Associations of maternal anthropometrics with newborn anogenital distance and the 2:4 digit ratio

Abstract

STUDY QUESTION

Are maternal anthropometrics associated with anogenital distance (AGD) and 2:4 digit ratio (2:4D) in newborns?

SUMMARY ANSWER

Select maternal anthropometrics indicative of obesity or increased adiposity are associated with elongated AGD in daughters.

WHAT IS KNOWN ALREADY

Excessive maternal weight or adiposity before or in early pregnancy may impact child reproductive, and other hormonally mediated, development. AGD and 2:4D are proposed markers of in utero reproductive development.

STUDY DESIGN, SIZE, DURATION

This study includes 450 mother/newborn dyads participating in the Illinois Kids Development Study (I-KIDS), a prospective pregnancy cohort from Champaign-Urbana, IL, USA. Participants included in the current study enrolled between 2013 and 2018.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Most mothers in this study were college-educated (82%) and non-Hispanic White (80%), and 55% were under- or normal weight before pregnancy. Pregnant women aged 18–40 years reported pre-pregnancy weight and height to calculate pre-pregnancy BMI. At 8–15 weeks gestation, we measured waist and hip circumference, and evaluated weight, % body fat, visceral fat level, % muscle and BMI using bioelectrical impedance analysis. Within 24 h of birth, we measured newborn 2nd and 4th left/right digits to calculate the 2:4D. In daughters, we measured AGD_AF (anus to fourchette) and AGD_AC (anus to clitoris). In sons, we measured AGD_AS (anus to scrotum) and AGD_AP (anus to base of the penis).

MAIN RESULTS AND THE ROLE OF CHANCE

Select maternal anthropometrics were positively associated with AGD in newborn daughters, but not sons. For example, AGD_AC was 0.73 mm (95% CI: 0.15, 1.32) longer for every interquartile range (IQR) increase in pre-pregnancy BMI and 0.88 mm (95% CI: 0.18, 1.58) longer for every IQR increase in hip circumference, whereas AGD_AF was 0.51 mm (95% CI: 0.03, 1.00) and 0.56 mm (95% CI: 0.03, 1.09) longer for every IQR increase in hip and waist circumference, respectively. Quartile analyses generally supported linear associations, but additional strong associations emerged in Q4 (versus Q1) of maternal % body fat and visceral fat levels with AGD_AC. In quartile analyses, we observed only a few modest associations of maternal anthropometrics with 2:4D, which differed by hand (left versus right) and newborn sex. Although there is always the possibility of spurious findings, the associations for both measures of female AGD were consistent across multiple maternal anthropometric measures, which strengthens our conclusions.

LIMITATIONS, REASONS FOR CAUTION

Our study sample was racially and ethnically homogenous, educated and relatively healthy, so our study may not be generalizable to other populations. Additionally, we may not have been powered to identify some sex-specific associations, especially for 2:4D.

WIDER IMPLICATIONS OF THE FINDINGS

Increased maternal weight and adiposity before and in early pregnancy may lengthen the female AGD, which warrants further investigation.

STUDY FUNDING/COMPETING INTEREST(S)

This publication was made possible by the National Institute for Environmental Health Sciences (NIH/NIEHS) grants ES024795 and ES022848, the National Institute of Child Health and Human Development grant R03HD100775, the U.S. Environmental Protection Agency grant RD83543401 and National Institute of Health Office of the Director grant OD023272. Its contents are solely the responsibility of the grantee and do not necessarily represent the official views of the US EPA or NIH. Furthermore, the US EPA does not endorse the purchase of any commercial products or services mentioned in the publication. This project was also supported by the USDA National Institute of Food and Agriculture and Michigan AgBioResearch. The authors declare no competing interests.

TRIAL REGISTRATION NUMBER

N/A.

pregnancy, obesity, anogenital distance, 2:4 digit ratio, sex-specific

Introduction

Over 50% of US women entering pregnancy are overweight or have obesity (Fisher et al., 2013; Branum et al., 2016), which is a major public health concern (Mitanchez and Chavatte-Palmer, 2018). Children of women with overweight/obesity have a higher risk of adulthood overweight/obesity (Godfrey et al., 2017; Chen et al., 2019) and neurodevelopmental or psychiatric disorders in childhood, including attention-deficit hyperactive disorder, cerebral palsy, autism spectrum disorder and anxiety (Edlow, 2017). In addition to these well-characterized consequences of maternal obesity, it is important to consider the impact of maternal obesity on child reproductive development. Limited studies suggest that maternal obesity is associated with earlier menarche in daughters (Keim et al., 2009; Deardorff et al., 2013), as well as earlier onset of puberty and decreased sperm DNA integrity in sons (Hakonsen et al., 2012). Given the recent declines in reproductive capacity in both men and women (Levine et al., 2017; Rossen et al., 2018), substantially more needs to be understood about the connection between maternal obesity and offspring reproductive health.

Anogenital distance (AGD, the distance between the anus and genitals) and the 2nd to 4th digit ratio (2:4D, the ratio of the length of the 2nd (index) finger divided by the length of the 4th (ring) finger) are proposed markers of reproductive development and, possibly, capacity (McIntyre, 2006; Liu et al., 2014). AGD and 2:4D are sexually dimorphic traits with biological females generally having a shorter AGD and larger 2:4D compared to males (Salazar-Martinez et al., 2004). Importantly, experimental and observational studies suggest both AGD and 2:4D are likely programmed in utero and may be sensitive to androgens (Galis et al., 2010; Auger et al., 2013; Dean and Sharpe, 2013) and estrogens (Manning et al., 2014) in the case of 2:4D. Thus, their developmental trajectory may provide insight into the hormonal milieu of the intrauterine environment. For example, experimental studies have shown that male offspring of pregnant rats exposed to anti-androgenic endocrine disrupting chemicals (EDCs), such as phthalates and pesticides have reduced AGDs (Wolf et al., 1999; Welsh et al., 2008), whereas pregnant rats exposed to testosterone gave birth to female pups with longer AGDs (Welsh et al., 2008). In newborn mice, the fourth digit has higher levels of nuclear androgen receptor (AR) and estrogen receptor α (ERα) compared to the second digit. Additionally, Ar deletion in the limbs of mice produced a larger 2nd to 4th digit ratio, whereas Erα deletion produced a smaller ratio (Zheng and Cohn, 2011), demonstrating that androgens and estrogens govern 2:4D development. In human infants, maternal pregnancy exposure to EDCs bisphenol A (BPA) or glyphosate was associated with shorter measures of AGD from the anus to fourchette (Barrett et al., 2017), longer measures of AGD from the anus to clitoris (Barrett et al., 2017; Lesseur et al., 2021) and a shorter right 2:4D (Arbuckle et al., 2018) in female, but not male infants. Because AGD and 2:4D may be programmed in utero, additional studies are needed to understand how other early-life developmental stressors, such as maternal overweight/obesity, impact these hormonally mediated markers.

