Timing of Antiretroviral Therapy

Abstract

Background

Combination antiretroviral therapy (cART) initiation during pregnancy reduces the risk of perinatal human immunodeficiency virus (HIV) transmission; however, studies have suggested that there may be unintended adverse consequences on birth outcomes for selected cART regimens.

Methods

We analyzed adverse birth outcomes among a prospective cohort of 1307 pregnant women with HIV in Dar es Salaam who initiated cART during the first or second trimester of a singleton pregnancy. Our primary analysis compared birth outcomes by gestational age at cART initiation among these women initiating cART in pregnancy.

Results

Among women who initiated cART in pregnancy, there was no relationship of gestational age at cART initiation with the risk of fetal death or stillbirth. However, women who initiated cART before 20 weeks of gestation compared with after 20 weeks had increased risk of preterm birth (risk ratio [RR], 1.30; 95% confidence interval [CI], 1.03–1.67) but decreased risk of small-for-gestational age birth (RR, 0.71; 95% CI, .55–.93).

Conclusions

With increasing use of cART preconception and early in pregnancy, clinicians should be aware of the benefits and potential risks of cART regimens to optimize birth outcomes.

In a cohort of women with HIV initiating combination antiretrovirals for the first time during pregnancy, we observed increased preterm birth, decreased small-for-gestational age birth, and no differences in fetal death between early versus late treatment initiators.

Combination antiretroviral therapy (cART) during pregnancy and lactation dramatically decreases the risk of perinatal transmission of human immunodeficiency virus (HIV). Between 2010 and 2018, expanded access to cART through Option B+ has been estimated to avert 1.4 million child HIV infections [1]. However, there is evidence that some cART regimens may have unintended adverse effects on some birth outcomes [2–6]. As a result, continued reductions in maternal to child transmission HIV infection through expanded coverage of effective cART and addressing adverse birth outcomes are priorities to improve child health and meet global HIV goals.

There is growing evidence on the complex relationship between HIV, timing of cART initiation, and adverse birth outcomes. Pregnant women with untreated HIV infection have an increased risk of fetal death, preterm birth, low birthweight, and small-for-gestational age (SGA) births compared with pregnant women who did not have HIV [7]. However, during the era of cART, several studies have observed that pregnant women on cART experience a greater risk of adverse birth outcomes than HIV-negative women [8], and that these adverse birth outcomes may vary by cART regimen and by the timing of treatment in pregnancy [9, 10]. Initiation of cART before conception has been shown to be associated with increased risk of preterm birth compared with initiation during pregnancy among women with HIV [10, 11], although this risk may vary by cART regimen [10, 12]. In addition, evidence on timing of cART before and after conception on the risk of SGA births (a marker of fetal growth restriction) is mixed, with some studies reporting no association [4, 5] and others reporting a higher risk for cART before conception [3]. There have been few studies that report on the relationship of cART timing with stillbirth [10]. A study in Botswana found that cART initiation before conception was associated with elevated risk of stillbirth compared with cART or monotherapy starting after conception [3], although there were no relationships observed in other cohorts comparing cART initiation before and after conception [3, 6, 13]. As a result, evidence suggests cART may increase the risk of preterm birth, but the relationship of cART and its timing of initiation with fetal loss and SGA remains unclear.

Prior studies on the timing of cART initiation with adverse birth outcomes have primarily focused on comparisons of initiation before versus after conception. Only 1 study has investigated timing within pregnancy for a tenofovir-based regimen and found no adverse birth outcomes associated with timing [14]. It is important to examine whether there is a relationship between the timing of cART initiation and adverse birth outcomes within pregnancy for more regimens of cART. We examined the relationship of timing of cART initiation with adverse birth outcomes in a prospective cohort of women who primarily initiated an efavirenz-based cART regimen in pregnancy in Tanzania. As a secondary analysis, we also examined the risks of adverse birth outcomes associated with preconception cART initiation compared cART initiation in pregnancy.

