Aims/hypothesis Low birthweight is a risk factor for type 2 diabetes. Most previous studies are based on cross-sectional prevalence data, not designed to study the timing of onset of type 2 diabetes in relation to birthweight. We aimed to examine associations of birthweight with age-specific incidence rate of type 2 diabetes in middle-aged to older adults over two decades. Methods Adults aged 30–60 years enrolled in the Danish Inter99 cohort in 1999–2001 (baseline examination), with information on birthweight from original birth records from 1939–1971 and without diabetes at baseline, were eligible. Birth records were linked with individual-level data on age at diabetes diagnosis and key covariates. Incidence rates of type 2 diabetes as a function of age, sex and birthweight were modelled using Poisson regression, adjusting for prematurity status at birth, parity, polygenic scores for birthweight and type 2 diabetes, maternal and paternal diabetes history, socioeconomic status and adult BMI. Results In 4590 participants there were 492 incident type 2 diabetes cases during a mean follow-up of 19 years. Type 2 diabetes incidence rate increased with age, was higher in male participants, and decreased with increasing birthweight (incidence rate ratio [95% CI per 1 kg increase in birthweight] 0.60 [0.48, 0.75]). The inverse association of birthweight with type 2 diabetes incidence was statistically significant across all models and in sensitivity analysis. Conclusions/interpretation A lower birthweight was associated with increased risk of developing type 2 diabetes independent of adult BMI and genetic risk of type 2 diabetes and birthweight. Graphical Abstract

Objective: Ectopic liver fat deposition, resulting from impaired subcutaneous adipose tissue expandability, may represent an age-dependent key feature linking low birth weight (LBW) with increased risk of Type 2 diabetes (T2D). We examined whether presumably healthy early middle-aged, non-obese LBW subjects exhibit increased liver fat content, whether increased liver fat in LBW associates with severity of dysmetabolic traits, and finally whether such associations may be confounded by genetic factors. Methods: Using 1H magnetic resonance spectroscopy, we measured hepatic fat content in 26 early middle-aged, non-obese LBW and 22 BMI-matched normal birth weight (NBW) males. Endogenous glucose production was measured by stable isotopes, and a range of plasma adipokine and gut hormone analytes were measured by multiplex ELISA. Genetic risk scores were calculated from GWAS data for birth weight, height, T2D, plasma cholesterol and risk genotypes for non-alcoholic fatty liver disease (NAFLD). Results: The LBW subjects had significantly increased hepatic fat content compared with NBW controls (P=0.014), and 20% of LBW versus no controls, had overt NAFLD. LBW subjects with NAFLD displayed widespread metabolic changes compared with NBW and LBW individuals without NAFLD, including hepatic insulin resistance, plasma adipokine and gut hormone perturbations as well as dyslipidemia. As an exception, plasma adiponectin levels were lower in LBW subjects both with and without NAFLD. Genetic risk for selected differential traits did not differ between groups. Conclusion: Increased liver fat content including overt NAFLD may be on the critical path linking LBW with increased risk of developing T2D in a non-genetic manner.