Although to our knowledge, no studies have evaluated associations of maternal obesity/adiposity with AGD and 2:4D, polycystic ovarian syndrome (PCOS) is one frequently studied maternal physiological stressor associated with AGD androgenization. Several previous epidemiologic studies have considered maternal BMI as a potential confounding factor for associations of PCOS with AGD (Barrett et al., 2018; Glintborg et al., 2019). However, maternal weight status or adiposity should also be evaluated as an independent predictor of fetal reproductive development. Preliminary epidemiological evidence shows that compared to being normal weight, maternal obesity is associated with lower placental estradiol synthesis (Gomes et al., 2014) and higher blood testosterone levels in women carrying males (Maliqueo et al., 2017). These observations are likely due to the hormonal activity of adipose tissue, as it is a major site of estrogen and androgen metabolism (Belanger et al., 2002; Kershaw and Flier, 2004). Given the complex interplay between hormones and adipose tissue, and the increased prevalence of overweight/obesity among reproductive-age women, our primary objective was to evaluate associations of pre- and early-pregnancy maternal anthropometric measures with AGD and 2:4D. Importantly, we considered several measures of body size, shape and composition to account for potential differences in associations by adipose tissue distribution and amount.

Materials and methods

Illinois Kids Development Study (I-KIDS) recruitment and enrollment

Pregnant women in this study were participants in Illinois Kids Development Study (I-KIDS), an ongoing prospective pregnancy and birth cohort investigating the impact of prenatal environmental chemical exposures on early offspring neurodevelopment. Details about recruitment and enrollment have been extensively described elsewhere (Pacyga et al., 2021). Briefly, women were recruited at their first prenatal appointment from two obstetric clinics in Champaign-Urbana, IL, USA. Interested women were contacted by I-KIDS staff. Eligible women were ≤15 weeks pregnant, 18–40 years old, fluent in English, carrying one fetus, not in a high-risk pregnancy (self-reported as diagnosed by their doctor), living within a 30-min drive to the University of Illinois campus and not planning to move out of the area before the child’s first birthday. Enrolled women provided written informed consent according to the Institutional Review Board at the University of Illinois. This current study includes the first 450 mother-newborn dyads who enrolled in I-KIDS between December 2013 and August 2018 and had at least one pre- or early-pregnancy anthropometric measure and measures of newborn AGD and/or 2:4D.

Collection of sociodemographic and lifestyle information at enrollment

We conducted a home visit between 8 and 15 weeks gestation to obtain information about sociodemographic, health and lifestyle characteristics. Specifically, women reported their age, race/ethnicity, highest level of educational attainment, marital status, employment status, smoking in the first trimester, parity and any prior PCOS diagnosis. Women also reported the race/ethnicity, education and employment status of their partner. Additionally, women completed the 10-item Perceived Stress Scale (PSS), which has been validated for use in pregnancy to assess psychological stress (Cohen et al., 1983; Karam et al., 2012). At 8–15 and 32–40 weeks gestation, participants also provided information about their diet during the previous 3 months via a semi-quantitative food frequency questionnaire (FFQ) adapted from the Block-98 FFQ (NutritionQuest, Berkeley, CA, USA) and validated in pregnant populations (Bodnar and Siega-Riz, 2002; Boucher et al., 2006; Laraia et al., 2007). Maternal diet information was used to calculate the Alternative Healthy Eating Index 2010 (AHEI-2010) in early and late gestation. AHEI-2010 is an 11-component (maximum of 110 points) diet quality index based on foods/nutrients predictive of chronic disease risk where a higher score reflects better diet quality (McCullough et al., 2002). We used the mean of early- and late-pregnancy AHEI-2010 to estimate maternal diet quality in pregnancy.

Collection of maternal pre- and early-pregnancy anthropometric data

Self-reported weight (lbs) and height (ft and in) were used to calculate pre-pregnancy BMI (kg/m² = (lb/in²) × 703), which has been validated against BMI calculated from measured weight and height (Tomeo et al., 1999). At 8–15 weeks gestation, trained I-KIDS researchers conducted triplicate measures of waist and hip circumference (cm) in accordance to standard Center for Disease Control protocols (Centers for Disease Control and Prevention (CDC), 2011). We also used a bioelectrical impedance analysis (BIA) scale (Bio-Impedance Omron HBF-510W) to measure the following in triplicate: total body fat (%), muscle (%), visceral fat (level), weight (lbs) and BMI (kg/m²). The BIA scale calculated BMI from self-reported height and measured weight. BIA has been validated during pregnancy, and early-pregnancy measures ensure that fluid volume and the fetus do not interfere substantially with estimates of maternal body composition (Van Loan et al., 1995; Larciprete et al., 2003; Loy et al., 2013). We used the mean of the triplicate waist and hip circumference, as well as BIA measures. Additionally, we calculated the waist/hip ratio, which has been deemed to be better than BMI for predicting body fat distribution in pregnant women (Brown et al., 1996; Berends et al., 2009; Salem et al., 2012). We also calculated the waist/height ratio and the body fat/muscle ratio, which have been associated with adverse maternal and newborn health outcomes (Yoo, 2016; Gamboa-Gomez et al., 2019).