METHODS

Study Design

In this analysis, we used data from a randomized, double-blind, placebo-controlled trial of vitamin D₃ supplementation conducted among pregnant women with HIV in Dar es Salaam, Tanzania (ClinicalTrials.gov Identifier NCT02305927) [15, 16]. Pregnant women were recruited from the 5 large antenatal care clinics that provided prevention of mother-to-child transmission services. Women were eligible for the trial if they (1) were at least 18 years of age, (2) were HIV positive, (3) were receiving or initiated cART at the time of enrollment, (4) were 12 to 27 weeks gestation as defined by self-reported date of last menstrual period, and (5) had albumin-adjusted calcium levels within the normal physiologic range (≤2.6 mmol/L). All participants provided written consent for participation. Study participants were randomized to receive either oral 3000 IU vitamin D₃ supplements or placebo to be taken daily from randomization until 12 months postpartum. The institutional review boards at the National Institute of Medical Research, the Harvard T.H. Chan School of Public Health, Muhimbili University of Health and Allied Sciences, and the Tanzania Food and Drug Authority approved the trial. Trial enrollment began in June 2015 and follow-up was completed in September 2019.

Tanzania adopted Option B+ in 2013, meaning that all pregnant and breastfeeding women with HIV were prescribed lifelong cART, regardless of CD4 T-cell count or HIV disease severity [17]. The first-line cART regimen during the trial was tenofovir (TDF), lamivudine (3TC), and efavirenz (EFV). At enrollment, study participants received a full clinical examination from study physicians, during which their HIV disease stage was assessed according to World Health Organization (WHO) criteria. Maternal height and weight were measured by trained study nurses. Study nurses administered a questionnaire to collect information on sociodemographic characteristics and self-reported comorbidities. For women who initiated cART, a CD4 T-cell count was obtained (FACSCalibur Flow Cytometer; Becton Dickinson, San Jose, CA).

Women were seen at study visits every 4 weeks during pregnancy. Information on miscarriages and stillbirths were obtained from hospital records and through maternal report on fetal breathing and movement when records were not available (2.2% of records). Study nurses and midwives attended deliveries, or if the birth occurred outside Dar es Salaam, relevant information was obtained from the mother or clinic staff by telephone. Study staff or hospital nurses obtained birthweight to the nearest 5 grams using a digital scale.

Exposure and Outcome Definitions

Date of cART initiation was obtained from HIV clinic records, and the timing of ART initiation was defined in relation to the date of conception, as estimated by self-reported last menstrual period (LMP), and gestational age for those who initiated cART during pregnancy.

For outcomes, miscarriage was defined as a fetal death before 20 weeks of gestation, early stillbirth as a fetal death occurring between 20 and 27 weeks of gestation, late stillbirth as fetal death occurring between 28 weeks of gestation and 37 weeks, and term stillbirth as fetal death after 37 weeks. Preterm birth was defined as a birth occurring at a gestational age less than 37 weeks based on LMP, and very preterm birth was defined as a as birth before 32 weeks. Low birth weight (LBW) was defined among live-born infants as weighing less than 2500 grams. Size-for-gestational age was calculated using the INTERGROWTH-21(st) standards [18]. If gestational age was greater than 300 days, a 300-day gestation was assumed for INTERGROWTH calculations. Small-for-gestational age births were defined as livebirths below the 10th percentile of birthweight for their gestational age and sex, whereas very small-for-gestational age (VSGA) births were defined as livebirths with weight below the 3rd percentile for gestational age and sex.

Statistical Analysis

The primary analysis was conducted among pregnant women who initiated cART during pregnancy with a singleton gestation and available birth outcome data. Log-binomial regression models were used to evaluate the association of the timing of cART initiation (gestational age in weeks) with fetal death (miscarriage or stillbirth) using relative risks. When log-binomial models did not converge, overdispersed Poisson regression models were used. Based on our a priori statistical analysis plan, timing of initiation was also modeled as <20 weeks of gestation versus ≥20 weeks of gestation. We conducted sensitivity analyses excluding fetal deaths before 20 weeks and 28 weeks to assess the potential contribution of selection bias.

