For the past two months, there has been rising controversy over the extent to which certain diabetes drugs called GLP-1 inhibitors may cause pancreatitis and pancreatic cancer. Although links to pancreatitis are actually not new, one of two recent papers suggested pre-cancerous cell growth may be cause for concern, prompting both the FDA and the European Medicines Agency to start investigations.
Now, a watchdog group is calling for a reassessment of the entire class of drugs after analyzing adverse events data that was reported to the FDA and finding distinctly higher odds that GLP-1 drugs are generating reports for these illnesses compared with a control group consisting of older drugs, such as metformin, that are used to treat diabetes.
Specifically, the analysis found the odds were roughly 25 times higher for the following drugs: Merck’s Januvia (MRK), which is the biggest seller in this group; Onglyza, which is sold by Bristol-Myers Squibb (BMY) and AstraZeneca (AZN); Byetta, which is also marketed by Bristol-Myers; the Tradjena treatment sold by Eli Lilly (LLY) and Boehringer Ingelheim, and Novo Nordisk’s Victoza.
“I think the future of the whole class is in question,” says Thomas Moore, senior scientist for drug safety and policy with the Institute for Safe Medication Practices, which published the data in its QuarterWatch report. While he says further study should be undertaken, “if results are confirmed in a broader patient population, it raises questions about the entire class of drugs.”
In reaching this conclusion, ISMP examined 1,723 serious adverse events reported to the FDA between July 1, 2011, and June 30, 2012 for all five of the GLP-1 drugs, which mimic a hormone called GLP-1 to stimulate natural insulin production. There were 831 cases of pancreatitis; 105 case of pancreatic cancer; 32 cases of thyroid cancer and 101 cases indicating a hypersensitivity reaction.
When breaking down the adverse event reports, ISMP found that after adjusting for differences, the odds that a report would be filed indicating pancreatitis was 28.5 times higher for the two injectable drugs – Byetta and Victoza – compared with older diabetes meds. By contrast, the odds of such a report being filed in connection with the pills were 20.8 times higher.
Overall, the odds the GLP-1 drugs would cause an adverse event report of pancreatitis to be filed was 25.6 times higher than for the metformin or sulfonulyreas. Similarly, the injectable drugs were associated with reports of thyroid cancer, while the pills were not, according to ISMP. Athough all of the GLP-1 yielded hypersensitivity reports, the odds that Victoza would do so was nearly 8 times higher than the older meds and the only one to show a statistically significant difference.
“There are some important signals here,” says Moore, who maintains the analysis is the most comprehensive analysis to date of adverse events for this class of drugs. “The hypersensitive issue needs more than AE reports to validate, but it’s a pretty big signal. And our analysis also provides preliminary evidence of the risks of the (injectables) could be higher than the oral agents” (here is the ISMP report). He added that the results “build on” what was seen in the recent studies that prompted regulatory probes. In one study, which was published in Diabetes, researchers found the drugs caused “marked” cell proliferation and damage, and displayed a potential for eventually transforming into cancer. However, the study was small – researchers examined the pancreas of 20 deceased human organ donors with type 2 diabetes.
The other study, which was published in JAMA Internal Medicine, examined insurance records for more than 2,500 diabetics between February 2005 and December 2008, and found that patients hospitalized with pancreatitis were twice as likely to have taken either Januvia or Byetta than a control group of type 2 diabetics who did not have pancreatitis. The other drugs were not analyzed (read more here and here is another interesting angle). This study generated considerable pushback from not only the drugmakers, but also Wall Street analysts and the American Association of Endocrinologists and the American Diabetes Association over concerns that physicians may change treatment practices and objections to the study design. They also noted pancreatitis was not a new issue.
But when the next study appeared in Diabetes, the dissension was muted, although the Public Citizen watchdog group called for the FDA to ban the class of drugs (read here). Not surprisingly, several of the drugmakers last night defended their medicines and noted the ISMP analysis had limitations, since the FDA adverse event reporting system does not demonstrate causality. The FDA database is “subject to reporting biases and (is) often limited by a lack of information concerning important variables such as confirmation of diagnosis, specific patient characteristics and co-morbidities, duration of a patient’s disease, prior drug exposures, and concomitant medication use. The authors themselves caution that their analysis ‘should be interpreted in light of the known limitations of a reporting system that does not collect data systematically,’ “ a Bristol-Myers spokesman writes us.
He adds that “direct comparison of event rates of different agents generated from this type of analysis should be interpreted with caution, as the authors themselves acknowledge. These data need to be put into context with data from clinical trials and epidemiology studies, which are better suited to assess risk.”
And a Novo Nordisk spokesman sent us this: “We have reviewed the totality of safety information available to us, and remain confident in the safety profile of Victoza. We continue to work closely with the FDA to provide an on-going assessment of Victoza’s risk-benefit profile.”
And a Boehringer spokeswoman wrote this: "Numerous factors can increase spontaneous adverse event reporting. Comparison of reporting rates across products is not methodologically supported due to biases that cannot be adequately controlled... The FDA’s Adverse Event Reporting System specifically states that data should not be used to compare findings across compounds. Products may be approved for different indications in patient populations that are inherently different, and therefore, the outcomes across these patients may also be different."
"...It is not appropriate to compare new users of a product with patients who have been treated with a product for a longer period of time as it introduces critical bias due to depletion of susceptibles (healthy survivor bias); the initially treated group of patients has a higher risk of adverse events than that after some years of market presence because patients with side effects may, for example, discontinue treatment," she adds.
Think I'll stick with LowCarb !
Graham
