August 28th, 2012

‘Rather Than … FAME, Give Me Truth’


In the FAME 2 randomized trial, patients with functionally significant coronary arterial stenoses, as determined by fractional flow reserve (FFR) measurements, who underwent percutaneous coronary intervention (PCI) in combination with the best available medical therapy were less likely to experience death, MI, or urgent revascularization than were patients who received such medical therapy alone. Recruitment was halted prematurely — after only 888 of the 1632 planned patients were enrolled — because of a lower rate of urgent revascularization in those receiving PCI plus medical therapy; the incidence of death or MI was similar in the two groups.


PCI + Medical Therapy

Medical Therapy Alone

P value

Composite of death, MI, or urgent            revascularization








Myocardial infarction




Urgent revascularization





Does this settle the issue that ischemia should guide revascularization? Consider these points:

1. The Data and Safety Monitoring Board halted the study on the basis of differences in the weakest component (urgent revascularization) of a composite endpoint in a setting where referral bias may have influenced the outcome. Interestingly, no formal rules for stopping the study prematurely were specified. How, then, was study termination decided? On average, trials that are stopped prematurely for benefit overestimate that benefit by 30%.

2. Half the patients undergoing urgent revascularization had no objective evidence of ischemia (i.e., elevated biomarkers or ECG changes). Is this really “urgent revascularization” or simply a bias to revascularize subjects with abnormal FFR (vs. those with normal FFR)?

3. The post-PCI diagnosis of MI was established using CK-MB levels (above 5 or 10 times the upper reference limit), whereas the diagnosis of MI in the setting of ACS was established with either CK-MB or troponin levels. Doesn’t this minimize the rate of peri-PCI MI?

4. The sponsor was “involved in the collection and source verification of the data.” What does this mean?

5. Although the study was designed to follow patients for 2 years, the mean follow-up was only 7 months. Come on now — not only was the study terminated prematurely (when only slightly more than half the planned subjects were enrolled), but the follow-up was truncated as well?   

6. Demonstration of a pressure gradient across an epicardial coronary stenosis (i.e., abnormal FFR) is not synonymous with ischemia. Instead, ischemia is a consequence of diminished myocardial tissue perfusion and can be identified by ECG changes, regional myocardial perfusion abnormalities during scintigraphy, wall-motion abnormalities on echocardiography, or abnormal tissue perfusion or metabolism on PET imaging. Given that none of these tests was performed in conjunction with the FFR measurements, we cannot know whether an FFR of 0.68 (the mean value in the study) was associated with ischemia. In addition, it’s difficult to compare the FAME 2 patients with the COURAGE patients (85% of whom had documented ischemia).

7. On the basis of the FAME II data, one would perform PCI in 100 patients to prevent 9 “urgent revascularizations,” only 4 of which are for ischemia (i.e., positive biomarkers or ECG changes) — without reducing the incidence of MI or death. This seems to be a long (and expensive) run for a short slide.

Henry David Thoreau asked for truth rather than FAME (or love or money for that matter). What truth do you take from this study? Should we be performing PCI in all stable CAD patients who have abnormal FFR?

See the review by Howard Herrmann on Journal Watch.

10 Responses to “‘Rather Than … FAME, Give Me Truth’”

  1. Q1. Does this settle the issue that ischemia should guide revascularization?
    A1. No! Whether the most important
factor related to clinically important outcome in management of
patients with stable CAD is the presence and extent of inducible
ischemia remains unaddressed by the FAME II trial.

    Q2. Should we be performing PCI in all stable CAD patients who have abnormal FFR?
    A2. To answer this question, a nonrandomized comparison between those with FFR less than 0.80 + OMT (one of the randomized cohorts) vs. those with FFR greater than 0.80 + OMT (registry cohort) might be instructive. The incidence of urgent revascularization in these treatment groups is 11.1% (49/441) vs. 2.4% (6/166). This translates into a sensitivity of 92%, specificity of 29%, PPV of 11%, NPV of 98%, AUC of 0.75. Thus, an FFR less than 0.80 has modest prognostic utility in guiding PCI in stable CAD patients. Stated another way, nearly 90% of FFR-guided PCI were likely unnecessary!