Among live births, we used log-binomial regression to analyze the association between timing of cART initiation and adverse birth outcomes: LBW, birth weight <2000 grams, preterm birth, very preterm birth, SGA, and VSGA. We chose to include birth weight <2000 grams as an outcome, because very LBWs (<1500 grams) were rare in the study population. Overdispersed Poisson regression models were again used in cases of no convergence. We also used linear regression models to analyze the association between timing of cART initiation and continuous birth weight in grams and continuous gestational age in weeks. A multinomial logistic regression model was used to further investigate the relationship with of cART initiation timing with term and appropriate SGA (term-AGA), term-SGA, preterm-AGA, and preterm-SGA livebirths.

Multivariable models controlled for maternal CD4 T cell (<350, 350–500, ≥500 cells/μL), WHO disease stage (1, 2+), self-reported history of hypertension, any alcohol use (in the last month), body mass index at randomization (<18.5, 18.5–24.9, 25.0–29.9, ≥30.0 kg/m²), parity (0–1, 2–3, 4+), maternal age (18–24, 25–34, ≥35 years), maternal education (no formal education, primary, secondary, advanced), marital status (married, cohabitating, single), wealth quintile, clinic site, and whether they received vitamin D or placebo. A missing indicator was used to account for those who were missing maternal CD4 count. For all other covariates, complete case analysis was performed due to low levels (<5% for each covariate) of unavailable covariate information.

A secondary analysis was conducted to investigate the association between preconception cART and pregnancy outcomes. For this analysis, timing of cART initiation was modeled in 3 categories (preconception, <20 weeks of gestation, and ≥20 weeks of gestation). Log-binomial regression and linear regression were used to evaluate the association between this exposure and all pregnancy and birth outcomes controlling for the same set of covariates noted above for the primary analysis. Sensitivity analyses evaluated the robustness of the primary analysis to inclusion of multiple gestation pregnancies. These sensitivity analyses used similar methods to the primary analysis but used compound symmetry correlation structures to account for correlated outcomes between siblings. Analyses were conducted using R 3.5.1.

RESULTS

The parent trial enrolled 2300 pregnant women with HIV, with 2260 women (98.3%) having known birth outcomes. Among women with known birth outcomes, 2217 had singleton births (98.1%) of which 910 women initiated cART preconception (41.0%) and 1307 women initiated during pregnancy (59.0%) (Figure 1). Baseline characteristics of the population for women with singleton pregnancies that started cART preconception and during pregnancy are presented in Table 1.

Figure 1.

Study Population Flow Diagram.

Table 1.

Baseline Maternal Characteristics For Women With HIV With Singleton Pregnancies Stratified By cART Initiation Before Conception or During Pregnancy

