The present analysis investigated associations between birth weight and objectively measured food intake in adulthood among 65 healthy Indigenous Americans. We discovered a sex difference such that birthweight was negatively associated with adult food intake in males and positively associated with adult food intake in females. Lower birthweight males consumed more fat as adults, whereas higher birthweight females consumed more fat and protein.
It is important to understand how energy intake patterns may be impacted by differences in intrauterine development, given that small but chronic disruptions in energy homeostasis may compound and lead to excess adiposity over time (22), a major risk factor for the diseases whose relationship with birthweight has been extensively studied (23). Most human studies to date suggest that fetal undernutrition, of which low birthweight is a proxy, may lead to a heightened preference for foods rich in either fat or carbohydrates, although the effects differ by sex. Studies in low-birthweight children are conflicting, with two studies showing greater fat intake (one of which is only in boys) (20, 24), another study demonstrating greater sugar intake in girls only (25) and another lower fat intake in boys (21). Studies in adults reveal a more consistent pattern whereby intrauterine adversity or growth restriction leads to a greater intake of either high-fat (10, 11, 13) or high-carbohydrate foods (12), or a decreased intake of low calorie foods such as fruits and vegetables (14). Two such studies (11, 12) found a stronger effect in females compared to males. Importantly, all the studies in humans have measured intake using food frequency questionnaires or food diaries. While these are valuable tools to learn about food preferences and eating patterns, they are not appropriate methods to reliably assess energy intake due to an inability to precisely measure the estimated energy contents of food items and possible systematic underreporting from some individuals due to recall and/or social desirability biases (26). The present study avoids these methodological issues by measuring energy intake objectively, using a highly reproducible computerized vending machine paradigm (19) that has been shown to predict weight gain in free-living conditions (27).
Our hypothesis of greater ad-libitum energy intake among adults with lower birthweight is supported in males but not females. Lower birthweight males consumed greater energy from fat and greater energy overall (Fig. 1), supporting a heightened preference for palatable foods in males who were smaller at birth. The findings resemble those from prior studies which found a relationship between poor intrauterine nutrition or smaller birth size and greater fat intake in adulthood (10, 11, 13). However, we found the opposite relationship for females, as lower birthweight females tended to consume less energy overall, driven by a decreased consumption of foods high in fat and protein (Fig. 1). Such findings suggest there are sex-specific mechanisms moderating the relationship between birthweight and energy intake in adulthood.
Evidence for sexually dimorphic effects of developmental programming has been discussed at length (28), and these processes are believed to be driven by sex differences in epigenetic regulation, action of hormones, or placental development. Epigenome-wide analyses have revealed patterns of DNA methylation in newborns that are sex-specific (29), and DNA methylation is systemically altered in children born during nutritionally challenging conditions (30). Children who were exposed to the Dutch Hunger Winter during prenatal development showed sex-specific changes in DNA methylation patterns across multiple gene loci important for metabolism, including significantly greater methylation within the leptin gene in exposed males compared to unexposed sibling controls (31). Given leptin’s known roles in fetal development and appetite regulation (32, 33), sex effects on its expression within the intrauterine environment could differentially impact patterns of energy intake in adulthood.
Sex differences in placental structure and development could also underpin distinct programming effects, as the placenta directly mediates nutrient sensing and transfer between the mother and fetus (34). Gestational exposure to the Dutch famine was also associated with sex-specific differences in placental structure and function, with exposed males showing a relatively greater decrease in placental area compared to females (35), and only males showing an effect of exposure on the relationship of placental area to later development of hypertension (36). As with these differences, it is plausible that sex-specific effects of even mild nutritional deficiencies on placental anatomy and function have long-term consequences on mechanisms controlling energy homeostasis, and these may contribute to sex-specific programming effects on energy intake in adulthood.
Several limitations must be considered when interpreting the present findings. Firstly, the sample size is small and included only 65 individuals, yet the sample size was sufficient to demonstrate a relationship between energy intake and birthweight by sex. Secondly, birthweight is an indirect, surrogate marker of fetal development, and caution should be used when interpreting patterns associated with birthweight compared to more severe forms of intrauterine dietary restriction or famine exposure. Since more extreme birthweights (e.g., < 2500g) were not evaluated, it is unclear how results would differ at these extremes. However, multiple studies point to a continuous relationship between birthweight and metabolic outcomes in adults, suggesting programming effects occur across the spectrum of birthweights (4). An additional limitation is that we did not have information about early postnatal growth, a variable also believed to have long-term programming effects (37). Lastly, the study did not account for potential intrauterine exposure to diabetes, although prior research from our unit found no relationship between such exposure and objectively measured energy intake (38). Despite these limitations, the study has multiple strengths. Unlike previous analyses which adjusted for BMI, the present analysis accounts for independently confounding effects of fat-mass index and fat-free mass index as determined using dual-energy X-ray absorptiometry. This is the first study to demonstrate associations between birthweight and objectively measured energy intake in adulthood. The computerized vending machines provide robust and reproducible patterns of intake (19), and are thus more appropriate methods to detect precise patterns of macronutrient consumption compared to self-report instruments as used in prior studies.