Prediction and Prevention of Preterm Birth in Pregnant Women Living with HIV on Antiretroviral Therapy

Abstract

Introduction:

The rate of spontaneous preterm birth among pregnant women living with HIV on antiretroviral therapy (ART) is 3–4-fold higher when compared to HIV-negative women. The pathophysiology of preterm birth related to HIV or ART remain unknown, especially as women living with HIV are often excluded from preterm birth studies.

Areas Covered:

This review discusses the currently available evidence on the prediction and prevention of preterm birth in pregnant women living with HIV. A review of the literature was conducted of primary articles between 2005 and 2021 measuring the association or lack thereof between combination ART and preterm birth, as well as of other predisposing factors to preterm birth in women living with HIV, including cervical length, vaginal microbiome, and cervico-vaginal biomarkers.

Expert Opinion:

Further research into the effect of ART exposure on preterm birth risk is critical, and development of preterm birth predictive tools in this population should be a priority. Vaginal progesterone supplementation deserves further investigation as a therapeutic option to prevent recurrent preterm birth in pregnant women living with HIV. The ProSPAR study, a multi-center randomized controlled trial studying progesterone supplementation in pregnant women on protease inhibitor-based regimens, has been designed but is not yet recruiting patients.

1. INTRODUCTION

Preterm birth, defined as delivery occurring between 20⁺⁰ weeks and 36⁺⁶ weeks of gestation, occurs in 5 to 18 percent of births worldwide[1]. The incidence of preterm births varies among ethnic groups. In 2018 in the United States, 14.1% of live births in non-Hispanic Black populations were preterm, 9.7% of live births in Hispanic populations were preterm, and 9.1% of live births in non-Hispanic White populations were preterm[2]. Preterm birth continues to be a major cause of perinatal morbidity and mortality, especially in women living with HIV[3]. In the United States, preterm births account for 75% of perinatal mortality and more than half the long-term morbidity[4]. In the general population, preterm birth can be subdivided into two categories: spontaneous preterm birth (preterm labor and Preterm Premature Rupture of Membranes, PPROM), which is responsible for 70% of all preterm births[5], and iatrogenic preterm birth (30%), when delivery is medically indicated due to fetal distress or maternal compromise.

The rate of spontaneous preterm birth among pregnant women living with HIV is three-to-four-fold higher when compared to pregnant HIV-negative women[6,7]. In women living with HIV, factors associated with preterm birth include substance use, ethnicity, cigarette smoking, unsuppressed HIV-RNA viral load, a history of a prior spontaneous preterm birth, and the use of antiretroviral therapy (ART)[3]. While placental mitochondrial toxicity has been proposed as a causal pathway for some risk factors for preterm birth in some women living with HIV[8], the molecular basis of preterm birth in many women living with HIV remains unknown.

Management strategies for prevention of preterm birth include vaginal progesterone, 17 alpha progesterone caproate, and the use of cervical cerclage. Progesterone is a key hormone involved in pregnancy maintenance. The administration of progesterone antagonists has been demonstrated to enhance the onset of labor[9], and studies show that vaginal progesterone supplementation reduces the rate of spontaneous preterm birth in those with prior history of spontaneous preterm birth or with cervical shortening in the current pregnancy[10]. Thus, a decline in progesterone activity can lead to preterm delivery, with adverse pregnancy outcomes. Among women living with HIV, studies have shown that use of ART during pregnancy can reduce progesterone serum concentrations, with subsequent progression to preterm birth and other adverse pregnancy outcomes[11]. Therefore, it has been theorized that increasing serum progesterone levels in pregnant women living with HIV would reduce the prevalence of preterm birth. For pregnant women living with HIV with cervical shortening who are eligible for transvaginal cervical cerclage, adverse perinatal events associated with cervical insufficiency may be further reduced with placement of a cervical cerclage.

Women living with HIV have been rarely included in preterm birth trials because HIV itself is a potent risk factor for preterm birth. This complicates studying predictive and preventative factors of preterm birth in women living with HIV on ART. Our goal is to discuss the prediction and prevention of preterm birth in pregnant women living with HIV on ART.

2. METHODS

This is a narrative review of the risk factors and management options pertaining to prediction and prevention of preterm birth in women living with HIV. A literature search was conducted in the PubMed database between September 11th and October 7th, 2021 for studies of any design or in any setting, on the correlation of pertinent exposures or interventions to the rate of preterm birth in pregnant women living with HIV. The search terms included were: HIV; antiretroviral therapy (ART); preterm birth; prediction; prevention; pregnancy; cervical length; vaginal microbiota; cervico-vaginal biomarkers; machine learning; artificial intelligence; vaginal progesterone; 17-alpha hydroxyprogesterone caproate; cervical cerclage; cervical pessary. Results were limited to articles in English. Other citations were identified in references of available literature and from ClinicalTrials.gov. The results and discussion of the literature review are summarized by category: section 3 describes findings pertaining to prediction of preterm birth in pregnant women living with HIV, and section 4 describes findings pertaining to prevention of preterm birth in pregnant women living with HIV.

3. PREDICTION OF PRETERM BIRTH IN PREGNANT WOMEN LIVING WITH HIV

Proposed systems for quantitative prediction of preterm birth risk scoring typically calculate an additive score based on epidemiologic, historical, and clinical risk factors[12,13]. However, the positive predictive value of preterm birth of most risk scoring systems is low, in the range of 20–30%[14]. Because HIV itself is an independent risk factor for preterm birth, a unique set of factors are relevant for consideration, including ART exposure, cervical length, cervico-vaginal biomarkers, vaginal microbiota, and imaging modalities.