Covariate	Women Starting cART Before Conception (n = 910) n (%) or Mean ± SD	Women Starting cART During Pregnancy (n = 1307) n (%) or Mean ± SD
Maternal Age (Years)
18–24	83 (9.1%)	268 (20.5%)
25–34	508 (55.8%)	790 (60.4%)
≥35	319 (35.1%)	249 (19.1%)
Maternal Education
No formal education	117 (12.9%)	121 (9.3%)
Primary	551 (60.5%)	741 (56.8%)
Secondary	175 (19.2%)	352 (27.0%)
Advanced	67 (7.4%)	91 (7.0%)
Marital Status
Married	718 (79.0%)	957 (73.5%)
Single	26 (2.9%)	315 (24.2%)
Widowed or divorced	165 (18.2%)	30 (2.3%)
Parity
1	143 (15.7%)	398 (30.5%)
2–3	569 (62.5%)	736 (56.4%)
≥4	198 (21.8%)	171 (13.1%)
Body Mass Index (kg/m2)
<18.5	35 (3.9%)	37 (2.8%)
18.5–24.9	415 (45.7%)	591 (45.3%)
25.0–29.9	297 (32.7%)	422 (32.4%)
≥30.0	161 (17.7%)	254 (19.5%)
Self-reported history of hypertension	69 (7.6%)	76 (5.9%)
Alcohol use in the last month	101 (11.3%)	196 (15.3%)
Maternal CD4 T-Cell Count (Cells/mL3)
<350	97 (33.0%)	358 (48.1%)
350–500	83 (28.2%)	214 (28.8%)
≥500	114 (38.8%)	172 (23.1%)
WHO HIV Disease Stage
I	671 (73.7%)	1231 (94.2%)
II	87 (9.6%)	52 (4.0%)
III	137 (15.1%)	21 (1.6%)
IV	15 (1.6%)	3 (0.2%)
cART Regimen
TDF/3TC/EFV	891 (98.1%)	1299 (99.5%)
Other	17 (1.9%)	7 (0.5%)
Weeks gestation at cART initiation		17.2 ± 4.8
Trimester of cART Initiation
First (≤13 weeks)	–	223 (17.1%)
Second (>13 weeks)	–	1084 (82.9%)
Initiation Before or After 20 Weeks
<20 weeks	–	916 (70.1%)
≥20 weeks	–	391 (29.9%)

Abbreviations: cART, combination antiretroviral therapy; EFV, efavirenz; HIV, human immunodeficiency virus; SD, standard deviation; TDF, tenofovir; WHO, World Health Organization; 3TC, lamivudine.

The primary analytic study population included 1307 singleton gestation pregnancies among women who initiated cART during pregnancy and had birth outcome data (Figure 1). Among women who initiated cART in pregnancy, 99.5% of women initiated cART on a TDF/3TC/EFV regimen (Table 1), 223 (17%) initiated cART in the first trimester, and 1084 (83%) initiated in the second trimester. The majority of women had WHO HIV Stage I disease (94%). Among women initiating cART in pregnancy, the cumulative incidence of fetal death was 6.6% and among livebirths 24.6% were born preterm (<37 weeks), 19.6% had SGA <10th percentile, and 10.0% were born with LBW (<2500 grams) (Supplemental Table).

The unadjusted and adjusted associations of cART initiation timing in pregnancy with fetal death are presented in Table 2. There was no observed association between gestational age at cART initiation and fetal death. There was also no association with stillbirth defined by fetal death after 20 or 28 weeks gestation.

Table 2.

Association of cART Initiation Timing With Fetal Death, Stillbirth and Preterm Delivery Among Women With HIV With Singleton Pregnancies Who Initiated cART in Pregnancy (n = 1307)

Outcome	Gestational Age at cART Initiation	Events/Number in Group (%)	Unadjusted Risk Ratioa (95% CI)	Adjustedb Risk Ratioa (95% CI)
Fetal death (miscarriage or stillbirth)	Continuous gestation (weeks)	82/1307 (6.3%)	0.99 (0.95–1.03)	0.99 (0.94–1.03)
	≥20 weeks gestation	19/391 (4.9%)	Ref	Ref
	<20 weeks gestation	63/916 (6.9%)	1.42 (0.88–2.39)	1.49 (0.90–2.57)
Stillbirth >20 weeks	Continuous gestation (weeks)	77/1302 (5.9%)	1.00 (0.96–1.05)	1.00 (0.95–1.05)
	≥20 weeks gestation	19/391 (4.9%)	Ref	Ref
	<20 weeks gestation	58/911 (6.4%)	1.31 (0.81–2.22)	1.37 (0.83–2.37)
Stillbirth >28 weeks	Continuous gestation (weeks)	57/1275 (4.5%)	1.00 (0.95–1.06)	1.00 (0.95–1.06)
	≥20 weeks gestation	15/386 (3.9%)	Ref	Ref
	<20 weeks gestation	42/889 (4.7%)	1.22 (0.70–2.23)	1.26 (0.71–2.38)

Abbreviations: cART, combination antiretroviral therapy; CI, confidence interval; HIV, human immunodeficiency virus; Ref, reference category, .

