Third trimester vitamin D status is associated with birth outcomes and linear growth of HIV-exposed uninfected infants in the United States

Abstract

Background:

Vitamin D status in pregnancy may influence the risk of prematurity, birth size, and child postnatal growth but few studies have examined the relationship among pregnant women living with HIV.

Methods:

We conducted a prospective cohort study of 257 HIV-infected mothers and their HIV-exposed uninfected infants who were enrolled in the 2009–2011 Nutrition sub-study of the Surveillance Monitoring for ART Toxicities (SMARTT) Study. HIV-infected pregnant women had serum 25-hydroxyvitamin D (25(OH)D) assessed in the third trimester of pregnancy and their infants’ growth and neurodevelopment were evaluated at birth and approximately one year of age.

Results:

The mean third trimester serum 25(OH)D concentration was 35.4 ± 14.2 ng/mL with 15% of women classified as vitamin D deficient (<20 ng/mL) and 21% as insufficient (20–30 ng/mL). In multivariable models, third trimester vitamin D deficiency and insufficiency were associated with −273g (95% CI: −450, −97) and −203g (95% CI: −370, −35) lower birth weights compared to vitamin D sufficient women, respectively. Maternal vitamin D deficiency was also associated with shorter gestation (mean difference −0.65 weeks; 95% CI: −1.22, −0.08) and lower infant length-for-age z-scores at one year of age (mean difference: −0.65; 95% CI: −1.18, −0.13). We found no association of vitamin D status with infant neurodevelopment at 1 year of age.

Conclusion:

Third trimester maternal vitamin D deficiency was associated with lower birth weight, shorter length of gestation, and reduced infant linear growth. Studies and trials of vitamin D supplementation in pregnancy for women living with HIV are warranted.

Introduction

The majority of pregnant women living with HIV in high-income countries, and now globally through expansion of the Option B+ approach, receive combination antiretroviral therapy (cART) which can reduce the risk of mother-to-child transmission of HIV to <1%.¹ As a result of the success in increasing cART coverage in pregnancy, the global number of HIV-exposed uninfected (HEU) children born each year is expected to reach over 1.5 million.² In some contexts, HEU children experience a higher risk of poor birth outcomes, linear growth, and neurodevelopment compared to their HIV-unexposed peers.^3–5 These poor outcomes may be to attributable to a combination of factors including exposure to HIV and antiretroviral drugs in utero, poorer health and nutritional status of the mother during pregnancy and breastfeeding, deficits in infant immune responses, and socioeconomic factors related to living in an HIV-affected household.^4,6

Vitamin D has well established effects on bone health and calcium homeostasis; however, more recent evidence suggests that vitamin D status in pregnancy may also influence the risk of preeclampsia, gestational diabetes, prematurity, birth weight, postnatal growth, and risk of childhood asthma.^7–9 The vast majority of published studies on vitamin D in pregnancy, however, are limited to populations of HIV-uninfected mothers.⁷ In particular, the immunomodulatory and anti-inflammatory effects of vitamin D may affect maternal and child health differently in the context of HIV.¹⁰ In terms of evidence on prenatal vitamin D among HIV-infected pregnant women, a cohort study among HIV-infected Tanzanian pregnant women who did not receive cART determined that low vitamin D status in the second trimester (25-hydroxyvitamin D (25(OH)D) concentration < 32 ng/mL) was associated with increased risk of maternal HIV progression and incidence of infant stunting and underweight.^11,12 In addition, Jao and colleagues found that HIV-infected women on cART in Latin America with severe vitamin D deficiency in pregnancy (25(OH)D <10 ng/mL) had increased risk of preterm birth.¹³ A small cross-sectional study of pregnant women living with HIV in the US found that 75% had cord blood plasma 25(OH)D concentrations <20 ng/mL with Black, non-Hispanic women having greater risk for low levels than White women.¹⁴ To the best of our knowledge, no studies have examined the relationship of prenatal vitamin D status with birth outcomes, growth and neurodevelopment among HIV-exposed infants in high-income settings.

We therefore performed a prospective cohort study of HIV-infected pregnant women residing in the US and their HEU infants. We examined the relationship of third trimester vitamin D status with birth outcomes and infant growth and neurodevelopment outcomes at approximately 1 year of age. These results are intended to inform whether trials and studies of vitamin D supplementation in pregnancy should be pursued among HIV-infected women in the US and similar settings to improve birth outcomes, growth and neurodevelopment of HEU infants.

