The hypothesis of “developmental origins of adult disease” was first stated by Barker (Barker and Osmond, 1986) in the 20th century, which mainly explained that the adverse influences in the early developmental period could cause permanent changes in physiology and metabolism, which finally results in an increased risk of disease in adulthood. Thus, birth weight (BW) is widely used as an indicator of exposure during the intrauterine period and early life development (Peck et al., 2003; Scharf et al., 2016). Numerous observational studies provided evidence for the correlation between reduced BW and increased risk of adult diseases, such as type 2 diabetes mellitus (T2DM) (Carlsson et al., 1999; Whincup et al., 2008), coronary heart disease (CHD) (Ferrie et al., 2006; Morley et al., 2006), hypertension (Eriksson et al., 2000a; Tamakoshi et al., 2006), and stroke (Eriksson et al., 2000b). In particular, it is worth noting that only women demonstrated an increased risk of T2DM and CHD with a raised BW in a recent sex-specific binary meta-analysis (Knop et al., 2018), indicating that BW is more acceptable in predicting the correlation between several traits and diseases in women. Indeed, early observational studies provided controversial evidence supporting the association between BW and female-related traits, including female-only body mass index (BMI) (Zhao et al., 2012; Jelenkovic et al., 2017), reproductive hormones (estradiol [E₂] (Jasienska et al., 2006; Espetvedt Finstad et al., 2009), testosterone (Ruder et al., 2011), anti-Mullerian hormone [AMH] (Dior et al., 2021)), menarche (Juul et al., 2017; Fan et al., 2018), menopause (Tom et al., 2010; Bjelland et al., 2020), and female-specific diseases (polycystic ovaries syndrome [PCOS] (Cresswell et al., 1997), endometriosis (Olšarová and Mishra, 2020), and leiomyomata (Wise et al., 2012)), which can influence women’s reproductive health and life expectancy. However, whether the identified correlation between BW and these female-related traits represents a truly causal relationship remains uncertain because of bias, pleiotropy, or common confounders during prenatal life (Ruiz-Narváez et al., 2014; Kahn et al., 2017; Lawlor et al., 2017).

Two-sample Mendelian randomization (TSMR), a novel and popular analysis tool, was used to estimate the causal inference in observational studies, which avoided all possible and potential biases from confounding factors. The fundamental theory of TSMR is that during the period when gametes were formatted and combined, the alleles of genetic variants were segregated randomly based on Mendel’s law, which led to their independence with confounding factors such as the environment, age, and sex. To some extent, this implies that the TSMR results are stable and convincing.

In recent years, Mendelian randomization (MR) studies provided evidence of a positive association between lower birth weight (LBW) and T2DM (Huang et al., 2019) and stroke (Wang et al., 2020), a negative association with chronic kidney disease (Yu et al., 2020), and no relationship with asthma (Zeng et al., 2019). Furthermore, no related or specific reports focused on women’s health and diseases exist. Here, a large TSMR analysis was conducted to comprehensively estimate the causality of BW on eight related traits and three common reproductive endocrine diseases in adulthood. Our results remained statistically significant and robust after validating the heterogeneity, sensitivity, and horizontal pleiotropy.

The present MR study clarified the genetic association between BW and female-related traits in the largest sample size of the European population. In this study, we found a positive effect of lower BW on bio-T, whereas LBW demonstrated an adverse effect on SHBG level. In contrast, we also found a causal effect of BW on BMI and lower BW on menarche but no detrimental effects of LBW on female-specific diseases. To the best of our knowledge, this is the first study to examine the likely causal relationship between BW and female traits and diseases based on hereditary information.