Measurement of newborn AGD and 2:4D

We visited mothers and newborns in the hospital ∼24 h after delivery. Newborn sex was recorded at birth. To calculate the 2nd to 4th digit ratio, trained researchers measured the 2nd and 4th digits of the infants’ left and right hands in triplicate from the fold at the base of the finger to the tip using a flexible paper ruler. We used digital calipers (Mitutoyo Model 500-195-20) to measure newborn AGD in triplicate using previously published protocols (Sathyanarayana et al., 2010). In daughters, the ano-clitoral AGD (AGD_AC) represents the distance from the center of the anus to the clitoral hood, whereas the ano-fourchette AGD (AGD_AF) represents the distance from the center of the anus to the posterior fourchette. In sons, the ano-penile AGD (AGD_AP) represents the distance from the center of the anus to the cephalad insertion of the penis, whereas the ano-scrotal AGD (AGD_AS) represents the distance from the center of the anus to the base of the scrotum. Additionally, we measured body length (cm) in triplicate using a measuring mat. In the few cases where we were unable to measure body length at the hospital (n = 27), we used body length information obtained from hospital crib cards, if available. We used the mean of the triplicate newborn measures in statistical analyses.

Collection and processing of urine samples for phthalate/replacement biomarker quantification

In sensitivity analyses, we accounted for potential confounding of our associations by phthalates, which are non-persistent endocrine-disrupting chemicals that are known obesogens and were previously shown to be associated with AGD (Yaghjyan et al., 2015; Santos et al., 2021). Urine collection, processing and storage protocols for phthalate metabolite assessment were extensively described elsewhere (Pacyga et al., 2021). Urinary phthalate metabolite concentrations were quantified at the CDC laboratories using established protocols (Silva et al., 2007).

Statistical analysis

Of the 450 mother/newborn dyads available for statistical analysis, the final sample size for each analysis differed depending on each exposure/outcome relationship and is outlined in Supplementary Fig. S1. We did not have BIA data for the earliest study participants and some women opted out of this analysis, thus we evaluated associations of early-pregnancy BIA measures with AGD/2:4D in a smaller sample of women (73.3% of our analytic sample). To select covariates, we first used the prior literature and our own data to generate a directed acyclic graph (DAG) (Textor et al., 2016; Supplementary Fig. S2) and determined the minimum sufficient adjustment set of covariates needed to evaluate our proposed associations (VanderWeele and Robins, 2010). Second, we reviewed splines to help discern an appropriate functional form of our covariates. Third, we assessed correlations between covariates to test for potential multicollinearity issues, but observed that all covariates were only weakly-to-moderately correlated (r < 0.5, data not shown). In final covariate-adjusted statistical models, we included maternal age (≥30, <30 (reference) years), race/ethnicity (other race/ethnicity, non-Hispanic White (reference)), parity (1, ≥2, 0 prior children (reference)), mean gestational AHEI-2010 (continuous), perceived stress score (≥11.5, <11.5 points (reference)) and smoking status in the first trimester (yes, unknown, no (reference)). We specified models that additionally included marital and employment status, but found that associations were unchanged with inclusion of these variables (data not shown). All maternal sociodemographic and birth characteristics are reported as n (%) or median (min, max). The distribution of maternal anthropometrics, as well as the distribution of newborn AGD and 2:4D for each sex are reported as median (25th, 75th percentile). We used the Kruskal–Wallis test to evaluate whether newborn AGD and 2:4D are sexually dimorphic in our sample.

We used linear regression models to evaluate the associations of 12 maternal anthropometrics with newborn AGD and 2:4D. We apriori stratified these models by newborn sex because of the sexually dimorphic nature of newborn AGD and 2:4D (Galis et al., 2010). Maternal anthropometrics were evaluated continuously and by quartiles. We checked regression models for non-constant residual variance to ensure model assumptions were met and did not transform newborn AGD and 2:4D since they were normally distributed. For maternal anthropometrics modeled continuously, the resulting β-estimates and 95% CIs were transformed to represent a change in AGD (in mm) for every one interquartile range (IQR) increase in maternal anthropometric (IQR values are reported in Supplementary Table SI). Given the small 2:4D in newborns, our tables present the β-estimate and 95% CI of 2:4D multiplied by 100. When maternal anthropometrics were modeled in quartiles, the resulting β-estimates and 95% CIs represent the change in AGD (in mm) or 2:4D ratio in newborns of women in quartiles two (Q2), three (Q3) and four (Q4) compared to quartile one (Q1) for each maternal anthropometric. We tested for linear trends across quartiles (P_trend) by assessing separate linear regression models that treated the ordinal maternal anthropometric variables as continuous.

We also conducted several sensitivity analyses. First, based on prior studies reporting associations of phthalates with newborn AGD (Swan et al., 2015; Dorman et al., 2018) and those characterizing phthalates as obesogens (Kim and Park, 2014), we evaluated associations of maternal anthropometrics with newborn AGD and 2:4D that additionally accounted for maternal phthalate exposure. To approximate total phthalate exposure (SumPhthalates) in pregnancy, we molar summed 14 urinary phthalate metabolites as follows: SumPhthalates = (MEHP/278) + (MEHHP/294) + (MEOHP/292) + (MECPP/308) + (MCOP/322) + (MiNP/292) + (MCNP/336) + (MCPP/252) + (MBzP/256) + (MEP/194) + (MBP/222) + (MHBP/238) + (MiBP/222) + (MHiBP/238), with each metabolite defined elsewhere (Pacyga et al., 2021). Due to having a skewed distribution, we natural log-transformed SumPhthalates. Second, several studies accounted for gestational age (Kizilay et al., 2021), birth weight (Jain et al., 2018) or weight-for-length z-score (Swan et al., 2015) when evaluating infant AGD as an outcome. In sensitivity analyses, we divided AGD and 2:4D by body length rather than including it as a covariate in statistical models because birth size could be on the causal pathway for our proposed associations. Given the small ratio of AGD or 2:4D to newborn length, our tables present the β-estimate and 95% CI of length-normalized AGD and 2:4D multiplied by 100. Finally, because of the known co-occurrence of PCOS with overweight/obesity, we excluded women with a PCOS diagnosis to understand whether they influenced our associations.