Methods

Study Population

This study includes data from a prospective cohort of 257 HIV-infected pregnant women and their HEU infants who were enrolled in the Nutrition sub-study (R01HD060325) of the parent Surveillance Monitoring for ART Toxicities Study (SMARTT) protocol of the Pediatric HIV/AIDS Cohort Study (PHACS). The sub-study recruited pregnant women during 2009–2011 at 15 clinical sites in the U.S., including Puerto Rico.¹⁵ HIV-infected pregnant women were sequentially enrolled in the study during the second or third trimester of pregnancy and HEU infants were enrolled at birth. HEU infants were followed-up at approximately 1 year of age (9–15 months) for anthropometric and neurodevelopment assessment. The SMARTT protocol and the Nutrition Sub-study were approved by the Institutional Review Board at each recruitment site, at the University of Miami Human Subjects Research Office, and at the Harvard T.H. Chan School of Public Health. Written informed consent was obtained from the mother for herself and her child.

Data Collection

Sociodemographics, maternal IQ and pregnancy data

At the study visit that occurred after delivery, and as part of the SMARTT protocol, participating mothers were interviewed to obtain data on sociodemographics and assess self-reported substance use during pregnancy. Data on pregnancy history were abstracted from the medical record, including: pre-pregnancy weight (kg) and height (cm), pregnancy and birth complications, gestational age, ART use by trimester, first and last CD4 count and HIV viral load (RNA) in pregnancy, and substance use by trimester of pregnancy. The participant was asked about supplement use in the last 30 days at the time of the first recall. Pre-pregnancy body mass index (BMI) was calculated as kg/m². A full scale maternal IQ was assessed by study staff with the Wechsler Abbreviated Scale of Intelligence at the child’s one year of age visit.¹⁶

Third trimester serum vitamin D assessment

Blood was drawn in the third trimester of pregnancy and serum stored at −70° C until the samples were sent to the laboratory of Armando Mendez at the University of Miami Diabetes Research Institute for analysis of 25-hydroxyvitamin D (25(OH)D). 25(OH)D was measured by a competitive protein binding assay couple with electrochemiluminescence detection on a Roche Cobas 6000 analyzer and utilizing manufacturer’s reagents (Roche Diagnostics, Indianapolis, IN). This method has been standardized against LC-MS/MS which in turn has been standardized to the National Institute of Standards and Technology (NIST) standard and is well correlated with LC-MS/MS assay.^17,18 The assay limit of detection was 3 ng/ml and inter and intra-assay coefficients of variability were < 4% and < 6.5%, respectively.

Birth outcomes

Gestational age at birth was determined by standardized obstetric procedures with dating based on ultrasound in 43% of pregnant women, both ultrasound and clinical findings in 18% using the best obstetric judgement, clinical findings alone in 35%, and 4% were based on last menstrual period or other methods. Preterm birth was defined as gestational age < 37 weeks. Birth weight (g) was abstracted from medicals record and the study team used standard techniques to measure length in triplicate within 14 days of birth.¹⁹ Small-for-gestational age (SGA) was defined as weight < 10^th percentile for gestational age by sex using the WHO Intergrowth standards.²⁰ Birth length z-scores for age and sex were calculated using WHO reference values for term infants and the Fenton reference for preterm infants.^20,21 Apgar scores at 1 and 5 minutes were abstracted from medical records.

Infant growth and neurodevelopment

HEU infants were followed up at 12 months of age ± 3 months. Infant weight, length, and head circumference measurements were assessed in triplicate using standardized protocols. Length-for-age (LAZ), weight-for-length (WLZ), and weight-for-age (WAZ) z-scores were calculated using WHO child growth standards.²² Trained psychologists administered the Bayley Scales of Infant and Toddler Development – Third Edition (BSID-III).²³ The BSID-III was only available in English. Infants whose mother/primary caregiver was not able to complete the assessment in English were excluded. The Puerto Rico site did not participate in BSID-III assessments. The cognitive, motor, and language scales were administered through direct interaction with the infant while the socioemotional and adaptive behavior scales were administered as face-to-face interviews with the mother or primary caregiver. BSID-III domain scores were age-standardized with a mean score of 100 and standard deviation of 15.

Statistical Methods

We defined vitamin D deficiency as 25(OH)D concentrations <20 ng/mL, insufficiency 20–30 ng/mL, and sufficiency as >30 ng/mL.²⁴ First, we examined risk factors for low vitamin D status (either deficiency or insufficiency, 25(OH)D <30 ng/mL) using log-binomial regression models to obtain prevalence ratio estimates.²⁵ We examined the following risk factors for low vitamin D status: maternal age, ethnicity, maternal income, region of residence, third trimester CD4 T-cell count, third trimester HIV-1 RNA copies/mL, third trimester cART regimen, maternal smoking, season of vitamin D blood draw, and vitamin D supplement use. Due to the small sample size and collinearity for some variables, we constructed a parsimonious multivariable model using stepwise variable selection with an entry p-value of 0.20 and retained variables with a p-value < 0.15. We also examine risk factors for 25(OH)D < 20 ng/mL as a sensitivity analysis.