    Of course, this comes with all the caveats of nonrandomized comparison with only a random 50% of registry cohort data available!

    Bottom line, truncation of enrollment compounded by truncation of follow-up translates into an UNINFORMATIVE trial result!

  2. Steven Greer, MD says:

    Nice post and comments. I think Steve Greer is influencing you all to be finally skeptical!!!!

    Seriously, early stoppage of trials was a theme we championed several years ago. Nice to see it catching on in the clinical academic community. That “Jig is up”

  3. And then the ESC press release today blared: “FAME-II showed that fractional flow reserve (FFR) should become the standard of care for treating most patients with stable coronary artery disease and significant coronary narrowings.” This is not exactly based on the science since FAME-II did not evaluate the incremental value of FFR. Will people skip the article, hear about the release, and assume it to be true. I hope not.

  4. Well, there was a pretty reasonable study called FAME (the original) which demonstrated that if you are going to do a PCI in patients with multivessel disease, that FFR-guided PCI leads to better outcomes (less death or MI) and lower short-term and long-term costs. So, I think we do actually know that FFR is a good way to guide our PCI procedures.

    But I agree with many that FAME-2 doesn’t really do much to address the question the investigators originally set out to address- whether all lesions with FFR <0.80 need to be revascularized. And that's unfortunate.

  5. David,
    While FAME demonstrated that FFR-guided PCI trumps non-FFR-guided PCI (the so-called oculodilatory reflex), FAME II demonstrates that FFR-guided PCI might not offer any major advantage over OMT. So why even bother with measuring FFR? I guess the argument boils down to what is the clinically relevant metric for FFR – high NPV or modest PPV?

  6. I think we are in agreement, Sanjay. That’s why I made my last comment that it is a real shame that FAME-2 was terminated early. It was an important opportunity to address the other side of the coin, and it has been lost forever.

  7. Ajay J Kirtane, MD, SM says:


    1. We are not privy to the DSMB decision-making, but the critical issue here is that “urgent revascularization” in the trial was not simply urgent revascularization. In fact, the endpoint was “unplanned hospitalization leading to urgent revascularization” (see supplementary appendix) rather than vice versa. I even suspect that one of the reasons the endpoint was termed “urgent revascularization” in the manuscript may have been due to the NEJM word limits (of course this is just speculation). When adjudicated in this manner, I can see the DSMB’s concern with allowing an excess of patients to essentially be re-admitted for suspected ACS leading to another invasive procedure. I like others indeed wish that the trial had continued so we could have seen whether these events were MI’s or not, but in reality, the study was UNDERPOWERED to look at death/MI alone, and the trial would have been very positive for the primary endpoint, which is why I suspect the DSMB stopped the study. We can argue all we want about the study design and the choice of endpoint, but the decision to stop the study (given the actual constraints of the design) is actually not all that unreasonable.

    2. Bias is certainly possible, but in the post-COURAGE era (somewhat facetiously) aren’t we supposed to treat people with ischemia-producing stenoses medically? Look at the registry arm of the trial (patients with severe lesions that were FFR negative)… these patients had angiographically tight lesions that were left alone and didn’t come back for unplanned hospitalizations!

    3. The universal definition of MI has recently been revised to address the oversensitivity of the initial definition. There are several studies (in fact, Dr. Cohen recently published one in the Archives of Internal Medicine) demonstrating that lower levels of CK elevation post PCI (without other sequelae) are far less meaningful.

    4. I took the sponsor portion to mean that the sponsor conducted the monitoring, which is not uncommon for sponsor-run studies.

    5. NO!! The follow-up has not been truncated. Data to 7 months was shown because of the early termination; It would of course have been nice to see the presentation of longer term data; the problem is the crossover that has occurred since the early termination.