^aLog-binomial regression models were used to analyze the association between gestational age at cART initiation and dichotomous outcomes.

^bAdjusted for maternal CD4 (<350, 350–500, ≥500 cells/mL³, Missing), World Health Organization disease stage (1, 2+), self-reported history of hypertension, alcohol use (none in the last month or in the last month), body mass index (<18.5, 18.5–24.9, 25.0–29.9, ≥30.0 kg/m²), parity (0–1, 2–3, 4+), maternal age (18–24, 25–34, ≥35 years), maternal education (no formal education, primary, secondary, advanced), marital status (married, cohabitating, single), wealth quintile, site (Mnazi Moja, Mabagala Round Table, Mbagala Rangi Tatu, Buguruni, and Tabata), and vitamin D or placebo. A missing indicator was used for maternal CD4.

Table 3 presents the association between cART initiation timing in pregnancy and other birth outcomes among livebirths. There was no association between gestational age at cART initiation with continuous birthweight, risk of LBW, or birthweight less than 2000 grams. Initiation of cART during pregnancy but before 20 weeks of gestation was associated with a 30% increase in the risk of preterm birth, controlling for covariates compared with cART initiation at or after 20 weeks (adjusted risk ratio [aRR], 1.30; 95% confidence interval [CI], 1.03–1.67). In contrast, cART initiation before 20 weeks compared with cART initiation at or after 20 weeks was associated with a 29% decrease in the risk of SGA birth after controlling for covariates (aRR, 0.71; 95% CI, .55–.93) (Table 3). We also examined the relationship of timing of cART initiation with the risk of term-SGA, preterm-AGA, and preterm-SGA births using multinomial regression (Supplemental Table 2). Combination ART initiation before 20 weeks of gestation decreased the risk of term-SGA births (aRR, 0.68; 95% CI, .49–.96) compared with term-AGA births. Although not statistically significant, women who initiated cART before 20 weeks had greater risk of both preterm-AGA (adjusted odds ratio [aOR], 1.30; 95% CI, .94–1.80) and preterm-SGA births (aOR, 1.27; 95% CI, .32–5.04) relative to term-AGA.

Table 3.

Association of cART Initiation Timing With Live Birth Outcomes Among Women With HIV With Singleton Births Who Initiated cART in Pregnancy (n = 1225)

