Introduction
Cytomegalovirus (CMV) is a common infection in humans. The typical disease course is subclinical in at least 90% of primary or non-primary infections in immunocompetent hosts. However, following infection, CMV persists indefinitely in human hosts [
1]. In developing countries, there is near-universal CMV sero-prevalence from infancy [
2]. On the other hand in developed countries CMV sero-prevalence increases with age, with infection rates of 36%, 50% and 91% among 6–11, 30- and 80-year-olds, respectively [
3].
Even in immunocompetent hosts the significance of CMV infection is increasingly implicated in comorbidities, since CMV seropositive individuals have systemic immune-dysregulation associated with increased risks of cardiovascular diseases and pneumonia [
4‐
6]. CMV infection is even more important in immunocompromised hosts such as individuals with human immunodeficiency virus (HIV) infection and graft recipients, in whom CMV can cause severe disease such as pneumonitis [
7] and meningitis [
8]. It remains an important co-infection in HIV-infected adults even among those on successful combination antiretroviral therapy (cART) with good viral control, and has been associated with increased morbidity and mortality, including the development of non-AIDS-defining illnesses (9;10). CMV reactivation or disease may be under-reported in high CMV/HIV sero-prevalence settings where access to CMV screening and treatment are limited due to poverty, compounded by the unavailability of such services in public health care facilities.
CMV reactivation often occurs during pregnancy, more frequently in HIV-infected women including those on cART [
1,
2]. In our clinical setting, infection is nearly universal with 99.6% of pregnant women showing chronic (non-primary) CMV infection [
11]. However, preconception CMV immunity provides limited protection as intrauterine transmission can occur in women who are sero-immune prior to pregnancy [
12]. CMV has been shown to replicate at the uterine-placental interface, potentially impairing vascular remodelling thereby occluding blood flow causing local hypoxia and other adverse foetal outcomes [
13].
CMV is the leading infectious cause of congenital abnormalities and permanent disabilities. CMV-MTCT rate is 2% in developed countries with up to 15% of the infected new-borns being symptomatic, presenting with microcephaly and intrauterine growth restriction [
14]. Furthermore, up to 15% of the initially asymptomatic infants progress to develop long term morbidities such as non-genetic sensorineural hearing loss, severe motor developmental and visual impairment as well as varying degrees of adverse neurodevelopmental outcomes, including cerebral palsy, seizures and intellectual disability [
15]. CMV also remains a risk factor for childhood acute lymphocytic leukaemia [
16]. Despite all these health problems and/or disabilities, there is poor awareness of congenital CMV (cCMV) among women of reproductive age in both developed and developing countries [
17‐
19]. Furthermore, the exact pathogenesis remains largely unknown but could be due to a complex interplay between, placental, foetal, maternal and viral factors [
20].
The role of maternal CMV replication on adverse pregnancy outcomes and infant health remains poorly understood in SSA with near-universal adult CMV infection and high HIV-1 sero-prevalence. Furthermore, in the same setting breastfeeding is the norm regardless of maternal HIV status. It is plausible that maternal CMV reinfection and/or reactivation may be contributing to the relatively higher HIV-1 vertical transmission rates and other adverse pregnancy outcomes common in this setting, but this has not been adequately investigated. In this pilot study we investigated the association between antenatal plasma CMV-DNA levels and HIV-1 vertical transmission events, plasma HIV-1-RNA load and pregnancy outcomes in a subgroup of pregnant women ≥ 20 weeks gestational age participating in the University of Zimbabwe Birth Cohort Study (UZBCS).
