Expert Blog

Diabetes Agents and Carcinogenesis

Irl B. Hirsch, MD
It is difficult to forget the dramatic controversy that occurred in the summer of 2009 when four observational studies[1] were published in the journal Diabetologia three of which suggested an association between insulin glargine and certain forms of cancer.  [2] While this particular relationship is still not completely resolved, the bulk of evidence thus far does not support this initial concern. Still, those publications had a greater impact: they pointed us to appreciate that both diabetes and obesity are independent risk factors for certain cancers. Most epidemiologic analyses have shown cancer is more common in people with diabetes (prostate cancer is the exception) while BMI appears to be a risk for many types of cancer in both men and women.[3] 
It therefore is not surprising that many are asking a simple question: with so many new diabetes agents, could any of them promote malignancy? In the past year we’ve learned of new concerns with pioglitazone and bladder cancer, resulting in new warnings.[4] At this year’s European Association for the Study of Diabetes (EASD) in Lisbon, this topic was as hot as the weather, culminating in a debate on the last day. In this debate, Dr. Peter Butler, reviewing his recently published analysis of the FDA data base,[5] noted an increased risk of both pancreatitis and pancreatic cancer with both sitagliptin and exenatide (both GLP-1 therapies). He supported this conclusion with animal studies and cell-culture studies. His opponent in the debate, Dr. Michael Nauck, argued that currently available data do not support that GLP-1 therapies increase the risk of cancer, adding that the biology underlying this possibility is not plausible because these types of malignancies take much more time to develop. 
Which side is correct? Even in the debate Dr. Butler agreed it is too early to confirm if there is a real cancer signal here or not. What I know is that this is not a topic that will be going away any time soon. Even in its most conservative mood, moreover, it would be impossible for the FDA to delay approval for a diabetes drug to wait for data proving an increased risk of cancer for any single agent. It would simply take too long. Furthermore, from a pure numbers point of view, the greatest enemy by far is still cardiovascular disease (CVD), and it is quite possible some of the newer agents will result in a benefit in CVD risk over time. Although some will say this is wishful thinking on my part, in the end we will require a better benefit-risk analysis than the information we have now. 
For argument’s sake, however, let’s take a theoretical example. A new GLP-1 agonist is found to reduce the risk of CVD events by 25% in a population where 50% will die from heart disease. That same drug increases the risk of cancer by 25% in a population where the risk of death from cancer is 5%. In this theoretical example, would you take your chances with this drug (especially if you could screen for the cancer)? Add to this the fact that we know that metformin protects against most types of cancer, and in this example we learn that this GLP-1 agonist with metformin mitigates all risks of cancer. While prompting an interesting discussion, this example is not nearly as complex as the actual science, which will hopefully someday become clearer. 
I am quite curious about whether this analysis makes sense to others and whether they think new concerns about carcinogenesis are warranted.