Setting

Between April 2015 and October 2016, we enrolled pregnant women (aged ≥18 years) at their first antenatal care (ANC) visit in a prospective cohort study, at a single large public sector primary care facility (Gugulethu Midwife Obstetric Unit (MOU)) in a low-income high HIV-prevalence subdistrict of Cape Town, South Africa.

Study design

The overall prospective, observational design includes two ‘nested’ groups of pregnant women:

• Group 1: the overall population of pregnant women (≥18 years) seeking ANC services at Gugulethu MOU during an 18-month period; within this group, a subset of women thought to be ≤24 weeks’ gestation based on history or examination (clinical GA) underwent US scan by a research sonographer for more accurate gestation estimation; enrolled into observation group.

• Group 2: all pregnant WLHIV seeking ANC who are ≤24 weeks’ gestation at US at their first ANC visit, regardless of current ART use at the first ANC visit (nested within group 1). Enrolled into longitudinal cohort with data collection through questionnaires, clinical assessments and phlebotomy spanning pregnancy to early infancy.

  This study design enables quantification of the risk of adverse birth outcomes in the overall cohort, as well as the more detailed group 2 also enabling investigation of the consequences of the immune response following ART initiation in pregnancy for onset of labour and PTD.

Patient public involvement

No patient involvement.

Routine care services

As part of routine ANC services, GA was estimated based on date of last menstrual period (LMP) and symphysis-fundal height (SFH) by public sector midwives. All women without a previous HIV diagnosis underwent HIV testing, with universal ART eligibility. WLHIV conceiving while on ART continued their current regimen throughout pregnancy; regimens included non-nucleoside analogue reverse transcriptase inhibitors (NNRTIs) such as efavirenz (EFV) or protease inhibitor (PI, predominantly used after failure of first-line therapy). For women initiating ART in pregnancy, a fixed-dose combination of tenofovir (TDF)+emtricitabine (FTC)+EFV was used throughout.

Recruitment

Following screening of all women attending their first ANC visit, those ≥18 years were eligible and approached to participate in the study. Following screening, ineligible women were referred back to their ANC clinics in line with the Western Cape Department of Health’s healthcare model. Women who agreed to participate had their routinely collected LMP-based GA and SFH-based GA reviewed by the counsellor; women estimated to be ≤24 weeks were referred for a research US scan for formal pregnancy dating by a research sonographer blinded to the midwife assessment. WLHIV who were ≤24 weeks’ gestation on US were then recruited into a nested cohort (group 2); half of these had initiated lifelong ART prior to conception and half initiated ART during the pregnancy.

All participants provided written informed consent prior to study participation, with reconsenting of mother–infant pairs at the first postpartum visit for paediatric follow-up. Consent for study participation included data abstraction from routine clinical records through the pregnancy and postpartum period.

Participant baseline characteristics

A total of 4431 women registered for ANC during the study recruitment period, of whom 4111 (93%) were screened for the study and 3972 (90%) enrolled; all delivered by May 2017 (figure 1). Main reasons for being screened out were under-age, referrals from Basic Antenatal Clinics or not being interested. Of the enrolled women, 2517 (63.4%) were HIV-negative and 1455 (36.6 %) WLHIV (table 1); 2199 (55.4%) were referred for US based on their clinical GA, 1327 (60.3%) were HIV-negative and 872 (39.7%) living with HIV.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Cohort profile. AGA, appropriate-for-gestational age; ART, antiretroviral therapy; GA, gestational age; PTD, preterm delivery; SGA, small-for-gestational age.

Median age at enrolment was 28 years (IQR 23–32), with HIV-negative women younger than WLHIV women, and those initiating ART preconception older than WLHIV initiating ART during the pregnancy. In line with age differences, WLHIV were of higher gravidity than HIV-negative women, but the difference in parity was small; women who initiated ART prepregnancy were of a higher gravidity than those initiating during the pregnancy. A quarter (25%) of all women were overweight, while over half (55%) were obese, with little or no difference between groups. Having previously had a PTD was more common among WLHIV than HIV-negative women. Mild anaemia was relatively common in all groups, especially in women initiating ART during the pregnancy (table 1). Overall, 3479 (87.6%) women had gestation estimated by LMP, 2327 (58.6%) by SFH and 2334 (58.8%) by US; with estimated median GA at enrolment visit varying by assessment method (table 1).

