In 2007, the Food and Drugs Administration (FDA) requested that information about acute pancreatitis should be included in the product label for the GLP-1 receptor agonist, exenatide. This followed review of 30 post-marketing reports of pancreatitis in patients receiving this drug: in six patients symptoms had begun or worsened after a dose increase; twenty-one patients were admitted to hospital and in 22 cases symptoms improved after drug withdrawal. In three cases, symptoms recurred after exenatide was reintroduced. Another way of looking at the data would be to point out that 9/30 (30%) weren’t admitted to hospital and in only 17 cases (57%) was the serum amylase reported – not typical of a diagnosis of ‘pancreatitis’, in the UK at least. Most would also be very wary of ‘re-challenging’ their patients with a drug they felt was responsible for this serious condition.
The response of the pharmaceutical industry was to review their clinical trial programmes and to extract data for pancreatitis in cohorts with type 2 diabetes, typically from insurance claims databases. The results from both of these lines of investigation were reassuring with similar rates of pancreatitis, irrespective of the use of incretin-based drugs (including both GLP-1 RAs and DPP-4 inhibitors).
The story was resurrected in 2011 when Elashoff and colleagues published a paper reporting a six-fold increase in pancreatitis in patients taking incretin-based therapies. They also went on to suggest that, based on their interpretation of histology from animal models, asymptomatic pancreatitis in patients exposed to incretin therapies would manifest as a longer term increase in pancreatic cancer. The paper was controversial for two major reasons: first it was published on-line by the journal Gastroenterology but then quickly withdrawn, leading to suggestions of improper pressure from the pharmaceutical industry; second, it used an analysis of the FDA Adverse Event Reporting System (AERS) to calculate the incidence of pancreatitis. The latter is an issue because the AERS is made up of a mixture of spontaneous reports from health care professionals (such as physicians, pharmacists, nurses) and consumers (e.g. patients, family members, lawyers). There is typically insufficient detail to allow for a reasonable evaluation of the event and there tends to be a bias in reporting against both newer agents and publicised side-effects. Finally, and most problematic, the FDA states that the AERS data cannot be used to calculate the incidence of an adverse event in the U.S. population.
All new diabetes medications are now subjected to large cardiovascular safety studies, a mandate from the FDA which followed on from the Rosiglitazone controversy (which also began in 2007). As a result, almost every incretin drug is being tested in a large randomized, placebo-controlled trial and the results of these will begin to emerge as early as 2014. Tens of thousands of patients have been recruited into these studies and one might have imagined that safety issues would have been put on hold pending their publication.
But this was not to be. On 14th March 2013, the FDA issued a Drug Safety Communication (1) stating that it was investigating reports of a possible increased risk of pancreatitis and pre-cancerous findings from incretin drugs. The Communication stated that FDA was evaluating unpublished findings by a group of academic researchers that suggest an increased risk of pancreatitis and pre-cancerous cellular changes. They went on to state that “these findings were based on examination of a small number of pancreatic tissue specimens taken from patients after they died from unspecified causes. FDA has asked the researchers to provide the methodology used to collect and study these specimens and to provide the tissue samples so the Agency can further investigate potential pancreatic toxicity associated with the incretin mimetics.” The website also highlights an e-publication in JAMA (2) which supports the increased risk of hospitalisation due to pancreatitis in patients on incretin-based therapies.
So what are clinicians to make of these comments? There is no doubt that the incretin class of drugs are popular; low hypoglycaemia rates and weight neutrality make the DPP-4 inhibitors an attractive alternative to sulphonylureas in primary care whilst the weight loss potential of GLP-1RA injectables is a viable alternative to insulin initiation. There is also no doubting the immense pharmaceutical company pressure to promote these agents as well as a much smaller scientific lobby which seeks to have them withdrawn from clinical use; whilst the motives for the former are obvious, those for the latter are more opaque. The position of the regulators, which include the FDA is clearly difficult, but to issue statements based on data which have not been peer-reviewed (even by themselves) seems a little extreme, to the point of being odd….
For the moment, I will continue with my current use of incretin-based therapies, aware that safety issues remain for all new diabetes drugs but cognisant that teasing out the risk-to-benefit ratios of therapies is not straight-forward. Witness the uncertainties around insulin and sulphonylurea use that persist despite over half-a-century of their use.
2) Singh S, Chang HY, Richards TM, Weiner JP, Clark JM, Segal JB. Glucagonlike peptide 1-based therapies and risk of hospitalization for acute pancreatitis in type 2 diabetes mellitus: a population-based matched case-control study. JAMA Intern Med 2013 Feb 25:1-6. doi: 10.1001/jamainternmed.2013.2720. [Epub ahead of print].
Professor Steve Bain