Conversations in Cardiology: In the ISCHEMIA Era, What Role for FFR?

Morton Kern, MD, often engages his colleagues via email in brief, informal dialogue on clinically relevant topics in cardiology.

Morton Kern

Morton Kern, MD, of VA Long Beach Healthcare System and University of California, Irvine, often engages his colleagues via email in brief, informal dialogue on clinically relevant topics in interventional cardiology. With permission from the participants, TCTMD presents their conversations for the benefit of the cardiology community. Your feedback is welcome—feel free to comment at the bottom of the page.


David Cox, MD (Brookwood Baptist Health, Birmingham, AL), asks:

I need help with another question posed by many of the docs at my site. While I agree fully with Sunil Rao that 70% of what most of us do is ACS, and that was my practice in Pennsylvania and North Carolina, I find myself in a situation in Birmingham with my hospital on a hill surrounded by a number of PCI hospitals who get most of the ACS.

Consequently and surprisingly, about 80% of this practice is elective stable ischemic heart disease (SIHD) PCI with much less ACS.

The conundrum exists, of course, like the patient I saw yesterday in office whose chest pain has heated up the past few weeks. He has gone from angina 2x/week to now daily angina with minimal exertion and some pain at rest. I wanted to admit. He will come in electively this AM.

1. Count him as crescendo angina/Class III, and this doesn’t become a ISCHEMIA patient.

2. The question at hand is, let’s say he did have a recent nuclear scan with moderate ischemia and Class II angina, what is the role  of FFR after ISCHEMIA?

3. Are all the data in FAME and FAME 2 now superseded by this trial? I know not all patients with SIHD needed FFR, but the good majority did.

4. If you do an FFR and it’s positive in a patient who met entrance criteria for ISCHEMIA (let’s leave the truly asymptomatic patients aside) and the FFR is done well and is positive . . . then what?

Kern replies:

Dave is struggling with the approach to SIHD in his area. Here is my response to Dave:

I'm not sure why you're struggling so much.

  1. The crescendo angina lets you treat as needed. Of course, escalate medical therapy and if still symptomatic, stent. ISCHEMIA trial said you will reduce medication use and give better symptom-free life.
  2. If the nuclear test is positive and you believe it, you don't have to use FFR. If not, FFR can be a more lesion-specific measure of ischemia than stress testing.
  3. This is not true. Both studies still identify who will do well. For FAME, FFR-guided stenting in patients will do better than those stented based on angiography alone. For FAME 2 those with positive FFR and undergo PCI will have fewer events than positive FFR treated only medically. This occurred regardless of status: ACS versus stable angina.
  4. The positive FFR entitles you to make a decision on how best this patient wants to live—more medicine or revascularization (stenting) for reduced symptoms? In my view, the ISCHEMIA trial did not change my thinking very much. Do we believe treating ischemia is a good thing? I still do, first with meds then with stents.

I don't think I missed the main point of the ISCHEMIA trial. Colleagues?

Jeffrey Moses, MD (NewYork-Presbyterian/Columbia University Irving Medical Center, New York, NY), replies:

Good summary, but I would add three concepts.

  1. The issue with the ischemia in the trial was likely the terrible imprecision of the myocardial perfusion imaging in identifying and localizing ischemia. Especially in multivessel disease, FFR supersedes this terrible test and its data stands on its own in the FAME trials.
  2. The key lesson for the trial is that even with over 40% of the two groups having the same treatment (20% unrevascularized in the active group and 23% revascularized in the med group). The symptom relief was superior in even the mildly symptomatic (ie, monthly angina), similar to FAME 2.
  3. As a bonus, spontaneous MI was reduced (with widening differences over the follow-up), something FAME 2 observed as well.

So the conversation with the patient can be built around these facts.

Bonnie Weiner, MD (Saint Vincent Hospital, Worcester, MA), replies:

I agree with all your comments, Mort. The critical issue here is that in the stable patient, it has not been about MI/death prevention but quality of life (QoL). We know that patient compliance and tolerance with complex medical therapy regimens outside of a clinical trial is less than optimal. The importance of patient preference and patient/doctor discussion remains the important message.