Methods
Design of University of Zimbabwe birth cohort study (UZBCS)
The UZBCS aims to investigate the role of maternal comorbidities, including co-infections with persistent viruses such as HIV, CMV, hepatitis B virus, and other infectious or non-communicable diseases including maternal nutritional status on pregnancy outcomes, infant mortality, development, immunity and health. By design, approximately 50% of the expecting women the UZBCS are HIV-infected. Briefly, at enrolment all women answered a structured questionnaire aiming at a comprehensive clinical, socio-demographic, environmental and household characterization of the research participants. Further, socio-economic information comprising employment status, family monthly income, money set aside for food were recorded including general household food security. Date for cART commencement and compliance were recorded in HIV-infected women. According to Zimbabwean national guidelines, all HIV-infected women should be on lifelong cART (Option B +), but in practice pregnant women often do not receive a timely HIV diagnosis. The current standard of care for all Zimbabwean HIV-infected pregnant women to prevent HIV MTCT consists of (non)–nucleoside reverse transcriptase inhibitors; TENOLAM-E (Tenofovir, Lamivudine and Efavirenz. Mothers breastfeed as long as they wish to. However, exclusive breastfeeding is encouraged during the first six months of life.
All HIV exposed infants are commenced on Cotrimoxazole prophylaxis until they stop breastfeeding or they test HIV-PCR positive, whichever occurs first. Upon detection of HIV-1 infection, infants are immediately commenced on cART, usually comprising the protease inhibitors Lopinavir/ritonavir and nucleoside reverse transcriptase inhibitors, Zidovudine and Lamivudine.
The UZBCS recruited 600 HIV-infected and 600 HIV-uninfected pregnant women at least 20 weeks of gestation from February 2016 until June 2019. Mother-infant dyads will be followed for 2 years, with an expected study completion date in June 2021.
Selection of participants at enrolment into UZBCS
Potential participants for the cohort were identified during routine antenatal care visits at any one of the four out of a total of twelve City of Harare Polyclinics, situated in the South-western high-density areas of Harare namely, Kuwadzana, Rujeko, Budiriro and Glenview. Residents of these communities are of relatively poor socio-economic status. Follow-up visits of the mother-infant dyads were performed at delivery, within 10 days of life and 6, 10, 14 and 24 weeks postpartum. Date and mode of delivery, birth weight and length, feeding practices and information on the general health and development of infants were also collected.
Selection of participants for CMV analysis from UZBCS participants
Data and bio-samples were available from the first 527 mothers recruited consecutively from February 2016 and August 2017, of which 280 were HIV-infected. Due to financial limitations, CMV-DNAemia measurements were only possible in a subgroup of pregnant women. We first selected all mothers who transmitted HIV to their infants by 6 months of age (n = 11) as MTCT cases. A total of 120 HIV-infected but non-transmitting mothers were randomly selected as controls by a computer Pseudo Random Number generated algorithm. Three of these women gave birth to live twins. Depending on respective analyses’ outcomes we referred to this group as 120 HIV-infected but non-transmitting mothers or 123 infant-mother dyads. An additional 46 HIV-uninfected controls and their infants were also randomly selected using the same procedure.
Blood collection and analysis
Ten millilitres of maternal venous blood was collected at enrolment. Maternal HIV diagnosis was done using qualitative rapid immunochromatographic assays, SD Bioline HIV-1/2 3.0 (Standard Diagnostics Inc., Kyonggi-do, South Korea) and Abbott’s Determine® HIV-1/2. Western Blot was the tie breaker for any indeterminate test results.
Full blood counts were determined from whole ethylenediaminetetraacetic acid (EDTA) blood samples using a Mindray© Haematology 3-part differential, 16 parameters BC3600 Analyser (Shenzhen, China). For diagnosis of anaemia in pregnancy, the World Health Organisation (WHO) definition was used (haemoglobin < 11.0 g/dL) [
21].
Absolute CD4+ T-lymphocyte counts in EDTA blood samples were enumerated within a maximum of 6 h after sample acquisition for all HIV-infected mothers using a Partec Cyflow counter (Cyflow, Partec, Munster, Germany).
Maternal CMV/HIV detection and quantification
For analysis of HIV-RNA and CMV-DNA loads, blood was centrifuged for 5 min at 3000 rpm and the plasma was aliquoted in appropriately labelled cryo-vials and stored at − 80 °C until testing.
