Background Increasing evidence suggests that some antidiabetic medications may improve markers of abnormal liver structure and function in people with non-alcoholic fatty liver disease. We investigated this association in older people with type 2 diabetes.

Methods 474 participants, aged 63–79 years, from the Edinburgh Type 2 Diabetes Study, a large, randomly-selected population of patients with Type 2 diabetes, underwent assessment on two occasions. At baseline, liver ultrasonography was undertaken and antidiabetic medications (metformin, sulphonyureas and thiazolidinediones) were recorded. Liver ultrasonography was repeated approximately 3 years later. Hepatosteatosis was graded as either normal or “fatty”, and the change between examinations as regression, no change or progression. χ² for trend was used to analyse the association.

Results Mean follow-up was 2.7 years. 9.5% (n=45) of participants progressed, 13.9% (n=66) regressed and 76.6% (n=363) remained the same. Thiazolidinedione use was significantly higher among participants whose hepatosteatosis regressed (5.0%, 11% and 19.7% in participants with progression, no change and regression respectively, p=0.02). Similar figures for metformin use were 55.0%, 67.0% and 67.2% respectively (p=0.28) and for sulphonyurea use were 15.0%, 34.9% and 31.1% respectively (p=0.19).

Conclusion In a large sample of patients with type 2 diabetes representative of patients with type 2 diabetes in general, thiazolidinedione use was associated with regression of hepatosteatosis. This is consistent with emerging evidence from clinical trials suggesting that there may be a role for thiazolidinediones beyond glucose control in patients with type 2 diabetes. Further analyses are warranted to explore the potential mechanism underlying this association.