Please note that the text below is not a full transcript and has not been copyedited. For more insight and commentary on these stories, subscribe to the This Week in Cardiology podcast on Apple Podcasts, Spotify, or your preferred podcast provider.

In This Week’s Podcast

For the week ending November 5, 2021, John Mandrola, MD comments on the following news and features stories.

Correction

First a correction on last week’s podcast. I mentioned a research letter published in the Journal of the American Heart Association that reported on an increasing incidence of Takotsubo cardiomyopathy.

An academic cardiologist who is more familiar with the data source —the National Inpatient Sample (NIS) data base — told me I may have missed the boat. I was struck by the increased incidence, but my colleague pointed out that there were many reasons NIS could show an increase in incidence. For instance, he wrote, it could all be due to coding practice. With the introduction of electronic medical records (especially EPIC), it has become much easier for people to select a very specific billing code for Takotsubo than it was previously. In addition, he suspected a lot of the increase is just greater recognition of a condition that has been with us since the dawn of time.

The bottom line is that tracking procedures in NIS or related databases is reliable. Tracking pure diagnoses is highly dubious. I still think we are seeing more of this condition, but that bias of mine got in the way of proper critical appraisal of the study.

By the way, I really appreciate this sort of feedback. Please keep it coming. This note came through a direct message on Twitter but you can also reach me via comments on theheart.org Medscape Cardiology or just plain old gmail.

Fractional Flow Reserve Percutaneous Coronary Intervention

The New England Journal of Medicine published yesterday the results of the FAME-3 RCT of fractional flow reserve (FFR)-guided percutaneous coronary intervention (PCI) vs coronary artery bypass graft (CABG) in patients with multivessel coronary artery disease (CAD).

FFR is a surrogate marker of a stenosis. It’s a physiologic measure. A catheter measures a pressure right before the blockage and right after. If there is no blockage, the FFR would be 1; if it was totally occluded it would be nearly 0, and if the FFR is less than 0.8, it is felt to be significantly flow limiting.

The therapeutic fashion is that the physiologic measurements provided by FFR help guide the doctor in telling which blockages should be stented and which should be left alone. Visual inspection of a blockage on an angiogram with contrast isn’t accurate enough, goes the thinking. It’s an enticing concept because it makes sense. (The phrase, “makes sense” should always raise your suspicions.) First, the background: We have three trials — SYNTAX, FREEDOM, and BEST — that reveal that CABG is better than PCI in patients in patients with multivessel disease. But these were old trials that used older stents, and many were done without the benefit of FFR. So, a better brand of PCI might be as good as CABG. Thus, the FAME-3 trial tested whether FFR-guided PCI with current generation stents was as good as CABG — was FFR-guided PCI noninferior (NI) to CABG?

Before I tell you the results, I want to say that this is actually a great use of noninferiority testing, because PCI clearly has a shorter and easier recovery than CABG. If it was as good as CABG, most patients would choose the less-invasive easier procedure.

FAME-3 used an efficacy endpoint of major adverse cardiac events (MACE) – death, myocardial infarction (MI), stroke, or repeat revascularization. The NI margin was set at 1.65, meaning PCI could be 64% worse and still deemed NI —a pretty wide margin.

• 1500 patients at 48 centers were randomly assigned to stent vs CABG. On average, patients in the stent arm got 3.7 stents and those in the CABG arm got 3.4 bypasses.

• At 1 year, 10.6% in the FFR-Guided PCI arm had a MACE vs 6.9% in the CABG arm.

• The absolute increase of 3.7% in the PCI arm translated to a hazard ratio (HR) of 1.5 or 50% worse. You might say, that’s NI because 1.5 is less than 1.65. But you can’t use the point estimate of the HR. In NI trials you have to consider the upper bound of the 95% confidence interval (CI) of the HR. The HR of 1.5 had a CI that went from 1.1 (10% worse) to 2.2 (more than 2-fold worse). Since the worst-case scenario, the upper bound, was greater than the NI margin of 1.65, we say the FFR-guided PCI was not found to be NI to CABG.