We then examined the prospective association of third trimester vitamin D status with birth outcomes and then infant growth and neurodevelopment outcomes at approximately 1 year of age. Linear regression models were fit to examine the relationship of vitamin D status with means of continuous outcomes (birth weight, birth weight z-score, birth length z-score, gestation duration, LAZ, WLZ, WAZ, HCAZ and BSID-III cognitive, motor, language adaptive behavior and socioemotional domain scores). These models were implemented using generalized estimating equation (GEE) models to provide robust variances. Log-binomial models were used to estimate relative risks for the binomial outcomes of preterm birth, SGA, and Apgar score of less than 7 at 1 and 5 minutes. An Apgar score of 7–10 is considered normal with scores 0–6 than indicating increased risk of death and other adverse outcomes.²⁶ Multivariable models included adjustment for maternal age (<25, 25–29.9, 30–34.9, ≥35 years), maternal race/ethnicity (Black non-Hispanic, White non-Hispanic, Hispanic, multiracial/other), maternal completion of high school (yes/no), maternal income <$10,000 per year (yes/no), region of residence (NY/NJ/Chicago, South, West, Puerto Rico), maternal pre-pregnancy BMI (<25, 25–29.9, 30–34.9, ≥ 35), third trimester CD4 T-cell count (<350, 350–499, ≥500 cells/mm³), 3rd trimester HIV-1 RNA viral load >400 copies/ml (yes/no), third trimester maternal smoking (yes/no) and child sex (male/female). Multivariable models for Bayley Scales of Infant Development scores were additionally adjusted for maternal IQ (<80, 80–95, ≥95). Due to potential differences in cut-offs to define vitamin D status among individuals of African descent, we present a sensitivity analysis restricted to Black non-Hispanic pregnant women.²⁷ We also conducted sensitivity analyses using a 25(OH)D cut-off of >40 ng/mL. All P values were 2-sided with a p-value < 0.05 considered statistically significant. Statistical analyses were performed using the SAS v 9.4 (SAS Institute Inc, Cary, NC).

Results

A total of 318 pregnant women were initially considered eligible and consented into the nutrition sub-study of the SMARTT. A total of 61 women were excluded from this analysis; 4 due to twin pregnancy, 19 due to no dietary assessment, 15 due to not having a vitamin D assessment, 20 due to having a vitamin D assessment after 37 weeks gestation, and 3 for other reasons. Therefore our study population for this analysis consisted of 257 HIV-infected mother-HEU infant pairs who had maternal serum 25(OH)D assessed between 27.0–37.0 weeks gestation (mean 32.0 weeks gestation). The analysis of BSID-III scores included 170 participants; an additional 87 participants were excluded including 27 participants who resided in Puerto Rico and were not eligible for BSID-III testing as well as 60 mainland US residents who were not assessed.

Characteristics of the study population are presented in Table 1. In terms of ethnicity, 55% of mothers were non-Hispanic Black, 35% were Hispanic, 7.1% were non-Hispanic white, and 3.1% were multiracial or of other ethnicity. Most women had an income <$10,000 per year (61.8%). In terms of HIV disease severity and treatment status in the third trimester, 20% of mothers had a CD4 T-cell count <350 cells/mm³, 13% had a HIV-1 RNA concentration >400 copies/mL, 96% of women received cART during the third trimester. The mean (SD) third trimester serum 25(OH)D concentration was 35.4 (14.2) ng/mL, with a range of 5–79 ng/mL. A total of 38 (15%) women were classified as being vitamin D deficient (<20 ng/mL), 55 (21%) insufficient (20–30 ng/mL) and 164 (64%) were vitamin D sufficient (>30 ng/mL).

Table 1:

Characteristics of 257 HIV-infected pregnant women who had vitamin D status assessed in the third trimester of pregnancy

	Frequency (%)
Maternal age (years)
<25 years	87 (33.9)
25 - 29.9 years	64 (24.9)
30 - 34.9 years	59 (23.0)
≥35 years	47 (18.3)
Maternal race/ethnicity
Black, non-Hispanic	140 (54.9)
White, non-Hispanic	18 (7.1)
Hispanic	89 (34.9)
Multiracial/other	8 (3.1)
Mother did not complete high school	83 (32.3)
Maternal income <$10,000 per year	144 (61.8)
Location of Residence
New York-New Jersey- Chicago	84 (32.7)
South	106 (41.3)
West	40 (15.6)
Puerto Rico	27 (10.5)
Language spoken in the home
English only	168 (65.9)
Any Spanish	69 (27.1)
Other	18 (7.1)
Maternal IQ
<80	57 (37.3)
80-94	51 (33.3)
≥ 95	45 (29.4)
Maternal pre-pregnancy BMI (kg/m2)
<25	72 (35.0)
25-29.9	50 (24.3)
30-34.9	38 (18.4)
≥ 35	46 (22.3)
Third trimester CD4 T-cell count (cells/mm3)
<350	52 (20.2)
350-499	50 (19.5)
≥500	103 (40.1)
Third trimester HIV-1 RNA >400 copies/ml	34 (13.2)
Third trimester cART	244 (96.1)
Protease inhibitor-based cART	218 (84.8)
Tenofovir-containing cART	122 (47.5)
Third trimester maternal smoking	30 (11.7)
Season of vitamin D blood draw
Spring (Mar 21st-Jun 20th)	87 (33.9)
Summer (Jun 21st-Sept 20th)	65 (25.3)
Fall (Sept 21st-Dec 20th)	63 (24.5)
Winter (Dec 21-Mar 20th)	42 (16.3)
Vitamin D supplement use
Daily supplements containing ≥ 400 IU	189 (73.5)
Daily supplements containing < 400 IU	45 (17.5)
No supplements containing vitamin D	23 (9.0)
Maternal vitamin D status
Deficient 25(OH)D <20 ng/mL	38 (14.8)
Insufficient 25(OH)D 20-29.9 ng/mL	55 (21.4)
Sufficient 25(OH)D ≥ 30 ng/mL	164 (63.8)