Statistical analyses were conducted in SAS software, version 9.4 (SAS Institute Inc, Cary, NC, USA). Based on recommendations from the American Statistical Association, rather than considering P-values for linear regression models, we assessed patterns in the direction and magnitude of associations, as well as the range of the 95% CI to identify potentially meaningful results (Wasserstein and Lazar, 2016; Amrhein et al., 2019). We did not adjust for multiple comparisons (Rothman, 1990).

Results

Participant characteristics

Characteristics of I-KIDS mother/newborn dyads are reported in Table I. Briefly, the median age of women was 30 years (min = 18, max = 40). Most participants were non-Hispanic White (80%), college-educated (82%), did not have PCOS (95%) and did not report smoking in the first trimester (96%), and nearly half were having their first child (49%). The median (min, max) AHEI-2010 was 55.8 (28.1, 82.8) and PSS was 11.5 (0.0, 39.0). Most fathers were non-Hispanic White (78%), college-educated (71%) and employed (97%) (Table I). Newborn sex was evenly distributed in this sample. The median (min, max) birth weight and gestational length were 3482.7 g (2239.6, 4887.5) and 39.4 weeks (36.0, 42.1), respectively.

Table I

Characteristics of I-KIDS analytic sample (n = 450).

Characteristic	n (%)1 or median (min, max)
MATERNAL CHARACTERISTICS
Maternal age2	30.0 (18.0, 40.0)
Race/ethnicity2 (missing = 1)
Non-Hispanic White (reference)	360 (80.0)
Others3	89 (19.8)
Education
Some college or less	84 (18.7)
College graduate or higher	366 (81.3)
Marital status
Married	396 (88.0)
Unmarried	54 (12.0)
Annual household income (missing = 3)
<$60 000	127 (28.2)
$60 000–$99 999	175 (38.9)
≥$100 000	145 (32.2)
Employment status
Unemployed	64 (14.2)
Employed	386 (85.8)
Parity2
0 children (reference)	222 (49.3)
1 child	146 (32.5)
≥2 children	82 (18.2)
Smoking in the first trimester2
No (reference)	399 (88.7)
Yes	17 (3.8)
Missing	34 (7.5)
Mean pregnancy diet quality2 (missing = 18)	55.8 (28.1, 82.8)
Perceived stress score2 (missing = 6)	11.5 (0.0, 39.0)
Polycystic ovarian syndrome
Yes	21 (4.7)
No	429 (95.3)
NEWBORN CHARACTERISTICS
Newborn sex2
Female	230 (51.1)
Male	220 (48.9)
Birth weight (g)	3482.7 (2239.6, 4887.5)
Size for gestational age (missing = 30)
SGA	20 (4.4)
AGA	341 (75.8)
LGA	59 (13.1)
Gestational age (weeks)	39.4 (36.0, 42.1)
Preterm birth status
<37 weeks	10 (2.2)
≥37 weeks	449 (97.8)
PATERNAL CHARACTERISTICS
Paternal race/ethnicity (missing = 1)
Non-Hispanic White	352 (78.2)
Others	97 (21.6)
Paternal education (missing = 2)
Some college or less	130 (28.9)
College graduate or higher	318 (70.7)
Paternal employment status (missing = 2)
Unemployed	12 (2.7)
Employed	436 (96.9)

1

Percentages may not add up to 100% due to missing data.

2

Covariates included in statistical models.

³

Includes American Indian or Alaska Native, Asian, Black or African American, Hispanic White, Native Hawaiian or other Pacific Islander and others. AGA, appropriate for gestational age; I-KIDS, Illinois Kids Development Study; LGA, large for gestational age; PSS, perceived stress score; SGA, small for gestational age.

Distribution of maternal pre- and early-pregnancy anthropometrics

Pre-pregnancy and early-pregnancy median weights were 150.0 lbs and 156.2 lbs, respectively. Median pre- and early-pregnancy BMI were 24.4 kg/m² and 25.7 kg/m², respectively (Table II). Only 2% of women were underweight prior to pregnancy (<18.5 kg/m²), 53% were normal weight (18.5–24.9 kg/m²), 23% were overweight (25.0–29.9 kg/m²) and 22% were obese (≥30 kg/m²). In early pregnancy, median hip and waist circumference were 104.6 cm and 82.0 cm, respectively, and median waist/hip and waist/height ratios were 0.8 and 0.5, respectively. Median total body fat, visceral fat level, muscle mass and total body fat/skeletal muscle ratio were 39%, 6, 26% and 1.5, respectively.

Table II

Distribution of maternal anthropometric characteristics.

Maternal anthropometric	n1	Median (25th, 75th percentile)
Pre-pregnancy weight (lbs)	450	150.00 (131.00, 176.00)
Early-pregnancy weight (lbs)	330	156.17 (137.27, 183.57)
Pre-pregnancy BMI (kg/m2)	450	24.44 (21.81, 29.12)
Early-pregnancy BMI (kg/m2)	330	25.70 (22.93, 29.97)
Early-pregnancy hip circumference (cm)	431	104.57 (98.00, 113.00)
Early-pregnancy waist circumference (cm)	431	82.00 (76.00, 91.57)
Early-pregnancy waist/hip ratio	431	0.79 (0.75, 0.83)
Early-pregnancy waist/height ratio	431	0.49 (0.46, 0.55)
Early-pregnancy body fat (%)	330	38.52 (32.67, 44.37)
Early-pregnancy visceral fat (level)	330	6.00 (4.00, 7.00)
Early-pregnancy muscle (%)	330	26.12 (24.03, 28.50)
Early-pregnancy body fat/muscle ratio	330	1.46 (1.13, 1.82)

1

Some women are missing data on select anthropometrics because they were either the earliest study participants or they opted out of the bioelectrical impedance (BIA) analysis.

