Our study showed that maternal obesity and maternal metformin treatment during obese pregnancy caused long-term increases in offspring adiposity in an age- and sex-specific manner. Male Ob offspring showed increased adiposity, upregulated proinflammatory gene expression in gWAT, and hepatic lipid accumulation at 12 months of age. Adiposity in aged male Ob-Met offspring was associated with more severe adipocyte dysfunction (worse inflammation and early-life hyperplastic response), leading to hepatic steatosis and lipid accumulation that was correlated with adipose tissue and metabolic variables. While female Ob offspring did not show gWAT or liver dysfunction, female Ob-Met offspring had increased adiposity, gWAT inflammation and hyperplasia, hepatic steatosis, hyperinsulinaemia and hyperleptinaemia. Table 4 summarises these findings.
Our work demonstrates clear sex differences in the programming of obesity and fatty liver by maternal obesity, while also highlighting the importance of age to the development of programmed phenotypes. The WAT expansion in Ob male offspring was associated with increased proinflammatory gene expression at 12 months (not present at 8 weeks ), in accordance with ageing promoting a proinflammatory environment . In contrast, female Ob offspring were protected against the development of obesity, gWAT inflammation and hepatic abnormalities. Sexual dimorphism is often described in the context of developmental programming, with the male sex generally being more vulnerable to programmed effects . This could be related to biologically underpinned differences, such as male mice growing more rapidly in utero, differential effects of sex steroids or male mice ageing faster than female mice .
Metformin is widely used in the developed world to treat GDM  and has proven beneficial for treatment of other pregnancy indications including PCOS and pre-eclampsia [9, 32, 33]. It is an attractive alternative to insulin in populations where access to insulin is limited or subject to financial barriers  and many women prefer metformin to insulin as it does not require injection . Our experimental model simulates certain features of human GDM, including impaired glucose tolerance in pregnancy [11, 16] that is no longer present after weaning , and our dosing protocol leads to maternal circulating metformin concentrations similar to those in humans (with equilibration to the fetal circulation) . This is consistent with detection of known metformin transporters in murine and human placenta [11, 36]. However, the initiation of metformin treatment around conception may more closely resemble treatment of pregnant women with type 2 diabetes or PCOS and our offspring data are therefore also relevant to metformin use in those clinical contexts. The difference in timing of WAT development between rodents and humans could be a limitation to clinical relevance. However, adipocyte lineage commitment largely occurs in gestation in both humans and rodents, and thus may be similarly influenced by intrauterine metformin . Importantly, our data parallel outcomes of human trials. Individual RCTs for GDM and PCOS pregnancies have shown increased adiposity in young metformin-exposed offspring compared with insulin and placebo groups, respectively [8, 9, 38], and these findings have been confirmed by a meta-analysis of trials in GDM .
The maternal metformin intervention did not correct the adiposity observed in 12-month-old male offspring exposed to maternal obesity but induced a more inflammatory gWAT phenotype. This was reflected by presence of CLS and upregulation of M1 (Itgax, Tnf) and migratory markers (Ccl2) alongside the macrophage marker Adgre1, indicating recruitment of proinflammatory macrophages to hypertrophied WAT. After macrophage infiltration, inflammation can be propagated by dysfunctional adipocytes and activated macrophages , potentially explaining the most severe inflammation in Ob-Met male offspring. While Con and Ob male offspring showed significant increases in adipocyte number between 8 weeks and 12 months of age, male Ob-Met offspring had more adipocytes at 8 weeks but failed to elicit compensatory hyperplasia with ageing. This suggests their limit of WAT hyperplastic expansion capacity is reached by young adulthood (further supported by WAT depot weights being lower than in male Ob offspring at 12 months). The restricted WAT expandability resulted in ectopic lipid deposition in the livers of male Ob-Met offspring. This may result from direct effects of metformin on adipocyte progenitors in utero. Metformin decreases maturation and differentiation of mouse and human pre-adipocytes in vitro  and directly inhibits adipocyte lineage commitment . Suppression of the number of adipocyte progenitors by metformin in utero could therefore explain the restricted WAT expandability, premature reliance on hyperplastic adipose tissue expansion, and hepatic lipid deposition in metformin-exposed male offspring. Increased mesenteric adiposity, liver weight and hepatic expression of lipogenic genes was previously reported in 20-week-old male and female offspring of metformin-treated chow-fed dams .