    6. FFR has been well-corrlated with non-invasively defined ischemia in multiple studies. In fact, one could argue that it is better at detecting regional ischemia than conventional (non-corelab or non-quantatative) methodologies. I think we can be sure that these patients had ischemia, and likely to a greater extent than those enrolled in COURAGE.

    7. As far as the run for the slide… it all depends on who we are treating. If it were me and I was already in the lab (assuming I was there for a good reason), why take the greater chance of urgently coming back via the ER for a repeat procedure with medical therapy? As was clearly demonstrated yet again (in FAME 2), if I had symptoms – I’m assuming that was why I would be in the lab – I clearly would feel better (and on less anti-anginal medication) with PCI.

    Bottom line: Whether you believe that the blinding issue completely negates the study results or not (I frankly think it’s a stretch to throw out the entire study because of that), PCI appeared to be a SAFE alternative to make patients feel better and keep them out of the ER/hospital.

  8. 1. The DSMB comprised highly experienced and influential experts. Nonetheless, we would all have benefited from a detailed discussion on the decision-making by the DSMB in the NEJM paper. What drove their decision to prematurely truncate the trial: the primary composite endpoint or the urgent revascularization component?

    2. Is it inconceivable in an ‘unblinded’ trial in which both patients and physicians have foreknowledge of whether a PCI was performed or not that ‘unplanned hospitalization leading to urgent revascularization’ might be biased in favor of the FFR + PCI treatment group? This is especially relevant given that nearly half of the urgent revascularizations were driven by symptoms alone without objective evidence of ischemia or injury! How did the investigators account for this potential bias?

  9. 1. We’d all benefit from more transparency and/or explanation regarding the DSMB decision to terminate the trial prematurely. Since early study termination tends to overestimate benefits, we’re left with wondering the true magnitude of benefit (if any) of FFR-determined PCI vs OMT.

    2. The issue of bias is difficult to ignore — and more difficult to control — in a study where the physician and patient know the results of a test. Especially if there is a preconception by either that PCI is the treatment of choice for “ischemia.” Specifically, the patient with an abnormal FFR and chest pain is more likely to be hospitalized for PCI by their physician than the patient with normal FFR and chest pain.

    3. The use of different criteria for establishing MI due to PCI vs ACS makes it difficult to compare rates of MI — which is one of the endpoints — in the 2 circumstances.

    4. A better explanation of the sponsor’s involvement would go a long way.

    5. Premature termination of the study (i.e., halting enrollment of patients) does not mean that follow-up needs to be truncated…..especially when the short follow-up favors one particular treatment arm (i.e., PCI).

    6. Although many patients with an abnormal FFR may have ischemia, the 2 don’t necessarily go hand-in-hand. We all take care of patients who are relatively sedentary (or well treated on OMT) who have minimal increase in myocardial oxygen demands throughout their day and do not experience ischemia (despite an abnormal FFR). Equating an abnormal FFR with ischemia is tempting…..but not accurate.

    7. I have a different perspective. If offered a procedure that doesn’t reduce my risk of MI or death and for which there is a 90% chance I’d not need it for symptom relief…..I’ll wait.

  10. Mani Prasad Gautam, MD says:

    I think the premature termination of the study had made the FAME II less meaningful. The use of strict prespecified criteria such as STEMI or NSTEMI for urgent revascularization would have made the study more meaningful. On the one side the incidence of MI (major indication for PCI in usual practice) and death (end result in the absence appropriate management) is not different significantly, on the another side, more patient in OMT group were taken for revascularizationsame. There seems alot biasness in slecting the subjects for PCI from OMT group without any objective evidence. It would have been informative if they have reported frequency of repeated hospitalization because of cardiac causes, incidence of arrythmias and functional status of cardiac complaints and also the some functional markers that denote deterioration of cardiac function from echocardiographic or exercise testing in these groups. Still, I am not sure whether it is better to have angioplasty in asymptomatic CAD subjects with significant FFR (hemodynamically), if should for what benefit.