Outcome	Gestational Age at cART Initiation	Mean ± SD or Events/Number in Group (%)	Unadjusted Mean Difference or Risk Ratioa (95% CI)	Adjustedb Mean Difference or Risk Ratioa (95% CI)
Birth weight (g)	Continuous gestation (wks)	3132 g (±526 g)	2.6 g (−3.6 to 8.7 g)	2.3 g (−4.0 to 8.6 g)
	≥20 weeks gestation	3141 g (±495 g)	Ref	Ref
	<20 weeks gestation	3127 g (±539 g)	−13.9 g (−78.1 to 50.4 g)	−2.2 g (−67.8 to 63.7 g)
Low birth weight (<2500 g)	Continuous gestation (wks)	96/1222 (7.9%)	0.97 (0.93–1.01)	0.97 (0.93–1.01)
	≥20 weeks gestation	25/370 (6.8%)	Ref	Ref
	<20 weeks gestation	71/852 (8.3%)	1.23 (0.81–1.94)	1.15 (0.73–1.86)
Birth weight <2000 g	Continuous gestation (wks)	20/1222 (1.6%)	1.02 (0.93–1.13)	1.03 (0.93–1.15)
	≥20 weeks gestation	5/370 (1.4%)	Ref	Ref
	<20 weeks gestation	15/852 (1.8%)	1.30 (0.51–3.97)	1.53 (0.50–6.10)
Gestational age (weeks)	Continuous gestation (wks)	38.7 wks (±3.3 wks)	0.06 (0.02–0.10)	0.06 wks (0.02–0.10)
	≥20 weeks gestation	39.2 wks (±3.2 wks)	Ref	Ref
	<20 weeks gestation	38.5 wks (±3.3 wks)	−0.78 wks (−1.17 to −0.38)	−0.77 wks (−1.17 to −0.36)
Preterm live birth (<37 weeks gestation)	Continuous gestation (wks)	293/1225 (23.9%)	0.98 (0.96–1.00)	0.98 (0.96–1.00)
	≥20 weeks gestation	74/372 (19.9%)	Ref	Ref
	<20 weeks gestation	219/853 (25.7%)	1.29 (1.03–1.63)	1.30 (1.03–1.66)
Very preterm live birth (<32 weeks gestation)	Continuous gestation (wks)	43/1225 (3.5%)	1.01 (0.95–1.07)	1.01 (0.95–1.08)
	≥20 weeks gestation	11/372 (3%)	Ref	Ref
	<20 weeks gestation	32/853 (3.8%)	1.27 (0.67–2.60)	1.30 (0.65–2.82)
SGA (<10th percentile)	Continuous gestation (wks)	217/1225 (17.7%)	1.02 (0.99–1.04)	1.02 (0.99–1.05)
	≥20 weeks gestation	78/372 (21%)	Ref	Ref
	<20 weeks gestation	139/853 (16.3%)	0.78 (0.61–1.00)	0.71 (0.55–0.93)
VSGA (<3rd percentile)	Continuous gestation (wks)	88/1225 (7.2%)	0.99 (0.95–1.03)	0.99 (0.95–1.04)
	≥20 weeks gestation	27/372 (7.3%)	Ref	Ref
	<20 weeks gestation	61/853 (7.2%)	0.99 (0.64–1.55)	0.87 (0.56–1.40)

Abbreviations: cART, combination antiretroviral therapy; CI, confidence interval; g, grams; HIV, human immunodeficiency virus; Ref, reference category,; SD, standard deviation; SGA, small-for-gestational age; VSGA, very small-for-gestational age; wks, weeks.

^aLog-binomial regression models were used to analyze the association between gestational age at cART initiation and dichotomous outcomes. Linear regression models were used to analyze the association between gestational age at cART initiation and birthweight and gestational age. For these linear outcomes, mean change in weight or weeks of gestational age are reported.

^bAdjusted for maternal CD4 (<350, 350–500, ≥500 cells/mL³, missing), WHO disease stage (1, 2+), self-reported history of hypertension, alcohol use (none in the last month or in the last month), body mass index (<18.5, 18.5–24.9, 25.0–29.9, ≥30.0 kg/m²), parity (0–1, 2–3, 4+), maternal age (18–24, 25–34, ≥35 years), maternal education (no formal education, primary, secondary, advanced), marital status (married, cohabitating, single), wealth quintile, site (Mnazi Moja, Mabagala Round Table, Mbagala Rangi Tatu, Buguruni, and Tabata), and vitamin D or placebo. A missing indicator was used for maternal CD4.

We also conducted secondary analyses that compared birth outcomes for pregnant women who initiated on cART preconception to women who initiated cART in pregnancy before and after 20 weeks. There were no relationships between preconception cART and risk of fetal deaths or stillbirth compared with those who initiated in pregnancy after 20 weeks (Supplemental Table 3). In addition, there were no associations between preconception cART initiation and preterm birth, LBW, SGA, or other adverse outcomes among livebirths except for continuous gestational age (Supplemental Table 4). However, cART-initiated preconception was associated with a half-week decrease in gestational age (−0.54 weeks; 95% CI, −.95 to −.12) compared with initiation of cART in pregnancy after 20 weeks of gestation. In our sensitivity analyses including 58 twin births, the estimates remained consistent across cART exposures for all birth outcomes (Supplemental Tables 5 and 6).