David Cohen, MD (University of Missouri-Kansas City), replies:

Here are my thoughts:

  1. A patient with crescendo angina/Class III does not become a ISCHEMIA patient. 

    Technically, this is correct. But the truth is that the difference between Class II and Class III angina is pretty subtle, and I wouldn't hang my hat on it. The main reason why patients with accelerating angina were excluded from ISCHEMIA was not medical but more related to a desire to keep the rate of crossover to cath down among patients randomized to the conservative arm.
  2. If he did have a recent nuclear scan with moderate ischemia and Class II angina, what is the role of FFR after ISCHEMIA?

    To me, the main role of FFR after ISCHEMIA is the same one it had before ISCHEMIA—to identify which lesions in patients with significant angina do NOT need revascularization (ie, the key take-home messages from both DEFER and FAME). I agree with Mort that if the lesion causing the ischemia is clear from precath testing, there’s no need to do FFR/iFR/etc to reconfirm. But if there is ambiguity, a key benefit of intracoronary physiology is sorting out which lesions are truly hemodynamically significant.
  3. Are all the data in FAME and FAME 2 now superseded by this trial?

    FAME is still as important as it was before ISCHEMIA—use physiology to identify which lesions to treat and which to leave alone in patients with multivessel disease requiring revascularization (ie, those with unsatisfactory symptom control). IMHO, the potential late spontaneous MI benefit in ISCHEMIA reinforces the same finding in FAME 2 (which previously seemed like a bit of an outlier). However, there is still no evidence that PCI saves any lives in the stable population, and the late MI benefit needs to be balanced against the early excess MI rate from PCI.
  4. If you FFR and it’s positive in a patient who met entrance criteria for ISCHEMIA, then what?

    Then you can either treat with optimal medical therapy (OMT) alone or offer revascularization, depending on the patient's preference—balancing the side effects from obligatory dual antiplatelet therapy (for some duration) versus those of antianginal therapy. 

    In my opinion, the main messages from ISCHEMIA are that not every patient with moderate-to-severe ischemia on functional testing needs to go to the cath lab. The cath lab should be reserved mainly for the following situations:
  • ACS: always
  • Unsatisfactory symptom control (for whatever reason): with an eye toward complete functional revascularization
  • Resolving diagnostic dilemmas: especially evaluating left main disease and sorting out the cause of ischemia in patients with conflicting findings from noninvasive testing (eg, significant ischemia with only moderate CAD)

Kirk Noel Garratt, MD (Christiana Care, Newark, DE), replies:

Jeff’s summary is a great distillation of the takeaways from ISCHEMIA. We’ve been treating SIHD patients chiefly for symptom improvement for more than a decade, and this study confirms the appropriateness of our practice.

I thought the most provocative finding from ISCHEMIA was the reduction in spontaneous MIs. Let’s remember that the big knock on COURAGE was the concern that the most at-risk patients were never enrolled. Presumably, those would be the folks at greatest risk of events like MI. Now that we’ve got a trial of SIHD that absolutely enrolled some high-risk people, we see an MI benefit from PCI and CABG. Sunil Rao, Tim Henry, and a couple others reminded me this was also observed in FAME 2 and COMPLETE—somewhat different populations but a signal of MI reduction benefit with revascularization nonetheless. We’ll have longer follow-up, but for now, I’m comfortable telling patients that a lower risk of spontaneous MI is a potential benefit of early revascularization.

David R. Holmes Jr, MD (Mayo Clinic, Rochester, MN), replies:

I agree entirely with KNG. It is part of a strong data set.

Farouc Jaffer, MD, PhD (Massachusetts General Hospital, Boston), replies:

Great questions and insights. Would also offer that a spontaneous, unpredictable MI is a greater clinical stress to patients and physicians compared to a periprocedural MI occurring in-house. 