There were 1455 WLHIV in group 1; 718 (49.3%) were ≤24 weeks on US, of whom 551 were enrolled into group 2 (figure 1). The likelihood of inclusion into group 2 did not differ by baseline characteristics (table 2). In comparison to group 1 participants, group 2 participants differed in age, gravidity, parity and previous PTD, likely driven by the HIV-negative women (table 3). In multivariable regression allowing for HIV infection, the only difference that persisted between these groups was age (table 4). When group 2 women were compared with other WLHIV (not enrolled in group 2), they were slightly older, more likely to have normal haemoglobin levels (≥110 g/L) and lower GA (table 3). In multivariable regression the only difference between these groups that persisted was GA, which was a group 2 inclusion criterion (table 4).

Of the 551 WLHIV in group 2, 261 (47%) initiated before pregnancy and 290 (53%) initiated during the pregnancy (table 5). Women who initiated ART during the pregnancy were on average younger, and of lower gravidity. Overall, three-quarters of group 2 women were overweight or obese, with little difference by ART status. Of the women who initiated ART during the pregnancy, the majority (64%) had tested HIV positive in the index pregnancy; the rest had previously tested positive although were not on ART at conception. In line with local and WHO treatment guidelines, most women (91%) were on a regimen of two nucleoside reverse transcriptase inhibitors (NRTIs) (TDF +FTC), plus NNRTI EFV. PI usage in this cohort was low, at only 9% of women who initiated pre-pregnancy and 1% of women who initiated during the pregnancy. Median CD4 count was 433 cells/µL overall, 527 cells/µL in women who initiated ART before pregnancy and 373 cells/µL in women initiating at the first antenatal visit. There were no differences in smoking or drug usage between these two groups of women, but women who initiated during the pregnancy were more likely to have ever consumed alcohol or consumed in the last 30 days (table 5).

Study follow-up

At baseline, all women (group 1) had clinical and medical history, routine first ANC visit physical examination, screening tests and GA assessment data collected via abstraction of the Maternity Case Record (MCR) booklet (table 5), which is a standardised patient-held maternity record used by all facilities providing maternity services to record clinical data from the antenatal through to postpartum period, including labour. The MCR also serves as a referral letter, thus serving as a link between antenatal and labour care. In addition, the National Health Laboratory Services database was searched for CD4, viral load results and other laboratory values not recorded in the MCR. Further follow-up for women in group 1, not eligible for group 2, was through data abstraction of the MCR following discharge from the postnatal ward (MCR retained at delivery facility). Data were abstracted from follow-up ANC notes, clinical notes during labour and newborn assessments (table 5).

Women in group 2 participated in up to eight scheduled study visits, from the start of ANC through to 12 months post partum. Women on ART from before pregnancy had three antenatal visits at <24 weeks, 28 and 34 weeks; women who initiated ART during the pregnancy had an additional study visit 2 weeks after the ART initiation (which in most women took place on the same day, or close to, the first study visit). Following delivery, women were reconsented for infant participation, and study visits were conducted <7 days, 10 weeks, 6 months and 12 months post partum. At all study visits, data were collected on maternal health (HIV care and ART use, clinical care and intercurrent clinical history). Other procedures included physical examinations (anthropometry, blood pressure measurement), phlebotomy (50 mL) for storage of plasma, peripheral blood mononuclear cells (PBMC) and RNA; a follow-up US was conducted at 28 weeks to assess fetal growth (table 5).

At postpartum study visits, additional data were collected on infant health (including infant feeding and intercurrent clinical history) and physical examination of infants was conducted (basic anthropometry). At the 12-month visit, developmental assessment was conducted using the Ages and Stages questionnaire22—a general developmental screening tool testing five key areas: personal social, gross motor, fine motor, problem solving and communication skills. Infant specimen collection included Dried Blood Spot sampling and storage at 10 weeks study visit, and phlebotomy (2 mL) for measurement of immunological functioning and antibody responses to routine childhood immunisations (rotavirus and measles) at 12-month study visit. In addition, data on infant health status, including vaccinations, chemoprophylaxis use (including nevirapine and cotrimoxazole) and routine HIV PCR testing, was abstracted from Road-to-Health Booklets—patient-held booklet taken to all clinical and immunisation visits used to monitor infant growth and development until age 5 years (table 6).

Further active follow-up of the women and their infants will occur at school-going age and beyond to chart growth, morbidity and development, as well as changes in family circumstances. It is envisaged that subsequent to this, longer term follow-up will be passive through the use of routinely collected data. The Western Cape Provincial Department of Health’s public-sector patient administration systems all share unique health identifier23; data relating to participants health service contacts, health conditions and health outcomes for specific conditions will be obtained from the Provincial Health Data Centre, which consolidates person-level clinical data across government services.24

Data collection

An overview of the main included data collection instruments is presented in table 6, covering self-reported information on clinical history, ART use and adherence, and medical events; as well as information obtained from routinely collected data.