3.1. There is conflicting evidence associating antiretrovirals with preterm birth in pregnant women living with HIV

The beneficial effects of ART for preventing perinatal transmission are indisputable. However, the literature contains conflicting findings on the association between ART exposure and preterm birth. Combination ART exposure in utero has been associated with an increased rate of preterm delivery, particularly with regimens that contain protease inhibitors (PI)[15,16]. The European Pregnancy and Paediatric HIV Cohort Collaboration (EPPICC) Study Group reported no association between nucleoside reverse transcriptase inhibitor (NRTI) therapy and preterm birth[17]. However, a 2006 prospective cohort study (n=999) found that combination ART with a PI was associated with an increased risk of preterm birth (OR 1.9, 95% confidence interval [CI] 1.1–3.0)[15]. In a retrospective analysis of the Mma Bana Study randomized clinical trial (RCT)[18], Powis et al found that PI-based combination ART initiated in the third trimester of pregnancy was associated with a two-fold higher odds of a preterm delivery compared with triple NRTI-based combination ART[19]. In France, combination ART compared with zidovudine monotherapy was significantly associated with prematurity (adjusted odds ratio [aOR] 1.69, 95% CI 1.38–2.07; P < 0.01), and the prematurity rate was higher with ritonavir-boosted PI therapy than with non-boosted PI therapy started during pregnancy (14.4% vs 9.1% P = 0.05; adjusted hazard ratio [aHR] 2.03, 95% CI 1.06–3.89)[20].

Similar results were reported in the 2016 PROMISE RCT conducted in resource-limited settings that compared two protease inhibitor-containing combination ART regimens. Preterm birth <37 weeks was found to be more frequent in the zidovudine-based ART arm than with zidovudine alone (20.5% vs. 13.1%, P<0.001), and tenofovir-based ART was associated with higher rates of very preterm delivery before 34 weeks than zidovudine-based ART (6.0% vs. 2.6%, P=0.04)[21]. Analyses of the PROMISE RCT have reported women receiving ZDV-based and tenofovir (TDF)-based ART had significantly higher odds of preterm birth <37 weeks compared with zidovudine (ZDV) alone (aOR 1.68; 95% CI 1.10–2.57; and 2.71, 95% CI 1.39–5.29)[22]. In the AmRo study, combination ART exposure at <13 weeks of gestation was associated with a 44% preterm delivery rate compared with 21% when ART was started at or after 13 weeks of gestation, and 14% rate in controls. The ART-associated increase in preterm delivery rate was reportedly mainly seen after first trimester ART use, with no evidence of an increased risk of preterm delivery in women initiating ART between 24–28 weeks[23]. A study in Botswana (n=9,504) found that combination ART exposure that was initiated prior to pregnancy was significantly associated with preterm delivery (aOR 1.2; 95% CI 1.1–1.4) compared with all other pregnant women living with HIV. Compared with women initiating zidovudine (ZDV) alone in pregnancy, initiating combination ART in pregnancy was also significantly associated with increased odds of preterm birth (aOR 1.4; 95% CI 1.2–1.8)[24].

A prospective 2016 study (n=3314) found that compared with ZDV monotherapy, combination ART exposure prior to pregnancy was associated with higher risks of preterm birth (relative risk [RR] 1.23, 95% CI 1.04–1.47)[25]. In a retrospective cohort study (n=1557), iatrogenic preterm delivery was significantly associated with the use of combination ART during the second half of pregnancy (OR 6.2, 95% CI 1.4–26.8). In this same study spontaneous preterm delivery was not found to be significantly associated with ART exposure (OR 0.55, 95% CI 0.18–1.68)[26]. Analysis of Swiss data prospectively collected by the Swiss Mother and Child HIV Cohort Study (MoCHiV) and the Swiss HIV Cohort Study (SHCS) (n=1180) reported that odds ratios for prematurity in women receiving combination ART were 1.8 (95% CI 0.85–3.6) and 2.5 (95% CI 1.4–4.3) compared with women not receiving ART during pregnancy (P=0.004), and found that confounding by maternal risk factors or duration of ART exposure was not an explanation for the effects of ART on prematurity in women living with HIV[27].

Other analyses have found conflicting results. A prospective cohort study (n=777) reported that in adjusted analyses, combination ART with PI was not significantly associated with spontaneous preterm birth (OR 1.22; 95% CI 0.70–2.12)[28]. A retrospective cohort study (n=3074) reported that the risk of extremely to very preterm birth was 2.33 (95% CI 1.39 to 3.92) times as great in those who never initiated ART compared with those who did at any point before 27 weeks, and that among women on ART before delivery, ART initiation before conception was associated with the lowest preterm birth risk[29]. A meta-analysis of 14 studies (n=11,224) found that ART exposure overall during pregnancy was not associated with an increased risk of preterm birth (OR 1.01; 95% CI 0.76–1.34) and did not find a significant increased risk of preterm birth with regimens containing PI (OR 1.24, 95% CI 0.76–2.02). However, they did report that initiation of combination ART before pregnancy or in the first trimester resulted in increased odds of preterm birth (OR 1.71, 95% CI 1.09–2.67) compared with therapy initiation in the second trimester and beyond[30]. A recent 2020 analysis of a Canadian cohort (n=631) over a time period of 1997–2018 showed that women living with HIV were at higher risk of spontaneous preterm birth and that, among those with HIV, cumulative combination ART use was in fact associated with a lower risk of spontaneous preterm birth (HR 0.98, 95% CI 0.96–0.99)[6]. There was furthermore no association between exposure to ART in the first trimester and spontaneous preterm birth in this study (HR 0.77, 95% CI 0.48–1.23), and for each week of combination ART exposure, the risk of spontaneous preterm birth reportedly decreased by 2%. Notably, this study did not compare its subjects to an HIV-negative population, but rather compared the data on pregnant women living with HIV with the general population. A recent 2021 prospective study of 488 singleton deliveries over a similar time period (1995–2018) also found a decreased risk of preterm birth among women with HIV receiving combination ART than among those receiving no ART (7% vs. 26%, adjusted RR 0.19, 95% CI 0.08–0.44)[31]. While this study may have been limited by small sample size as well as self-report of pregnancy outcomes, ART exposure, and adherence, it suggested the possibility that ART may prevent preterm birth in a longitudinal prospective US cohort.