William Fearon, MD (Stanford University, CA), replies:

I agree completely with Jeff’s comments. ISCHEMIA in fact reinforced much of what FAME 2 showed: in patients with stable CAD, compared with medical therapy, FFR-guided PCI reduced spontaneous MI at 5-year follow-up (same as revascularization in ISCHEMIA) at a cost of a small bump in peri-PCI MI (same in ISCHEMIA), decreased urgent revasc (revascularization significantly decreased unstable angina in ISCHEMIA), improved quality of life and angina relief (same as ISCHEMIA), without a change in overall mortality (same is ISCHEMIA). ISCHEMIA did not show a relationship between the degree of ischemia on stress imaging and outcomes, but I think this is simply a reflection of the lack of precision of the noninvasive tests (as mentioned by Jeff); numerous previous studies have shown a clear gradation between FFR values (degree of ischemia) and risk for events in medically treated patients. Of note, one-quarter of patients in ISCHEMIA had treadmill tests alone.

Since 35% of patients in ISCHEMIA were asymptomatic, it will be interesting to see how these patients fare as a subgroup. The forest plot suggested that more symptomatic patients received a greater benefit over asymptomatic. However, this substudy from FAME 2 suggests that symptomatic patients randomized to medical therapy were “protected” by their symptoms because they crossed over to PCI and avoided hard events. On the other hand, asymptomatic patients with low FFR randomized to medical therapy had increased death/MI compared with those who received PCI, presumably because they had no warning system.

George Vetrovec, MD (VCU Pauley Heart Center, Richmond, VA), replies:

At least one unanswered question for the ISCHEMIA trial is the impact of completeness of revascularization on the results. Dr. Judith Hochman noted that this was still awaiting analysis. There is significant data that shows the completeness of revascularization, ie, the effectiveness of alleviating ischemia is an important factor in PCI outcome. PCI’s limitations in the most complex disease are often the reason that surgical revascularization is more effective. Until we know those results, I think we are somewhat limited in being certain of the outcome results. Even later follow-up will also provide important late results on MI prevention and, I believe, for completeness of revascularization. However, the current completeness of revascularization data should be available hopefully by the time of publication or for an upcoming scientific meeting like the ACC.

Keith Oldroyd, MBChB (Golden Jubilee National Hospital, Glasgow, Scotland), replies:

Great discussion. There is also a potentially important issue in relation to complete relief of ischemia or, if you prefer, functionally complete revascularization. For various reasons, we all know that significant numbers of patients are left with potential ischaemia even after revascularization, whether by PCI or CABG. FAME 3 will shed some further light on this—it should finally complete recruitment next month.

Gregg W. Stone, MD (Icahn School of Medicine at Mount Sinai, New York, NY), replies:

We plan on reporting the impact and implications of complete versus incomplete and anatomic and functional revascularization at the ACC in 2020.

Ajay Kirtane, MD (NewYork-Presbyterian/Columbia University Irving Medical Center), replies:

All great points. Dave, regarding your patient, one thing to remember about QoL is that the number need to treat to render a patient with weekly angina completely “angina-free” with the invasive approach versus the conservative one is three. Also, only 29% of patients in ISCHEMIA had angina that started or became more frequent in the last 3 months, with the median angina frequency being monthly.

Beyond your patient, some take homes from the trial for me:

  1. The likelihood that a low-symptom patient with SIHD “drops dead” without left main CAD is exceedingly low, which is very reassuring (and I think a novel finding). This allows TIME to figure out how best to treat the individual patient.
  2. The invasive strategy (and therefore presumably revascularization) was associated with a DURABLE improvement in symptoms (remember one of the main critiques of COURAGE was that the effect was gone at 3 years). This was DESPITE the fact that in ISCHEMIA 20% never got revascularized and 23% crossed over, as Jeff pointed out.
  3. I now feel comfortable telling patients that (especially if we do the cath/revascularization procedures without periprocedural complications) there may be a small reduction in later MI with an invasive approach. The reduction in spontaneous MI in ISCHEMIA is remarkable (and also runs counter to “COURAGE-wisdom”) but is consistent with FAME 2, COMPLETE (with limitations of a different population), and a prior meta-analysis by Sripal Bangalore that included COURAGE. We don’t yet know how much of this was from CABG and/or PCI, and complete versus incomplete revasc, and ALSO revasc concordant with noninvasive ischemia or not. But the finding is still notable.

Neal Kleiman, MD (Houston Methodist Hospital, TX), replies:

Another parallel question is adequacy of revascularization. Based on ULTIMATE and DEFINE-PCI, I’d want to know how often stents were optimized and would also be concerned about not truly eliminating the ischemia.

