Maternal body burden of cadmium and offspring size at birth

Abstract

Increasing evidence suggests an inverse association between cadmium (Cd) and size at birth, potentially greatest among female neonates. We evaluated whether greater maternal body burden of Cd is associated with reduced neonatal anthropometry (birthweight, birth length, head circumference, and ponderal index) and assessed whether these associations differ by infant sex. The analytic sample for the present study (n=396) was derived from a subcohort of 750 women randomly drawn from among all participants (N=4,344) in the Omega Study, a prospective pregnancy cohort. Creatinine-corrected Cd in maternal clean-catch spot urine samples (U-Cd) was quantified by inductively coupled plasma mass spectrometry. Continuous log₂-transformed Cd (log₂-Cd) and U-Cd tertiles (low <0.29 µg/g creatinine, middle 0.29–0.42 µg/g creatinine, high ≥0.43 µg/g creatinine) were used in multivariable linear regression models. Females had reduced birth length with greater U-Cd tertile, whereas males birth length marginally increased [β(95% CI) females: low=reference, middle= −0.59 cm (−1.37, 0.19), high=–0.83 cm (−1.69, 0.02), p-trend=0.08; males: low=reference, middle=0.18 cm (−0.59, 0.95), high=0.78 cm (−0.04, 1.60), p-trend=0.07; p for interaction=0.03]. The log₂-Cd by infant sex interaction was statistically significant for ponderal index [p=0.003; β(95% CI): female=0.25 kg/m³ (−0.20, 0.70); male= −0.63 kg/m³ (−1.01, −0.24)] and birth length [p<0.001; β(95% CI): female= −0.47 cm (−0.74, −0.20), male= 0.32 cm (0.00, 0.65)]. Our findings suggest potential sex-specific reversal of Cd’s associations on birth length and contribute to the evidence suggesting Cd impairs fetal growth.

INTRODUCTION

Cadmium (Cd) is widely used in the production of batteries, pigments, plastics, and other commercial products. Cd has long been recognized as an occupational health hazard, but less is known about the adverse effects of environmental Cd exposure among the general population, particularly in early life (ATSDR et al., 2012). Ingestion of food containing Cd and inhalation of tobacco smoke are the primary routes of Cd exposure in the general population (Satarug and Moore, 2004). Cd exposure leads to renal damage (Kido et al., 2004), cardiovascular disease (Tellez-Plaza et al., 2013), and osteoporosis (Alfvén T, 2000). Calcium, zinc, and iron deficiencies, as well as diets low in protein can augment Cd absorption (Kippler et al., 2007; Kippler et al., 2009). Lower iron stores in reproductive age women (Järup and Akesson, 2009) and substantial pregnancy-related increases in iron requirements (Bothwell, 2000) may increase Cd absorption during pregnancy (Akesson et al., 2002; Kippler et al., 2009). Following ingestion or inhalation, Cd is absorbed and not readily excreted (Satarug and Moore, 2004). Though Cd is primarily stored in the liver and kidneys the placenta also accumulates Cd (Kido et al., 2004; Satarug and Moore, 2004). This may contribute to decreased uteroplacental blood flow (Chertok et al., 1984) or disrupt the synthesis and metabolism of placental hormones, leading to fetal growth restriction (Stasenko et al., 2010).

Several previous studies have observed inverse associations between gestational exposure to Cd (characterized using measures in cord blood, placenta, maternal blood, or urine) and birthweight (Al-Saleh et al., 2014; Fréry et al., 1993; Galicia-García et al., 1997; Johnston et al., 2014; Kippler et al. 2012a; Llanos and Ronco, 2009; Nishijo et al., 2002; Salpietro et al., 2002; Shirai et al., 2010; Sun et al., 2014; Tian et al., 2009; Xu et al., 2015), birth length (Nishijo et al., 2004; Tian et al., 2009; Zhang et al., 2004), and head circumference (Kippler et al. 2012a; Nishijo et al., 2004), although others have observed no associations, (Klapec et al., 2008; Odland et al., 2004; Osman et al., 2000), inverted U-shaped (Kippler et al., 2012b), or positive associations(Bloom et al., 2015).

The largest studies to date each included more than 1,000 maternal-infant pairs and observed inverse associations between Cd and birthweight (Al-Saleh et al., 2014; Johnston et al., 2014; Kippler et al. 2012a), but many prior studies included fewer than 80 maternal-infant pairs (Galicia-García et al., 1997; Kippler et al., 2010; Llanos and Ronco, 2009; Nishijo et al., 2002; Nishijo et al., 2004; Salpietro et al., 2002; Shirai et al., 2010; Zhang et al., 2004). Kippler et al. observed decreasing birthweight with increasing maternal urinary Cd among female neonates only (Kippler et al. 2012a). Effect modification by infant sex has not been well investigated, and control for confounding has been inconsistent. In particular, co-exposure to arsenic, which has been suggested to reduce birthweight (Rahman et al., 2009; Xu et al., 2011), should be accounted for in geographic areas where Cd and arsenic commonly co-occur, such as Bangledesh (Kippler et al. 2012a) or areas of the United States with historic copper smelting activities (e.g. Puget Sound) (King County Environmental Health Services, 2010).

