|Home | About | Journals | Submit | Contact Us | Français|
Cytomegalovirus is highly prevalent worldwide and an important opportunistic pathogen in human immunodeficiency virus (HIV)–infected individuals. The effects of cytomegalovirus infection on HIV-exposed infants are poorly understood. We conducted a systematic review to assess the relationship between cytomegalovirus and HIV infections among HIV-exposed infants. Limited evidence suggests that HIV-induced immunosuppression in the mother increases the rate of congenital cytomegalovirus infection, while maternal antiretroviral therapy may reduce it. Limited information exists on the direction of the relationship between cytomegalovirus and HIV transmission among HIV-exposed infants. Only 2 studies have addressed this temporal sequence of events, and they suggest that cytomegalovirus can lead to subsequent HIV infection in HIV-exposed infants. Most evidence suggests that early cytomegalo-virus infection accelerates HIV disease progression in infants. Gaps remain in understanding the role that cytomegalovirus infection plays in HIV-exposed infants. Decreasing cytomegalovirus transmission prenatally and in infancy might further decrease HIV transmission and lead to better health among HIV-exposed infants.
Cytomegalovirus (CMV) infection is highly prevalent worldwide. In human immunodeficiency virus (HIV)–infected individuals, CMV is an important pathogen, even though the advent of effective antiretroviral therapy (ART) has made opportunistic infections with CMV less common. CMV seropositivity is still associated with increases in non–AIDS-related events and non–AIDS-related death among HIV-infected individuals . CMV may also play a role in HIV disease progression [2–8].
CMV can be transmitted in utero, during the intrapartum period, and postnatally via breastfeeding, and horizontally, through contaminated secretions such as saliva or urine. Congenital disease can be a source of considerable morbidity, whereas postpartum infection is typically asymptomatic in healthy, full-term infants . However, in resource-limited settings, CMV has been suggested as a cause for morbidity and decreased growth in infants of HIV-infected mothers, even in HIV-uninfected infants [10, 11]. Additionally, symptomatic perinatal CMV infections have been described in infants exposed to but uninfected with HIV (hereafter, “HIV-exposed-uninfected infants”) . This population of infants is increasing in magnitude worldwide because of the successes in decreasing mother-to-child transmission (MTCT) of HIV .
CMV has also been hypothesized as a cofactor in the transmission of HIV from mother to child. There are several potential pathways that could mediate such an effect. CMV and HIV can infect the same cell types, with direct coinfection leading to enhanced HIV replication. CMV can also enhance other coreceptor pathways for HIV entry, transactivate viral long terminal repeats, and cause inflammation and immune activation, all of which can enhance HIV replication in vitro [2–8, 14, 15].
While there is evidence of an association of in utero CMV and HIV infections in infants, the temporal sequence of transmission events is not well understood. Whether the burden of infant CMV infection is higher among infants of HIV-infected mothers, particularly in the era of effective ART, is also unclear. In addition, the role of CMV as a cofactor in HIV disease progression during infancy is not well characterized. We conducted a systematic review of the published scientific literature to assess the relationship between CMV and HIV infection in infants.
We investigated 5 questions in this systematic review. For each question, the study inclusion criteria are shown.
First, we sought studies that investigated whether HIV-exposed infants are more likely to acquire CMV infection, compared with HIV-unexposed infants. Studies were included if they determined infant CMV infection by 6 months of age and compared CMV infection rates by infant HIV-exposure status.
Second, we searched for studies that considered the effect of prenatal/postnatal antiretroviral exposure on transmission of CMV infection to the HIV-exposed infants. Studies were included if they reported information on maternal ART during pregnancy and/or breastfeeding, determined infant CMV infection status and the timing of infection, and compared infant CMV infection rates by maternal antiretroviral status.
Third, we looked for research that evaluated whether HIV-infected infants are more likely to acquire CMV infection, compared with HIV-exposed-uninfected infants. Studies were included if they determined infant HIV infection status in the first 6 weeks of life, determined infant CMV infection, and compared CMV infection rates by HIV status.
