The pancreas is small and irregular in shape in people with type 2 diabetes. If these abnormalities are caused by the disease state itself rather than being a predisposing factor, remission of type 2 diabetes should restore normal pancreas morphology. The objective of this study was to determine whether changes in pancreas volume and shape occurred during 2 years of remission.

For this post-hoc analysis, we included a subset of adult participants of the Diabetes Remission Clinical Trial (DiRECT), who had type 2 diabetes and were randomly assigned to a weight management intervention or routine diabetes management. Intervention group participants were categorised as responders (HbA 1c <6·5% [48 mmol/mol] and fasting blood glucose <7·0 mmol/L, off all anti-diabetes medication) and non-responders, who were classified as remaining diabetic. Data on pancreas volume and irregularity of pancreas border at baseline, 5 months, 12 months, and 24 months after intervention were compared between responders and non-responders; additional comparisons were made between control group participants with type 2 diabetes and a non-diabetic comparator (NDC) group, who were matched to the intervention group by age, sex, and post-weight-loss weight, to determine the extent of any normalisation. We used a mixed-effects regression model based on repeated measures ANOVA with correction for potential confounding. Magnetic resonance techniques were employed to quantify pancreas volume, the irregularity of the pancreas borders, and intrapancreatic fat content. β-cell function and biomarkers of tissue growth were also measured.

Between July 25, 2015, and Aug 5, 2016, 90 participants with type 2 diabetes in the DiRECT subset were randomly assigned to intervention (n=64) or control (n=26) and were assessed at baseline; a further 25 non-diabetic participants were enrolled into the NDC group. At baseline, mean pancreas volume was 61·7 cm 3 (SD 16·0) in all participants with type 2 diabetes and 79·8 cm 3 (14·3) in the NDC group (p<0·0001). At 24 months, pancreas volume had increased by 9·4 cm 3 (95% CI 6·1 to 12·8) in responders compared with 6·4 cm 3 (2·5 to 10·3) in non-responders (p=0·0008). Pancreas borders at baseline were more irregular in participants with type 2 diabetes than in the NDC group (fractal dimension 1·138 [SD 0·027] vs 1·097 [0·025]; p<0·0001) and had normalised by 24 months in responders only (1·099 [0·028]). Intrapancreatic fat declined by 1·02 percentage points (95% CI 0·53 to 1·51) in 32 responders and 0·51% (−0·17 to 1·19) in 13 non-responders (p=0·23).

These data show for the first time, to our knowledge, reversibility of the abnormal pancreas morphology of type 2 diabetes by weight loss-induced remission.