The goal of the present study was to evaluate whether greater maternal body burden of Cd impairs fetal growth as measured by infant anthropometry (birthweight, birth length, head circumference, and ponderal index), and to evaluate whether these associations differ for male versus female neonates. We hypothesized that maternal urinary cadmium is inversely associated with birth anthropometrics and that these associations are potentially more pronounced among female neonates.

METHODS

Study Setting and Study Population

The Omega Study, a large prospective cohort study at the Center for Perinatal Studies at Swedish Medical Center in Seattle, WA, was designed to investigate risk factors for pregnancy complications (Qiu et al., 2011). Subjects were recruited from multiple prenatal care clinics affiliated with Swedish Medical Center and Tacoma General Hospital (1996–2008). English speaking women initiating prenatal care at a study clinic prior to 20 weeks gestation were eligible for participation. Women who were less than 18 years of age, planning to deliver at a non-study hospital, or not intending to carry the pregnancy to term were excluded. The analytic sample for the present study was derived from a subcohort of 750 women randomly drawn from among all participants (N=4,344) in the Omega Study. All procedures and study protocols were approved by the institutional review boards at the University of Washington and the study hospitals, and all participants provided written informed consent.

Women with a history of renal disease (n=7), pre-existing diabetes mellitus (n=17), chronic hypertension (n=33), multiple fetal births (n=27), and known complications of pregnancy [preeclampsia (n=27), gestational diabetes (n=44), preterm delivery, miscarriage, or stillbirth (n=115), infant with congenital anomaly (n=22)], and women that did not provide a urine sample during pregnancy (n=18) were excluded from analyses. World Health Organization (WHO) guidelines suggest that creatinine concentration may be used to identify spot urine samples that are too dilute (<30 mg/dL) or too concentrated (>300 mg/dL) to provide valid estimates of the concentration of the urinary chemical of interest (WHO, 1996). We excluded 125 women who had overly dilute urine. Aside from their low urinary creatinine level, the characteristics of these women were similar to those remaining in the study (data not shown). The exclusions described above are not mutually exclusive; some women were excluded for multiple reasons. The final analytic population included 396 maternal-infant pairs. Women in the analytic population were similar to women in the full Omega Cohort in terms of demographics and other key covariates (data not shown).

Data Collection

Information on sociodemographic characteristics, reproductive and medical histories, lifestyle factors such as alcohol and tobacco use, and maternal anthropometry was collected by trained interviewers using a structured questionnaire shortly after study enrollment (15 weeks gestation, on average). Maternal medical records were abstracted to ascertain pregnancy outcomes, covariate data related to pregnancy complications and relevant characteristics during pregnancy. Participants also completed a self-administered, validated, and semi-quantitative food-frequency questionnaire (FFQ) describing nutritional intake over the prior six months (Patterson et al., 1999) (three months before pregnancy through the first three months of pregnancy, on average).

Cadmium and Arsenic Assessment

At 15 weeks gestation, on average (standard deviation=2.8, interquartile range=13–17 gestational weeks), clean-catch spot urine samples were collected in polyethylene containers, promptly separated into 2 mL aliquots, and stored at −80 °C until analysis. Urine concentrations of Cd and total arsenic were quantified using a validated method of inductively coupled plasma mass-spectrometry (ICP-MS) following published protocols (Heitland, 2004) at Metametrix Clinical Laboratory, a Clinical Laboratory Improvement Amendments (CLIA) certified facility in Duluth, GA. Briefly, urine samples were shaken and 1 mL was acidified with 1% HNO₃ (100 µl). An internal standard solution containing scandium, rhodium, and germanium (500 µl) was added. Samples were diluted to 5 mL with deionized water. Polyatomic interference was minimized through the use of ICP-MS with a dynamic reaction cell (PerkinElmer SCIEX Elan DRC II with ESI SC-4, FAST Autosampler). The accuracy of ICP-MS was checked by conducting proficiency testing using urine reference material (New York Toxic / Trace Elements in Urine Event #1 2012). The limits of detection for urinary Cd and total arsenic were 0.12 and 3.0 µg/g creatinine respectively, and the coefficients of variation (CV’s) were less than 10–15%. (Urinary creatinine concentration was assessed using a commercially available kit (Genzyme Diagnostics, Catalog # 221-30/# 221-50) with improved Jaffe Reaction.

The distribution of creatinine-corrected, urinary Cd concentration (U-Cd) was right- skewed. We used a log₂-transformation (log₂-Cd) to diminish the influence of extreme values on the regression coefficients. The majority of the women in the analytic population (84%) had U- Cd greater than the limit of detection (LOD=0.12 µg/g creatinine). Appropriate to our observed distribution of Cd, we substituted LOD/(√2) for women with U-Cd values below the LOD (Croghan and Egeghy, 2003; Glass and Gray, 2001). Maternal U-Cd was also categorized into low (<0.29 µg/g creatinine), middle (0.29–0.42 µg/g creatinine), and high (≥0.43 µg/g creatinine) tertiles.

