An Interview With Dr Pamela Douglas

; Pamela S. Douglas, MD


October 22, 2015

This feature requires the newest version of Flash. You can download it here.

Editor's Note:
Dr Pamela Douglas—Ursula Geller Professor of Research in Cardiovascular Disease at the Duke Clinical Research Institute and principal investigator for the PROMISE trial, which compared CT angiography vs functional testing (exercise ECG, nuclear stress, stress echocardiography) in patients with chest pain, and of the PLATFORM trial, which added virtual fractional flow reserve (FFR) to CT angiography in a similar patient population—is interviewed at Transcatheter Cardiovascular Therapeutics (TCT).

Robert A. Harrington, MD: Hi. I'm Bob Harrington, from Stanford University. I'm here at TCT in San Francisco, where I saw my friend and colleague, Mark Hlatky, presenting some economic and quality-of-life data from a clinical study that had been led by Pam Douglas, dealing with FFR measurements on CT scans.

I'm pleased to be with Pam today. Pam is professor of medicine and a cardiologist at Duke University and runs the imaging study group at the Duke Clinical Research Institute. Pam, thanks for joining us here.

Pamela S. Douglas, MD: You're welcome. It's a pleasure to be here.

Noninvasive FFR

Dr Harrington: Let's just dive right into it, but before we go too deep, let me ask: Can you describe how we utilize cardiac CT to derive FFR?

Dr Douglas: FFR, as you well know, is a time-honored technique that is usually thought of as being done only in the cath lab. It's something that we do with a flow wire, and we look at pressure differentials under conditions of hyperemia.

Dr Harrington: Which typically involves adenosine infusions, and maximum hyperemia measure; you have a wire across the lesion, so it's an invasive procedure.

Dr Douglas: Exactly—and not a simple one, either. It's been proven to reduce the need for revascularization and improve outcomes when revascularizations are performed only when there's a hemodynamically significant lesion. We know that from the FAME studies,[1,2] as well as others.

Dr Harrington: Basically, it's taking anatomical information and adding physiologic information on top of it, and the studies have suggested thus far that the addition of physiology actually matters. Is that fair?

Dr Douglas: That's exactly right, and it's one of the few things that we have in medicine where you actually do better by the patient (improve outcomes) and save money because you do fewer interventions.

Dr Harrington: As our mutual friend Dan Mark would say, that is the unusual situation in the cost-effectiveness literature.

Dr Douglas: That's where we all want to be, though, so we keep trying.

Dr Harrington: That's the invasive FFR, but now we have the ability to do noninvasive FFR.

Dr Douglas: Noninvasive FFR takes a standard set of CT angiography images acquired in the usual way: It segments those images and creates a 3D model of them. Then, through computational fluid dynamics (a complex data analysis that requires supercomputers), it simulates conditions of hyperemia in that heart and with that amount of myocardium that's perfused by the vessel to come up with pressure and flow data that mimic the invasive data. These data are presented in exactly the same way, such that an FFR CT ≤ 0.8 is likely to indicate ischemia, whereas a higher one (> 0.8) is unlikely to be hemodynamically significant.

Dr Harrington: A couple of things to note—one of which is that I'm from Stanford, and the technology was originally developed by some Stanford faculty members. I have no interest in the company.

Second, your point is that this is computational; this is modeling. How good is it compared with the invasive gold standard?

Dr Douglas: That's a great question, and it's been validated in three studies[3,4,5] compared with invasive FFR. It does quite well, but it's not perfect. It has sensitivities and specificities in the 80%-85% range, which is about as good as we get from most of our noninvasive diagnostic tests.

Dr Harrington: In fact, better than some.

Dr Douglas: It's considerably better than some. When you look at all the research that's been done with our usual stress testing techniques—including stress nuclear, which we feel is the most accurate—only MR comes close, but it's still not as good at predicting invasive FFR as the noninvasive FFR.

What's also important is that it doesn't seem to be bothered by calcification. I know one of the big problems with CT is that once you get a significant calcium load in those coronary arteries, there's so much blooming artifact that you can't tell what's going on in the lumen. The modeling techniques that are performed are perturbed a little, but not very much, by the presence of calcium, so even when you think you've had an inadequate CT scan, you can usually get a pretty good FFR CT.

