Spiriva and heart attack risk: new safety kerfuffle

Last year, PulmCCM reported on the TIOSPIR safety trial comparing the Spiriva dry-powder HandiHaler against the Respimat mist-delivery device. TIOSPIR showed no difference in all-cause mortality or composite cardiovascular risk endpoints between either Spiriva preparation. But this week, a group of drug safety researchers report their granular analysis of TIOSPIR data shows the Respimat device (available in Europe) is associated with a three-fold increased risk of heart attack compared to the HandiHaler (the U.S. device).

Boehringer Ingelheim’s tiotropium (Spiriva) has been a mainstay of treatment for chronic obstructive pulmonary disease (COPD) since its FDA approval in 2004. Spiriva improves daily shortness of breath, and also prevents exacerbations of COPD.

Like other blockbuster drugs, Spiriva has also been dogged periodically by questions of safety. Observational postmarketing studies raised questions about increased cardiovascular risk (e.g., heart attacks) with tiotropium. The FDA closely analyzed the pooled data from randomized trials, especially UPLIFT (which should be less prone to bias) and concluded that overall, Spiriva did not cause cardiovascular events.

Questions lingered about a potential increased risk from the Respimat relative to the HandiHaler, but the massive TIOSPIR (17,000 patients) trial had seemed to put this question to rest.

But after examining the data, Yoon K. Loke, Sonal Singh and Curt Furberg wrote a letter to the editors of the New England Journal of Medicine, pointing out that patients using the tiotropium Respimat (mist; European) were ~3.5 times more likely to have fatal heart attacks (p=0.04), compared to those using the HandiHaler (dry powder; U.S.). There was a trend toward a 37% increased frequency of any heart attack (fatal or nonfatal) in the Respimat arm (p=0.05). This possible signal was smaller in the lower-dose Respimat (2.5 vs 5.0 μg). This was not new information: the absolute event rates are in tables 2 and 4 of the original article. However, the relative risks and statistical significance for fatal myocardial infarction were not included in the article, and coming from these noted drug safety researchers and advocates, it prompted a publication and formal response.

TIOSPIR randomized 17,135 COPD patients to double-blind (double-dummy) Respimat (at 2.5 or 5 μg ) or Handihaler (18 μg) tiotropium regimens. It showed noninferiority of both Respimat groups for all-cause mortality relative to HandiHaler and no difference in the number of overall cardiovascular events (examined in a composite endpoint including stroke, transient ischemic attack, myocardial infarction, or sudden cardiac death).

That’s a lot of convincing statistical power, but Loke et al make the case that “focusing on overall mortality or composite cardiovascular end points may mask differences in cause-specific mortality, such as from myocardial infarction.”

Robert Wise, the lead TIOSPIR author, counters that “because of the multiple comparisons among the three treatment groups and the several enumerated causes of death, it seems possible that this is a spurious finding.” Also, “because overall mortality was not increased with the use of Respimat, it would be necessary to conclude that any increase in mortality from one cause of death was offset by a reduction in other causes of death.”

A separate letter to the editor by Katia M.C. Verhamme et al suggested that chronic kidney disease could have conferred the cardiovascular risk associated observationally with the Respimat. Tiotropium is cleared by the kidneys; patients with significant renal impairment were excluded from TIOPSIR, precluding any conclusions there.

Considering things in absolute terms, there were only 3 fatal heart attacks (adjudicated) in the HandiHaler arm, and ~10 in each of the Respimat arms (out of ~400 deaths among ~5,700 patients in each arm). There were 52 total MIs with HandiHaler, vs. ~71 in each Respimat arm -- about a 1 to 1.2% incidence in an at-risk population, and a theoretical number needed to harm of ~330 to observe one additional MI with Respimat, if the signal were real.

An aside: Drs. Furberg and Singh have long been watchful, statistically savvy, and sometimes vocal skeptics of safety data from major drug industry sponsored trials. In 2004, the FDA went so far as to block Dr. Furberg from presenting at a meeting regarding the cardiovascular safety of Bexxtra (a COX-2 inhibitor like Vioxx) -- after appointing him to their panel -- apparently because of his public remarks about the drug's safety. Dr. Furberg was recused for an "intellectual conflict of interest" -- while 7 of the 14 advisors on the June 2004 panel, including the chair, received waivers for financial conflicts of interest. (The following year, the FDA instructed Pfizer to withdraw Bexxtra from the market.)

The numerical increase in heart attack risk seen here is certainly enough to merit concern and close observation in postmarketing studies. But with the small numbers of events and the negative overall result of a massive randomized trial, this is likely going to remain a case of “My SAS said, your SAS said” for the foreseeable future.

"By no means should these findings be ignored," Dr. Stephen Rennard of the University of Nebraska Medical Center commented recently. "I'm not sure the debate on safety is ever closed ... for any medication, no matter how long it has been around."

Correspondence: Tiotropium and the Risk of Death in COPD. N Engl J Med 2014; 370:480-483.