In summary, the data is conflicting regarding the association between combination ART and preterm birth risk in pregnant women living with HIV (Table 1). Analyses exploring the role of timing of ART initiation or duration of ART in relation to preterm birth risk are prone to selection bias because pregnancies delivered preterm have less opportunity to start ART in the later weeks of pregnancy. Careful monitoring of this population for preterm birth should be a priority while maintaining ART for suppression of viral load per current guidelines.

Table 1:

Association between antiretroviral therapy and preterm birth

Study	Type of study	Antiretroviral therapy	Sample size	PTB Rate (<37 weeks)	ARR/AHR/AOR of PTB on ART [95% CI]	Limitations
Venkatesh et al, 2021 [31]	Prospective cohort	Monotherapy cART No ART	37 – monotherapy 188 – cART 47 – no ART	24.3% 29.7% 48.9%	ARR 0.19 [0.08–0.44] of cART vs. no ART	Small sample size; ART exposure based on patient self-report and did not include measures of medication adherence; unable to classify the phenotype of PTB (spontaneous vs. medically indicated).
Albert et al, 2020 [6]	Cohort	cART	516 – all combinations of ART exposure during pregnancy	20.3%	HR 0.98 [CI 0.96–0.99] for cART exposure in pregnancy vs. no exposure HR 0.77 [0.48–1.23] for exposure to ART in the first trimester vs. no exposure	Notably, this study did not compare its subjects to an HIV-negative population, but rather compared the data on pregnant women living with HIV with the general population.
EPPICC Study Group, 2019 [17]	Retrospective cohort	ZDV/3TC TDF/XTC ABC/3TC LPV/r Other PI	5122 – ZDV-3TC 1122 – TDF-XTC 711 – ABC-3TC 5558 – LPV/r 934 – other PI	10.1% 10.5% 9.3% 10.3% 8.9%	AORs relative to ZDV-3TC: 0.97 [0.73–1.28] for ABC-3TC 1.06 [0.83–1.35] for TDF-XTC AOR relative to other PI: 1.17 [0.75–1.82] for LPV/r	Exclusion of 148 pregnancies with <2 weeks of ART and of 495 pregnancies with ART switch may have introduced selection bias; study dataset dominated by the UK/ Ireland and Ukraine; did not have information available on important confounders.
Venkatesh et al, 2019 [22]	RCT (PROMISE Trial)	ZDV monotherapy ZDV-based ART TDF-based ART	1467 – ZDV alone 1454 – ZDV ART 412 – TDF ART	13.0% 19.9% 18.9%	AORs relative to ZDV alone: 1.68 [1.10–2.57] for ZDV ART 2.71 [1.39–5.29] for TDF ART	Secondary analysis; unable to differentiate spontaneous vs medically indicated PTB
Fowler et al, 2016 [21]	RCT (PROMISE Trial)	ZDV monotherapy LPV/r/ZDV/3TC LPV/r/TDF/FTC	1411 – ZDV alone 1406 – ZDV ART 335 – TDF ART	13.1% 20.5% 18.5%	ZDV ART relative to ZDV alone: P<0.001 TDF ART relative to ZDV alone: P=0.10	TDV-based ART assignment did not commence until October 2012 (“Period 2”)
Sibiude et al, 2012 [20]	Prospective cohort	PI/r Non-boosted PI	1066 – PI/r 187 – non-boosted PI	14.4% 9.1%	AOR 1.69 [1.38–2.07] of cART vs. ZDV alone AHR 2.03 [1.06–3.89] of PI/r vs. non-boosted PI	France study dataset; much of the increase in premature birth occurred during 1997–1999, a period during with boosted PI therapy was not yet predominant; possibility that it was the efficacy rather than type of ART correlating with PTB
Chen et al, 2012 [24]	Retrospective cohort	NVP/ZDV/3TC LPV/r/ZDV/3TC ZDV monotherapy	2851 – NVP-ZDV-3TC 312 – LPV/r/ZDV/3TC 3762 – ZDV alone	24.7% – all cART 14.2% – ZDV	AOR 1.2 [1.1–1.4] for cART initiated prior to pregnancy vs. all other women with HIV AOR 1.4 [1.2–1.8] for cART initiated in pregnancy vs. initiating ZDV alone in pregnancy	Botswana study dataset; did not distinguish NRTI vs PI group preterm birth data; some variables in obstetric records self-reported; retrospective with possibility of confounding; women receiving cART were sicker.
Powis et al, 2011 [19]	RCT (Mma Bana Study)	ABC/ZDV/3TC (NRTI group) LPV/r/ZDV/3TC (PI group)	263 – NRTI group 267 – PI group	11.8% 21.4%	AOR 2.02 [1.25–3.27] for PI group vs. NRTI group	Botswana study dataset may not be generalizable.
Kourtis et al, 2007 [30]	Meta-analysis of 14 studies (n=11,224)				OR 1.01 [0.76–1.34] for ART exposure overall during pregnancy vs. no exposure OR 1.24 [0.76–2.02] for regimens containing PI vs. regimens without PI OR 1.71 [CI 1.09–2.67] for cART initiated before pregnancy or in the first trimester vs. therapy initiated in the second trimester and beyond	Many of the analyzed studies had relatively small numbers of subjects, a retrospective design, and/or a lack of control over other factors that could influence the rates of adverse pregnancy outcomes such as previous PTB, smoking, recreational drug use, other STIs, and advanced disease with immunosuppression.
Cotter et al, 2006 [15]	Prospective cohort	Monotherapy cART without PI cART with PI	492 – monotherapy 373 – cART without PI 134 – cART with PI 338 – no therapy	23.9% 29.6% 36.6% 26.2%	AOR 1.8 [1.1–3.0] for cART PI vs. any other therapy combination	Did not report or distinguish gestational age of ART initiation.