Characteristics associated with low third trimester vitamin D status

The association of maternal sociodemographic and pregnancy factors with low vitamin D status (25(OH)D <30 ng/mL) in the third trimester are presented in Supplemental Table 1. In multivariable models, mothers who resided in New York-New Jersey-Chicago (locations of similar latitude) and the South had 3.99 (95% CI: 1.41, 11.29) and 3.12 (95% CI: 1.10, 8.83) times the risk of low vitamin as compared to those residing in the Western US (p-values <0.01). In addition, women who did not take nutritional supplements containing vitamin D had 1.80 (95% CI: 1.01, 3.20) times the risk of low vitamin D as compared to mothers that took supplements containing ≥400 IU vitamin D / day. Only 8 of the 23 (34.8%) women who did not take vitamin D containing supplements had 25(OH)D concentrations >30 ng/mL. Among women who took daily supplements containing <400 IUs vitamin D, 26 of 45 (58.8%) had 25(OH)D concentrations >30 ng/mL. Women whose blood draw was in the Fall/Winter seasons appeared to have increased prevalence of low vitamin D status in a univariable analysis (PR: 1.48; 95% CI: 1.07, 2.04), but the results did not reach statistical significance in multivariable models (PR: 1.37; 95% CI: 0.91, 2.03; p=0.13). Of note, 0 of 18 white non-Hispanic women in the sample had low vitamin D status which made relative risks incalculable. In two sample-comparisons using the Fisher’s exact test, white non-Hispanic women had significantly lower prevalence of 25(OH)D <30 ng/mL as compared to both Black non-Hispanic (p<0.01) and Hispanic women (p= 0.01). In a sensitivity analysis, risk factors for 25(OH)D levels <20 ng/mL included not taking nutritional supplements containing vitamin D (PR: 2.67; 95% CI: 1.21–5.91) and a HIV viral load >400 copies / mL (PR: 2.44; 95% CI: 1.15–5.21).

Vitamin D status and birth outcomes

Univariable and multivariable analyses of third trimester vitamin D status with birth outcomes are presented in Table 2. In multivariable models, both third trimester vitamin D deficiency and insufficiency were associated with 273 g (95% CI: −450, −97) and 203 g (95% CI: −370, −35) lower birth weight as compared to vitamin D sufficiency, respectively. Vitamin D deficiency and insufficiency were also associated with lower birth weight z-scores (p values: 0.01 and 0.02, respectively). In addition, third trimester vitamin D deficiency and insufficiency were associated with −0.61 (95% CI: −1.13, −0.08) and −0.51 (95% CI: −0.94, −0.08) lower birth length z-scores. Vitamin D deficiency was associated with −0.64 weeks shorter gestation (95% CI: −1.26, −0.02) as compared to vitamin D sufficiency, but there was no significant association for insufficiency (mean difference: −0.19 weeks; 95% CI: −0.69, 0.37). In univariable models, vitamin D deficiency was associated with increased risk of an Apgar score of 7 or less at 1 minute (relative risk: 2.86; 95% CI: 1.08, 7.55); however, the results did not remain statistically significant in multivariable models.

Table 2.

Association of third trimester vitamin D status among HIV-infected pregnant women with birth outcomes (n=257)