Total CMV nucleic acids were extracted from 200 µL plasma using the QIAamp MinElute Virus Spin Kit (Qiagen, Hilden, Germany). Then, 60 µL of total viral nucleic acids including CMV-DNA were eluted and immediately stored at − 80 °C for subsequent testing. CMV-DNAemia detection and quantification was performed by quantitative PCR (qPCR) using the RealStar CMV PCR kit v1.0 (Altona Diagnostics, Hamburg, Germany). CMV-DNA was quantified using the QuantStudio 3 Real-Time PCR System (Applied Biosystems, CA). The detection limit was 1 copy per mL.
For HIV, viral nucleic acids were extracted from 1 mL maternal baseline plasma and HIV-1-RNA quantified using an automated TaqMan Roche Amplicor 1.5 Monitor Test (Cobas AmpliPrep/Cobas TaqMan, Roche Diagnostics, Branchburg NJ), according to the manufacturer’s instructions. The detection limit was 20 copies per mL.
Assessment of adverse birth outcomes
Adverse birth outcomes included preterm birth (< 37 weeks of gestation), low Apgar scores at 5 min (< 7) and low birth weight (LBW, < 2500 g), weighed within the first hour of life, before significant postnatal weight loss had occurred. Head circumference and birth lengths were also measured in centimetres. Z-score transformation of infant birth weight and birth length followed normal distribution of WHO reference data of 2006. Transformation was performed using the R package “zscorer”.
Early infant HIV diagnosis
Venous blood (EDTA) or heel prick blood spots were collected and preserved on 903 protein saver card (number 105311018) at delivery, 10 days and 6, 10, 14 and 24 weeks. Every infant blood sample of each study visit was processed and stored as dried blood spots and plasma at − 20 and − 80 °C, respectively for further evaluation, including HIV and CMV assessments. Infants’ HIV infection was detected using a qualitative 1.5 Roche Amplicor HIV-1 proviral DNA PCR kit (Roche Diagnostics Incorporation, Branchburg, New Jersey).
Data management and statistical analysis
Data were entered and managed using Research Electronic Data Capture (REDCap v 8.0, © 2020);
http://www.redcap.uzchs.ac.zw/redcap/. Quality assurance on the accuracy of data entry included independent double entries and verification in cases of discrepancies.
Parameters from groups of HIV-infected/-uninfected and/or CMV-DNA positive (viremic) and negative (aviremic) women were compared using the Kruskal Wallis test, Mann–Whitney U test, Chi-squared test and Fisher exact test where appropriate. For all serology positive test results, undetectable levels of HIV-1-RNA load or CMV-DNA were assigned the value zero. For multivariate linear and logistic regression analyses, the following predictors: CMV-DNAemia of > 50 copies per mL, HIV-1-RNA load, years since HIV diagnosis, ongoing cART use duration and maternal age, all at the time of enrolment were tested.
For infant outcomes, all infants (including twins) were included, and for each infant the respective maternal parameters were used (resulting in usage of the same maternal parameters for both twins). For maternal outcomes (e.g. prediction of maternal HIV-1-RNA load) the mothers with twin deliveries were counted once.
Automated parameter elimination for optimization of the Akaike information criterion using the “glmulti” R package was done. Due to limited statistical power, we did not consider interaction between predictors. All calculations were performed in R Studio 1.2.5001.
Discussion
In this pilot study, we used data from UZBCS to test the association between antenatal plasma CMV-DNAemia and HIV-MTCT in a case–control design of 131 HIV-infected pregnant women and their 134 HIV exposed infants (including 3 sets of twins). Our results indicate that pregnant women ≥ 20 weeks gestational age with CMV-DNAemia > 50 copies per mL had an odds ratio of approximately 4 for HIV-1 vertical transmission within the first 6 months of life.
In our pilot study, CMV-DNAemia did not differ regarding HIV status and cART use, suggesting CMV reactivation as a potential problem in pregnant women regardless of HIV status. These results contrast with previous findings where CMV reactivation occurred more frequently in HIV-infected but non-pregnant American women, even among those on cART [
22]. Furthermore, no significant association between maternal age and CMV-DNA load was apparent in our pilot study, most likely due to the much younger age of pregnant women of our cohort, in contrast to a Brazilian study that observed higher CMV-DNA loads in relatively older women [
23]. Generally, differences may potentially be related to distinct environmental exposures including presence of other co-infections in SSA compared to Western countries.