Abbreviations: PTB – preterm birth; ARR – adjusted relative risk; AHR – adjusted hazard ratio; AOR – adjusted odds ratio; ART – antiretroviral therapy; 95% CI – 95% confidence interval; cART – combination antiretroviral therapy; ZDV – zidovudine; 3TC – lamivudine; TDF – tenofovir disoproxil; XTC – emtricitabine; ABC – abacavir; LPV/r – lopinavir/ritonavir; PI – protease inhibitor; RCT – randomized controlled trial; PI/r – ritonavir-boosted protease inhibitor; NVP – nevirapine; NRTI – nucleoside/nucleotide reverse transcriptase inhibitors

3.2. Cervical length in the prediction of preterm birth

Among predictors of preterm birth, a history of prior preterm birth and a short cervix have been described as the two most powerful predictors[32]. Although the cervix is measured at various times in the first and early second trimester, cervical length measurement is best correlated with predicting preterm birth between 18–24 weeks gestational age[33–35]. Cervical length measurements in the first trimester of pregnancy, at the time of nuchal translucency examination, or early in the mid trimester, are each less accurate for predicting preterm birth. The mean cervical length in the mid trimester of pregnancy is approximately 35 mm (range 27–43 mm)[36,37]. The risk of spontaneous preterm birth is directly proportional to decreasing cervical length and increasing gestational age[36].

3.3. Cervical length assessment and preterm birth risk in pregnant women living with HIV on ART

In the studies from which reference ranges of cervical lengths were determined, pregnant women living with HIV were rarely included because these women are at increased risk for preterm birth. Although a short cervix in the mid-trimester is a powerful predictor of preterm birth as previously discussed, mid-trimester cervical length shortening was not found to be a predictor of preterm birth in a cohort of pregnant women living with HIV on ART in Botswana[38]. Similar findings were reported in the Zambian Preterm Birth Prevention Study (ZAPPS), a prospective cohort study of pregnant women living with HIV on ART in which there was no evidence that HIV infection was associated with a short cervix[39]. Due to a limited number of pregnant women living with HIV on ART in the Botswana study, and unavailability of CD4 counts and HIV-RNA viral load data for some pregnant women living with HIV in previous studies, it was difficult to explore differences in cervical length based on type of ART regimen, timing of ART initiation during pregnancy (first, second or third trimester of pregnancy), CD4 counts (<200, 200–350, 350–1000, >1000), and HIV-RNA viral load during pregnancy[38]. More studies are needed to determine the association between cervical length and preterm birth risk in pregnant women with HIV on ART using multivariable logistic regression models to estimate the association between cervical length and the risk of preterm birth. It will also be important to assess for effect-modification by type of ART, timing of ART initiation, CD4 count, and HIV-RNA viral load. These cervical length studies should exclude women who have undergone loop electrosurgical excision procedures (LEEP), which can present as a short cervix, and are prone to cervical insufficiency[40], which further increases the risk of developing preterm birth in women living with HIV.

3.4. Vaginal microbiome in women living with HIV on ART and preterm birth risk.

As first reported by Ravel et al, the prevalence of vaginal microbiomes varies across populations[41]. In asymptomatic, otherwise healthy women, several community state types of vaginal microbiota exist. The majority of vaginal microbiota are often dominated by species of Lactobacillus, but others are composed of a diverse array of anaerobic microorganisms[42]. The predominant vaginal microbiome has been noted to differ among women who tested positive for human papillomavirus (HPV) relative to HPV-negative women, with variation in terms of several metabolites, including biogenic amines, glutathione, and lipid-related metabolites[43]. Emerging research has found that pregnancy alters the vaginal microbiome profile. Placental metabolism during pregnancy leads to a marked increase in circulating estrogen concentrations, with a consequent reduction in vaginal bacterial diversity and promotion of Lactobacillus species dominance[44]. There is increasing evidence that some vaginal microbiomes are associated with an increased risk for spontaneous preterm birth, and the prevalence of these microbiomes varies across populations[45]. To understand the underlying mechanism of preterm birth, Kindinger and colleagues studied the interaction between vaginal microbiome, cervical length, and preterm birth risk, and demonstrated that dominance of Lactobacillus in the vaginal microbiome at 16 weeks of gestation was a risk factor for preterm birth[46]. Studies have associated Lactobacillus crispatus with a lower risk of preterm birth in populations predominantly of European descent[14], and this finding has been replicated in Africans[47].

HIV infection has been demonstrated to be a risk factor for vaginal dysbiosis during pregnancy[48]. As reported by Short et al in 2020, pregnant women living with HIV on ART had a lower prevalence of L. crispatus-dominant vaginal microbiota group (15% vs 54%) than HIV uninfected pregnant women, and higher prevalence of two L. iners-dominant groups (36% vs 9% and 15% vs 5%) and mixed anaerobes (21% vs 0%)[49]. Across the second and third trimesters in pregnant women living with HIV, the L. iners-dominant and mixed anaerobe vaginal microbiota groups were associated with preterm birth and with increased local inflammation, as measured by cervicovaginal fluid cytokine concentrations. In another vaginal microbiome study in pregnant women living with HIV by Gudza-Mugabe et al, HIV infection was independently associated with diverse vaginal community state types (P = 0.005) as well as preterm birth (P = 0.049) in multivariable analyses[50]. At this time, whether interventions designed to favorably alter the vaginal microbiome might modify the risk of preterm birth has not been studied in pregnant women living with HIV on ART. Understanding the potential of ART to influence the vaginal microbiome is another important gap in our current understanding for the population of pregnant women living with HIV.