Sex-specific distribution of newborn AGD and 2:4D

Median (25th, 75th percentile) AGD lengths and 2:4D for sons and daughters can be found in Table III. As expected, both AGD measures were longer in sons than in daughters at birth. However, we did not observe meaningful differences in the 2:4D between newborn daughters and sons in either hand.

Table III

Distribution of newborn anogenital distance and 2:4 digit ratios.

	Daughters		Sons
Newborn measure	n	Median (25th, 75th percentile)	n	Median (25th, 75th percentile)	P-value1
AGDAF/AGDAS(mm)	167–228	9.56 (7.31, 11.45)	140–194	19.13 (16.03, 22.18)	<0.0001
AGDAC/AGDAP(mm)	166–226	34.20 (31.45, 35.88)	138–192	44.80 (41.47, 47.87)	<0.0001
Right 2:4D	169–229	0.94 (0.92, 0.97)	150–205	0.93 (0.91, 0.96)	0.14
Left 2:4D	168–229	0.95 (0.92, 0.98)	149–206	0.95 (0.93, 0.97)	0.83

1

Kruskal–Wallis test was used to evaluate differences in the distribution of newborn measures by sex. 2:4D, 2:4 digit ratio; AGD, anogenital distance; AGD_AF, AGD measured from the anus to fourchette in females; AGD_AS, AGD measured from the anus to scrotum in males; AGD_AC, AGD measured from the anus to clitoris in females; AGD_AP, AGD measured from the anus to penis in males. ‘n’ represent sample sizes for infants whose mothers had at least one anthropometric measure. Specific sample sizes for each newborn measure can be found in Supplementary Fig. S1.

Associations of pre- and early-pregnancy maternal anthropometrics with AGD in newborn daughters

In newborn daughters, select maternal body size measures were positively associated with AGD_AF and AGD_AC (Table IV, Fig. 1 and Supplementary Table SII). When modeled continuously, each IQR increase in pre- and early-pregnancy weights were associated with 0.7 mm (95% CI: 0.2, 1.1) and 0.6 mm (95% CI: 0.1, 1.2) longer AGD_AF, and 1.0 mm (95% CI: 0.3, 1.6) and 1.0 mm (95% CI: 0.2, 1.9) longer AGD_AC, respectively. These associations were consistent in quartile analyses (Fig. 1 and Supplementary Table SII). For instance, in newborn daughters of mothers in Q4 of pre- and early-pregnancy weight, AGD_AF was 1.2 mm (95% CI: 0.1, 2.4; P_trend = 0.01) and 1.3 mm (95% CI: 0.1, 2.6; P_trend = 0.02) longer, whereas AGD_AC was 2.4 mm (95% CI: 0.9, 4.0; P_trend = 0.003) and 2.4 mm (95% CI: 0.5, 4.3; P_trend = 0.02) longer, respectively, compared to daughters of women in Q1. Positive associations of maternal pre- and early-pregnancy BMI with AGD_AF were similar, but less robust compared to weight, such that each IQR increase in pre- and early-pregnancy BMI was associated with 0.4 mm (95% CI: 0.0, 0.8) and 0.4 mm (95% CI: 0.0, 0.9) longer AGD_AF, whereas AGD_AC increased by 0.7 mm (95% CI: 0.1, 1.3) and 0.7 mm (95% CI: 0.0, 1.4), respectively (Table IV). Only AGD_AC was meaningfully longer at the upper quartile of pre- and early-pregnancy BMI (Fig. 1 and Supplementary Table SII).

Figure 1.

Associations of maternal anthropometrics in quartiles with anogenital distance (AGD) in newborn daughters. (A) ADG_AF and (B) AGD_AC. Linear regression models accounted for age, parity, race/ethnicity, smoking in the first trimester, diet quality and perceived stress. β-estimates and 95% CIs represent the mm change in AGD at maternal anthropometric quartiles 2 (Q2), 3 (Q3) and 4 (Q4) compared to the first quartile (Q1). ADG_AF, anogenital distance measured from the anus to fourchette; AGD_AC, anogenital distance measured from the anus to clitoris.

Table IV

Associations of continuous maternal anthropometrics with newborn anogenital distance (AGD) and 2:4 digit ratios.