DISCUSSION

In this prospective cohort study, we found that pregnant women with HIV who initiated cART before 20 weeks of gestation were at increased risk of preterm birth but lower risk of SGA birth compared with those initiating cART after 20 weeks. There was no evidence that timing of cART initiation was associated with the risk of fetal death or stillbirth. In secondary analyses, we found that preconception cART was associated with shorter gestation duration compared with initiation in pregnancy after 20 weeks gestation, but there was no relationship with other adverse birth outcomes.

Although we found a high rate of fetal death in the study population, we found no relationship between the timing of cART initiation and the risk of fetal death and stillbirth. The rate of third trimester and intrapartum stillbirth in our study among 2217 births to women who initiated ART before or during pregnancy was 4.4%, which is greater than the estimated 2.2% in the general population of Tanzania [19]. This high rate is consistent with literature showing higher rates of stillbirth and other adverse outcomes among women on cART with HIV compared with women without HIV across multiple cART regimens [8, 12, 20–24]. Our finding of no relationship of timing of cART initiation within pregnancy with the risk of fetal death is consistent with studies that found no association between cART timing and stillbirth among women with HIV [6, 14]. However, our finding of no association of preconception cART with the risk of stillbirth is in contrast to a study in Botswana that found a significantly higher rate of stillbirth among pregnant women who initiated cART preconception compared with other pregnant women with HIV including those with no treatment and those who received ART monotherapy [3]. Two potential reasons for the difference between our preconception findings and the findings from Botswana are the comparison group and the cART regimens. First, the comparison group used in the Botswana study included those who initiated cART during pregnancy in comparison to those on monotherapy or no cART, whereas our comparison group included only women who received cART preconceptionally. Second, different cART regimens may account for different results, because 87% of the women in the Botswana study were on a zidovudine (ZDV)/3TC/nevirapine (NVP) regimen compared with the first-line TDF/3TC/EFV regimen in our Tanzanian cohort [3]. Tenofovir-based regimens have been shown to be associated with significantly lower risk of stillbirth compared with ZDV/3TC/NVP in a meta-analysis of clinical trials [9]. As a result, the relationship of cART initiation timing with the risk of fetal death remains unclear and may differ by cART regimen as well as other population factors.

There is a relatively large evidence base that has linked cART initiation before conception with increased risk of preterm birth [10]. Our study further demonstrates that initiating cART before 20 weeks of gestation was associated with increased risk of preterm birth compared with initiating after 20 weeks. Prior studies from Botswana and Tanzania have also found elevated rates of preterm birth associated with nucleoside reverse-transcriptase inhibitor (NRTI)-based regimens [3, 4, 23]. Our study is consistent with these results, and we found elevated rates of preterm birth associated with early exposure to an NRTI-based regimen. Several mechanisms have been proposed that may explain the associations between cART and preterm birth. These include modification of the cytokine environment by cART leading to inflammation [25, 26], increased incidence of pre-eclampsia through immune reconstitution [27], and low progesterone levels due to atypical placental development [28, 29]. As a result, earlier and longer exposure to cART during pregnancy seems to be associated with an increased risk of preterm birth.