Antenatal plasma CMV-DNAemia of > 50 copies per mL was a significant predictor for HIV vertical transmission (OR 4.02, CI 1.09–15.30,
p = 0.035), which is in agreement with previous findings [
24]. Similarly, in a clinical trial of South African and American cART naïve pregnant women, urinary CMV levels were significant risk factors for CMV and HIV transmission to infants [
25].
Thirty six (36) % of the HIV vertical transmission occurred early, either in utero or intrapartum among cART naïve or mothers on cART for < 30 days. Some previous studies showed much higher rates of intrauterine HIV infection of about 70% (
26,
27) versus just 9% observed in our study. The difference could be that these previous studies assessed HIV vertical transmission in infants with an already confirmed CMV infection.
CMV has been shown to enhance placenta susceptibility and replication of HIV-1, and may facilitate in utero HIV transmission [
28]. In line with these findings, the presence of CMV DNAemia in HIV-infected mothers might be the underlying reason behind our earlier observation of pregnant women who transmitted HIV to their infants despite having low antenatal HIV-1-RNA-loads of < 20 copies/mL [
29].
We also observed higher CMV-DNA levels in pregnant women with baseline absolute CD4 counts < 200 cells/µl, in line with other reports of CMV-DNA load as a marker for immunosuppression and elevated HIV-1-RNA load [
30].
In a North American study, the risk of infant hearing deficit increased with higher antenatal plasma CMV-DNA of ≥ 17,000 copies/mL [
31]. However, there is still insufficient local/African data to define reliable viral load cut-offs to indicate disease severity or justify the need for CMV-specific antiviral treatment. Our data indicate that screening for CMV-DNA in pregnant women in SSA may provide useful information regarding infant prognosis.
Secondary analyses indicated higher rates of deliveries before 37 weeks gestation in HIV-infected CMV viremic women, although without statistical significance probably due to small sample size (
p = 0.063 in the complete model,
p = 0.083 after variable elimination, Table
4). This trend is not surprising since CMV infection has been shown to interfere with placental development, critical for the maintenance of a healthy pregnancy [
32]. Preterm deliveries may occur due to CMV-induced maternal immune activation and systemic inflammatory caused by CMV reactivation in pregnancy [
9].
We observed an association with lower birth weight (> 2500 g) which also failed to reach statistical significance, probably due to the limited power of our pilot study. In other studies, maternal CMV reactivation affected placental function, leading to intra-uterine growth retardation [
32], which might lead to lower infant length for age and head circumference later in life.
Our pilot study has several strengths and limitations. Extensive clinical characterization of HIV-infected pregnant women with different duration of exposures to cART compared to their HIV-uninfected counterparts is a unique strength of our study. Further, all research participants reside in highly similar environmental conditions in high-density areas of Harare resulting in an unusually homogenous study population. Limitations include the modest number of participants for which CMV-DNA levels were available and our pilot study may be underpowered for other potential adverse infant outcomes. Maternal HIV-RNA-load, the most significant factor determining HIV vertical transmission was assessed once at enrolment and no additional measurements of maternal HIV -1-RNA load are available until at exit. cCMV infections and trends are yet to be assessed in all the infants born of 600 HIV-infected and 600-uninfected pregnant women of the UZBCS, to investigate whether early HIV infection predisposes infants to CMV infection or the other way round. Further, our non-interventional study design does not allow us to distinguish whether CMV is an independent risk factor for HIV-1-MTCT or directly responsible for adverse outcomes, so further mechanistic studies are needed in this regard.
Acknowledgements
The authors would like to thank the research participants of the UZBCS and the research support team for their commitment, Professor C Dandara, Mrs Doreen Mhandire (University of Cape Town), Dr G Kandawasvika, Dr P Kuona, Paediatrics and Child Health UZ-CHS, Ms P Chandiwana, Mrs P Muzire (Research Support Centre, UZ-CHS), Ms H Mataramvura, Ms E Mazengera, Mr N Taremeredzwa, Sr M Ngoweni and Professor RBL Gutsire (Department of Immunology).
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.