Infant anthropometry- Information on infant birthweight (g), birth length (cm), and head circumference (cm) was abstracted from infant medical records. Ponderal index, a measure of leanness in newborns used to quantify asymmetric fetal growth restriction, was calculated as: [birthweight (kg)]/ [birth length (m)]³ (Landmann et al., 2006).

Statistical Analysis

We examined the frequency distribution of maternal sociodemographic characteristics, medical and reproductive histories according to tertile of U-Cd and infant sex. Multivariable linear regression was used to model the relation between continuous log₂-Cd or tertile of maternal U-Cd and each continuous outcome [birthweight (g), ponderal index (kg/m³), birth length (cm), and head circumference (cm)]. The models adjusted for the following confounders based on a priori knowledge of their associations with exposure and outcome (Al-Saleh et al., 2014; Järup and Akesson, 2009; Kippler et al., 2013; Kippler et al., 2009; Satarug and Moore, 2004; Shirai et al., 2010): continuous maternal age (years), parity (nulliparous/multiparous), continuous maternal pre- pregnancy BMI, maternal race (non-Hispanic white: yes/no), gestational age (continuous weeks plus quadratic term). However, we did not adjust for gestational age in the ponderal index analysis, as ponderal index remains fairly steady at term (Landmann et al., 2006) and was confirmed to be steady at term in this cohort (data not shown). For the birth length analysis, maternal height (m) was substituted for maternal pre-pregnancy BMI, as it tends to be strongly associated with birth length (Bisai, 2010; Shirai et al., 2010). Additional potential confounders were assessed as described below. Environmental exposure to inorganic arsenic can decrease birthweight (Hopenhayn et al., 2003; Llanos and Ronco, 2009; Rahman et al., 2009; Xu et al., 2011), but inorganic arsenic appears not to be associated with birth length and head circumference (Rahman et al., 2009; Shirai et al., 2010). We conducted an exploratory analysis to assess the impact of confounding by arsenic exposure on the Cd-birthweight and Cd-ponderal index associations. Total arsenic in urine reflects both organic and inorganic species of arsenic. Ideally, we would quantify speciated arsenic, because the toxicologically relevant form is unclear, and recent ingestion of seafood increases the level of organo-arsenic species in urine (Orloff et al., 2009). In order to partially account for expected higher levels of organic arsenic species in the urine of women that regularly consume seafood, we further adjusted for self-reported fish consumption habits (servings of fish and seafood per week derived from food frequency questionnaire; n=370, mean±sd: 1.4 ±1.3) in models containing creatinine corrected total urinary arsenic (U-As). Because only 6.3% of participants in the full Omega Study cohort are current smokers, we did not a priori adjust for smoking status in the main analyses. However, we tested the effect of smoking during pregnancy (yes/no) and included it in the final birthweight and ponderal index models. We also conducted a sensitivity analysis restricted to never smokers. We assessed the impact of nutritional factors including, weight gain during pregnancy, calcium, iron, and zinc intake on the effect estimates. The Institute of Medicine (IOM) guidelines were used to classify total weight gain during pregnancy based on pre-pregnancy BMI as below optimal, optimal, or above optimal (IOM et al., 2009). Daily caloric intake (kcal/day) and dietary intake of calcium, iron, and zinc (mg/day) were estimated using food composition tables from the University of Minnesota Nutrition Coding Center Nutrient Database (Nutrition Coordinating Center, Minneapolis, MN) (Schakel, 2004) and the FFQ. Low dietary intake of micronutrients was designated based on the IOM recommended dietary allowance (RDA) for calcium, iron, and zinc during pregnancy (IOM Panel on Micronutrients, 2001). We calculated birthweight z-scores, a measure of birthweight for gestational age standardized to United States reference data (Oken et al., 2003), and repeated the analysis with birthweight z-score as an outcome.

An interaction term was added to each of the final models to formally assess the joint effect of U-Cd (either log₂-Cd or tertile) and infant sex on outcomes of interest. We conducted trend tests using the median values of the distribution within tertiles of maternal urinary exposure as the score variable (Greenland, 1995).

As there are no specific recommendations for creatinine-based exclusions among pregnant women, we performed a sensitivity analysis in which we relaxed the urinary creatinine exclusion criteria. Data from NHANES suggest that for women the cutoff of <30 mg creatinine/dL urine may be inappropriate, because it is common for women to have low urinary creatinine (Barr et al., 2005). In our sensitivity analysis, we excluded women with urine samples that were too concentrated (>300 mg/dL) according to both the guidelines of the WHO and the NHANES based recommendations, but included all women with dilute urine (creatinine <30 mg/dL)(Barr et al., 2005; WHO, 1996). We also repeated analyses with creatinine added to the model as a covariate rather than using creatinine-standardized U-Cd.

All multivariable linear regressions used robust standard error estimates (Huber, 1967; White, 1980) and the alpha level of 0.05 to define statistical significance. All analyses were completed using Stata 13 (StataCorp 2013) or SAS 9.4 statistical software (SAS Institute Inc. 2012).

RESULTS

Women in the study population were predominately married, Non-Hispanic white, nulliparous, and had education beyond high school (Table 1). The geometric mean of U- Cd was not different for women who delivered female (0.31 µg/g creatinine; 95% CI 0.28, 0.34 µg/g creatinine) versus male infants (0.31 µg/g creatinine; 95% CI 0.28, 0.34 µg/g creatinine). Average birth anthropometrics were not significantly different across tertiles of maternal U-Cd among women who delivered male or female infants (Table 1). Regardless of infant sex, women in the low tertile for U-Cd tended to be younger (Females: mean=31.7, Males: mean=32.3 years) than women in the higher two tertiles (Table 1). Among women who delivered female infants only, those in the highest tertile of U-Cd had lower average maternal pre-pregnancy BMI (mean=21.8 kg/m²) and a smaller proportion of nulliparous women (49.2%) than the lower two tertiles (Table 1). Maternal total U-As and U-Cd were weakly but significantly correlated in our population (Spearman's rho =0.19; p<0.01), and total urinary arsenic was retained in the multivariable models for birthweight and ponderal index.

Table 1.

Maternal and infant characteristics of the study population by tertile of maternal urinary cadmium and infant sex

	Female Neonates Tertiles of Maternal Urinary Cadmium (µg/g creatinine)			Male Neonates Tertiles of Maternal Urinary Cadmium (µg/g creatinine)
	<0.29	0.29– 0.42	≥0.43	<0.29	0.29–0.42	≥0.43
	n=68	n=62	n=63	n=67	n=72	n=64
Mean±SD
Infant anthropometrics
Birthweight (g) (n=396)	3475 ±452	3522 ±403	3392 ±470	3642 ±459	3570 ±472	3682 ±366
Birth length (cm) (n=390)	51.2 ±2.6	50.8 ±2.2	50.5 ±2.4	51.6 ±2.6	51.8 ±2.2	52.3 ±2.3
Head circumference (cm) (n=387)	34.6 ±1.7	34.7 ±1.7	34.3 ±1.6	35.1 ±1.7	35.2 ±1.6	35.3 ±1.7
Ponderal index (kg/m3) (n=390)	25.9 ±3.5	26.9 ±3.5	26.6 ±4.3	26.5 ±3.0	25.8 ±3.6	25.9 ±3.0
Maternal age years	31.7 ±4.7	33.4 ±4.2	34.5 ±4.7 *	32.3 ±4.4	32.8 ±3.8	33.6 ±4.1
Pre-pregnancy BMI (kg/m2)a	24.5 ±5.1	23.2 ±3.6	21.8 ±3.1 *	23.9 ±6.9	23.4 ±3.9	23.4 ±4.0
Gestational week of urine collection	15.3 ±2.8	15.4 ±2.7	15.1 ±2.9	15.4 ±2.9	15.2 ±3.0	15.1 ±2.7
n(%)
Maternal race/ethnicity
Non- Hispanic white	59 (86.8)	57 (91.9)	52 (82.5)	58 (86.6)	56 (77.8)	51 (79.7)
Other	9(13.2)	5 (8.1)	11 (17.5)	9 (13.5)	16 (22.2)	13 (20.3)
Maternal education
Post high school	65 (95.6)	59 (95.2)	54(85.7)	59 (88.1)	65 (90.3)	60 (93.8)
High school or less	1 (1.5)	1 (1.6)	3 (4.8)	3 (4.5)	3 (4.2)	2 (3.1)
Missing (n=19)
Marital Status
Married	58 (85.3)	56 (90.3)	51 (81.0)	56 (83.6)	63 (87.5)	60 (93.8)
Unmarried	10 (14.7)	6 (9.7)	12 (19.1)	11 (16.4)	9 (12.5)	4 (6.3)
Parity			*
Nulliparous	49 (72.1)	37 (59.7)	31 (49.2)	37 (55.2)	38 (52.8)	38 (59.3)
Parous	19 (27.9)	24 (40.3)	32 (50.8)	30 (44.8)	34 (47.2)	26 (40.6)
Family history of diabetes b			*
Yes	6 (8.8)	10 (16.1)	14 (22.2)	6 (9.0)	14 (19.4)	7 (10.9)
No	62 (91.2)	52 (83.9)	49 (77.8)	61 (91.0)	58 (80.6)	57 (89.1)
Family history of hypertension c
Yes	30 (44.1)	23 (37.1)	23 (36.5)	31 (46.3)	37 (51.4)	24 (37.5)
No	38 (55.9)	39 (62.9)	40 (63.5)	36 (53.7)	35 (48.6)	40 (62.5)
Smoking status d
Never	50 (73.5)	41 (66.1)	40 (63.5)	42 (62.7)	50 (69.4)	45 (70.3)
Former	12 (17.7)	15 (24.2)	14 (22.2)	13 (19.4)	12 (16.7)	11 (17.2)
Current	4 (5.9)	4 (6.5)	3 (4.7)	7 (7.5)	4 (5.6)	3 (4.7)
Missing (n=20)
Alcohol during pregnancye
Consumed none	49 (72.1)	44 (71.0)	39 (61.9)	49 (73.1)	44 (61.1)	48 (75.0)
Consumed any	19 (27.9)	18 (29.0)	24 (38.1)	18 (26.9)	28 (38.9)	16 (25.0)
IOM weight gain guidelines f
Below optimal	9 (13.2)	8 (12.9)	6 (9.5)	2 (3.0)	3 (4.2)	4 (6.3)
Optimal	21 (30.9)	22 (35.5)	27 (42.9)	24 (35.8)	20 (27.8)	17 (26.6)
Above optimal	38 (55.9)	32 (51.6)	28 (44.4)	41 (61.2)	48 (66.7)	42 (65.6)
Missing (n=4)
Infant delivered by C- section			*
Yes	21 (30.9)	13 (21.0)	10 (15.9)	14 (21.0)	24 (33.3)	23 (35.9)
No	47 (69.1)	49 (79.0)	53 (84.1)	53 (19.1)	48 (66.7)	41 (64.1)
GM (95% CI)
Dietary micronutrient intake g
Calcium (g)	1.1 (1.0, 1.3)	1.1 (1.0, 1.3)	0.9 (0.7, 1.1)	1.1 (0.9, 1.2)	1.0 (0.9, 1.2)	1.1 (0.9, 1.2)
Total iron (mg)	14 (12, 15)	13 (11, 14)	12 (10, 15)	12 (10, 14)	12 (11, 14)	12 (11, 14)
Zinc (mg)	12 (10, 14)	12 (10,13)	10 (8, 13)	11 (10,12)	11 (10, 12)	11 (9, 12)
Missing (n=35)
Maternal Urinary Measures
Cadmium µg/g creatinine	0.14 (0.12, 0.16)	0.35 (0.34, 0.36)	0.64 (0.58, 0.71)	0.14 (0.12, 0.16)	0.35 (0.34, 0.36)	0.60 (0.56, 0.66)
Total Arsenic µg/g creatinine	17 (12, 22)	19 (15, 24)	26 (20, 34)*	19 (15, 23)	18 (15, 23)	21 (18, 26)
Creatinine mg/dL	81 (69, 94)	87 (78, 97)	78 (67, 91)	79 (69, 92)	91 (80, 103)	78 (68, 88)