Female offspring exposed in utero to metformin exhibited an adiposity phenotype characterised by hyperinsulinaemia, hyperleptinaemia, gWAT inflammation and ectopic lipid deposition. The excessive WAT expansion in Ob-Met female offspring resulted from hyperplasia rather than hypertrophy. Sexual dimorphism in ageing dynamics might explain why, unlike the male offspring, female Ob-Met offspring retained the ability to elicit hyperplasia between 8 weeks and 12 months. This compensatory hyperplasia was insufficient to cope with expansion demands, as evidenced by the hepatic steatosis and increased proinflammatory signature of WAT in female Ob-Met offspring. The increase in CLS-surrounded adipocytes indicates increased adipocyte death in female Ob-Met offspring WAT . The presence of hyperinsulinaemia in female Ob-Met offspring at 12 months of age is distinct from the previously reported phenotype at 8 weeks, where no difference was observed . Since obesity and increased fat mass are strongly associated with these phenotypes , the abnormalities in insulin homeostasis in 12-month-old female Ob-Met offspring are likely secondary to the development of obesity. The increased adiposity in female Ob-Met offspring could result from alterations in energy expenditure, energy intake or nutrient assimilation in the intestines (alterations in BAT thermogenic capacity and intestinal microbiota have been reported following early-life metformin exposure [15, 43]). We found no detectable differences in food intake in 11-month-old offspring of either sex, and did not assess energy expenditure.
A striking finding of this study is that adiposity was more strongly induced by in utero metformin in aged female offspring and accompanied by hyperinsulinaemia in female but not male Ob-Met offspring. Most previous rodent studies investigating metformin interventions reveal similar outcomes in both sexes, although few studied offspring until 12 months of age. Since our study period coincides with the onset of oestropause , recent loss of oestrogen’s protective anti-inflammatory and pro-adipogenic effects on WAT  might have exacerbated the phenotype in female Ob-Met offspring, while intervention effects in male offspring may be masked by additional programming effects induced by maternal obesity. Metformin may also affect ageing dynamics in exposed offspring in a sex-specific manner, as suggested by a study in which repeated metformin injections were administered to neonatal mice . We know of only one other study that described clear sex differences, with increased susceptibility in female offspring: prenatal metformin treatment to genetically obese neuropeptide Y-overexpressing dams caused adiposity and glucose intolerance in female but not male offspring at 7 months of age . Others reported sex-specific timing of in utero metformin intervention: in lean glucose-tolerant pregnancy there were beneficial effects on insulin homeostasis in young adult metformin-exposed male offspring, and these effects weakened with age; in female offspring improvements in metabolic function only appeared at 15 months .
To our knowledge, our model is the most clinically relevant model of metformin intervention during maternal diet-induced glucose-intolerant pregnancy that is currently reported, with treatment dose and maternal serum concentrations comparable with those in human studies . The main strength is the long-term follow-up of both male and female offspring, providing causal evidence for the development of obesity beyond young adulthood. This is especially pertinent as most phenotypes emerge after 6 months of age, a common maximum endpoint in developmental programming studies. This prolonged follow-up revealed age-sensitive sexually dimorphic effects of maternal obesity and metformin intervention. Although offspring were followed up until 12 months of age, the lifespan of mice is around 2 years and therefore this equates to middle age . Since the age of offspring heavily influenced metabolic outcomes, it is important that future studies assess effects beyond 12 months.
Metformin treatment during a pregnancy complicated by GDM has clear short-term beneficial effects beyond glycaemic control, including decreased gestational weight gain (benefiting a subsequent pregnancy by preventing excessive interpregnancy weight gain ), lower incidence of pre-eclampsia, and improved neonatal outcomes [4, 49]. However, the findings of offspring adiposity and fatty liver resulting from maternal metformin exposure are concerning, as both childhood and adult obesity are an increasing problem worldwide . Therefore, the relative short-term benefits and potential adverse long-term metabolic effects must be weighed against one another. It is vital that the outcomes investigated in the current study are addressed in human trials, as we cannot exclude species-specific differences, or that the intervention might have different effects depending on clinical indication or timing of metformin prescription. Offspring follow-up beyond childhood is therefore crucial in human clinical trials.
Metformin exposure in utero during diet-induced obese pregnancy increased metabolic risk factors in a sex- and age-dependent manner. Our work highlights the importance of following up offspring of both sexes throughout the life course, in addition to immediate effects on mother and fetus, and illustrates the complexity of balancing short-term benefits of therapeutic agents that cross the placenta vs any long-term metabolic risks. Alternative treatment regimens or formulations that retain maternal benefits but limit fetal exposure to metformin might be promising areas of future research.