Fourth, we sought studies that addressed whether early CMV infection increases the risk of subsequent MTCT of HIV. Studies were included if they determined infant CMV infection in the first 6 months of life and determined infant HIV infection status and the timing of infection.
Fifth, we looked for investigations that considered whether infants coinfected with HIV and CMV have faster HIV disease progression during infancy. Studies were included if they included children perinatally infected with HIV who were <1 year old, tested for CMV infection in the first year of life, included at least 1 measure of HIV disease progression as an outcome (including death), and made comparisons by CMV status.
Medline, Embase, CINHAL, CAB Abstracts, Global Health, Web of Science, FedRIP, LILAC (PAHO/WHO), WHOLIS, and the Cochrane Library databases were searched for the following terms: (prenatal OR postnatal OR perinatal OR congenital OR Infant OR fetus OR newborn OR fetal OR baby OR babies) AND (cytomegalovirus OR CMV OR HCMV) AND (HIV OR “human immunodeficiency virus”). Results were limited to studies published in English. No restrictions on publication year were used; the search was last conducted on 30 March 2015.
All abstracts and titles were screened for relevance to each of the 5 review questions. Only publications reporting results from original studies were included. The full text of all relevant articles was reviewed by 2 authors (K. E. A. and A. P. K.) independently, to determine whether they met eligibility criteria for any questions assessed in this systematic review. The third author (S. R. E.) resolved discrepancies. For each included study, the following data were abstracted: (1) objective, (2) study design, (3) population, (4) results, (5) methodologic strengths, and (6) methodologic weaknesses. Fisher exact P values were used to compare proportions for studies that did not provide a statistical test or did not specify the statistical test.
The search results yielded 1186 articles. After removing duplicate references and screening titles and abstracts, we reviewed the full text of 102 articles. There were 6 discrepancies resolved by the third reviewer. In total, 19 studies met the eligibility criteria for at least 1 question of the systematic review.
Five studies met the eligibility criteria to be included in this assessment [10, 16–19]. A cross-sectional study of neonatal admissions in Zambia examined the prevalence of congenital CMV infection, determined by detection of CMV DNA in saliva, urine, or serum or detection of immunoglobulin M (IgM) antibody in serum in the first 3 weeks of life . Congenital CMV infection was detected in 11.4% (9 of 79; 95% confidence interval [CI], 6.1%–20.3%) of HIV-exposed neonates, compared with 2.1% (6 of 293; 95% CI, .8%–4.6%) of unexposed neonates. Maternal HIV infection was independently associated with an increased odds of congenital CMV infection, compared with no maternal infection (odds ratio [OR], 6.66; 95% CI, 2.13–20.88; P = .001). As this study was not population-based—the point of entry was hospital admission of a sick neonate—results may not be generalizable. An Italian study used CMV-specific polymerase chain reaction (PCR) analysis to examine cord blood obtained using the dried blood spot technique from newborns of women with and those without HIV infection. The study found that 3 of 187 HIV-exposed infants (1.6%) had congenital CMV infection, compared with 0 of 372 infants born to HIV-uninfected women (P = .04) . A limitation of this study was that it only tested cord blood samples, which could underestimate CMV prevalence. A study in Zambia found a significant difference in the proportion of infants with high serum CMV DNA loads at 6 months: 6 of 67 HIV-exposed infants (9%), compared with 7 of 235 HIV-unexposed infants (3%), had a CMV DNA load of >50 copies/mL (P = .044) .