CI=confidence interval, GM=Geometric Mean, SD=standard deviation

^*p-trend<0.05 for regression analysis of continuous covariates or p<0.05 for analysis of variance of log₂-transformed cadmium by categorical covariates

^a.Body mass index (BMI)

^b.Any maternal primary or secondary relative with a diabetic condition

^c.Any maternal primary or secondary relative with hypertension

^d.Maternal self-reported smoking of tobacco cigarettes during the study pregnancy

^e.Maternal self-reported alcohol consumption during the study pregnancy

^f.Institute of Medicine (IOM) guidelines for weight gain during pregnancy (Institute of Medicine 2009)

^g.Low intake designated by consuming less than the IOM daily recommended dietary allowance for pregnant women: calcium (1 g), iron (27 mg), zinc (11 mg)(Institute of Medicine Panel on Micronutrients 2001)

Among all infants, we observed small and statistically insignificant decreases in birthweight, ponderal index, and birth length with increasing maternal U-Cd (both continuous log₂-Cd and tertiles of U-Cd) (Table 2). For each doubling of maternal U-Cd we observed a 0.63 kg/m³decrease in ponderal index among male infants (95% CI −1.01 kg/m³, −0.24 kg/m³), but not among female infants (0.25 kg/m³; 95% CI: −0.20 kg/m³, 0.70 kg/m³; p-for-effect modification=0.003). However, no clear trend in ponderal index was observed across tertiles of maternal U-Cd among either female (p-trend=0.96) or male infants (p-trend=0.15). For each doubling of maternal U-Cd we observed a 0.47 cm decrease in birth length among female infants (95% CI −1.01 cm, −0.24 cm) and a 0.32 cm increase in birth length among male infants (95% CI: 0.00 cm, 0.65 cm; p-for-effect modification <0.001). Among female neonates, we observed reduced birth length with increasing tertile of maternal U-Cd [β(95% CI): low=reference; middle= −0.59 cm (−1.37, 0.19 cm); high=–0.83 (−1.69, 0.02 cm); p-trend=0.08]. An increase in birth length across increasing tertiles of maternal U-Cd was observed among male neonates [β(95% CI): low=reference; middle=0.18 cm (−0.59, 0.95 cm); high=0.78 cm (–0.04, 1.60 cm); p-trend=0.07]. The interaction between tertile of maternal urinary Cd and sex of neonate for the birth length analysis was statistically significant (p=0.03) (Table 2).

Table 2.