When the trialists reported harder endpoints of death, MI, and stroke, the rates were 7.3% vs 5.2% — a 40% increase in the FFR-guided PCI arm. As expected, the rates of major bleeding, arrhythmia, and acute kidney injury were higher in the CABG arm. CABG will always have more immediate post-op risk.

When you look at the components of the endpoint, there were no outliers. Pretty much each of them was higher in the PCI arm. Recall that these were 1-year results. That PCI could not match CABG at 1-year is quite remarkable. Recall that the benefits of CABG accrue over time. The hope was that FFR would lead to more judicious stenting, which would be enough to equal CABG, were clearly not born out from the evidence.

Interventional cardiologist Ajay Kirtane Tweeted out that his take-home was that “for 3VD, gotta take patient off the table (which we have been doing). Differences will be greater with longer FU.” He added that that physiology is just one part of optimized PCI; 12% imaging in study aiming to match CABG doesn’t cut it!

His point here was that imaging was used in only 12% of the PCI patients. Thus, use of imaging would improve stenting results. Imaging may allow for better stent placement and that may lead to less need for repeat procedures. Maybe. Imaging-guided stenting as a way to get it to measure up to CABG is a reasonable hypothesis and it surely could be tested in the same way as FFR. I would offer two take-homes of this trial, one similar and one slightly different than my colleague Dr. Kirtane.

• FAME-3 had a quite lax NI margin, it used a fair definition of MI, and it clearly showed CABG is the better strategy. If CABG is better at 1 year it will surely be even better at 3 and 5 years. So, yes, patients with multivessel disease should come off the table and CABG should be recommended.

• My second take home pertains to the lessons learned from the failure of FFR: I see FAME-3 as totally predictable because a neutral Martian view of FFR would be, how in the world did this procedure get accepted as standard operating procedure?

A 2019 editorial from David Brown and Andrea Soares in the International Journal of Cardiology describes the four fallacies of FFR:

• The first fallacy is that a focal obstruction is on the direct path to death and MI. We learn very early in our training that, in general, high-grade stenoses are less likely to rupture and cause MI and death than mildly obstructive younger lesions.

• The second fallacy is that microvascular disease is not relevant to patients with angina. Of course it is, and FFR measures only the epicardial stenoses.

• The third fallacy of FFR is that it improves outcomes through targeted lesion selection. One interpretation of the original FAME trial was that PCI reduced death and MI because fewer stents were placed in the FFR arm.

• The fourth fallacy of FFR is that it improves outcomes relative to optimal medical therapy (OMT). Recall that in FAME-2, FFR beat medical therapy, but the entire reason was higher rates of urgent revascularization. Well, FAME-2 was unblinded; patients in the non-stent arm knew they were untreated. Chris Rajkumar and colleagues call this subtraction anxiety, so what do you think happens when a patient with a blockage left untreated has 10 seconds of left pinky pain—yes, they go for a repeat revascularization.

In sum, FFR as a standard practice is a classic example of therapeutic fashion established on dubious grounds. The intent was beneficent—to make PCI better; but FAME-3 merely builds on all the previous less-than-robust evidence for FFR.

ISCHEMIA Trial Challenged

A group of interventional cardiologists, publishing in the interventional journal, JACC-Cardiovascular Interventions, have made an argument using observational data, that one of the most important clinical trials of this generation, the ISCHEMIA trial, does not apply to the majority of patients who get revascularization in the real world. Reading between the lines: ISCHEMIA is not practice-changing, and its neutral results only apply to a small proportion of patients.

Here are some quotes made by study first author Saurev Chatterjee in Sue Hughes’ excellent recap:

“While ISCHEMIA was a very well conducted trial, our results show that it only applies to about one third of stable CAD patients undergoing intervention in US clinical practice in the real world.

… ISCHEMIA did not really answer the question for 67% of stable IHD in current US practice.

We may be abIe to defer PCI in these patients, but we don’t know that from the ISCHEMIA trial, as these patients were not included in the trial,

We believe that patients who do not fall into the ISCHEMIA population need a nuanced individual approach, taking into account symptom burden and patient preferences.”

