Thiazolidinedione drugs and the risk of cancer in patients with diabetes: nested case-control studies using a primary care database

Talk Code: 

The problem

Type 2 diabetes is a condition which is becoming more common as the population ages and obesity levels rise. Several studies have established that people with diabetes have an increased risk of developing cancer, including colorectal, breast and pancreatic cancer. Thiazolidinedione drugs, also known as glitazones, are used to treat diabetes, and have been prescribed in large quantities, for example in 2013 there were more than 1.3 million prescriptions in England. There have been concerns though that these drugs might increase the risk of cancer, and in particular some studies have reported increased risks of bladder cancer for pioglitazone, leading to its suspension in some countries. Other research however suggests thiazolidinediones may have anti-cancer properties. Overall the evidence is limited and few studies have accounted for smoking and other potential confounding variables. The aim of this study was to investigate associations between thiazolidinedione drugs and the risks of different cancers in people with diabetes.

The approach

A series of nested case-control studies was carried out using the QResearch primary care database. A cohort was assembled of patients with a diagnosis of diabetes and registered between 1997 to 2013, and within this cohort cases with a diagnosis of cancer were identified. We selected cases with 12 of the most common types of cancer - bladder, breast, colorectal, gastro-oesophageal, kidney, liver, lung, melanoma, pancreatic, prostate, non-Hodgkin lymphoma, and uterine cancer, and matched each case to five controls. Information was extracted on all prescriptions for diabetes drugs up to one year before the cancer diagnosis. Conditional logistic regression was used to estimate odds ratios for each cancer type associated with thiazolidinedione use adjusting for potential confounding variables.


There were 1791 bladder cancer cases, 2386 breast cancer, 3401 colorectal cancer, 726 kidney cancer, 653 liver cancer, 2921 lung cancer, 574 melanoma, 755 non-Hodgkins, 1430 gastro-oesophageal, 919 pancreatic, 3393 prostate and 760 uterine cancer cases. The proportions of cases prescribed thiazolidinediones before diagnosis ranged from 9.8% for lung cancer to 14.5% for melanoma.There were no significant associations between any use of thiazolidinediones and the individual cancers. The adjusted odds ratio for bladder cancer was 0.90 (95%CI 0.75 to 1.09). In an analysis of duration of use the adjusted odds ratio for bladder cancer for 36 months or more use of thiazolidinediones was 0.67 (95%CI 0.47 to 0.95, P=0.016 for trend), for kidney cancer it was 0.60 (95%CI 0.35 to 1.05, P=0.019 for trend).


This study has found no evidence that thiazolidinediones are associated with an increased risk of any of twelve common cancers, although there was some evidence of reduced risks of bladder and kidney cancer with longer durations of use. This is reassuring in light of recent safety concerns.


  • Carol Coupland, University of Bristol, Bristol, UK
  • Trevor Hill, University of Bristol, Bristol, UK
  • Peter Brindle
  • Yana Vinogradova, University of Bristol, Bristol, UK
  • Julia Hippisley-Cox, University of Bristol, Bristol, UK