News
CHICAGO – Something smells fishy about the results from Russia and Georgia in the TOPCAT trial, according to a study reappraisal conducted by the trial’s leaders.
TOPCAT (Treatment of Preserved Cardiac Function Heart Failure with an Aldosterone Antagonist) was a randomized, double-blind, placebo-controlled trial of spironolactone for the treatment of heart failure with preserved ejection fraction (HFpEF) in 3,445 patients in six countries. The primary outcome was negative, as presented at last year’s AHA scientific sessions and later published (N. Engl. J. Med. 2014;370:1383-92). However, a new post hoc analysis casts doubt on the validity of the results reported from Russia and Georgia, countries that contributed 49% of TOPCAT participants, Dr. Marc A. Pfeffer reported at the American Heart Association scientific sessions.
Dr. Marc A. Pfeffer
The patient outcomes reported from Russia and Georgia were spectacularly at odds with those reported from the United States, Canada, Argentina, and Brazil. Upon careful scrutiny, it appears likely that many Russian and Georgian patients either did not actually have HFpEF or were not taking their spironolactone. And if the results from the two Eastern European countries are put aside, then spironolactone markedly reduced the rate of the primary composite outcome – cardiovascular death or hospitalization for management of heart failure – in the 1,767 study participants in the Americas.
Indeed, the primary composite outcome occurred in 27.3% of spironolactone-treated subjects in the Western Hemisphere during a mean 3.3 years of follow-up, for an event rate of 10.4 per 100 patient-years, compared with rates of 31.8% and 12.6 per 100 patient-years in placebo-treated controls. That translates to a highly significant, 18% relative risk reduction (P = .026) in the primary outcome in spironolactone-treated HFpEF patients in the Americas. Cardiovascular mortality was reduced by 26% and heart failure hospitalization was reduced by 18%, according to Dr. Pfeffer, professor of medicine at Harvard University, Boston.
A post hoc analysis such as this would ordinarily be viewed as hypothesis-generating and nondefinitive. But this is a special situation, according to the cardiologist, who noted that guidelines offer no recommendations for the treatment of HFpEF, which now accounts for roughly 50% of all cases of heart failure.
“HFpEF is a growing part of the heart failure syndrome; it’s a frustrating part of the heart failure syndrome. And if we can improve the prognosis of the 40%-50% of people with symptomatic HFpEF by stating that our observation in the Americas was that spironolactone was associated with reduced cardiovascular deaths as well as hospitalizations for heart failure, then this should be taken into account. Since we don’t have other things we can do for these patients, I bring this to your attention,” Dr. Pfeffer said.
Among the major tip-offs that the Russian/Georgian TOPCAT data were dodgy was the post hoc finding that all-cause mortality, irrespective of treatment, was 21.8% in the Americas but a mere 8.4% in Eastern Europe.
“When I look at anyone’s clinical trial, if I want to ask about the severity of illness, I go to all-cause mortality. And here it was markedly different,” he observed.
Indeed, life-table analyses showed that the Russian/Georgian HFpEF subjects had a life expectancy typical of the region’s general population, whereas death rates for HFpEF enrollees from the Americas were several-fold higher than expected for age- and gender-matched controls.
Discussant Dr. Judith S. Hochman issued the usual caveats about the hazards of drawing conclusions from post hoc analyses of overall negative clinical trials, but then went on to agree with Dr. Pfeffer’s conclusions.
“I conclude, as does Dr. Pfeffer, that it’s reasonable to try mineralocorticoid receptor antagonists for symptomatic HFpEF patients with anticipated risks similar to those enrolled in the Americas. ... This is a growing condition and we have no other treatments,” said Dr. Hochman, professor and associate director of cardiology at New York University.
A key factor in her thinking was the mechanistic plausibility of the differential subgroup treatment effects, she added. For example, hyperkalemia – a well-known side effect of spironolactone – was 3.5-fold more common in patients randomized to spironolactone than in placebo patients in the Americas, as would be expected; but hyperkalemia was equally infrequent in both treatment arms in Russia and Georgia. Conversely, hypokalemia was 49% less likely with spironolactone than placebo in the Americas, yet there was no difference in the incidence of this side effect according to treatment status in Eastern Europe.
Moreover, systolic blood pressure fell by a placebo-subtracted 4.2 mm Hg at 1 year in spironolactone-treated patients in the Americas but was unchanged in Russian and Georgian patients. And while doubling of creatinine levels to above normal range was 60% more common with spironolactone than placebo in the Americas, rates were identical in the two treatment arms in Russia and Georgia.