In contrast to the preterm risks, we found that that initiation of cART in the first 20 weeks of pregnancy was associated with reduced risk of SGA births, an indicator of fetal growth restriction, among women initiating cART at or after 20 weeks of pregnancy. Furthermore, preconception cART had a similar magnitude of reduced risk of SGA relative to initiation after 20 weeks gestation. The broader evidence base on timing of cART initiation and risk of SGA is inconclusive. Our results are not consistent with data from Botswana indicating that cART before conception was associated with increased incidence of SGA birth [3] or no increased incidence of SGA birth [14]. However, this may be due to differences in cART regimens and comparison groups between studies. In cohorts in the United States and Tanzania, no association was found between cART initiation before conception or during pregnancy with SGA compared with antiretroviral monotherapy [4, 10]. However, our result may be supported by some earlier cohort data collected during the introduction of antiretroviral in Tanzania and suggests that cART may reduce the negative effects of HIV on intrauterine growth, because although untreated women with HIV had significantly higher rates of SGA births, women with HIV on cART had similar rates as women without HIV [30]. Early cART may improve intrauterine growth through multiple mechanisms. Both pregnancy and untreated HIV increase energy requirements for women [31, 32]. Initiation of cART reduces viral load, which is correlated with reduced energy requirements [33]. Furthermore, treatment with cART can reduce the incidence of opportunistic infections during pregnancy, which can contribute to fetal growth restriction. Combination ART may also decrease viral load in the placenta [34], which may improve placental function and fetal growth. Therefore, there is biologic plausibility that earlier cART may improve fetal growth in some settings.

Despite the studies strengths including the large sample size and low loss to follow-up (<2%), our study has several limitations. Selection bias is possible in studies assessing cART timing, particularly from the exclusion of women who would have initiated on cART late in pregnancy but had an early delivery or loss before enrollment [35]. We would expect this to bias our estimate towards more harmful associations between early initiation and stillbirth. However, we performed additional sensitivity analyses to investigate outcomes that occurred later in pregnancy (≥20 and ≥28 weeks of gestation), and we still saw no effect of cART timing on stillbirth. Furthermore, misclassification of stillbirths and neonatal deaths is possible and may have caused some degree of bias if this misclassification was related to timing of cART initiation. In addition, it is possible that women who initiate cART earlier in pregnancy or preconception have more access to care or differ in other factors from those who start cART later in pregnancy. For example, we were unable to control for sexually transmitted infections and bacterial vaginosis in pregnancy, which may be associated with both cART timing and birth outcomes. However, we controlled for several important confounders including socioeconomic factors, WHO HIV disease stage, and other factors. CD4 T-cell counts were only available for 57% of the population, which adds to the potential for residual confounding. If women who initiated cART earlier tended to have more access to care, it is possible that the findings from this study may have underestimated the association of earlier cART initiation with preterm birth and may have overestimated the association with SGA. The risk of confounding by HIV disease severity may not be as large in the Option B+ era, because initiation is not based on disease stage or CD4 T-cell criteria.

Finally, we were unable to extrapolate information on other or newer regimens of cART because almost all study participants were on the same tenofovir-based regimen. It is important to note that newer dolutegravir-based regimens may have lower risks of preterm birth with earlier treatment; a trial found that these regimens were associated with lower risk preterm birth compared with an efavirenz-based regimen [36]. Research is needed to determine how the timing of newer cART regimens, including dolutegravir-based regimens, impact maternal and infant outcomes.

CONCLUSIONS

It is undisputed that cART during pregnancy improves women’s health and reduces perinatal HIV transmission; however, this study adds to the body of evidence that cART initiation timing is associated with some adverse birth outcomes. Our preconception secondary analysis results are consistent with recent guidelines on the timing of ART on preterm birth [37], and our results offer additional findings on timing of cART within pregnancy. These findings highlight the importance of further monitoring and studying of the relationship of timing of cART initiation, including newer regimens, during pregnancy. With increasing use of cART preconception and early in pregnancy, clinicians should be aware of both benefits and risks in relation to adverse birth outcomes to effectively optimize cART regimens for maternal and child health.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Financial support. The trial was funded by the Eunice Kennedy Shriver National Institute of Child Health and Human Development (Grant R01 HD83113). C. P. D. was supported in part by the National Institutes of Health ([NIH] Grants K24DK104676 and 2P30 DK040561). M. K. Q. was supported by NIH under Award Number T32AI007358.

Potential conflicts of interest. All authors: No reported conflicts of interest.

All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest.