Medtronic (NYSE:MDT) has yet to release the full results of its Symplicity HTN-3 renal denervation clinical trial, but that hasn’t stopped rival device makers, analysts, physicians and other stakeholders from alternately speculating and lamenting about what went wrong and whether the technology will survive.
One cardiologist seems to find himself consistently fighting the mob, initially with attempts to curb what he correctly predicted as unrealistic expectations about Medtronic’s HTN-3 trial, and now in trying to stem the backlash as hopes are dashed and aggressive development programs are checked.
Boston Scientific (NYSE:BSX) and St. Jude Medical (NYSE:STJ) have recently announced both optimism and hesitation toward their own renal denervation development programs, and industry insiders have alternately called for determination and trepidation in pursuing the technology.
Dr. Darrel Francis last year published a research paper that flew in the face of the prevailing enthusiasm about renal denervation, which had some people talking about sustained blood pressure reductions of up to 30 mmHg, the equivalent of taking about 3 hypertension drugs per day. His research highlighted deeply entrenched biases in earlier hypertension clinical trials, which he said produced dramatic results that weren’t likely to turn up in more robust trials such as HTN-3, which included both blinding and a sham control arm.
Now that his predictions have mostly come true, Francis finds himself playing precisely the role that he expected to fulfill when the time came: he’s defending renal denervation.
In an email interview with MassDevice.com, Francis explained why the much-lamented results of Medtronic’s HTN-3 trial are no death knell for renal denervation and explained that even a relatively small benefit from the technology could make a big difference for patients.
MassDevice: There has been a lot of shock and sorrow over Medtronic’s Symplicity HTN-3 clinical trial findings. Are you disappointed in the results?
Dr. Darrel Francis: Yes, but not with the news that Symplicity 3 did not hit its planned endpoint, which was more demanding than what is asked of drugs. My disappointment is that the Medtronic press release was so widely summarized as "it did not work". That is a travesty.
MassDevice: You predicted that the HTN-3 results would fall far short of the market’s expectations, but now that you’ve been proven right you’re defending renal denervation. Why are you still optimistic about the technology despite the trial outcome?
DF: I don’t think I am optimistic. I think I am relatively steady in focusing on reliable numbers. It feels like everyone else has been rushing from side to side in the small lifeboat we all share. I hope they do not capsize us.
The news release, presumably enforced by financial regulators, was that it did not meet the primary endpoint.
David Kandzari’s Methods paper is characteristically crystal clear in logic and mathematics: Symplicity-3 focuses on the incremental effect in the denervation arm beyond the control arm. They were planning while around them there were shouts of "29!" and "31!" like excited trading at a fish-market, so they understandably set the bar a little too high. They aimed to achieve a lower boundary of the 95% confidence interval better than 5 mmHg That is to say, if the study yielded "11 mmHg (95% CI 6 to 16 mmHg)" it would be a roaring success, while if it was "9 mmHg (95% CI 4 to 14 mmHg)" it would be an abject failure.
How can anyone have their expectation of passing the primary endpoint disappointed, without (a) actually knowing what the primary endpoint was, or (b) having a rational basis for forming an expectation?
So now you know the primary endpoint target. Meanwhile our CONVERGE paper shows how to set an expectation. Individual readers can be bolder than our paper by dismissing data that appear fraudulent, and updating the arithmetic. You would thereby predict between 5 and 10 mmHg. With a headline result in that range, it is quite likely that the lower 95% CI will dangle below 5 mmHg, i.e. that the trial will be technically negative.
The January Symplicity-3 news fitted that expectation perfectly.
[Editor’s note: The CONVERGE report examined the differences in reported patient outcomes in different types of hypertension clinical trials, looking at unblinded trials in which patients are monitored via in-office readings and comparing them against blinded studies and those that rely on ambulatory blood pressure monitors. The authors, including Dr. Francis, concluded that:
"Discrepancies in drug trials between office and ambulatory blood pressure reductions disappear once double-blinded placebo control is implemented. Renal denervation trials may also undergo similar evolution. We predict that as denervation trial designs gradually improve in bias-resistance, office and ambulatory pressure drops will converge. We predict that it is the office drops that will move to match the ambulatory drops, that is, not 30, but nearer 13."]
MassDevice: One of the HTN-3 principal investigators anonymously told financial analysts that renal denervation failed as an approach to hypertension and that it’s "highly unlikely" to proceed. What do you think of that?
DF: We should consider very seriously the views of people with such expertise and knowledge of the actual results. The only caveat is that they might have been hoping so strongly for 30 mmHg that anything smaller seems catastrophic and topples them into unwarranted despondency. I wish him a quick recovery and continued high-quality research in the fine tradition of Symplicity-3.
MassDevice: Some rival device makers have speculated that Medtronic’s 1st-generation renal denervation system may be at fault and that their respective devices are sufficiently different that they’d perform differently in a similar trial. What do you think?
DF: They would say that, wouldn’t they?
If anyone takes that seriously perhaps you could tell them about my side-venture, the Francis-Benz supercar? In my marketing literature so far I have been pointing out how similar it is to the Mercedes-Benz, but now I would like to point out that my product has 4 times as many wheels if you look closely, and goes 3 times faster.
Of course I haven’t actually measured speed in the same way the Mercedes people measure it, with lasers and whatnot, as that is unethical and not practical. Instead I have just asked a bunch of people in a bar to watch it drive past and say how fast it goes, and they replied unanimously "Very fast! Much faster than a Mercedes, whose manufacturers didn’t even buy us a round of drinks, unlike you."