3.5. Cervico-vaginal biomarkers, cervical length, and preterm birth risk in women living with HIV on ART.

Previous studies have demonstrated that a number of vaginal biomarkers, including inducible heat shock protein Hsp70, tissue inhibitor of metalloproteinases (TIMP-1 and TIMP-2), matrix metalloproteinases (MMP-2 and MMP-8), sequestosome-1, and a2 isoform of vacuolar ATPase, predict cervical length[51]. The authors concluded that measuring TIMP-1 and lactic acid concentrations in vaginal secretions could predict microbiome composition and a short cervix. Cervicovaginal fetal fibronectin (fFN), a cervical biomarker, is useful for predicting preterm birth in asymptomatic high-risk women, with fFN values ≥50 ng/mL at 22 to 28 weeks of gestation having a sensitivity of 55% and positive predictive value of 27% for prediction of preterm birth <34 weeks[52]. A RCT published in 2007 compared second trimester and intrapartum antibiotics with placebo to prevent chorioamnionitis and reduce preterm birth in women living with HIV on ART, and they reported that fetal fibronectin concentrations at 28 (but not at 20–24 weeks) were associated with increased risk of preterm birth, and that high levels of fetal fibronectin were positively associated with perinatal transmission of HIV at 20–24 weeks but not at 28 weeks[53].

Prospective cohort studies have demonstrated that presence of cervical placental alpha-microglobulin-1 (PAMG-1) is associated with an increased likelihood of delivery within 7 days[54–56]. A recent meta-analysis of 17 studies involving 2590 women demonstrated that cervical PAMG-1 has high accuracy in predicting preterm birth within 7–14 days of testing in symptomatic pregnant women (64.4% sensitivity, 96.9% specificity, 0.97 SROC)[57]. When compared prospectively in one study to fetal fibronectin for prediction of preterm birth, PAMG-1 performed similarly at ruling out preterm birth, but was superior at ruling in preterm birth, with positive predictive values of 23.1% for PAMG-1 vs. 4.3% for fFN, and negative predictive values of 99.5% vs 99.6% respectively[54]. When compared prospectively in one study to cervical length <25 mm and number of contractions for prediction of preterm birth, PAMG-1 showed a higher accuracy rate for prediction of delivery within ≤48 hours and ≤7 days, with a positive predictive value for deliveries within ≤48 hours and ≤7 days greater for PAMG-1 than the positive predictive value for cervical length of <25 mm and number of contractions, and no significant difference in negative predictive value[55]. For deliveries before 37 weeks, the PAMG-1 test showed higher positive likelihood ratio than CL <25 mm, contractions of 12–17/hour, and contractions of ≥18/hour (10.24, 2.01, 1.30, and 5.12, respectively). Another study has also demonstrated superiority in predictive accuracy of PAMG-1 versus CL cutoffs of 20 mm, 15 mm, and 10 mm[58]. To date, the efficacy of PAMG-1 for predicting preterm birth has not yet been assessed in the population of pregnant women living with HIV on ART.

Other biomarkers which have been studied regarding prediction of preterm birth include two serum proteins, insulin-like growth factor-binding protein 4 (IGFBP-4) and sex hormone-binding globulin (SHBG), which are used in a test that is available for clinical use to predict preterm birth. The test was reported to predict preterm birth with a sensitivity of 75% and specificity of 74%[59]. Additionally, insulin-like growth factor-binding protein-1/alpha fetoprotein have been used in a test for diagnosis of prelabor rupture of membranes, with sensitivity and specificity (97.9% and 97.6% respectively) that exceeded that of traditional clinical assessment via nitrazine, ferning and pooling as well as higher than that of PAMG-1[60]. A systematic review of 72 observational studies of over 30 biomarkers, including inflammation-related biomarkers, placental protein/hormone-related biomarkers, angiogenesis-related biomarkers, coagulation-related biomarkers, genetic-biomarkers, and proteomic-related biomarkers, among nearly 90,000 women who were asymptomatic and at high-risk for preterm birth, concluded that none of these other biomarkers was clinically useful for predicting spontaneous preterm birth in asymptomatic women[61].

3.6. Artificial Intelligence (machine learning) for prediction of preterm birth

Artificial intelligence is being studied extensively for its potential to improve patient outcomes and reduce healthcare costs and workload for clinicians. Current uses of artificial intelligence and machine learning in obstetrics include as an adjunct to cardiotocography in predicting preterm labor; to create algorithms identifying asymptomatic women with short cervical length who are at risk of preterm birth; and potentially, using vast data capacity, to determine risk factors for preterm labor using multiomics and extensive genomic data[62,63]. Woolery and Grzymala reported in 1994 that machine learning had 53–88% accuracy in predicting preterm birth[64]. Different machine learning algorithms which have been studied, including support vector machine (SVM), decision tree, random forest, k-nearest neighbors (KNN), logistic regression, stochastic gradient boosting, and neural networks, and have been shown to predict preterm birth with a reasonable accuracy[65].