	Mean ± SD or n n / N (%)	Univariable mean difference or relative risk (95% CI)	p-value	Multivariable* mean difference or relative risk (95% CI)	p-value
Mean difference birth weight (grams)
Deficient <20 ng/mL	2939 ± 403	−159 (−308, −10)	0.04	−273 (−450, −97)	<0.01
Insufficient 20-29.9 ng/mL	2945 ± 535	−154 (−315, 6)	0.06	−203 (−370, −35)	0.02
Sufficient ≥ 30 ng/mL	3099 ± 514	Ref.		Ref.
Mean difference birth weight z-score
Deficient <20 ng/mL	−0.46 ± 0.85	−0.28 (−0.59, 0.03)	0.07	−0.47 (−0.84, −0.10)	0.01
Insufficient 20-29.9 ng/mL	−0.46 ± 0.97	−0.28 (−0.57, 0.02)	0.06	−0.37 (−0.68, −0.06)	0.02
Sufficient ≥ 30 ng/mL	−0.18 ± 0.97	Ref.		Ref.
Mean difference birth length z-score
Deficient <20 ng/mL	−0.36 ± 0.93	−0.51 (−0.86,−0.16)	<0.01	−0.59 (−1.01, −0.17)	<0.01
Insufficient 20-29.9 ng/mL	−0.29 ± 1.42	−0.45 (−0.87, −0.02)	0.04	−0.51 (−0.93, −0.09)	0.02
Sufficient ≥ 30 ng/mL	0.15 ± 1.28	Ref.		Ref.
Mean difference gestation duration (weeks)
Deficient <20 ng/mL	37.7 ± 1.6	−0.50 (−1.04, 0.04)	0.07	−0.65 (−1.22, −0.08)	0.03
Insufficient 20-29.9 ng/mL	37.9 ± 1.7	−0.25 (−0.75, 0.25)	0.32	−0.21 (−0.72, 0.29)	0.41
Sufficient ≥ 30 ng/mL	38.2 ± 1.5	Ref.		Ref.
Relative risk preterm birth (<37 weeks)
Deficient <20 ng/mL	8 / 38 (21.1)	1.11 (0.56, 2.23)	0.76	1.21 (0.49. 3.01)	0.68
Insufficient 20-29.9 ng/mL	12 / 55 (21.8)	1.15 (0.64, 2.09)	0.64	1.13 (0.54, 2.39)	0.74
Sufficient ≥ 30 ng/mL	31 / 164 (18.9)	Ref.		Ref.
Relative risk SGA (<10th percentile)
Deficient <20 ng/mL	7 / 38 (18.4)	1.44 (0.66, 3.14)	0.36	1.34 (0.47, 3.80)	0.58
Insufficient 20-29.9 ng/mL	10 / 55 (18.2)	1.42 (0.71, 2.83)	0.32	1.29 (0.56, 2.96)	0.55
Sufficient ≥ 30 ng/mL	21 / 164 (12.8)	Ref.		Ref.
Relative risk Apgar score < 7 at 1 min
Deficient <20 ng/mL	6 / 38 (15.8)	2.86 (1.08, 7.55)	0.03	2.19 (0.70, 6.91)	0.18
Insufficient 20-29.9 ng/mL	4 / 55 (7.3)	1.32 (0.42, 4.11)	0.64	0.88 (0.35, 2.26)	0.80
Sufficient ≥ 30 ng/mL	9 / 163 (5.5)	Ref.		Ref.
Relative risk Apgar score < 7 at 5 min
Deficient <20 ng/mL	1 / 38 (2.6)	1.07 (0.12, 9.32)	0.95	1.09 (0.12, 9.75)	0.94
Insufficient 20-29.9 ng/mL	0 / 55 (0)	-	-	-	-
Sufficient ≥ 30 ng/mL	4 / 163 (2.5)	Ref.		Ref.

SGA: small-for-gestational age

^*Multivariable models adjusted for maternal age (<25, 25-29.9, 30-34.9, ≥35 years), maternal race/ethnicity (Black non-Hispanic, White non-Hispanic, Hispanic, multiracial/other), mother did not complete high school (yes/no), maternal income <$10,000 per year (yes/no), region of residence (NY/NJ/Chicago, South, West, Puerto Rico), maternal pre-pregnancy BMI (<25, 25-29.9, 30-34.9, ≥ 35), third trimester CD4 T-cell count (<350, 350-499, ≥500 cells/mm³), 3^rd trimester HIV-1 RNA viral load >400 copies/ml (yes/no), third trimester maternal smoking (yes/no) and infant sex (male/female).

In a sensitivity analyses restricted to Black non-Hispanic women (Supplemental Table 2), vitamin D deficiency and insufficiency, defined by <20 ng/mL and 20–30 ng/mL cut-offs, remained significantly associated with lower birth weight, birth weight z-scores, and birth length as compared to vitamin D sufficiency (>30 ng/mL) (p-values <0.05). Vitamin D deficiency also remained associated with shorter duration of gestation among Black non-Hispanic women (p<0.01). We also found no differences in birth outcomes for pregnant women with 25(OH)D levels between 30–40 ng/mL and those with >40 ng/mL (Supplemental Table 3).