Specimen collection

To investigate the proposed hypothesis that immunological changes resulting from maternal ART exposure are associated with adverse birth outcomes, women enrolled into the follow-up cohort were intensively sampled, with repeated phlebotomy throughout pregnancy and the postpartum period for immunological investigations (table 7). Using samples from all antenatal plasma, inflammatory markers (C reactive protein, serum amyloid A and CCL10 (IP-10) in women are being measured. Further, following a nested case–control design in group 2 (n=90), investigations compare women who delivered preterm (PTD cases) or had SGA infants (SGA cases) and those from appropriate controls (term AGA) (matched for GA and ART status) (figure 1). Investigations include longitudinal quantification of plasma cytokine profiles, phenotypic and functional characterisation of regulatory T cells (Tregs), antigen-presenting cells and metabolites associated with mitochondrial functioning and lysophospholipids (figure 2). The combined studies of these immune parameters will inform understanding of ART use during the pregnancy on the areas of the immune system that have been shown to be critically involved in regulating maternal immune tolerance to the fetus and their associations with onset of labour and PTD.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Maternal and infant specimens. GA, gestational age; mTOR, mammalian target of rapamycin; PP, post partum.

At delivery, placentas and cord bloods were collected whenever possible, a scoring algorithm was developed which graded placentas and dictated specimen processing according to membrane completeness and time received in laboratory relative to delivery time (table 8). Using flow cytometry and tissue immunostaining techniques the following investigations will be conducted: examination of the effect of HIV infection/ART exposure on the phenotypic characteristics and functionality of placental macrophages (Hofbauer cells and decidual macrophages) at the maternal–fetal interface and placental Tregs and their association with adverse birth outcomes. Characterisation of cord blood Tregs and correlation of their frequency, function and phenotype with placental Tregs and birth outcomes (figure 2).

Study retention

Lost to follow-up was categorised based on the last visit before the woman was lost; in total 158 (29%) women were lost to follow-up (LTFU) (figure 3). Women lost before the postdelivery study visit (n=88), either experienced pregnancy losses (n=37), were no longer interested in participating (n=24) or relocated out of the study area (n=17). For women LTFU between delivery and the 6-month postpartum visit (n=24), reasons included relocation (n=8), not contactable (n=6), no longer interested in participating (n=5) and maternal/infant death (n=5) (figure 2). For women LTFU between 6 months and 12 months post partum (n=46), reasons included relocation (n=15), not contactable (n=26), no longer interested in participating (n=2) and maternal/infant death (n=3).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Loss to follow-up (LTFU) in group 2 cohort. ART, antiretroviral therapy.

There were no appreciable differences by ART status in women LTFU before postdelivery visit (RR 0.81, 95% CI 0.55 to 1.19). However, women who initiated ART before pregnancy were less likely to be LTFU between delivery and 6 months postpartum (RR0.44, 95% CI 0.19 to 1.04) and between 6 months and 12 months post partum (RR 0.58, 95% CI 0.33 to 1.02). No baseline characteristics were associated with LTFU.

GA assessment

In the overall cohort, 1787 women with live singleton births were included in the analysis of the association between HIV status and timing of ART initiation and PTD by GA assessment method used (LMP, measurement of SFH and US. Using US-GA, PTD risk was associated with maternal HIV infection and ART use, with WLHIV, on ART from before or early pregnancy, almost twice as likely to deliver preterm than HIV-negative women.25 A weaker association was observed when GA assessment was based on SFH; while with LMP-GA the difference by HIV status was minimal. We did not find any appreciable differences in the PTD risk for WLHIV by timing of ART initiation across all three assessment methods.25 Our findings (in both the overall cohort and in women with all three assessments) suggest that methods of GA assessment explain at least partially the heterogeneity of findings from previous studies on the association between ART use and adverse birth outcomes, suggesting that care should be taken when interpreting results from such studies.

Obesity

In the overall cohort, 2921 women with live singleton births were included in the analysis of the association between maternal body mass index and adverse birth outcomes. In a subset cohort the association between gestational weight gain (GWG) and adverse birth outcomes was examined. Maternal obesity was associated with increased likelihood of having high birthweight and large size for GA infants. In the subset cohort, GWG was associated with increased likelihood of spontaneous PTD and high birthweight infants.26 Obesity during the pregnancy is prevalent in this setting and appears associated with increased risk of adverse birth outcomes in both WLHIV and HIV-negative women.

Placental pathology

Preliminary analysis of placental histopathology from a subset of women enrolled in the prospective cohort showed significant associations between placental pathology and adverse birth outcomes: presence of focal infarction was associated with increased risk of LBW; the lower the weight of the basal plate weight lead to increased risk of LBW, PTD and SGA; and prolonged meconium exposure was associated with increased risk of SGA. These findings suggest that adverse birth outcomes are driven primarily by placental abnormalities which do not appear to be associated with the ART initiation timing.27 Immunofluorescence and immunohistochemistry staining were performed on these wax blocks to identify regulatory T cells along with macrophages. Further analysis is ongoing.

Within the placenta, investigation of the distribution of pro-inflammatory (M1) and anti-inflammatory (M2) placental macrophages at the maternal–fetal interface showed no differences in the tissue density of these macrophages within the decidual membranes and villous tissue according to timing of ART initiation. Data suggest that the Hofbauer cells (which are fetal macrophages) are not polarised into M1/M2 phenotypes but are rather ‘intermediate’ types.28

Strengths and limitations

Key strengths of the PIMS study include the recruitment of a large community-based cohort in an area of high HIV prevalence; the observational nature of the study provides good external validity of experiences of a public sector primary care population over pregnancy. A further strength lies in the use of a research sonographer for the GA assessment in women ≤24 weeks when US is highly reproducible and accurate (while routinely used clinical assessments are less reliable) which is particularly important when studying associations with adverse birth outcomes, as compromises in outcome ascertainment methods can affect the detection of the magnitude of associations.

Additionally, the maternal biological specimen from Group 2 at three or four times (depending on ART status group) throughout pregnancy and at delivery is an important strength because it enables immunological, metabolomic and placental investigations to inform understanding of mechanisms underlying adverse birth outcomes in in WLHIV. As pregnancy is a state of immunoregulation requiring tolerance of a semiallogeneic fetus, the assessment of placentas of enrolled women provides a unique opportunity to investigate the link between HIV, ART and adverse birth outcomes. Collection of infant specimens further strengthens the study as it is one of the first studies combining metabolomic and immunological assessments. This will provide an integrated model of the immune-metabolism association in HIV-exposed infants and the consequences of maternal metabolic dysregulation for the immune responses of the infant. Furthermore, the developmental assessments carried out in the infants provides the opportunity to consider the association between maternal immune function during the pregnancy and early childhood immunological and developmental outcomes.

PIMS has valuable subdesigns in addition to the observational study design with the overall cohort stratified by maternal HIV status, ART use, and related risk factors, with all details to be analysed by timing of ART initiation in line with the main PIMS objectives. One subdesign is the cohort study in women living with HIV, who have data collection through questionnaires, clinical assessments and phlebotomy spanning pregnancy to early infancy. This study design enables quantification of the risk of adverse birth outcomes in the overall cohort, as well as the more detailed group 2 group also enabling investigation of the consequences of the immune response following ART initiation in pregnancy for onset of labour and PTD. Another subdesign is the nested case–control study which will enable immunological investigations in women who did and did not delivery preterm/SGA infant. The ability to track patients using the Western Cape unique identifier across different health and laboratory services, enables the passive long term follow-up of the group 2 women and their infants; with available data including patient-level data (administrative, demographic and clinical), visit-level data (clinical observations and findings), laboratory tests and medication.

A limitation of the study is while HIV-negative women are included in group 1, providing a comparison group for birth outcome and various maternal characteristics, group 2 did not include HIV-negative or ART-unexposed comparator groups. As such the detailed immunological analyses over pregnancy are limited to WLHIV, with timing of ART initiation a main explanatory variable. Additionally, we do not have full detailed information on all group 1 women and were instead limited by routinely collected data in medical notes some information relating to maternal characteristics was based on self-report and thus subject to potential biases. In order to address this limitation, routinely collected data were also collected to confirm data on birth outcomes.

Findings to date

Using data collected from study participants living with HIV throughout pregnancy and first year of life, PIMS provides a valuable platform for answering a variety of research questions related to maternal and child health. In particular, PIMS well equipped to investigate temporal changes of immunology markers in women whose immune status is altered by HIV infection, and how ART initiated during the pregnancy affects immune responses. The relationship between these immunological changes with adverse birth outcomes as well as possible longer-term impact of exposure to ART in fetal and early life will be explored. Additionally, through use of the Western Cape Department of Health unique identifier further active and passive follow-up of mothers and their infants is planned at school-going age and beyond to chart growth, morbidity and development, as well as changes in family circumstances.

The PIMS investigators welcome new collaborations with other investigators interested in using study data and stored specimens. Interested investigators should contact M-LN (M.Newell@soton.ac.uk) and LM (Landon.Myer@uct.ac.za) to obtain additional information and discuss collaborative opportunities. Proposed projects and data analyses plans will be reviewed to ensure no overlap with planned projects and efficient use of data and specimens.