Recently, Raja et al designed a novel machine learning model termed a risk prediction conceptual model (RPCM) for the prediction of preterm birth in rural India, and reported performance accuracy of 90.9% with the support vector machine classifier[66]. The artificial neural network machine learning method has been described as comparable or superior to traditional methods such as logistic regression regarding the prediction of preterm birth[67,68]. In another recent retrospective cohort study of nulliparous women (n=112,963) with a singleton gestation in Ontario, Arabi Belaghi et al reported maximum area under the receiver operating characteristic curve (AUC) using artificial neural networks (AUC 65%, 95% CI 63–66%), with moderate prediction power in the second trimester by both logistic regression and machine learning approaches[69]. To date, there are no studies into the use of machine learning and other artificial intelligence methods for the prediction of preterm birth in women living with HIV on ART. Further investigation into this association is needed.

3.7. Preterm birth and social determinants of health in women living with HIV on ART.

The purpose of this review article is to focus specifically on objective risk factors which may be used clinically to assess the risk of preterm birth for patients living with HIV. However, given that socioeconomic disparities are known to be associated with morbidity with HIV[70], a discussion of preterm birth risk in patients living with HIV would be incomplete without addressing social determinants of health. Several studies investigating the role of social determinants of health, such as socioeconomic position, education, occupation, and income, in the development of preterm delivery have revealed that the lower the socioeconomic condition, the higher the rate of preterm delivery in pregnant women living with HIV on ART[71,72]. This association may be due in part to the decreased prenatal care, insufficient nutrition, anemia, drug abuse, and other maternal conditions that occur with increased incidence under poor socioeconomic conditions. Preterm delivery in pregnant women living with HIV on ART has also been significantly associated with inappropriate housing, low maternal education, and income level[73,74].

4. PREVENTION OF PRETERM BIRTH IN PREGNANT WOMEN LIVING WITH HIV ON ART:

Interventions used to reduce the risk of preterm birth include vaginal progesterone supplementation, 17-hydroxyprogesterone caproate (17-OHPC) supplementation, cerclage placement, or a combination of these treatment modalities. These interventions are discussed in further detail below, and their applicability for pregnant women living with HIV is examined.

4.1. Vaginal progesterone for preventing preterm birth in women living with HIV on ART

Progesterone is a key hormone involved in pregnancy maintenance. Several large RCTs and meta-analyses have demonstrated the efficacy of vaginal progesterone for the prevention of preterm birth and neonatal outcomes[75–77]. In a NEJM RCT published in 2007 of patients recruited from UK, Brazil, Greece, and Chile, Fonseca et al demonstrated that treatment with progesterone reduced the rate of spontaneous early preterm delivery[75]. In the PREGNANT trial, a phase III, multicenter RCT across 44 centers worldwide funded by the NICHD published in 2011, the administration of vaginal progesterone gel to women with a sonographic short cervix in the mid-trimester was found to be associated with a 45% reduction in the rate of preterm birth <33 weeks of gestation (the primary outcome) with improved neonatal outcome[76]. An individual patient meta-analysis including five RCTs by Romero et al in 2018 further demonstrated that vaginal progesterone administration was associated with a significant reduction in preterm birth at <33, <37, and <28 weeks of gestation, as well as with improved perinatal outcomes in pregnancies with a mid-trimester sonographic short cervix[77]. Most recently, the EPPPIC Group meta-analysis of individual participant data from 30 RCTs evaluating progestins for preventing preterm birth (n>11,600 participants), vaginal progesterone (RR 0.78, 95% CI 0.68–0.90), and 17-OHPC (RR 0.83, 95% CI 0.68–1.01) were both found to reduce preterm birth <34 weeks of gestation in high-risk singleton pregnancies with either a previous spontaneous preterm birth or cervical shortening (≤25 mm) in the current pregnancy[10].

Based on these data, as well as the increased risk of preterm birth in women living with HIV, vaginal progesterone has been studied as a therapeutic option to prevent recurrent preterm birth in pregnant women living with HIV on ART. In a pilot feasibility study, Price et al demonstrated successful uptake, adherence, and retention for a trial studying vaginal progesterone in pregnant women living with HIV on ART for the prevention of recurrent preterm birth[78]. Thus, a phase III RCT of vaginal progesterone for the prevention of preterm birth in pregnant women living with HIV is feasible, but while a protocol for this study has been published[79], no phase III/IV efficacy studies of vaginal progesterone have yet been conducted in pregnant women living with HIV on ART. Similarly, because ARTs have been conflictingly associated with preterm birth, Siou et al designed a multi-center RCT to study progesterone supplementation for pregnant women living with HIV on protease inhibitor-based regimens (the ProSPAR study)[80]. The group hypothesizes that pregnant women living with HIV on protease inhibitor regimens may benefit from vaginal progesterone supplementation, as studies have also suggested that protease inhibitor use during pregnancy is associated with a decline in progesterone levels[11]. This trial is not yet recruiting patients (ClinicalTrials.gov Identifier: NCT02400021).

4.2. 17-alpha progesterone caproate (17OHPC) for preventing preterm birth in women living with HIV on ART.

The use of 17-alpha hydroxyprogesterone caproate (17OHPC) for the management and prevention of preterm birth in pregnant women has been studied in two large multicenter RCTs, which have shown conflicting results. In pregnant women living with HIV on ART, the Improving Pregnancy Outcomes with Progesterone (IPOP), a phase III RCT of 17OHPC in pregnant women with history of recurrent preterm birth and HIV, yielded negative results regarding prevention of preterm birth[81]. In this RCT in pregnant women living with HIV on ART in Zambia by Price et al, pregnant women living with HIV at <24 weeks of gestation were randomly assigned to receive weekly antenatal 250 mg intramuscular 17OHPC or placebo[82]. Incidence of preterm delivery or stillbirth was similar among patients assigned to 17OHPC vs placebo (9% vs. 9%, relative risk 1.0, 95% CI 0.6–1.6; P=0.98)[82]. 17OHPC injections given to women in Zambia with HIV and no history of spontaneous preterm birth were not found to reduce the composite risk of preterm birth or stillbirth compared with placebo[83]. Thus, this evidence indicates that 17OHPC is unlikely to reduce spontaneous preterm birth or stillbirth among women whose risk derives solely from HIV infection.