	Daughters		Sons		Daughters		Sons
	AGDAF	AGDAC	AGDAS	AGDAP	Right 2:4D	Left 2:4D	Right 2:4D	Left 2:4D
Maternal anthropometric	β (95% CI)1				β (95% CI)1,2
Pre-pregnancy weight	0.66 (0.19, 1.13)	0.98 (0.32, 1.64)	0.02 (−0.89, 0.93)	0.28 (−0.69, 1.25)	0.28 (−0.45, 1.00)	−0.10 (−0.97, 0.77)	0.21 (−0.64, 1.06)	0.50 (−0.34, 1.35)
Early-pregnancy weight	0.65 (0.12, 1.18)	1.02 (0.19, 1.85)	0.09 (−1.05, 1.23)	−0.35 (−1.57, 0.86)	0.19 (−0.72, 1.10)	−0.29 (−1.41, 0.84)	0.07 (−1.04, 1.18)	−0.09 (−1.14, 0.96)
Pre-pregnancy BMI	0.40 (−0.02, 0.83)	0.73 (0.15, 1.32)	−0.19 (−1.06, 0.69)	0.07 (−0.82, 0.97)	0.24 (−0.44, 0.91)	0.00 (−0.81, 0.82)	0.01 (−0.85, 0.86)	0.30 (−0.58, 1.17)
Early-pregnancy BMI	0.42 (−0.03, 0.87)	0.71 (0.00, 1.41)	−0.40 (−1.55, 0.76)	−0.67 (−1.85, 0.52)	0.19 (−0.58, 0.95)	0.12 (−0.81, 1.06)	−0.15 (−1.25, 0.95)	−0.28 (−1.31, 0.74)
Early-pregnancy hip circumference	0.51 (0.03, 1.00)	0.88 (0.18, 1.58)	−0.01 (−0.98, 0.96)	0.32 (−0.66, 1.30)	0.15 (−0.64, 0.94)	0.32 (−0.59, 1.23)	−0.27 (−1.22, 0.68)	0.14 (−0.79, 1.06)
Early-pregnancy waist circumference	0.56 (0.03, 1.09)	0.73 (−0.05, 1.50)	0.35 (−0.73, 1.43)	0.63 (−0.45, 1.72)	−0.03 (−0.90, 0.84)	−0.20 (−1.24, 0.84)	0.27 (−0.77, 1.31)	0.43 (−0.60, 1.46)
Early-pregnancy waist/hip ratio	0.05 (−0.40, 0.50)	−0.24 (−0.89, 0.40)	0.53 (−0.43, 1.49)	0.55 (−0.43, 1.54)	−0.29 (−1.02, 0.44)	−0.66 (−1.52, 0.21)	0.77 (−0.16, 1.70)	0.52 (−0.43, 1.48)
Early-pregnancy waist/height ratio	0.30 (−0.23, 0.84)	0.50 (−0.29, 1.28)	0.12 (−0.81, 1.04)	0.36 (−0.59, 1.30)	−0.05 (−0.93, 0.83)	−0.09 (−1.14, 0.96)	0.07 (−0.86, 0.99)	0.21 (−0.72, 1.14)
Early-pregnancy visceral fat	0.52 (−0.25, 1.29)	1.01 (−0.18, 2.19)	−0.09 (−1.02, 0.83)	−0.33 (−1.30, 0.65)	0.14 (−1.15, 1.44)	0.92 (−0.68, 2.51)	−0.13 (−1.47, 1.20)	−0.35 (−1.63, 0.92)
Early-pregnancy body fat %	0.26 (−0.36, 0.89)	0.87 (−0.11, 1.84)	−0.13 (−1.44, 1.17)	−0.29 (−1.62, 1.03)	0.21 (−0.85, 1.27)	0.37 (−0.91, 1.64)	−0.17 (−1.39, 1.05)	−0.30 (−1.42, 0.82)
Early-pregnancy muscle %	0.03 (−0.60, 0.66)	−0.66 (−1.64, 0.32)	0.15 (−0.98, 1.27)	0.16 (−1.01, 1.33)	−0.14 (−1.19, 0.92)	−0.48 (−1.78, 0.83)	0.19 (−0.87, 1.26)	0.19 (−0.81, 1.19)
Early-pregnancy body fat/muscle ratio	0.20 (−0.41, 0.81)	0.70 (−0.24, 1.64)	−0.21 (−1.37, 0.96)	−0.39 (−1.64, 0.85)	0.21 (−0.82, 1.23)	0.28 (−0.99, 1.55)	−0.09 (−1.19, 1.01)	−0.13 (−1.16, 0.90)

1

Data are presented as the change (β-estimate) in AGD and 2:4D for every IQR increase in maternal anthropometric.

2

2:4D Δ and 95% CIs were multiplied by 100 to improve data reporting. Linear regression models evaluated sex-specific associations of maternal anthropometrics with AGD_AF, AGD_AC, AGD_AS, AGD_AP and 2:4D for the right and left hands. Models adjusted for age, parity, race/ethnicity, smoking in the first trimester, mean pregnancy diet quality and perceived stress. Bold indicates potentially meaningful relationships. Specific sample sizes for each association can be found in Supplementary Fig. S1. 2:4D, 2:4 digit ratio; AGD, anogenital distance; AGD_AF, AGD measured from the anus to fourchette in females; AGD_AS, AGD measured from the anus to scrotum in males; AGD_AC, AGD measured from the anus to clitoris in females; AGD_AP, AGD measured from the anus to penis in males.

In newborn daughters, select maternal body shape and composition measures were also associated with AGD_AF and AGD_AC (Table IV, Fig. 1 and Supplementary Table SII). Maternal hip and waist circumference were positively associated with AGD_AF (β: 0.5 mm, 95% CI: 0.0, 1.0; β: 0.6 mm, 95% CI: 0.0, 1.1, respectively). Hip circumference was also associated with 0.9 mm (95% CI: 0.2, 1.6) longer AGD_AC, but associations were less consistent for waist circumference (β: 0.7 mm, 95% CI: −0.1, 1.5) (Table IV). Quartile analyses were consistent with linear associations at the highest quartile of hip, but not waist, circumference with AGD_AF and AGD_AC (Fig. 1 and Supplementary Table SII). In models where visceral fat and total body fat were modeled continuously, there was a moderate, positive association with AGD_AC, but not AGD_AF (Table IV). In quartile analyses, AGD_AC was longer in daughters of women in Q4 of visceral fat (β: 2.9 mm, 95% CI: 0.5, 5.3, P_trend = 0.09) and total body fat (β: 2.0 mm, 95% CI: 0.1, 3.9, P_trend = 0.08) (Fig. 1 and Supplementary Table SII). Overall, we noted few meaningful differences from our primary linear regression findings in sensitivity analyses that additionally accounted for phthalate exposure, that normalized AGD to body length or that excluded women with PCOS (Supplementary Table SIII).

Associations of maternal pre- and early-pregnancy anthropometrics with AGD in newborn sons

When modeled continuously, maternal anthropometrics were not associated with AGD_AS or AGD_AP in newborn sons (Table IV). However, in quartile analyses, we observed moderate associations of total body fat with AGD_AP, but not AGD_AS. For example, compared to Q1, sons of mothers in Q2 of total body fat had 3.1 mm (95% CI: 0.6, 5.5; P_trend = 0.79) longer AGD_AP, which did not persist in women who were in Q3 or Q4. We did not observe meaningful differences between findings from sensitivity analyses and those from primary linear regression models (Supplementary Table SIII).

