April 9th, 2014
Putting TOPCAT in Perspective
CardioExchange’s John Ryan poses questions about the TOPCAT trial to its first two authors, Bertram Pitt and Marc Pfeffer, and to regular CardioExchange contributor Clyde Yancy. The study is published in The New England Journal of Medicine.
Researchers randomly assigned 3445 patients with symptomatic heart failure and a left ventricular ejection fraction >=45% to receive either spironolactone (15 to 45 mg daily) or placebo. At a mean follow-up of 3 years, incidence of the primary outcome — a composite of death from cardiovascular causes, aborted cardiac arrest, or hospitalization for heart failure — was statistically similar in the spironolactone and placebo groups (18.6% and 20.4%, respectively; hazard ratio, 0.89; 95% CI, 0.77–1.04; P=0.14), although the incidence of hospitalization for heart failure was significantly lower with spironolactone than with placebo (12.0% vs. 14.2%; HR, 0.83; 95% CI, 0.69–0.99; P=0.04). The incidences of all-cause mortality, all-cause hospitalization, MI, and stroke were similar in the two groups.
THE EXPERT PERSPECTIVES
John Ryan: Are you surprised that spironolactone failed to reduce incidence of the primary outcome?
Pitt & Pfeffer: We were surprised and disappointed by the result on the primary outcome. However, we did find a significant effect on one component of the primary outcome: hospitalization for heat failure.
Clyde Yancy: We should all be surprised because the hypotheses studied are sound, the conduct of the trial was exceptional, and the consistency of the efficacy of mineralocorticoid antagonism (MRA) has been unwavering. We ought to refrain from taking a binary approach to TOPCAT and simplifying the take-home message as “positive” (embrace the data) or “negative” (dismiss the trial). A much better approach is to declare that this was an informative trial that adds to our understanding of heart failure with preserved ejection fraction (HFpEF). In effect, we must respect the results, but we must also question everything in these data — especially the variation of disease as a function of geography — and notably, we should not hesitate to try again. The condition of HFpEF is of such heft now that it merits our most diligent efforts to discover a treatment strategy that modifies the natural history.
John Ryan: In the end, is the study underpowered? Many would consider the statistically nonsignificant difference to be clinically significant (NNT of about 50). Can we really conclude that spironolactone is ineffective in this group?
Pitt & Pfeffer: I do not believe the problem was overall underpowering. The problem, as I see it, is that about half of the patients (those from Russia and the Republic of Georgia) had a placebo event rate that was not compatible with what we understand from prior trials of HFpEF. Their placebo-group event rate is compatible with a risk factor such as hypertension without HF. Given the low event rate in Russia and Georgia, we were certainly underpowered in these countries. However, if you look at the results in the Americas (Canada, U.S., Argentina, Brazil), where the placebo-group event rate is compatible with that in prior HFpEF studies, spironolactone significantly reduced the primary outcome and its two major components: CV mortality and hospitalization for HF. However, this post hoc analysis is open to debate.
In the subset of patients randomized on the basis of elevated BNP, rather than a history of hospitalization (a major component of which was HF), spironolactone also resulted in a significant reduction in the primary endpoint. However, as we note in the paper, this was only one of several subgroups analyzed; therefore, the result is subject to chance.
Clyde Yancy: Is any clinical trial truly powered adequately? We make our best a priori approximation of power, but event rates are entirely predicated on the enrolled patient population and the comparative care in the control population. Ultimately, the signal here was simply not robust enough — but it was not absent. The HF hospitalization data are reasonable, and even more so when considered both in the North America cohort and again in the group stratified according to an elevated BNP. Moreover, the significant geographic variation (i.e., the Russian and Georgian cohorts) really speaks to a potential clinically important but not statistically significant advantage. Reducing heart failure hospitalizations, especially in HFpEF, is a good thing. Thus, in the interpretation of these data, perhaps we steer away from our purist inclinations and become more pragmatic. The clinical answer to be gleaned from these data is not a “no” vote for MRAs in HFpEF, but rather a reasonableness assessment that considers the safety of this approach with the likelihood of a favorable impact on HF hospitalizations. It’s not the home run we sought, and it won’t make the guidelines for heart failure today, but perhaps it’s good enough—at least for now. We’d be foolish to dismiss this approach altogether. In my own practice, I have and will continue to use MRAs in this setting.
John Ryan: The discussion mentions “the clinical challenges in diagnosing heart failure with a preserved ejection fraction.” What are those clinical challenges, and how can we overcome them when designing a clinical trial?
Pitt & Pfeffer: As we point out, the criteria for the diagnosis of HFpEF are rather subjective. It is likely that a number of patients with shortness of breath or dyspnea from causes other than HFpEF (e.g., obesity, COPD) may have been included. The baseline echo findings make it obvious that although many patients had mild diastolic dysfunction (a nonspecific finding that can occur in patients with hypertension without HF), others did not have any structural evidence of HF. I believe that the presence of an elevated BNP level helped to identify a subset of patients whose symptoms were due to HFpEF and who were at increased CV risk. As mentioned above, in that subset of patients, spironolactone seemed to work.
I believe we will have to let readers reach their own conclusions about whether spironolactone has an appropriate role in their patients with HFpEF. My personal view, for what it is worth, is that in patients with HFpEF and increased CV risk (e.g., those with elevated BNP or NT-proBNP), spironolactone is effective — and I would use it in my patients.
Clyde Yancy: This, I think, is the crux of the issue. For lots of reasons, we are inclined to be “lumpers” and think of diseases and patient cohorts in big buckets (e.g., hypertension, ACS, atrial fibrillation) — but for HFpEF (much like acute or hospitalized heart failure), we may need to be “splitters” and respect the very important within-group differences. How you get to HFpEF matters. We should let new science dictate how best to partition this patient population and then align future candidate therapies with targeted phenotypes.
Share your thoughts on the TOPCAT trial.