Observational analyses of data from population registeries have suggested that metformin may be associated with a decreased prevalence of malignancy. The ADOPT and RECORD studies both contain groups of subjects randomly allocated to metformin and rosiglitazone. This meta-analysis combines malignancy serious adverse events from ADOPT and RECORD in order to compare their incidence on metformin with that on rosiglitazone.

The objective of this meta-analysis is to compare the incidence of malignancy (excluding non-melanomatous skin cancers) reported as serious adverse events in the ADOPT and RECORD studies between subjects randomly allocated to treatment with metformin with those allocated to rosiglitazone.

Subjects from ADOPT that are included in this meta-analysis (those in the metformin or rosiglitazone groups) had the dose of their study medication increased to a maximum of 2g (metformin) or 8mg (rosiglitazone) if their fasting plasma glucose was greater than 140mg/dl.

Subjects from RECORD that are included in this meta-analsyis entered the RECORD study taking one of three sulfonylureas (glibenclamide, gliclazide or glimepiride). They were randomly allocated to metformin or rosiglitazone in a 1:1 ration to use as add-on treatment to the background of sulfonylurea. These subjects had the dose of their study medication increased to a maximum of 2.55g (metformin) or 8mg (rosiglitazone) if their HbA1c was greater than 7.0%.

5135 subjects with type 2 diabetes mellitus from the intention to treat (ITT) populations of 2 randomised, controlled studies (2576 metformin/2559 rosiglitazone)