We thank Yang and Chan (1) for their interest in our articles (2,3). They propose that our studies may be biased by use of time-varying exposures in Cox regression models.
If more severe hyperglycemia is associated both with increased cancer risk and with use of pioglitazone, and it is also affected by pioglitazone therapy (i.e., hyperglycemia is both a confounder and an intermediate variable), we agree that our use of the time-dependent Cox model may have led to an over-estimate of our hazard ratios for pioglitazone use (4).
Although current evidence suggests that diabetes is weakly to modestly associated with increased risk of cancer and cancer mortality in the general population (5), the data on an association between levels of hyperglycemia and cancer risk among patients with diabetes are more limited and less clear (5,6).
Several of our findings also suggested that hyperglycemia was not a risk factor for cancer within our cohort of diabetic patients. First, we did not see an association between increasing HbA1c levels at baseline and increasing risk of cancer at any site or an association between other diabetes medications and cancer risk. We also did not observe an increased risk of any cancer associated with ever versus never use of pioglitazone. Finally, we did not see an association with longer duration of pioglitazone use and risk of cancer at any site other than bladder.
We therefore believe that our use of time-varying Cox models was appropriate and that our hazard ratios for pioglitazone use are unlikely to be appreciably biased by hyperglycemia.
Article Information
A.F. has received research funding from Takeda and Eli Lilly and Co. J.D.L. has received other research funding from Takeda and GlaxoSmithKline and has served as a consultant to Takeda and GlaxoSmithKline. B.L.S. has received research funding from Takeda and has served as a consultant to GlaxoSmithKline and Eli Lilly and Co. L.H. has received other research funding from Merck, Genentech (Roche), and sanofi-aventis. No other potential conflicts of interest relevant to this article were reported.