Vitamin D status and infant growth and neurodevelopment

The association of prenatal vitamin D status with infant anthropometric growth outcomes and BSID-III domain scores are presented in Table 3. In multivariable models, HEU infants who mothers had third trimester vitamin D deficiency had −0.65 (95% CI: −1.18, −0.13) lower LAZ at 1-year of age as compared to those with vitamin D sufficient mothers (Supplemental Table 2). In a sensitivity analyses, vitamin D deficiency remained significantly associated with shorter infant LAZ among Black non-Hispanic mothers (p=0.03) and there no difference in infant LAZ between those with 25(OH)D concentrations 30–40 ng/mL and >40 ng/mL (Supplemental Tables 2 and 3). We found no statistically significant association of maternal vitamin D status with infant WLZ, WAZ, HCAZ or cognitive, motor, language, adaptive behavior, and socioemotional development at 1-year of age (p-values >0.05).

Table 3.

Association of third trimester vitamin D status among HIV-infected pregnant women with HIV-exposed uninfected infant anthropometry (n=257) and neurodevelopment outcomes (n=170) at 9-15 months of age

	Mean ± standard deviation	Univariable mean difference (95% CI)	p-value	Multivariable* mean difference (95% CI)	p-value
Infant anthropometry
Length-for-age z-score
Deficient <20 ng/mL	−0.18 ± 0.97	−0.34 (−0.73, 0.05)	0.09	−0.65 (−1.18, −0.13)	0.02
Insufficient 20-29.9 ng/mL	0.13 ± 1.39	−0.03 (−0.51, 0.45)	0.90	−0.33 (−0.73, 0.08)	0.11
Sufficient ≥ 30 ng/mL	0.16 ± 1.27	Ref.		Ref.
Weight-for-length z-score
Deficient <20 ng/mL	0.57 ± 1.12	0.06 (−0.38, 0.49)	0.80	−0.11 (−0.60, 0.39)	0.68
Insufficient 20-29.9 ng/mL	0.66 ± 1.16	0.14 (−0.27, 0.55)	0.50	0.20 (−0.21, 0.60)	0.34
Sufficient ≥ 30 ng/mL	0.52 ± 1.22	Ref.		Ref.
Weight-for-age z-score
Deficient <20 ng/mL	0.35 ± 0.92	−0.10 (−0.47, 0.28)	0.61	−0.35 (−0.82, 0.10)	0.13
Insufficient 20-29.9 ng/mL	0.54 ± 1.08	0.09 (−0.30, 0.47)	0.66	0.00 (−0.36, 0.36)	0.99
Sufficient ≥ 30 ng/mL	0.45 ± 1.22	Ref.		Ref.
Head circumference-for-age z-score
Deficient <20 ng/mL	0.35 ± 1.01	−0.21 (−0.60, 0.19)	0.30	−0.33 (−0.80, 0.14)	0.17
Insufficient 20-29.9 ng/mL	0.56 ± 1.29	0.01 (−0.43, 0.45)	0.96	0.06 (−0.35, 0.48)	0.77
Sufficient ≥ 30 ng/mL	0.55 ± 1.08	Ref.		Ref.
BSID-III age-standardized domain scores
Cognitive
Deficient <20 ng/mL	104.5 ± 11.9	1.28 (−3.64, 6.20)	0.61	2.33 (−2.53, 7.19)	0.35
Insufficient 20-29.9 ng/mL	102.8 ± 12.0	−0.40 (−5.08, 4.28)	0.87	0.52 (−3.43, 4.48)	0.80
Sufficient ≥ 30 ng/mL	103.2 ± 14.0	Ref.		Ref.
Motor
Deficient <20 ng/mL	98.8 ± 11.7	1.73 (−2.93, 6.39)	0.47	2.15 (−2.55, 6.85)	0.37
Insufficient 20-29.9 ng/mL	98.8 ± 10.4	1.70 (−2.31, 5.72)	0.41	1.50 (−2.45, 5.44)	0.46
Sufficient ≥ 30 ng/mL	97.1 ± 11.7	Ref.		Ref.
Language
Deficient <20 ng/mL	94.8 ± 11.5	1.23 (−3.43, 5.89)	0.60	2.33 (−2.53, 7.18)	0.35
Insufficient 20-29.9 ng/mL	95.0 ± 11.9	1.40 (−3.08, 5.88)	0.54	3.57 (−0.70, 7.83)	0.10
Sufficient ≥ 30 ng/mL	93.6 ± 12.4	Ref.		Ref.
Adaptive behavior
Deficient <20 ng/mL	93.8 ± 8.2	1.16 (−3.15, 5.46)	0.60	−1.28 (−6.64, 4.08)	0.64
Insufficient 20-29.9 ng/mL	94.4 ± 13.3	1.77 (−3.50, 7.04)	0.51	2.18 (−2.68, 7.04)	0.38
Sufficient ≥ 30 ng/mL	92.6 ± 16.0	Ref.		Ref.
Socioemotional
Deficient <20 ng/mL	100.0 ± 14.3	−0.36 (−6.53, 5.82)	0.91	1.84 (−4.95, 8.62)	0.60
Insufficient 20-29.9 ng/mL	99.7 ± 16.5	−0.62 (−6.91, 5.67)	0.85	1.38 (−4.71, 7.47)	0.66
Sufficient ≥ 30 ng/mL	100.4 ± 17.8	Ref.		Ref.

