New Study Refutes Pioglitazone and Bladder Cancer Link

Pam Harrison

December 03, 2014

An independent study carried out by a large research consortium has again found little evidence that cumulative exposure to pioglitazone causes bladder cancer in men or women.

The large, pooled multipopulation analysis was published online December 3, 2014 in Diabetologia by Daniel Levin (University of Dundee, Scotland) and colleagues.

"In this pooled, aggregate-level analysis across multiple populations in several different countries, we found no significant association between cumulative exposure to pioglitazone and bladder cancer when using methods to minimize allocation bias," they say.

Earlier this year, Takeda, the manufacturer of pioglitazone (Actos) announced 10-year data from a longitudinal cohort study that it said showed there was no risk for bladder cancer associated with the agent. This was regardless of the duration of pioglitazone use, the cumulative dose employed, or the time since initiating treatment with pioglitazone. These data have yet to be published, however.

Pioglitazone, a thiazolidinedione (TZD) for the treatment of type 2 diabetes, has been associated with a number of adverse effects, including fractures and fluid retention, and is contraindicated in patients with class 3 or 4 heart failure. But it has been the association with bladder cancer that has caused the most controversy.

A number of countries have placed warnings about an increased risk for bladder cancer on the packaging of pioglitazone, including the Food and Drug Administration (FDA). The US agency updated the pioglitazone label in 2011 to warn against starting the drug in patients with bladder cancer and to caution against using it in patients with a history of bladder cancer. France, Germany, and India have also variously banned or severely restricted the use of pioglitazone for this same reason.

"I have not seen the 10-year follow-up published yet, but clearly reports that do not find any association with bladder cancer should be reassuring," senior author of the new study, Dr Helen Colhoun (University of Dundee), told Medscape Medical News.

"But I think [regulatory agencies] may find our data useful, and I would hope they would consider it, perhaps, when the report Takeda refers to also emerges," she added.

But one researcher who has previously published work showing that pioglitazone is associated with an increased risk for bladder cancer, says he has reservations about the methodology employed in the new study. Dr Laurent Azoulay (McGill University, Montreal, Quebec), said he commends the authors for their unique attempt to control for confounding, but he doesn't believe they have succeeded in their aim.

Study Employed Numerous Means to Limit Bias

Mr Levin and investigators affiliated with the Scottish Diabetes Research Network Epidemiology Group and the Diabetes and Cancer Research Consortium collected prescription, cancer, and mortality data on 1.01 million persons across six populations in Canada and a number of European countries for over 5.9 million person-years of follow-up.

They defined all periods of exposure to thiazolidinediones, as well as to metformin, insulin, and sulfonylurea for each patient.

Out of this cohort, they identified 3248 cases of incident bladder cancer, 117 of which occurred among subjects who had ever been exposed to pioglitazone. Another 204 bladder neoplasms were also identified among subjects who had ever been exposed to pioglitazone by the end of follow-up.

The median follow-up of the cohort ranged between 4 to 7.4 years.

The authors point out that most observational pharmacoepidemiological studies are limited by allocation bias (ie, potential differences in the prior susceptibility to bladder cancer between those prescribed and those not prescribed a drug). This cannot be reliably overcome using simple statistical adjustments, they note.

For this reason they examined the effects of cumulative exposure as well as "ever" exposure, because the former "is less likely to be subject to allocation bias."

Overall, there was no evidence for any association between cumulative exposure to pioglitazone and bladder cancer in men or in women after adjustment for age, calendar year, diabetes duration, smoking, and any "ever" use of pioglitazone.

The rate ratio (RR) per 100 days of cumulative exposure to pioglitazone was 1.01 for men and 1.04 for women.

Similarly, no association was observed between exposure to another thiazolidinedione, rosiglitazone, and bladder cancer in men or women, with virtually identical RRs of 1.01 in men and 1.00 in women.

"We have not found any evidence to support previous reports of an association [between pioglitazone and bladder cancer]," they assert.

They also considered exposure to other glucose-lowering drugs in their analytical models — something many other studies have not done.

And they included data from a number of countries worldwide using the same methodology to analyze the data from each country, thereby increasing the size of the cohort while limiting heterogeneity.

"Our study combines data across countries — this is useful, as false associations due to bias can sometimes arise because of specific prescribing patterns within a given country. Looking across different [countries] means that one is less likely to jump to erroneous conclusions," Dr Colhoun told Medscape Medical News.

Statistical Reservations

Asked to comment, Dr Azoulay said the authors should be commended for trying to answer this question across different data sources, which is not often attempted. But he has some reservations about the methodology used.

"They modeled cumulative exposure [to TZDs] to see if every unit of exposure was associated with an increased risk of bladder cancer, which is perfectly fine," he said.

But they also adjusted for another variable — ever vs never use — in their model that essentially "is doing the same thing, defining exposure," he added. "Including these two variables in the same model is problematic," and using them both does not deal with confounding and instead, "likely biased the results," he observes.

"There are sophisticated ways to deal with confounding, but modeling exposure in this particular way does not take care of this in an observational study."

But Dr Colhoun contends that these two variables — ever-never exposure and cumulative exposure — "do not 'do the same thing,' as Dr Azoulay suggests."

By analyzing the numbers in this way, "we can study the joint effects of two or more drugs and we can use the covariate data better," she responded. "This type of analysis is appropriate for a disease like cancer, which takes years to develop." She noted that this method has been described in a prior publication (Diabetologia. 2012;55:2929–2937) and has been recommended by others too (Diabetes Care. 2014;37:e221-e222) "as a sensible approach."

Mr Levin reported he had no relevant financial relationships . Dr Colhoun has received research support from Roche, Pfizer, Eli Lilly, Boehringer Ingelheim, and AstraZeneca and membership in a speaker's bureau for and acted as a consultant for Pfizer. She is on an advisory panel for Sanofi, Pfizer, Novartis Pharmaceuticals, and Eli Lilly, and owns Roche stocks and shares. Disclosures for the coauthors are listed in the article. Dr Azoulay reported he had no relevant financial relationships.

Diabetologia. Published online December 3, 2014.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.