Recall that ISCHEMIA randomly assigned patients with stable CAD and moderate or severe ischemia on stress testing to initial invasive management with cardiac catheterization and revascularization plus OMT or to an initial conservative strategy with guideline-directed medical therapy alone. Patients with left main coronary artery stenosis (LMCA), left ventricular ejection fraction (LVEF) < 35, and end stage renal disease (ESRD) were excluded.

As I’ve covered on this podcast many times, the trial showed no statistical difference in death or MI between invasive vs conservative approaches and slightly more than two-thirds of patients with seriously positive stress tests and severe CAD did not even get an angiogram and still had similar rates of MI or death.

ISCHEMIA was done because the main criticism of prior trials — COURAGE and BARI-2D — that also found no benefit of PCI over and above OMT was that they randomized after catheterization and the scariest patients were not randomized. In other words, it was criticized on external validity grounds.

The question raised by the newest observational study is how applicable is ISCHEMIA? Using the National Cardiovascular Data Registry CathPCI Registry (NCDR) catheterization database, the investigators found that about 388,000 patients with stable CAD got PCI during the time ISCHEMIA was enrolling. But only one-third would have met inclusion criteria for ISCHEMIA and two-thirds would have been excluded.

The 388,000 patients were broken down into four groups:

• One-third who were ISCHEMIA-like.

• About 19% were high-risk based on LVD or LMCA or ESRD.

• About 17% who had PCI and would have been excluded from ISCHEMIA because they were too low risk

• And 31% who were unclassifiable because of lack of stress testing.

In other words, ISCHEMIA wasn’t that landmark because its external validity or applicability is so low. Thus, in the face of the third major trial showing that in stable CAD, initial revascularization does not reduce MI or death, we need not change our thinking on the clogged-pipe frame of stable CAD.

I could not disagree more. First, the direct problems with the methods of the study, then the more important criticism of the external validity argument.

The lead authors of ISCHEMIA wrote the editorial, which nicely outlines the unfairness of the four categories of patients in the observational analysis:

• First, you can’t count 19% of high-risk patients. Patients with LMCA and LV dysfunction were excluded from ISCHEMIA because we know previous trials have shown the benefit of revascularization over medical therapy.

• Second, you can’t count patients with ESRD, because ISCHEMIA enrolled these patients in ISCHEMIA-CKD study.

• Third, you shouldn’t be counting low-risk patients because it’s already been proven in COURAGE and BARI-2D that PCI did not reduce death or MI. So why are these patients getting PCI in the real world?

• Finally, the authors don’t have data on 31% of patients so they assume all would be excluded. That’s not fair either.

In their editorial, Drs Maron, Bangalore, and Hochman recalculate and find ISCHEMIA applies to about two-thirds, not one-third of patients who get PCI. I agree with them that the authors’ choices were dubious, and that ISCHEMIA remains highly applicable. Positive stress tests in my zip code go to the cath lab and in many/most cases, blockages get revascularized.

But the greater lesson here is what the authors are implying about external validity. The argument from the principal investigator seems to say that low external validity of ISCHEMIA is a reason not to abandon early angiography and intervention in most patients. This is backwards. Here is why. Trials always recruit and select for the best patients—the ones most likely to do well with an intervention. Trials represent a best-case scenario. I know of no procedure or drug that performs better in the real world than the rarified confines of a clinical trial. Every therapy does worse in the messy real world. So it goes with early revascularization. If early revascularization didn’t work in the best-case scenario of the ISCHEMIA trial, it would almost certainly do worse in the real-world, where patients have more co-morbidity, less intense follow-up, less medical adherence. If you say a nonsignificant trial has low external validity, this is an argument for doing less revascularization in the real world, not more. Because if it can’t win in a trial, it surely can’t win in the real-world.

Quantitative Flow Ratio

Earlier, I discussed the problems with FFR. What about a different way to assess the physiology of a coronary stenosis? At this year’s TCT conference, a group of researchers from Beijing reported on the use of something called quantitative flow ratio (QFR) assessment. QFR does not require use of specialized guide wire and is easily repeated multiple times during the procedure. Lancet published the FAVOR III China study.