Multivariable linear regression analyses of the association between maternal urinary cadmium and newborn anthropometrics

Urinary Cadmium (µg/g creatinine)	n	All Infants β (95% CI)	n	Female infants a β (95% CI)	n	Male infants a β (95%CI)	p-value
Birthweight (g) b
Continuous log2-Cd	349	−29(−70, 12)	169	−50(−109, 8)	180	−7(−65, 51)	0.311c
Low (<0.29)	125	0 (reference)	64	0 (reference)	61	0 (reference)
Medium (0.29–0.42)	119	−29(−130, 73)	56	−10(−153, 133)	63	−46(−191, 99)
High (≥0.43)	105	−12(−120, 95)	49	−61(−228, 106)	56	28(−112, 167)
p-trend d		0.809		0.328		0.603	0.538 e
Ponderal Index (kg/m3)f
Continuous log2-Cd	343	−0.18(−0.49, 0.13)	164	0.25(−0.20, 0.70)	179	−0.63(−1.01, −0.24)	0.003c
Low (<0.29)	121	0 (reference)	61	0 (reference)	60	0 (reference)
Medium (0.29–0.42)	118	−0.05(−0.93, 0.83)	55	0.79(0.23, −0.50)	63	−0.84(−2.03, 0.35)
High (≥0.43)	104	−0.27(−1.18, 0.64)	48	0.36(−1.00, 1.71)	56	−0.87(−2.02, 0.28)
p-trend d		0.559		0.964		0.154	0.157e
Birth Length (cm) g
Continuous log2-Cd	390	−0.08(−0.30, 0.14)	188	−0.47(−0.74, −0.20)	202	0.32(0.00, 0.65)	<0.001c
Low (<0.29)	131	0 (reference)	65	0 (reference)	66	0 (reference)
Medium (0.29–0.42)	133	−0.20(−0.75, 0.35)	61	−0.59(−1.37, 0.19)	72	0.18(−0.59, 0.95)
High (≥0.43)	126	−0.01(−0.61, 0.59)	62	−0.83(−1.69, 0.02)	64	0.78(−0.04, 1.60)
p-trend d		0.986		0.084		0.069	0.029e
Head Circumference (cm)h
(Continuous log2-Cd	387	−0.04(−0.21, 0.12)	185	−0.17(−0.39, 0.06)	202	0.07(−0.17, 0.31)	0.160c
Low (<0.29)	127	0 (reference)	61	0 (reference)	66	0 (reference)
Medium (0.29–0.42)	134	0.09(−0.29, 0.47)	62	0.10(−0.45, 0.65)	72	0.08(−0.46, 0.62)
High (≥0.43)	126	0.12(−0.27, 0.50)	62	−0.04(−0.61, 0.52)	64	0.26(−0.30, 0.82)
p-trend d		0.562		0.995		0.366	0.668 e

^a.Multivariable linear regression models included a product interaction term for maternal urinary cadmium by infant sex

^b.Controlling for maternal age, pre-pregnancy BMI, race/ethnicity, nulliparity, gestational age at delivery, smoking during pregnancy, infant sex, gestational week of urine sample collection, and maternal total urinary arsenic adjusted for fish consumption

^c.p-value for log2-Cd*sex interaction

^d.p-value for trend across tertiles of maternal urinary cadmium

^e.p-value for Cd tertile*sex interaction

^f.Controlling for maternal age, pre-pregnancy BMI, race/ethnicity, nulliparity, smoking during pregnancy, infant sex, gestational week of urine sample collection, and maternal total urinary arsenic adjusted for fish consumption

^g.Controlling for maternal age, maternal height, race/ethnicity, nulliparity, gestational age at delivery, infant sex, and gestational week of urine sample collection

^h.Controlling for maternal age, pre-pregnancy BMI, race/ethnicity, nulliparity, gestational age at delivery, delivery by Caesarean section, infant sex, and gestational week of urine sample collection

The overall pattern of observed associations between maternal U-Cd and fetal growth markers were generally the same for the sensitivity analyses including women with dilute urine (<30 mg creatinine/dL) (Supplemental Table S1) or including creatinine as a covariate in the model (Supplemental Table S2), although a decreasing trend in ponderal index across U-Cd tertiles was observed among male infants in both scenarios (Supplemental Table S1 p-trend=0.05 and Supplemental Table S2 p-trend=0.03). The pattern of effect estimates between maternal U-Cd and birth anthropometrics were similar when restricted to never smokers (Supplemental Table S3). The addition of nutritional factors to the birthweight, ponderal index and head circumference models did not substantially change the observed estimates. Upon adding control for maternal weight gain during pregnancy and dietary calcium intake to the birth length models, the inverse trend across tertiles of maternal U-Cd and birth length was attenuated among the female infants [β(95% CI): low=reference; middle=–0.55 cm (−1.51, 0.41 cm); high=–0.38 cm (−1.42, 0.65 cm); p- trend=0.64] (Supplemental Table S4). The pattern of results for birthweight z-scores was similar overall to the results for the birthweight analysis (Supplemental Table S5).

