve this, but only because, as I understand it, the interventional component requires FFR of all but visually certainly obstructive lesions, similar to FAME II. The imaging component suffers from the same inaccuracy of SPECT but of course the mandatory CTA helps with eliminating false positives (& Left main lesions) and thus enhances the overall accuracy of the noninvasive imaging. If the trial results prove to be useful, then to be translated into routine clinical use, then patients will need a combination of stress SPECT and CTA to achieve comparable results, which is not feasible for routine widespread use.

    5. Finally, is really needed is progress in accurate non-invasive detection of ischemia causing lesions (beyond SPECT) with PET imaging as Dr. Gould has demonstrated with absolute Coronary flow reserve in addition to relative flow reserve and of course reduction of artifacts common in SPECT and an ischemia trial to replicate the results of FAME II.

    Lastly I certainly agree that there is a very good reason for such a vigorous debate from ‘both sides of the aisle’ (interventionists and non-interventionalists) and this should certainly stimulate efforts to find an answer.