Associations of maternal pre- and early-pregnancy anthropometrics with 2:4D in newborn daughters

When modeled continuously, maternal anthropometrics were not associated with 2:4D in newborn daughters (Table IV). In quartile analyses, several associations emerged in Q2 and Q3 (compared to Q1), but not in Q4 (Supplementary Table SII). The right 2:4D tended to be smaller in daughters of mothers in Q3 of waist/height ratio (β: −0.02, 95% CI: −0.03, 0.00, P_trend = 0.91), but not in Q2 or Q4. However, several maternal anthropometrics were positively associated with the left 2:4D. For example, newborn daughters of women in the lower quartiles (Q2 and Q3) of pre- and early-pregnancy BMI had moderately larger 2:4D compared to those in Q1. Additionally, compared to daughters of women in Q1, left 2:4D was longer in daughters whose mothers were in Q3 of both hip circumference and waist/hip ratio. We did not observe meaningful differences between these findings and sensitivity analyses (Supplementary Table SIII).

Associations of maternal pre- and early-pregnancy anthropometrics with 2:4D in newborn sons

In newborn sons, maternal anthropometrics were not associated with 2:4D when modeled continuously (Table IV). In quartile analyses (Supplementary Table SII), compared to Q1, the right 2:4D was smaller in newborn sons of women in Q2 of visceral fat (β: −0.03, 95% CI: −0.06, 0.00, P_trend = 0.84), but larger in Q2 of muscle % (β: 0.03, 95% CI: 0.01, 0.05, P_trend = 0.72). Finally, compared to Q1, the right 2:4D was smaller in newborn sons of mothers in Q2 of waist circumference (β: −0.02, 95% CI: −0.04, 0.00, P_trend = 0.41), but larger in Q2 of waist/hip ratio (β: 0.02, 95% CI: 0.00, 0.04, P_trend = 0.28). As with other findings, we did not note meaningful differences between these linear regression results and those from sensitivity analyses (Supplementary Table SIII).

Discussion

Summary of major findings

Our findings suggest that pre- and early-pregnancy maternal anthropometric measures are consistently associated with alterations in newborn AGD. Associations were most prominent in daughters, where having higher maternal pre- and early-pregnancy weight, pre- and early-pregnancy BMI, hip circumference, waist circumference, body fat and visceral fat appeared to be important predictors of longer AGD_AC or AGD_AF. However, there were few associations between maternal anthropometrics and AGD in sons or the 2:4D in either sex. Importantly, in daughters, certain maternal measures of body composition and shape (visceral fat and hip circumference) appeared to be robust at predicting increases in AGD particularly at higher quartiles of adiposity.

Associations of maternal anthropometrics with newborn AGD differed by sex, with potential implications for women’s health

Our study suggests that pre- and early-pregnancy maternal obesity predominantly alters the development of the female AGD, which agrees with many prior studies reporting sex-specific impacts of various inutero exposures on AGD. For example, a 2013 study reported that maternal stressful life events were associated with longer AGD in daughters, but not sons, with similar patterns observed with maternal smoking (Barrett et al., 2013; Kizilay et al., 2021). Other studies have identified associations of EDCs, such as polychlorinated biphenyls and BPA, with reduced AGD in newborn sons, but not in daughters (Mammadov et al., 2018; Sheinberg et al., 2020). Interestingly, exposure to certain phthalates has been associated with both shorter and longer AGD_AC in newborn daughters, but only longer AGD_AP in sons (Arbuckle et al., 2018).

In the context of our findings, elongation of the female AGD may pose a concern for future reproductive health of women. One study in college-aged Spanish women observed a positive association between AGD_AF or AGD_AC and ovarian follicle count (Mendiola et al., 2012). Authors discussed that a large ovarian follicle count may be indicative of excess androgen exposure, which was also demonstrated in animal models (Steckler et al., 2005; Mendiola et al., 2012). A large ovarian follicle count is often related to rapid follicular recruitment, which may result in early ovarian reserve depletion and a shortened reproductive window marked by earlier onset of menopause (Broekmans et al., 2004; Wellons et al., 2013). An elongated AGD_AF in women has also been associated with elevated levels of testosterone (Mira-Escolano et al., 2014), and women with longer AGD_AF in adulthood were shown to experienced menarche at an earlier age compared to women with shorter AGD_AF (Barrett et al., 2015). Age at menarche is a predictor of reproductive health, such that early menarche is associated with fertility issues and decreased parity (Gillette and Folinsbee, 2012). Although no large-scale longitudinal studies have confirmed the persistence of AGD measures from birth to adulthood, additional studies are warranted to understand newborn AGD as a predictor of menarche and other markers of female reproductive health.

Associations of maternal anthropometrics with 2:4D differed by newborn sex

Our findings suggest that select maternal anthropometrics may be non-linearly associated with 2:4D development, and these associations are both hand- and sex-specific. Although no prior studies (to our knowledge) have evaluated associations of maternal obesity/adiposity with 2:4D in newborns, one study reported negative associations of maternal plasma testosterone with 2:4D in the left hand of newborn daughters (Ventura et al., 2013), further supporting the development of 2:4D in utero, as well as our hand-specific associations. Another study evaluated the association between maternal smoking and 2:4D in 8-year-old children and only observed a smaller 2:4D in the right hand of boys exposed to maternal smoking, compared to those not exposed (Rizwan et al., 2007). In addition to providing a window into the inutero environment, 2:4D is important to study because reversal or reduction of typical sex differences may be indicative of impaired reproductive capacity (Honekopp et al., 2007). Interestingly, we did not observe obvious differences in 2:4D between newborn sons and daughters in our study, despite previously reported larger 2:4D in newborn daughters compared to sons. Potentially because androgens and estrogens are needed to stabilize the ratio, both hand (Knickmeyer et al., 2011; Zheng and Cohn, 2011; Ventura et al., 2013) and sex differences in 2:4D may be inconsistent through the first two years of life (Knickmeyer et al., 2011). Furthermore, because sex differences in 2:4D are relatively small, our study may have been underpowered to detect them. Therefore, it may be important to consider the impact of maternal weight status and adiposity on children’s 2:4D later in life and in larger populations of newborns.