Dr Harrington: Which, as our viewers will know, is one of the critical things about CT angiography—because if you have a normal CT angiogram, the negative predictive value is really high. It's knowing what to do with people with intermediate lesions that we have struggled with, and so the addition of FFR could be really useful from a clinical perspective.

The early studies (as I understand this field, Pam) really concentrated on the technical aspects of trying to match up the noninvasive technology with the invasive gold standard.

Dr Douglas: Right, and validating it.

Dr Harrington: And validating it. Now we've moved to a different set of studies, which is trying to say how might we use this in the context of clinical decision-making?

Dr Douglas: Not just how might we use it, but what impact does it have? I think we know how we want to use it, but does it matter? Does it change what we do? Does it give us better information? Does it help us to risk-stratify better? Does it help us to direct care better? Is it a better, more efficient use of invasive techniques?

Dr Harrington: Which is really interesting. We're seeing this shift in imaging research, from concentrating on the engineering and technical aspects of things to an era where we're now trying to understand how to use it. With your leadership, last spring we saw the results of the PROMISE trial[6]; we're starting to have some insight into not just the technical performance but also the clinical performance.

Dr Douglas: We're starting to focus on that diagnostic population a little bit more. As cardiologists, we're used to dealing with established disease, and we're starting to understand that what we do in the cath lab is critically dependent on what happens before that, so we can't just leave that to somebody else.

Dr Harrington: We're also in an era of value. We can't say, "Let's get these four tests on everybody," and then in almost a heuristic way put it all together and make a decision. We've got to have evidence to guide our diagnostic pathway to utilize the resources we have efficiently.

Dr Douglas: Right. These are tens of millions of patients every year.

PLATFORM: Triaging Chest Pain Patients

Dr Harrington: These are big public health questions; this is not a thought exercise. This particular set of studies dealt with people with suspected ischemic heart disease or stable ischemic heart disease; this is a big population. What are the PLATFORM studies?

Dr Douglas: The PLATFORM study[7] (I'll just focus on the primary endpoint arm of the study) was an observational study conducted in Europe

Dr Harrington: Meaning nonrandomized.

Dr Douglas: Meaning nonrandomized, conducted in Europe in patients whose physicians planned to send them to the cath lab. About one half of them had had noninvasive tests, and about one half of them were going directly to cath lab—that was the decision-making.

Dr Harrington: If I think it's angina and they're on medication, I might well send them to the cath lab.

Dr Douglas: Just do one test. If you need to fix it, you can fix it—that's the rationale for that strategy. We took two sequential cohorts. One planned to go to the cath lab– 100% went to the cath lab. The other group, we sent them to the CT suite first.

Dr Harrington: Because the doctors had not yet decided to send them to the cath lab?

Dr Douglas: No, the decision had already been made that they would go to the cath lab. Once somebody put their penny down on what they wanted to do, we said, before you do that, let's see what would happen if instead of your usual care—which is catheterization—we do a CT scan. What would change in your processes of care?

All of those patients had CT scans, and on the basis of those results, catheterizations were canceled in 61%. The CT found that 70% had lesions ≥ 30%, so 30% didn't undergo any further tests.

Dr Harrington: You actually gave the information to the providers and let them make a decision. You did not tell them what to do; they made a decision. You provided them this information, and they didn't take them to the cath lab.

Dr Douglas: They said, I don't need to go to the cath lab. Either I don't have any lesion or it's not hemodynamically significant, and I don't need to worry about this person.

Dr Harrington: That's extraordinary.

Dr Douglas: It's a big number, 61%. We ended up with the same number of revascularizations in the two groups. I don't think we deferred a lot of people who needed revascularization. If you had 61% fewer revascularizations, you'd be worried about who you missed.

In the direct to cath cohort, the qualitative coronary angiograpy , which is a high bar (independent QCA or an invasive FFR) found that 73% of the patients in that cohort did not have obstructive disease. That's a lot. By the site read, it was a little lower and very consistent with what we published previously from the National Cardiovascular Data Registry NCDR cath percutaneous coronary intervention (PCI). In the CT-guided arm, it was 12%. That's dramatic—73% vs 12%. A huge effect that held true in every subgroup we could look at: men, women, diabetics, nondiabetics, low or high pretest probability.