DISCUSSION

In the current study, we observed that maternal U-Cd is associated with sex-specific differences in birth length and ponderal index. Maternal U-Cd was inversely associated with birth length among females and positively associated with birth length among males. Maternal U-Cd was inversely associated with ponderal index among males only; however, it is likely that the observed effect modification by infant sex of the relation of U-Cd with birth length is influencing the observed relations between maternal U-Cd and ponderal index. In addition, we observed modest, statistically insignificant, decreases in birthweight, ponderal index, and birth length among all neonates with increasing maternal U-Cd. The modest decreases in birth anthropometrics with increasing maternal U-Cd observed in our study are unlikely to be of clinical significance at the individual level. However, even incremental decreases in average size at birth may have adverse implications for infant mortality at the population level (Wilcox, 2001).

Our findings are broadly in line with prior research of Cd and birthweight. We observed a statistically insignificant decrease in birthweight corresponding to greater maternal log₂-Cd among all neonates, in general agreement with prior studies which observed inverse associations between maternal U-Cd (Kippler et al. 2012a; Shirai et al., 2010) or other measures of maternal Cd and weight at birth (Al-Saleh et al., 2014; Fréry et al., 1993; Galicia-García et al., 1997; Johnston et al., 2014; Llanos and Ronco, 2009; Nishijo et al., 2002; Salpietro et al., 2002; Sun et al., 2014; Tian et al., 2009; Xu et al., 2015). Levels of urinary Cd were slightly greater on average in our study (geometric mean=0.31 µg/g creatinine 95% CI: 0.28, 0.34) than those of women in the general US population as estimated by the National Health And Nutrition Examination Survey (geometric mean=0.25 µg/g Cr; 95% CI: 0.24, 0.27)(CDC, 2013). However, U-Cd was lower on average in our US based population compared to those in Bangladesh (median=0.63 µg/L)(Kippler et al. 2012a) and Japan (geometric mean= 0.77 µg/g creatinine) (Shirai et al., 2010), potentially suggesting that Cd’s effect on birthweight may be of greater concern for populations with higher average Cd exposure than the general population of the US. Additionally, only a few previous studies have accounted for arsenic co-exposure (Kippler et al. 2012a; Llanos and Ronco, 2009; Shirai et al., 2010) when assessing the relation between Cd and birthweight. Though arsenic may not be a relevant co-exposure in every geographic region, it was an important potential co-exposure for our study (King County Environmental Health Services, 2010).

Although we did not observe any statistically significant associations of Cd with continuous birthweight, our observation of more substantial decreases in birthweight among female infants are in general agreement with previous findings, as reductions in birthweight with increasing maternal U-Cd were more pronounced among female infants (Kippler et al. 2012a). Kippler et al. conducted a large prospective study in which sex-specific effects of Cd exposure on size at birth were suggested (Kippler et al. 2012a). Specifically, they observed a 45 g decrease in birthweight for each 1 µg/L increase in maternal urinary Cd among female neonates (95% confidence interval: −82.5, −7.3 g), and no clear impact of Cd on fetal growth impairment among males (Kippler et al. 2012a), which is similar in magnitude to the 50 g decrease we observed among female neonates with each doubling of maternal U-Cd (95% confidence interval:-109, 8 g).

We observed some evidence to support effect modification of the relation between maternal U-Cd and ponderal index by infant sex, with girls experiencing modest increases in ponderal index, whereas boys had decreased ponderal index with greater maternal U-Cd, though this may be due to the observed differences in Cd’s effect on birth length among male versus female infants. Only two studies have assessed maternal Cd during pregnancy and either newborn BMI or ponderal index. In a cross-sectional study (n=262) spanning 6 regions across the Russian and Norwegian arctic and subarctic regions, Odland et al. observed a statistically insignificant increase of 4.4 kg/m² in neonatal BMI with each µg/g increase in placental Cd (95% CI: −3.95, 12.75 kg/m²)(Odland et al., 2004). Al-Saleh et al. conducted a large cross-sectional study (n=1,578) among Saudi Arabian women, in which they assessed lead, cadmium and mercury in cord blood, placenta, and maternal blood (Al-Saleh et al., 2014). They reported that there was no association between ponderal index less than the 10^th percentile and Cd, but did not publish their estimates (Al-Saleh et al., 2014). Neither study assessed effect modification by infant sex.

For the Cd-birth length relation, we observed effect modification by sex, suggesting an inverse dose-response relation between maternal U-Cd and birth length among female neonates and a positive dose response association among male infants. Although small and not statistically significant, Kippler et al. previously observed an inverse relation between increasing maternal U- Cd (µg/L) and birth length among females (β= −0.14 cm; 95% CI: −0.36, 0.08 cm) and a direct relation among males (β=0.07 cm; 95% CI −0.18, 0.32 cm)(Kippler et al. 2012a). To the best of our knowledge, ours is the first study to identify potential sex-specific differences in ponderal index or birth length related to intrauterine Cd exposure.

