ORLANDO – findings from meta-analyses suggest.
In one meta-analysis of data from 21 prospective studies with a total of nearly 15.2 million women, 325,117 breast cancer cases, and a mean follow-up time of 8 years (nearly 33 million person-years), the risk of breast cancer was significantly greater among patients with diabetes than it was among patients without diabetes (summary relative risk, 1.11), Maria Bota reported at the annualof the American Diabetes Association.
However, there was substantial unexplained heterogeneity of results across the individual studies (I2 = 82%), said Ms. Bota, a faculty member at the International Prevention Research Institute, Lyon, France.
“When the analysis was restricted to the 12 studies that adjusted for [body mass index], the summary relative risk decreased to 1.09 and the heterogeneity also decreased to a moderate value of 32%; when the analysis was restricted to the 9 studies that did not adjust for BMI, the summary relative risk increased to 1.14 again, and the heterogeneity increased even more to 91%,” she said.
In an analysis that combined the results of the nine studies that did not adjust for BMI along with crude relative risks from studies that reported both crude and BMI-adjusted relative risks (17 studies in all), the SRR was 1.12, and heterogeneity among the studies was high at 84%.
Additionally, an analysis by menopausal status based on four studies that reported breast cancer in both pre- and postmenopausal women showed SRRs for breast cancer of 0.97 (a 3% decrease in risk) and 1.14 among diabetic vs. nondiabetic premenopausal women and postmenopausal women, respectively, she said, noting that heterogeneity was low (I2 = 0%) among the premenopausal breast cancer study groups and high (I2 = 70%) among the postmenopausal study groups.
The findings provide evidence for a moderately increased risk of breast cancer in women with T2DM, Ms. Bota said.
“However, the effect of the adjustment or lack of adjustment on the heterogeneity suggests that the higher risk of breast cancer in women with diabetes may not be due to diabetes itself, but to adiposity,” she said, adding that “this hypothesis is equally supported by our subgroup analysis according to menopausal status because we saw that the risk of breast cancer was only associated with diabetes in postmenopausal women and this pattern resembles the pattern of the risk of breast cancer associated with adiposity, which is also only increased in postmenopausal women.”
This study was limited by insufficient data for investigating the sources of heterogeneity, she said.
“Therefore we propose ... future pooled analyses based on individual data from good quality prospective studies in order to increase the study power and to do some detailed analysis of the links between adiposity, diabetes, and breast cancer,” she concluded, adding that new studies to examine those relationships only in premenopausal women are also needed
In a separate meta-analysis, she and her colleagues, including Peter Boyle, PhD, president of the International Prevention Research Institute, assessed the association between insulin treatment and breast cancer risk in patients with diabetes.
“The long-acting insulin analogues glargine and detemir have been shown in some studies to be associated with increased risk of breast cancer, and other studies have shown no association between the use of these two compounds and the risk of breast cancer,” Dr. Boyle said in a separate presentation at the ADA meeting. “It was important to sort out a little bit what was going on in the literature.”
Overall, the meta-analysis of data from 12 longitudinal cohort studies – including more than 6,000 cases of breast cancer – showed a slight increase in breast cancer risk in patients taking long-acting insulin (SRR, 1.13) with “a relatively reasonable” level of heterogeneity (I2 = 23%).
“But we see that the story is not that simple,” he said.
For example, some studies included only patients who were prescribed insulin for the first time after the study began (new users), some included only patients who were prescribed insulin before the study began (prevalent users), and some included both (ever users), which may have introduced bias in the results, he explained.
Studies of glargine included 4,168 breast cancer cases over a total of 1,418,743 person-years of observation, and studies of detemir included fewer than 2,047 breast cancer cases (not all studies reported case numbers). Among both new users of glargine and detemir, the SRR was 1.12, suggesting no real association between either glargine or detemir use and breast cancer, he said.
“One important take-home message is that you have to be careful that these pharmaco-epidemiological studies, even when working with the same database, may have conflicting results ... so we still need more robust standards in methods for [such] studies,” he concluded.
Ms. Bota reported having no disclosures. The study presented by Dr. Boyle was funded by Sanofi. Dr. Boyle is president of a charity that has received donations from Pfizer, Roche, Novartis, and Lilly.
SOURCE: Bota M. ADA 2018, ; Boyle P. ADA 2018, .