Dr Harrington: Did age have any influence?

Dr Douglas: No difference by age. No difference in the patients who had a functional test beforehand and those who didn't. It was consistent in a propensity-matched cohort to help address the problem of not being a randomized study.

Dr Harrington: That is the clinical observation. How do the economics attach to that?

Dr Douglas: There's one other piece to the clinical observation, and that's the safety piece. There were just two events in the whole study, and as it turns out, prognostically this was a pretty low-risk group. The events were a periprocedural myocardial infarction (MI) in a patient undergoing a needed intervention (proven by both hemodynamic and anatomical information), and an unstable angina event in a patient with left main triple-vessel disease who was awaiting coronary artery bypass grafting (CABG) in the UK system—both things that we know are part of the deal with coronary disease.

The economic results[8] showed that it cost $10,700 in the usual care group to take care of these patients, and $7300 in the FFR CT group because there were so many fewer catheterizations.

Dr Harrington: Does the $7300 include the cost of the CT?

Dr Douglas: It includes the cost of the CT, but it doesn't include the cost of the FFR CT, because there's no Medicare cost for that yet.

Dr Harrington: You did some modeling on this about varying the cost, as I understand it, of the FFR.

Dr Douglas: We modeled it as multiples of CT cost, and you had to get up to 20 times the cost of the CT before the costs for both groups became equal.

Dr Harrington: These data were not highly sensitive to minor perturbations of change here?

Dr Douglas: No. Even if you assumed the cost of the FFR CT was seven times the cost of the CT (I don't know who would set it at that), you still had a 20% reduction of costs.

Dr Harrington: Better clinical outcomes, less cost. That's the cost-effectiveness quadrant that Dan Mark always tells me you almost never end up in, because cost-effectiveness analyses take into consideration the added value of a higher-cost procedure to calculate a quality-adjusted life-year. In some ways, you don't even have to do that here, because you're spending less money.

Dr Douglas: You're dominant. Better outcomes, less cost.

Dr Harrington: Tell me about the quality-of-life piece.

Dr Douglas: The quality-of-life piece was a little different. We did not see a difference in the invasive arm. But in the planned noninvasive group, whom I haven't talked about up to now (to make it a little bit easier to understand), quality of life was better with FFR CT. It was a small population, not a big population. We're a little puzzled as to why that could be. We didn't dissect out the reasons. Very few of them went to the cath lab in usual care arm or FFR CT arm—very few interventions.

Dr Harrington: Any insights from the medication patterns or anything like that?

Dr Douglas: No. Maybe people thought they were having a fancy new test. They were having something better, and so they felt better about the care they were receiving. That's possible.

PROMISE/PLATFORM: Fitting It All Together

Dr Harrington: Still work to do on that. Fit this now into what you learned from the PROMISE study back at the American College of Cardiology (ACC) meetings. How do you recommend that we incorporate these results into the clinical care patients?

Dr Douglas: The story from PROMISE and PLATFORM together is really interesting. As a quick review, PROMISE was a 10,000-patient, National Institutes of Health (NIH)-funded trial that took the same patient population as PLATFORM—stable, suspected coronary artery disease who required testing—and randomized them to either regular CT or a functional test. Then we left everything else to usual care, to be very pragmatic. We found no difference in our primary endpoint, which is clinical outcomes, because the event rate was just 1% per year.

Dr Harrington: Hard clinical outcomes.

Dr Douglas: Hard clinical outcomes—just 1% per year, which was a brand-new finding in this population, and very important. It tells us that we need to look at processes of care and more intermediate endpoints than only the hard endpoints to really understand what we're doing with the chest pain evaluation.

In that study, there was no difference in the hard endpoints—death, MI, unstable angina—but we saw a difference in processes of care. We looked at the finding of number of catheterizations without obstructive disease (the primary endpoint in PLATFORM), and it was reduced by about 25% in the CT arm vs the usual care arm.

Dr Harrington: An absolute reduction of 25%.

Dr Douglas: Yes, absolute, and according to strategy—not just in the cath lab, but according to strategy.

Dr Harrington: Which is an important observation for the viewers. That's 25 catheterizations per 100 patients, just to put it in context.