We did not see evidence of an association between maternal U-Cd and head circumference. Findings from prior studies have been equivocal, with one study observing decreasing head circumference with increasing Cd (Nishijo et al., 2004), one observing increasing head circumference with increasing Cd (Nishijo et al., 2002), several suggesting no association (Al-Saleh et al., 2014; Johnston et al., 2014; Kippler et al., 2010; Osman et al., 2000; Shirai et al., 2010), another suggesting an inverse association between U-Cd and head circumference among female neonates only (Kippler et al. 2012a), and one observing an inverted U-shape between maternal Cd and biparietal diameter measured by fetal ultrasound (Kippler et al., 2012b).

Several biological mechanisms have been proposed to explain Cd-induced fetal growth impairment. Experimental evidence supports Cd’s placental toxicity and ability to inhibit transfer of nutrients to the fetal compartment through reduced uteroplacental blood flow (Chertok et al., 1984; Wier et al., 1990). Disruption of the synthesis, metabolism, or release of hormones by the placenta also occurs as Cd accumulates in placental tissue. Placental metabolism of glucocorticoids diminishes in the presence of Cd (Ronco et al., 2009), failing to shield the fetus from the growth-inhibiting properties of glucocorticoids. Likewise, Cd is associated with reductions in placental leptin synthesis (Stasenko et al., 2010), which has been linked to intrauterine growth restriction (Jaquet, 1998).

Given the observation of different relations between Cd and fetal growth among girls versus boys in the current study and prior literature (Kippler et al. 2012a), the previously described mechanisms incompletely define the effect of Cd on fetal growth. Likewise, part of the discrepancy in results across prior studies is potentially due to failure to account for sex-specific effects of Cd on fetal growth. Sex-specific Cd-birthweight associations may be accomplished by Cd-induced changes to epigenetic regulation of genes governing growth. In the presence of Cd-exposure, hypomethylation of DNA isolated from cord blood was observed predominantly among genes associated with organ development, morphology, and bone development among females versus global hypermethylation among males (Kippler et al., 2013). Additional studies have observed sex-specific effects of cadmium on methylation of growth associated genes in cord blood (Vidal et al., 2015) and placenta (Mohanty et al., 2015). The reason for the apparent reversal of the Cd-birth length association between sexes is not readily apparent, but may be explained somewhat by sex-specific differences in Cd induced bone damage. Experimental studies have suggested that the male skeleton is somewhat resistant to Cd induced bone damage, as evidenced by sex-specific differences in the impact of Cd exposure on bone mineral density, bone mineral content, and bone metabolism (Bhattacharyya, 2009; Brzóska and Moniuszko-Jakoniuk, 2005a; Brzóska and Moniuszko-Jakoniuk, 2005b). Epidemiologic studies have demonstrated that Cd diminishes bone mineral density and increases fractures and osteoporosis risk for both adult men and women; however, the effects tend to be more pronounced among women (Alfvén T, 2000; Engström et al., 2012). Although these findings do not explain the positive association observed between Cd and birth length among male infants in our study, they demonstrate that less Cd induced bone damage occurs among males than among females. Additional work in this area is necessary to clarify the observed associations.

Our study drew on several important strengths, including its prospective design and the availability of extensive information for a broad array of potential confounders within a well-characterized cohort. We were able to measure Cd and arsenic in maternal urine, using a robust, well validated, and sensitive method (ICP-MS) to quantify concentrations in maternal urine, and ours is the first study, to the best of our knowledge, to report sex-specific differences in ponderal index. There are also limitations that should be considered when interpreting our study results. Humans are exposed to complex mixtures of toxic substances every day, and as with all studies of environmental exposures there is an inherent difficulty in singling out the effect of Cd alone.

Although we were able to control for total arsenic, additional unmeasured or unknown co- exposures may contribute to the risk of impaired fetal growth. Likewise, we could not fully assess iron, zinc, and calcium status among the women in our population. Cd absorption increases in the presence of iron, zinc, and calcium deficiencies (Kippler et al., 2007; Kippler et al., 2009). While we were able to evaluate the associations of micronutrient intake to some degree, we could not fully account for interactions among toxic and nutrient metals. Due to the large number of comparisons made in our statistical analyses, some of the observed associations may be chance findings. External validity may be reduced since women in the Omega Study are generally Non-Hispanic white, married, and prosperous, reflective of the underlying population that utilizes Swedish Medical Center and Tacoma General Hospital. However, these women should represent a subpopulation within the general public with elevated exposure to Cd, as high income has been associated with increased body burden of Cd (McElroy et al., 2007).

CONCLUSIONS

Our study provides further evidence that associations between Cd and size at birth may be sex-specific. We observed divergent effects of Cd on birth length and ponderal index among boys and girls. Future research is warranted to replicate these findings in diverse populations and to further clarify the observed relations. This study reiterates the adverse impact of gestational exposure to Cd on fetal growth among female neonates and adds to the evidence suggesting that the male and female fetus may employ differing growth strategies in response to similar intrauterine insults.

Highlights.

• Cadmium levels in the general population potentially adversely affect size at birth

• Maternal urinary cadmium was inversely related to birth length among female infants

• For male infants, maternal cadmium was positively associated with birth length

• Maternal cadmium was also inversely related to ponderal index among male infants