DF: I am delighted. It is a great time to rethink what information we really want from a trial and how to get it without bias.
Readers might enjoy a paper explaining the main traps to avoid. It was put together by world authorities in cardiovascular clinical trials and interventions. It was an honor that they allowed me to distill their years of insight, and illustrate it with simple figures.
We should also not divert into pointless subgroup analyses looking for responders and non-responders, using office BP data. Mathematics can sometimes warn you in advance not to do things. Let us learn the lesson from cardiac resynchronization therapy, where ignoring that warning led to madness.
MassDevice: You had predicted that there would be a backlash when HTN-3 failed to support the excess hype for renal denervation. Have things gone the way you’d thought?
DF: No, I guessed wrong. I thought people would say "Oh, it is only 5-10 mmHg! First let’s get a posse together, to tar and feather the people who said it was 30, then let’s move on with more reliable research into this intervention."
Instead some people are saying "It doesn’t work" (implying zero effect). And others are outrageously casting aspersions on the Symplicity-3 people and the Symplicity denervation algorithm, which is the only one with many thousands of patients of safety history. I expected better behavior.
MassDevice: Can renal denervation survive if the treatment effect remains small?
DF: Sorry, dear friends, the world of treatment of hypertension is constructed of small effects, carefully stacked up one on another.
Were you thinking of all those shiny drug-company give-away leaflets talking about 15-20 mmHg effects? Do you believe them to be true? And if not, why have such an expectation of a new therapy?
A lasting effect of 5-10 mmHg from a single procedure is a huge benefit to patients, because it is similar to what you get from an additional drug that you have to keep taking daily. Patients find that difficult in real life. That is why they have resistant hypertension – because they are resistant to regularly taking any more (or sometimes any) drugs.
MassDevice: Dozens of other trials have reported much stronger effects for patients treated with renal denervation. What makes HTN-3 different?
DF: Because the dozens of trials are unblinded and people cannot help cheating because normal clinical practice depends on cheating. It is so ingrained that we don’t even call it cheating, we call it "using clinical judgment", or "using the most clinically representative value."
I have tested thousands of clinicians with a multiple-choice question presented in a lecture, and without exception they vote for cheating, and then laugh afterwards when it is pointed out to them what they have done. (I do not criticize their choice; I would do the same. The difference with me is that I recognize it as cheating).
Quality trumps quantity, when it comes to bias. Noise is different, doing large or numerous studies helps suppress it. But bias unfortunately gets accentuated when you just collect more data – we have written a whole paper on this depressing phenomenon.
MassDevice: Hypertension specialists have suggested, given the potentially small effect of renal denervation, that patients should simply take the 1 additional drug.
DF: Yes, if they want to. Patient preference should always come first.
But if you are prescribing more and more drugs and the pressure is not falling, then ask yourself how this can be.
Think of your patients in the emergency room or catheter lab. When you give them intravenous drugs that lower blood pressure, for example nitrates, how often does the blood pressure fail to drop? When it does fail to drop, what is the most likely explanation? Do you write up the patient as a rare genetic mutant? Personally, I 1st check the tubing, and look for a puddle on the floor.
Renal denervation is for patients who prefer it to actually taking more medications.
MassDevice: If, as you say, the HTN-3 press release is nothing to wring ones hands about, why all the commotion?
DF: People like to get hysterical. That is why people watch thrillers and go on roller coasters. That would explain why the most obviously incorrect claims from the early studies were the most frequently repeated by people who had not thought about the study designs.
This month’s amusement seems to be to arbitrarily proclaim it doesn’t work and compete in depths of misery one can sink to. At ACC 2014 the mood may swing up again.
MassDevice: What led you to conduct your hypertension clinical trial analysis?
DF: Credit goes to the far-sighted wisdom of Gregg Stone’s [Transcatheter Cardiovascular Therapeutics conference]. They invited me, for no particular reason, to attempt a seemingly futile debate against denervation for blood pressure in TCT 2012. In doing so, I pointed out that we had the effect size all wrong and it would be much less than what was then claimed.
People in the audience afterwards told me that my talk was "unusual" (a euphemism for bizarre) and enjoyable but it must be wrong because obviously the effect is 30 mmHg. That’s when I knew a published analysis would be useful for people to read.
From the conference hall of TCT I telephoned back to the U.K. to enlist the cleverest and hardest-working people I knew to conduct a formal analysis: James Howard and Alex Nowbar. My role was to "manage" them, i.e. take credit for their analytical genius.
MassDevice: Now that you’ve flexed your premonitory muscles with HTN-3, do you have any more mathematical analyses up your sleeve?
DF: We published last year that the effect of renal denervation on office blood pressure was genuine but overstated by a factor of approximately 3. That claim was variously described as untrue or unimportant by commentators. At ACC you will be able to check who was right.
Our next analysis is on the exciting science of bone marrow stem cell therapy for heart failure. Our DAMASCENE collaborative, which has been examining this closely, believes the positive results are a result of unintentional error. However, compared to the mild challenge in denervation, for cell therapy the situation is catastrophic.
As trial design and reporting improves, I predict that we will find that, contrary to numerous peer-reviewed publications, none of the currently-vaunted protocols for improving ventricular function have any efficacy at all. We hope to be able to publish our main analysis soon but, as you might imagine, there have been some legal threats.
Prof. Francis has received consultancy honoraria from Medtronic, but has no involvement in Symplicity-HTN-3 nor knowledge of its results other than from the press release and his mathematical analysis of the publicly available denervation data.
Editor’s note: This interview has been lightly edited for style and clarity.