^*Multivariable models adjusted for maternal age (<25, 25-29.9, 30-34.9, ≥35 years), maternal race/ethnicity (Black non-Hispanic, White non-Hispanic, Hispanic, multiracial/other), mother did not complete high school (yes/no), maternal income <$10,000 per year (yes/no), region of residence (NY/NJ/Chicago, South, West, Puerto Rico), maternal pre-pregnancy BMI (<25, 25-29.9, 30-34.9, ≥ 35), third trimester CD4 T-cell count (<350, 350-499, ≥500 cells/mm³), 3^rd trimester HIV-1 RNA viral load >400 copies/ml (yes/no), third trimester maternal smoking (yes/no) and infant sex (male/female). BSID-III score analyses additionally adjusted for maternal IQ (<80, 80-95, ≥95).

Discussion

In this prospective cohort study of pregnant women living with HIV in the US, we found that third trimester vitamin D deficiency (<20 ng/mL) and insufficiency (20–30 ng/mL) were present in 14.8% and 21.4% of women, respectively. Risk factors for low maternal vitamin D status (<30 ng/mL) included residence in New York – New Jersey – Chicago and the South and not taking nutritional supplements containing vitamin D. Black non-Hispanic and Hispanic race women were also at greater risk for low vitamin D status compared to white non-Hispanic women. In terms of birth size, both third trimester vitamin D deficiency and insufficiency were associated with lower birth weights, birth weight z-scores and birth length z-scores. Vitamin D deficiency was also associated with approximately 4.5 days shorter gestation duration. HEU infants born to mothers with vitamin D deficiency in pregnancy had significantly lower LAZ at 9–15 months of age as compared to infants born to mothers who were vitamin D sufficient. We found no association of maternal vitamin D status with other growth indicators and child neurodevelopment outcomes as assessed by the BSID-III.

Our study adds to the growing body of evidence that suggests that maternal vitamin D status can influence infant size, but extends these findings to HIV-infected pregnant women in the US. We determined that third trimester vitamin D deficiency and insufficiency were robustly associated with lower birth weights, birth weight z-scores and birth length z-scores. A recent meta-analysis of observational cohort studies of HIV-uninfected women found that pregnancy 25(OH)D concentrations <15 ng/mL were associated with a 131 g deficit in birth weight, while a meta-analysis of randomized controlled trials determined that vitamin D supplementation in pregnancy increased infant birth weight by 58 g.^7,28 As a result, we hypothesize that based on our study findings of a 273 gram deficit with prenatal 25(OH)D concentrations <20 ng/mL that the relationship of vitamin D status with birth weight may be stronger among HIV-infected women. The mechanisms by which maternal vitamin D status may influence fetal growth remains unclear, but the leading hypotheses include anti-inflammatory, placental vascularization, angiogenesis, and uterine blood flow pathways.^29–31

We also found that third trimester vitamin D deficiency was associated with approximately 4.5 days shorter gestation, which may partially explain our birth weight and length findings. A previous cohort study of HIV-infected pregnant women in Latin America enrolled in the NICHD International Site Development Initiative (NISDI) protocol determined that severe vitamin D deficiency (<10 ng/mL) in pregnancy was associated with nearly 5 times the risk of preterm birth as compared to vitamin D sufficiency (≥30 ng/mL).¹³ In concordance, observational studies among HIV-uninfected women have noted an association between low maternal vitamin D and preterm birth.³² In terms of mechanisms, the strong anti-inflammatory properties of vitamin D may increase gestation length and reduce the risk of preterm birth.³³ To the best of our knowledge, there is only one on-going randomized trial of vitamin D supplementation conducted among HIV-infected pregnant women which intends to examine birth outcomes (NCT02305927). This trial includes 2,300 HIV-infected women in Tanzania and will assess the impact on maternal progression of HIV, small for gestational age, and stunting in infancy.³⁴