Dr Douglas: The criticism of PROMISE was that the event rate didn't improve. The likelihood of going to the cath lab went up by about 50%, and the likelihood of revascularization went up about twofold; it doubled in the CT arm. You now have a trial with more accurate triage to the cath lab, which we like, but more procedures and no change in event rates. It's a little hard to put that together.

There are some good things and some bad things in there. We certainly don't want to do more procedures without benefit. Dan Mark did the cost analysis,[9] and it was even. There's no difference in cost in the two arms.

It's left us with a little bit of a puzzle. We've got a low-risk group; we know that now. CT didn't do any worse than functional testing, but it seemed to have the signal for extra procedures.

Dr Harrington: Extra resource usage.

Dr Douglas: Well, the resource use was the same or was not different in terms of a formal analysis, but extra procedures were done even though those procedures seemed to be more appropriate (higher-yield revascularization procedures). Not a clear picture. We took the cohorts of patients in PROMISE, who had both a CT and invasive angiogram within 90 days, and then we ran CT FFRs on those patients. It's just about 200 patients. We looked at what incremental information the FFR gave us over the regular, ordinary CT.

Dr Harrington: Now to be clear, that information was not available to the investigators during the study.

Dr Douglas: No; this is a post hoc analysis. We did it after we presented the main results of the trial. A couple of things: One, it better predicted the need for revascularization events. The hazard ratio is about twice the size, and it would have triaged people out of the cath lab who didn't need to be there.

Dr Harrington: So you would have taken away some of those catheterizations that we talked about earlier and even further increased the yield of the catheterization—the yield being interpreted as appropriate revascularization.

Dr Douglas: Right. And the likelihood of going to the cath lab and needing a revascularization, so it was both a better risk stratification and a better guide to care.

Dr Harrington: It fits well with what you've just told us from PLATFORM.

Dr Douglas: Very consistent, and it fits well with what we know about invasive FFR. It's exactly the kind of effect that we would expect to see, knowing what invasive FFR does.

Dr Harrington: As you know, the guidelines are beginning to incorporate FFR into the decision-making in the cath lab. We're working on the ISCHEMIA trial with Judy Hochman and Dave Maron, and we have recommendations about using invasive FFR in the trial. There is this movement toward combining anatomy with physiology, which I think it going to result in more appropriate revascularization procedures. At least, that's the hope.

Dr Douglas: One other piece in the PROMISE data is to take the catheterization cancellation rate in PLATFORM (61% of catheterizations cancelled) and apply that to the catheterization rate in PROMISE—which was problematic, because it was higher. We would reduce the catheterization rate from 12.2% to 4.7% (dramatically, obviously), and the functional arm rate was 8.1%. Now the catheterization rate, instead of being 50% higher, is about 40% lower. Now we would have fewer procedures and hopefully the same outcomes. We didn't test that prospectively, but that's what we would expect it to do.

Dr Harrington: That seems [to be] where we're going. Tell me, in the closing minute: Do we have noninvasive FFR available for use in clinical practice?

Dr Douglas: It's been approved by Medicare, actually back in the fall of 2014. You can't do it at home because it requires a supercomputer to do this, so it's a model where you upload the scans and you get it back in a matter of hours from HeartFlow (Redwood City, California).

Dr Harrington: From the population that we're studying of largely low-risk—maybe bordering up to moderate-risk, but largely low-risk people—that delay does not seem at all out of the ordinary.

Dr Douglas: No, and we looked at the timing of our noninvasive tests in PROMISE, and it was 7-10 days. An extra 6 hours on the end of that is not going to bother anyone.

Dr Harrington: It's not going to bother the patient or the practitioner, and the patient should have some peace of mind that they're getting incrementally more valuable pieces of information.

Dr Douglas: Absolutely.

Dr Harrington: Pam, thanks for joining us here and sharing this very interesting substudy from the PLATFORM study. Good luck with additional analyses of PROMISE, and maybe we can catch up at the American Heart Association (AHA) and hear about what you're presenting there.

Dr Douglas: Great. Thank you very much.

Dr Harrington: Thank you for listening, and hope you found this enjoyable and informative.

Disclosures: Pamela S. Douglas, MD, has disclosed the following relevant financial relationships:
Received research grant from: HeartFlow, GE Healthcare