A look at the available data on the link between insulin and cancer

This whole synthetic insulin–cancer (particularly breast cancer) association has me weirded out (recall I have type 1 diabetes). And so I did some digging around to figure out the best level of data on this issue. There is a nice summary on HemOnc Today, or you can go directly to the source and read the to-be-published articles here. Allow me to summarize two of the less publicized studies here (the remaining two I will summarize in a post later this week).

Colhoun et al. (the Scottish Diabetes Research Network Epidemiology Group): A retrospective study of diabetes patients crosslinked with the Scottish cancer registry. They used the diabetes registry to identify patients who were exposed to any insulin over a four-year period ending in 2005, and from that, they defined a "fixed exposure" cohort over only a four-month period. They then looked at who developed breast, colon, prostate, pancreas and lung cancers by insulin type (glargine vs. non-glargine). Their analyses were adjusted for age, date, prior cancer, gender and type of diabetes. To highlight a few points: nearly all of the glargine only users had type 2 diabetes, were older, weighed more, and had been on insulin a lot (like 10 times less) longer. There is a trend that those on insulin glargine alone are more likely to have been diagnosed with an incident cancer (RR=1.55-1.73 depending on the model used, with P values ranging from 0.057 to 0.038. In summary: not overwhelming, but concerning. Breaking it down by type of cancer diagnosed, only breast cancer was more common.
Currie et al.: Also retrospective cohort study of patients with type 2 diabetes only taken from The Health Information Network, which is a database of about 300 United Kingdom general medical practices. They specifically looked at people in the database starting in 2000, and analyzed cancer incidence by whether the patients were on insulin or not. This included 10,067 insulin-receiving patients (2,286 on glargine), who were compared to patients on oral hypoglycemics. Those on insulin had their diabetes longer and had more diabetic complications. Average HbA1c was 9.4% for the insulin patients. In this study, sulfonylureas and insulin both were associated with a higher risk of solid tumors. Insulin glargine did not appear any worse than any other insulin therapy.

A few thoughts:

The exposure to insulin in all these cohorts is relatively short, especially when cancer is the outcome, which is known to take many (10+) years from exposure to development for many cancers.
The analyses may not have adequately controlled for obesity or severity of diabetes, both of which are also associated with cancer risk.
Stage of cancers and outcomes are not known — could those on insulin be seen more often by health care providers, then be more likely to get recommended cancer screening, leading to a lead time bias?

I think the culprit in the end will be higher levels of systemic insulin from insulin resistance and obesity and not the drug itself, but a prospective trial is warranted. And, memo to the agency that will sponsor this kind of trial, please include patients with type 1 and type 2 diabetes! Patients with type 2 diabetes have a lot of oral and lifestyle options relative to those with type 1 diabetes.