This was an RCT comparing use of QFR (which isn’t widely available in US labs) to standard angiography in about 3800 patients with both stable and unstable coronary lesions. In the QFR group, QFR was measured in all coronary arteries with a lesion but PCI was performed only in lesions with a QFR of at least 0.80 or diameter stenosis of 90%. In the control arm, angiography was performed on the basis of visual inspection—old school. Patients and caregivers were blinded and the primary endpoint was MACE at 1 year.

The results clearly favored QFR:

• The composite of all-cause death, MI, or ischemia-driven revascularization, occurred in 5.8% of the QFR-guided group and 8.8% of the angiography-guided group (HR, 0.65; 95% CI, 0.51 – 0.83; P = .0004).

• Fewer MIs and fewer repeat procedures were the main drivers of the endpoint.

• BUT the pre-randomization plan was changed in 23.3% of patients with QFR and only 6.2% in the angiography group (P < .0001), mainly due to deferral of treatment of at least one vessel originally planned for PCI (19.6% vs 5.2%; P < .0001).

Translation: QFR led to substantially fewer PCIs being performed, less metal in the coronaries associated with fewer MACE. This is similar to the FAME-1 trial.

Both the editorialists and trial discussants pointed out that the problem with this strategy and study was that the new kid on the block, QFR, was not compared with the standard FFR. The problem with that argument is that the standard, FFR, doesn’t lie on strong evidentiary grounds. As a neutral Martian, fewer stents seem to be connected with better outcomes.

MI as a Surrogate

A team led by Dr. David Brown from Washington University in St. Louis, first author Kevin O’Fee, has published one of the most important papers of 2021. I wrote a column on this paper.

The question is, how good a surrogate marker of death or cardiovascular (CV) death is MI? What I mean by a surrogate is an outcome that is on the causal pathway of death. MI seems to fit that description. A treatment that reduces MI will likely reduce death. We need surrogates because over the course of a 2- to 5-year trial, not that many patients die. When event rates are low, you need huge trials, which are expensive.

When I trained, heart attacks were devastating. An occluded artery led to scar then heart failure (HF) or ventricular tachycardia (VT) or both. If you had a treatment that reduced MI, that treatment would surely lengthen life. But times change. Brown and his team wondered how well MI reduction correlated with survival.

They searched PubMed and RCTs over the decades that tested therapies for both MI reduction, CV death, and all-cause death. They found 144 RCTs and each trial acted as a data point. MI reduction measured in odds ratio was on the x-axis, death reduction on the y-axis and then they drew a correlation line.

• They found almost no correlation between MI reduction and death reduction.

• The type of trial (primary or secondary prevention), the date of the trial (pre- or post-2000) and the length of trial did not matter: nonfatal MI still did not come close to meeting the criteria for surrogacy for CV or all-cause death.

This tome is a huge finding. Trial after trial I have discussed over the years have had MI as an important endpoint. Often MI is a driver of a composite endpoint. But these authors have shown that we should be darn cautious about surrogate outcomes.

You might wonder why MI reduction does not serve as surrogate marker. Three reasons:

• First, we now detect smaller MIs and these smaller MIs are subjected to better therapies. As our ability to diagnosis and treat improves, MI becomes less devastating.

• Second, multimorbidity. As patients live longer and with more co-morbid conditions, ischemia represents a smaller proportion of their causes of death.

• Third: troponin assays increasingly pick up type 2 MIs, which are due to supply and demand mismatch. This differs from type 1 MIs where plaque rupture is the cause. Type 2 Mis are treated by treating the underlying cause, so therapies directed at plaque rupture will have no effect on the increasingly common type 2 MI.

Take a look at this study and my commentary. The finding that MI is not a reliable surrogate for death has huge implications

I have come to believe that evidence needs an expiration date. This study deserves attention because it reminds us to question prevailing dogma, changes the way we assign value to our therapies, and helps plan future trials.