4.3. Cervical cerclage and pessary for preventing preterm birth.

In women with previous history of spontaneous preterm birth, singleton gestation, and cervical length <25 mm, cerclage has been found to significantly prevent preterm birth and perinatal mortality and morbidity, and is recommended in clinical practice[84,85]. A 2017 meta-analysis of nine RCTs of cerclage versus no cerclage in singleton pregnancies (n=2415 women) reported that placement of a cerclage reduced preterm birth <34 weeks (RR 0.77, 95% CI 0.66–0.89) as well as preterm birth <37 weeks (RR 0.80, 95% CI 0.69–0.95)[86]. Women living with HIV are eligible for cerclage if they meet the general criteria for optimal candidacy; there is no relative contraindication.

Placement of a cervical pessary to prevent cervical shortening is not currently recommended for pregnant women with a short cervix[87]. Multiple RCTs have been published in the last several years, with conflicting results. As reported by Goya et al in 2012, in a RCT of cervical pessary use in pregnant women with a cervical length of 25 mm or less, spontaneous delivery before 34 weeks gestation was significantly less frequent in the pessary group than in the expectant management group[88]. However, as reported by Nicholaides et al in 2016, in a RCT of cervical pessary use in pregnant women with a cervical length of 25 mm or less, cervical pessary did not result in a lower rate of spontaneous early preterm delivery than the rate with expectant management[89]. In meta-analyses of RCTs comparing use of a cervical pessary with standard care in asymptomatic patients at high risk for preterm birth (CL≤25 mm), pessary use did not reduce spontaneous preterm birth <28, 34, or 37 weeks or improve perinatal outcome[90,91]. However, these data cannot be extrapolated to women living with HIV, and the role of pessary for prevention of preterm birth in this population individually has not been studied.

5. CONCLUSION

The rate of spontaneous preterm birth among pregnant women living with HIV on ART is 3–4-fold higher when compared to pregnant HIV-negative women. The literature contains conflicting evidence regarding whether exposure to combination ART may increase risk of preterm birth. Other factors which may enable prediction of preterm birth risk in women living with HIV include cervical length measurements, the composition of the vaginal microbiome, cervico-vaginal biomarkers, artificial intelligence (machine learning), and social determinants of health. The efficacy of vaginal progesterone supplementation for the prevention of preterm birth and adverse neonatal outcomes in at-risk women is well-established, and deserves further investigation as a therapeutic option to prevent recurrent preterm birth in pregnant women living with HIV on ART, particularly investigation of novel proposed delivery formulations such as nanosuspension. The evidence to date indicates that intramuscular 17-alpha hydroxyprogesterone caproate is unlikely to reduce spontaneous preterm birth in pregnant women living with HIV on ART. Cervical cerclage is an option for reducing the risk of preterm birth in women living with HIV with a short cervix. A feasibility study of vaginal progesterone for preterm birth prevention in women living with HIV on ART was successful, and the ProSPAR study, a multi-center RCT studying progesterone supplementation for pregnant women living with HIV on protease-inhibitor based regimens, has been designed but is not yet recruiting patients. We need to continue to facilitate pharmacologic research in pregnant women to help in answering critical questions about the effects of antiretroviral medications in pregnant women living with HIV [92–100].

6. EXPERT OPINION

Pregnant women living with HIV have historically been rarely included in preterm birth trials because of the role of HIV itself as a risk factor for preterm birth, and therefore several research gaps exist in the prediction of preterm birth in women living with HIV on ART (Table 2). Prediction of preterm birth in women living with HIV currently relies on clinical metrics such as cervical length measurement and understanding of maternal risk factors such as prior history of preterm birth. Cervical length screening is not currently routinely performed for all pregnancies; however, due to the potentially increased risk of preterm birth in pregnant women living with HIV on ART, a policy of universal cervical length screening in these women might be a reasonable recommendation. HIV infection has been demonstrated to be independently associated with diverse vaginal microbiome community state types that are associated with increased risk of preterm birth.⁵⁵ Thus, whether characterization of patients’ vaginal microbiome state types might serve clinically as a predictive factor for preterm birth warrants further study. Further investigation is also necessary into whether interventions that favorably alter the vaginal microbiome could modify the risk of preterm birth. The predictive value of cervico-vaginal biomarkers such as fFN and PAMG-1, which have been associated with increased risk of preterm birth in the general population, also deserves further investigation regarding potential for routine screening of preterm birth risk in pregnant women living with HIV on ART.

Table 2:

Predictors of spontaneous preterm birth and research gaps in pregnant women living with HIV.

PTB Prediction Method	Research Gap(s) in pregnant women living with HIV
Biophysical methods
Transvaginal cervical length	Due to a limited number of women living with HIV, and unavailability of CD4 and HIV RNA data for some participants in previous trials, it is difficult to explore differences in cervical length based on ART regimen, timing of initiation of ART in pregnancy, CD4 cell count, or HIV RNA.
Biochemical methods
Vaginal microbiome	HIV infection has been demonstrated to be independently associated with diverse vaginal microbiome community state types that are associated with increased risk of preterm birth. Whether interventions designed to favorably alter the vaginal microbiome might modify the risk of preterm birth in pregnant women living with HIV has not been studied. Understanding the potential of antiretroviral therapies to influence the vaginal microbiome is another important gap in our current understanding.
Vaginal biomarkers	The efficacy of biomarkers such as fFN, PAMG-1, and insulin-like growth factor-binding protein for prediction of preterm birth or prelabor rupture of membranes has not been assessed in the population of pregnant women living with HIV.
Mathematical models
Machine learning and artificial intelligence	No current studies exist in women living with HIV on ART.