Our research results on infant BW and adult BMI are similar to those of most existing studies. Rogers (2003) and Zhao et al. (2012) provided good evidence of an association between high BW and subsequent BMI and an increased risk of overweight in young adults. In addition, Jelenkovic et al. (2017) and Liao et al. (2020) suggested a positive association between high BW and a later high BMI. For each individual, a 1.0 kg of BW increased, accompanied with a 0.33 or 0.9 kg/m² of BMI increase in adulthood, respectively (p < 0.001). Moreover, factors such as genetics, development, and environment could result in individual variations in the concentrations of reproductive hormones. The association between lower BW and hyperandrogenism has been confirmed by almost all published evidence. Petraitiene et al. (2020) and Ruder et al. (2011) stated that small for gestational age or reduced BW girls demonstrated lower SHBG levels (p < 0.05) but higher concentrations of androstenedione, testosterone (T) (p < 0.05), dehydroepiandrosterone sulfate, and free androgen index (p < 0.01). Some studies showed that premature adrenarche results in an increased insulin response and hyperandrogenism in later adulthood (Szathmári et al., 2001; Schulte et al., 2016; Novello and Speiser, 2018). However, the relationship between lower BW and E₂ levels remains controversial. Ruder et al. (2011) and Espetvedt Finstad et al. (2009) found an inverse association between BW and levels of E₂, while Tworoger et al. (2006) and Sydsjö et al. (2019) demonstrated no differences in E₂ levels between LBW women and controls. Although no direct evidence exists to prove the positive relationship between BW and E₂, we might estimate an association between the ponderal index at birth/birth size and E₂ (Jasienska et al., 2006; Espetvedt Finstad et al., 2009). Furthermore, our research was the same with that of Kerkhof et al. (2010) and Sydsjö et al. (2019), concluding that BW did not affect AMH concentrations. However, Dior et al. (2021) reported a significant association between lower BW and reduced AMH levels in 32-year-old women who were identified after adjusting for confounders (β = 0.18, p < 0.05).

It is known that both genetic and environmental factors, such as smoking, body fat content, exposure to endocrine-disrupting chemicals, and BW, may result in early AAM (Behie and O'Donnell, 2015; Epplein et al., 2010; Żelaźniewicz et al., 2020; Adair, 2001). In our MR study, a surprising result was confirmed as well as in numerous observational studies. Moreover, Fan et al. (2018), Juul et al. (2017) , and Morris et al. (2010) discovered that lower BW in infancy may increase the risk of early menarche (p < 0.001). However, other studies found that no or an inverse relationship was found between BW and AAM (Sorensen et al., 2013; Sydsjö et al., 2019). Moreover, conclusions on LBW and ANM have not yet been unified, and positive (Alexander et al., 2014; Bjelland et al., 2020; Goldberg et al., 2020), or inverse (Tom et al., 2010), and even no relationship (A.Treloar et al., 2000) exist.

PCOS, endometriosis, and leiomyoma were regarded as the main female endocrine diseases that may affect reproduction. Although, we did not identify any causal effects of LBW on these three diseases. Few studies considered BW as an independent risk factor related to PCOS (Fulghesu et al., 2015), and they suggested that the high risk of PCOS and related traits are because of high BW (Cresswell et al., 1997; Michelmore et al., 2001). Almost all studies confirmed a correlation between LBW and endometriosis (Olšarová and Mishra, 2020). The studies of Gao et al. (2019), Gao et al. (2020) and Borghese et al. (Borghese et al., 2015) supported the fetal origins hypothesis of endometriosis (hazard ratio [HR] = 1.35, 95% CI = 1.08–1.67; OR = 1.5, 95% CI = 1.0–2.3, p < 0.05); even after adjusting for confounding factors, the results still remained (risk ratio = 1.3, 95% CI = 1.0–1.8, p < 0.05) (Missmer et al., 2004). Furthermore, the results of Aarestrup et al. (2020) and Wolff et al. (2013) studies did not reach statistical significance and confirmed this relationship. Last, limited evidence exists of an association between leiomyomas and BW (Wise et al., 2012).

The current study exhibited a few strengths. The major preponderance was the MR design, which cut down remaining confounders and reverse causality and, thereby, improved the causal inference in associations of lower BW with female-related traits. To avoid false-positive results and bias, we selected different sourced populations to eliminate overlapping, and two-sample MR analysis was performed via SNPs to analyze the causal relationship from exposure to outcomes. Next, the random allocation of individual genetic variation during gamete binding is used as an IV; thereby, MR analysis can largely avoid the influence of confounders, artificial errors, and bias and provide high-quality evidence. This MR analysis included sufficient samples and only included European participants to improve the dependability of the results. To conclude, the MVMR analysis was vital for exploring the direct correlation between LBW and female outcomes.

Next, inevitably, the present study demonstrates several limitations. First, GWAS were obtained only from European individuals in this study, whose results are not representative of other races or geographic areas. Second, parts of the summary dates were incomplete because of the privacy policy and long application period, which caused a lost partial population; in addition, the BW datasets obtained contained both males and females, which may lead to collider bias (Fry et al., 2017). To conclude, the present study mainly focused on the causal role of LBW on female-specific traits, but the underlying mechanisms remain to be elucidated.

In summary, this analysis demonstrated that BW is positively associated with BMI in adulthood. In addition, LBW exhibits causal effects on decreased SHBG levels, increased bio-T levels, and early AAM.