Potential biological drivers of our findings

Although we were unable to consider hormonal mediation in our study, adiposity-driven hormonal disruption is an important potential biological mechanism of our observed relationships. Obesity in pregnant women has been associated with decreased activity in the hypothalamic–pituitary–adrenal axis, such that pregnant women with obesity have lower cortisol, estriol, estradiol and progesterone compared to normal weight women (Stirrat et al., 2016). As was previously discussed, the development of AGD and 2:4D are both sensitive to maternal estrogens and androgens, so hormonal changes in women with higher body weight and adiposity could have sex-specific deleterious impacts on the development of the reproductive tract. As previously discussed, maternal PCOS is an excellent example of maternal androgen-mediated disruption in AGD development (Barrett et al., 2018). While PCOS and obesity/metabolic syndrome co-occur, the cause-effect direction of the relationship between obesity and PCOS is complex (Barber et al., 2019; Li et al., 2019). Given our findings in women without PCOS, future studies evaluating the impact of maternal PCOS on child AGD may wish to consider weight or adiposity as mediators of these relationships and could also consider whether associations of maternal PCOS with child AGD differ depending on maternal body size or composition.

Despite the biological plausibility of our findings, it is important to note that reproductive characteristics like age at menarche, age at menopause and pregnancy characteristics are generally similar from mother to daughter (Pouta et al., 2005; Steiner et al., 2008). With regards to the 2:4D, studies have also observed strong similarities in the ratio between children and their parents (Voracek and Dressler, 2009; Banyeh et al., 2021). Furthermore, a cross-sectional study in Polish children aged 6–13 reported a positive association of right hand 2:4D with their own hip circumference, waist circumference and BMI (Kobus et al., 2021), suggesting that obesity and 2:4D may ‘track’ together. However, no studies (including ours) have considered whether women with obesity also have longer AGD, which contributes to their daughters’ AGD, or whether their own 2:4D has also been disrupted during a sensitive developmental window. Thus, future studies may need to account for heritability when evaluating whether maternal adiposity acts as an endocrine disruptor and creates a milieu that contributes to the elongation of AGD in their newborn daughters or to changes in the 2:4D in their daughters and sons.

Strengths and limitations

This study had important strengths and limitations. Although it may have been underpowered to detect some associations stratified by fetal sex, trained researchers measured our exposures and outcomes in triplicate, thereby improving precision. While the homogenous nature of our sample limits generalizability of our findings, our study design allowed us to evaluate an extensive panel of pre- and early-pregnancy maternal anthropometrics. Although there is always the potential for residual confounding (e.g. by maternal physical activity, which we did not collect in our study), we made decisions to include covariates using a DAG that was developed using a priori considerations and relevant literature. Additionally, we conducted sensitivity analyses to include more widely studied determinants of AGD, such as phthalates and PCOS, and our associations of interest did not significantly differ when we accounted for these factors. Though we are unable to establish causality, our associations were temporal and biologically plausible. Finally, because no prior studies have evaluated the associations between maternal obesity and infant reproductive health predictors, we are unable to compare our findings to prior literature. However, our findings provide an important new avenue of research related to maternal obesity/adiposity and child reproductive health.

Conclusions

To our knowledge, this is the first study to evaluate associations of maternal anthropometrics with AGD and 2:4D in newborns. Results from our study suggest that higher maternal weight and BMI may alter the development of AGD in daughters. Additionally, we observed strongest associations with AGD in daughters at the highest quartiles of hip circumference and visceral fat, suggesting that these early-pregnancy measures may better represent maternal adiposity than traditional body composition measures (i.e. BMI). As we discussed, elongation of the female AGD could be an early indication of future reproductive capacity, such as ovarian reserve. Given the increasing prevalence of overweight/obesity and emerging infertility among reproductive age women, future studies in other populations are needed to corroborate our findings. Additional studies may also consider investigating whether maternal hormonal disruption or weight change across pregnancy mediate the observed associations, and may also consider evaluating the role of maternal AGD or 2:4D in the observed associations. This knowledge will provide a better understanding of the mechanism by which maternal obesity impacts offspring reproductive development. Finally, understanding early life indicators of offspring reproductive health could spur future interventions that target the populations at risk.

Data Availability

The data underlying this article are available in the article and in its online supplementary material.

Authors’ roles

M.E.K. contributed to the statistical analyses, interpretation of the data and drafting of the paper. D.C.P. assisted with statistical analyses and data interpretation, while providing critical revisions to the draft. J.C.G. assisted with statistical analyses and draft revisions. J.A.F. and S.L.S. contributed to the study design and data collection, and provided and approved draft revisions. R.S.S. conceived the study, assisted with data interpretation, provided essential draft revisions and gave approval for publication.

Funding

This publication was made possible by the National Institute for Environmental Health Sciences (NIH/NIEHS) grants ES024795 and ES022848, the National Institute of Child Health and Human Development (NICHD) grant R03HD100775, the U.S. Environmental Protection Agency grant RD83543401 and National Institute of Health Office of the Director grant OD023272. Its contents are solely the responsibility of the grantee and do not necessarily represent the official views of the US EPA or NIH. Furthermore, the US EPA does not endorse the purchase of any commercial products or services mentioned in the publication. This project was also supported by the USDA National Institute of Food and Agriculture and Michigan AgBioResearch.

Conflict of interest

The authors declare no conflicts of interest.