Abbreviations: PTB – preterm birth; ART – antiretroviral therapy; fFN – fetal fibronectin; PAMG-1 – placental alpha microglobulin-1

There is conflicting data regarding the association between combination ART and the incidence of preterm birth in pregnant women living with HIV on ART. Analyses exploring the role of timing of ART initiation or duration of ART in relation to preterm birth risk are prone to selection bias because pregnancies delivered preterm have by definition less opportunity to start ART in the later weeks of gestation. Combination ART, particularly protease inhibitors, has been shown in multiple RCTs to be associated with significantly higher odds of preterm birth than with monotherapy, mainly seen with ART initiation in the second or third trimester of pregnancy versus women who initiated ART before 24–28 weeks[20,23]. However, a large meta-analysis found that ART exposure during pregnancy was not associated with an increased risk of preterm birth and did not find a significantly increased risk of preterm birth with regimens containing protease inhibitors, and reported increased odds of preterm birth with ART initiation earlier in pregnancy[29]. Regardless, the beneficial effects of ART for preventing perinatal transmission are indisputable, and therefore careful monitoring of this population of patients for preterm birth should be a priority while maintaining ART for suppression of viral load per current guidelines. Further investigation into the use of artificial intelligence (machine learning) algorithms for predicting preterm birth in women living with HIV on ART may help to further clarify these data.

Research gaps also exist in the prevention of preterm birth in pregnant women living with HIV on ART (Table 3). The efficacy of vaginal progesterone supplementation for the prevention of preterm birth and adverse neonatal outcomes in women with a sonographic short cervix is well-established. Among women living with HIV, studies have shown that use of ART during pregnancy can reduce progesterone serum concentrations, with subsequent progression to preterm birth and other adverse pregnancy outcomes[11]. Therefore, progesterone supplementation deserves further investigation as a therapeutic option for preterm birth risk reduction in women without a shortened cervix whose preterm birth risk derives solely from living with HIV. A feasibility study to this end was successful, and the ProSPAR study, a multi-center RCT studying progesterone supplementation for pregnant women living with HIV on protease-inhibitor based regimens, has been designed but is not yet recruiting patients[80]. Recent data indicate that intramuscular 17-alpha hydroxyprogesterone caproate is unlikely to reduce spontaneous preterm birth or stillbirth among women whose risk derives solely from HIV infection[82]. Cervical cerclage is an option for reducing risk of preterm birth in women living with HIV with shortened cervical length; cervical pessary has not been studied for prevention of preterm birth in the population of women living with HIV.

Table 3:

Prevention of spontaneous preterm birth and research gaps in pregnant women living with HIV.

PTB Prevention Method	Research Gap(s) in pregnant women living with HIV
Progesterone supplementation
Vaginal progesterone	A feasibility study of vaginal progesterone supplementation as a therapeutic option for preterm birth risk reduction in pregnant women living with HIV was successful, and the ProSPAR study, a multi-center randomized controlled pilot trial studying progesterone supplementation for pregnant women living with HIV on protease inhibitor-based regimens, has been designed but is not yet recruiting patients.
Intramuscular 17-alpha hydroxyprogesterone caproate	Data from Zambia indicates that intramuscular 17-alpha hydroxyprogesterone caproate is unlikely to reduce spontaneous preterm birth or stillbirth among women whose risk derives solely from HIV infection.
Cervical support methods
Cervical cerclage	Women living with HIV are eligible for cerclage if they meet the general criteria for optimal candidacy, including shortened cervical length.
Cervical pessary	The role of pessary for prevention of preterm birth in pregnant women living with HIV has not been studied.

Abbreviations: PTB – preterm birth

ARTICLE HIGHLIGHTS.

• Preterm birth continues to be a major cause of perinatal morbidity and mortality, especially in women living with HIV. The rate of spontaneous preterm birth among pregnant women living with HIV is three to four-fold higher when compared to pregnant women without HIV.

• For patients living with HIV, a unique set of risk factors are relevant for prediction of preterm birth, including ART exposure, cervical length, cervico-vaginal biomarkers, vaginal microbiota, and imaging modalities.

• There is conflicting data regarding the association between combination ART and preterm birth risk in pregnant women living with HIV. Pertinently, analyses exploring the role of timing of ART initiation or duration of ART in relation to preterm birth risk are prone to selection bias because pregnancies delivered preterm have less opportunity to start ART in the later weeks of pregnancy. Careful monitoring of this population for preterm birth should be a priority while maintaining ART for suppression of viral load per current guidelines.

• Several large randomized clinical trials and meta-analyses have demonstrated the efficacy of vaginal progesterone for the prevention of preterm birth and neonatal outcomes. A multi-center randomized controlled pilot trial has been designed studying progesterone supplementation for pregnant women living with HIV on protease inhibitor-based regimens (the ProSPAR study). The group hypothesizes that pregnant women living with HIV on protease inhibitor regimens may benefit from vaginal progesterone supplementation, as studies have also suggested that protease inhibitor use during pregnancy is associated with a decline in progesterone levels. This trial is not yet recruiting patients (ClinicalTrials.gov Identifier: NCT02400021).

• The evidence indicates that 17-alpha progesterone caproate is unlikely to reduce spontaneous preterm birth or stillbirth among women whose risk derives solely from HIV infection. Cervical cerclage is an option for reducing risk of preterm birth in women living with HIV with shortened cervical length. Cervical pessary has not been studied for prevention of preterm birth in the population of women living with HIV.