David Kandzari, MD (Piedmont Heart Institute, Atlanta, GA), replies:

My two additional takeaway messages from ISCHEMIA:

  1. I am evermore reminded by ISCHEMIA of the disconnect between patient-reported and MD-reported symptoms. Even among a Class I/II angina population, the median SAQ was 80 with a range or 70 to 100 (essentially asymptomatic).
  2. We need to rethink trials that place such an overemphasis on the credibility of noninvasive stress testing and hold it to a seemingly uncontested standard. I’m anxious to see the correlation of noninvasive testing with anatomy, but the fact that many were excluded because of no significant disease by CT angiography is telling. 

Augusto Pichard, MD (MedStar Washington Hospital Center, Washington, DC), replies:

In addition, most of them had 50% lesions by CT, with some 70%. We know physiology has proven most of these do not need intervention! No wonder the PCI arm ended up similar (in some respects) to OMT.

Cohen replies:

I don't think we know what the lesion severity was by CT, Gus. We know the patients had to have at least one 50% lesion to qualify for the study and about half had "3Vdz" by this definition. But the CT results were blinded to the investigators, and I haven't seen any data on the actual lesion severity (yet).

John Hirshfeld Jr, MD (Penn Medicine, Philadelphia, PA), replies:

I’m waiting for later follow-up.

Note that all the Kaplan-Meier curves cross at 1-2 years’ follow-up and continue to diverge. With the follow-up available at the time of this report, P values are in the 0.2-0.3 range (so not “significant"). Only half of the cohort followed up at 3 years and one-third at 4 years.

It will be interesting to see what happens to this trend in the next several years.

Hopefully whatever effect there is will not be obscured by too much crossover from the conservative group to the revascularization group (already at 23%).

Cohen replies:

Let's hope that the extension gets funded by the NIH—otherwise, there won't be longer follow-up.

Also, one thing that has not been appreciated is that fully half of the crossovers occurred AFTER a primary endpoint had already occurred (I don't know the details, but presumably in most cases, the patient presented with a NSTEMI and underwent PCI during the hospitalization). Given the sequence, those types of crossovers actually have no effect on the primary endpoint Kaplan-Meier curves (although they could lead to some erosion of the antianginal benefit). So from the standpoint of CV death/MI, the effective crossover rate is only around 10-11%, which isn't bad for a 4-year trial.

Larry S. Dean, MD (UW Medicine, Seattle, WA), replies:

Thanks to David for starting the conversation. Very interesting to read everyone’s take on the trial, and it will be helpful as our patients begin to ask the questions we know will come. I guess my main concern is all we have so far is what’s been presented without the publication.

Kirtane replies:

But Dave: either of those things is “failing the strategy.” Whether the patient crosses over OR has an MI event (because death is low), both are not great.

Cohen replies:

Yes, but most people worry about crossover because it dilutes the treatment effect. In the case of an MI preceding a revasc, the event is counted correctly and there’s no dilution.

Stone replies to Cohen:

Only 8% of patients of the 28% who underwent angiography by 4 years in the conservative arm “crossed over” for nonadherence to the protocol (ie, without a valid reason – a protocol violation). The remainder had valid reasons such as development of refractory angina or an actual or suspected endpoint event. Hard to argue any of these crossovers affected the absolutely neutral mortality outcomes or the markedly positive QoL benefits (100% certainty of effect by Bayesian in patients with baseline angina).

Stone replies to Hirshfeld:

It is difficult to describe the overall treatment effect when the curves are crossing. The HR is not completely accurate given the nonproportional hazards, and Cox models are invalid. There were actually statistically fewer events in the invasive arm at 4 years—the CV death/MI risk difference was -2.2% (95% CI -4.4%, -0.1%). However, given the upfront risk of periprocedural MI, the total event-free time in both groups was not different.

The Bottom Line (According to Kern):

The ISCHEMIA trial reinforces some of our beliefs in the need for better anti-ischemic medical therapy prior to revascularization in SIHD patients and that absent objective ischemic findings FFR still applies to decision-making. What may be in question is what will be the best noninvasive ischemic testing: FFRCT, PET, or something else? The longer-term follow-up of ISCHEMIA will be warmly welcomed to this debate.