In addition, we found that prenatal vitamin D deficiency was associated with a relatively large 0.65 z-score deficit in infant length at 1 year of age. Further, vitamin D deficiency also appeared to be associated with a 0.33 z-score deficit in infant head circumference at 1-year, which is also an indicator of skeletal growth, but the results did not reach statistical significance. Previous cohort studies examining the relationship of prenatal 25(OH)D concentrations with child linear growth and head circumference have yielded mixed results. A cohort study of HIV-uninfected pregnant women in the UK, the Southampton Women’s Survey and the Avon Longitudinal Study of Parents and Children, found no association between maternal 25(OH)D in pregnancy with child height at 9 years of age.³⁵ In contrast, a cohort study of HIV-uninfected women in the US found that prenatal vitamin D levels <30 ng/mL were associated with deficits of 0.13 in LAZ and 0.20 in HCAZ during the first year of life.³⁶ A recent randomized trial of prenatal vitamin D supplementation among HIV-uninfected pregnant women found no effect on child growth. As a result, the relationship of maternal vitamin D status with growth may be stronger in infancy as compared to later childhood or vary by maternal HIV status or setting. In terms of mechanisms, maternal vitamin D status can impact fetal bone development which has been shown to affect infant skeletal growth.³⁷ In support of this mechanism, we found robust associations of prenatal vitamin D status with birth length z-scores and the magnitude of the association was similar to that for infant LAZ at one year of age. Nevertheless, extra-skeletal mechanisms are also possible given that maternal vitamin D status in pregnancy can influence the infant immune system and modification of gene expression which can also affect child linear growth.^38,39

In addition, while we found that maternal vitamin D status was strongly associated with lower birth weight and birth weight z-scores, there was no association of vitamin D status with WLZ or WAZ at 1-year of age. As a result, catch-up weight gain in early infancy appears to have eliminated the differences in weight seen at birth. These findings are in line with a cohort study of HIV-uninfected mothers in the US that also found that maternal 25(OH)D levels <30 ng/mL in pregnancy were associated with lower birth weights but there was no difference in infant BMI at 12 months of age.³⁶ Accelerated weight gain in infancy, which has been documented among HEU infants in US, may increase the risk of later life obesity.^40,41 Therefore, the pattern of rapid catch-up weight gain experienced by HEU infants born to mothers with vitamin D deficiency or insufficiency in our study may be of concern. A recent study in the US among HIV-uninfected women determined that prenatal 25(OH)D <15 ng/mL was associated with increased child BMI and waist circumference at 6 years of age.⁴² Long-term follow-up studies of prenatal vitamin D status with child and adult adiposity and chronic disease risk are needed.

We did not identify a relationship between vitamin D status in pregnancy with infant neurodevelopment as assessed by the BSID-III at 1 year of age. Nevertheless, our sample size of 257 mother-HEU infant pairs is underpowered to detect the small effect sizes that are likely for nutritional exposures in pregnancy (~0.1 standard deviations or 1.5 points in age standardized BSID-III scores).⁴³ However, we did find that vitamin D deficiency in pregnancy was associated with lower birth weight and decreased gestation duration, both of which have evidence for a relationship with poor child development outcomes.^44,45 The few observational cohort studies that have examined the relationship of vitamin D status in pregnancy with child development in HIV-uninfected populations have found mixed results.⁴⁶ Studies in Spain and Australia found vitamin D deficiency in pregnancy was associated with lower child cognitive scores, while studies in the UK, Denmark and India determined no association.⁴⁶ In addition, a study of HIV-uninfected women in the US found a small positive association between maternal and cord blood 25(OH)D concentration and child IQ at 4–7 years of age.⁴⁷ No randomized trials of vitamin D supplementation in pregnancy have assessed child neurodevelopment outcomes to date.⁴⁶ As a result, the relationship of prenatal vitamin D status with child neurodevelopment remains unclear.

This study has several limitations. First, geographic region of residence was highly correlated with ethnicity (i.e. 100% of Puerto Rico residents were Hispanic or multiracial) and therefore it was not possible to completely disentangle these potential risk factors for low vitamin D status. Region of residence likely captured aspects of both ethnicity and sun exposure and therefore resulted in better fit in stepwise selection models over ethnicity alone. In addition,due to the observational design of the study we cannot rule out residual or unmeasured confounding. As a result, randomized controlled trials of vitamin D supplementation will be needed to determine a causal effect. Further, the relatively small sample size likely led to inadequate power to detect modest effect sizes on child neurodevelopment as well as binomial outcomes like preterm birth and SGA. In addition, there is evidence that use of serum 25(OH)D status in the absence of vitamin D binding protein may overestimate vitamin deficiency in Black populations and require altered cut points to define vitamin D status.²⁷ Nevertheless, our results were robust to using 25(OH)D cut-offs of <20 ng/mL and 20–30 ng/mL to define deficiency and insufficiency among Black non-Hispanic women. In addition, relatively few women (n=88) in our study had circulating 25(OH)D levels >40 ng/mL which are required to optimize production of 1,25(OH)2D during pregnancy and therefore we may have underestimated the magnitude of risk for poor birth, growth and development outcomes associated with low vitamin D levels.⁴⁸

Overall, this study expands the growing evidence that vitamin D status in pregnancy can influence birth outcomes and infant growth to HEU infants in a high-income setting. Therefore vitamin D supplementation in pregnancy may reduce the high risk of preterm birth, low birthweight and linear growth faltering experienced infants born to HIV-infected mothers. Randomized controlled trials of vitamin D supplementation for HIV-infected pregnant women at risk for vitamin D deficiency or insufficiency in the US and diverse settings are warranted.