Two other studies yielded different results. A cohort of infants (with or without HIV exposure) recruited into a trial of micro-nutrient-fortified complementary foods in Zambia provided information on rates of CMV infection, as screened by serum CMV DNA at age 6 months . The infants were representative of the region. CMV DNA was detected in the serum of 55 of 120 HIV-exposed infants (45.8%) and 152 of 393 HIV-unexposed infants (38.7%); the difference was not statistically significant. A study from Brazil examined rates of congenital CMV infection among HIV-exposed infants, compared with unexposed infants, from a CMV-immune, low-income population; only 8.7% of HIV-infected mothers had an AIDS-defining condition, and none had late-stage HIV infection . No difference was observed in rates of congenital CMV infection, defined as CMV detection in the urine by PCR or culture by age 15 days, by HIV-exposure status (2.7% vs 2.9% in infant with and those without HIV exposure, respectively). Of interest, perinatal CMV infection (defined as detection of viruria during age 1–3 months) was detected in fewer HIV-exposed infants than HIV-unexposed infants (7.9% vs 39.4%; P < .001); most infants (93.9%) in the latter group breastfed, whereas only 5.9% of HIV-exposed infants breastfed.
Most of the evidence on the frequency of congenital CMV infection among HIV-exposed infants accumulated prior to the combination ART (cART) era [20–22]. However, 5 recent studies shed some light on this question and met the criteria to be included in this review [12, 20, 23–25]. The large French Perinatal Cohort Study showed lower rates of congenital CMV infection in HIV-exposed-uninfected infants in the cART era (1.2%), compared with the pre-cART era (3.5%), particularly if cART began in the first trimester (P = .004); among HIV-infected infants, however, rates of congenital CMV infection remained high in the cART era . A study from the United States did not show a change in congenital CMV rate by maternal cART use; most mothers started cART after the first trimester in this study . Consistently, a study of 367 HIV-exposed infants in the United States  found that congenital CMV infection was associated with higher maternal HIV load at the start of prenatal care (P = .02) and with maternal HIV diagnosis during pregnancy/delivery (P = .03); thus, cART would have been started later in pregnancy or not at all in these women. Data from sub-Saharan Africa are limited. A study of 748 newborns of HIV-infected mothers from South Africa (96% received prenatal antiretrovirals) collected saliva from newborns at a median age of 1 day and tested specimens for CMV via PCR analysis. The study reported a congenital CMV infection rate of 2.9% (22 infants); no association of congenital CMV infection was observed with length or type of maternal antiretroviral prophylaxis. However, low maternal CD4+ T-cell count (<200 cells/μL) during pregnancy was associated with congenital CMV (adjusted OR, 2.9; 95% CI, 1.2–7.3) .
Rates of perinatal/early postnatal CMV infection (defined as a positive culture result in the first 6 months of life) were decreased in the cART era, compared with the pre-cART era (8.9% vs 17.9%; P < .01) in a US cohort of 414 HIV-exposed infants ; maternal ART was associated with a decreased odds of perinatal/early postnatal CMV (OR, 0.21; 95% CI, .07–.63). Furthermore, the likelihood of symptomatic perinatal/early postnatal CMV infection (symptoms included splenomegaly, lymphadenopathy, and hepatomegaly) was increased in infants whose mothers had not received cART, compared with those whose mothers were receiving cART (41% vs 6% of CMV-infected infants; P < .05) .
The effect of cART on breast milk CMV load was examined among 69 HIV-infected, lactating Malawian women . There was an association between milk HIV-1 RNA and CMV DNA load. However, milk CMV load was similar in women who did and those who did not receive ART in this small sample, leading the authors to postulate that the impact of maternal ART on the magnitude of infant CMV exposure may be limited. Of interest, there was an inverse relationship between milk CMV load and infant growth .
Eleven studies met the criteria to be included in this assessment (Table 1) [12, 18, 21–24, 26–30]. Ten of these addressed congenital CMV infection [12, 18, 21–24, 26, 28–30]. One early study in the United States examined data from 154 infants born during 1988–1995 . Congenital CMV infection (defined as infants who tested positive for CMV within 3 weeks of birth) was more common in HIV-infected infants, compared with HIV-exposed-uninfected infants (Table 1). Additionally, the first positive HIV test result was noted to be at an earlier mean age in infants with congenital CMV infection, compared with those without congenital CMV infection (8.8 vs 30.1 days), suggesting a higher frequency of in utero HIV infections among infants with congenital CMV infection, but this difference was not statistically significant (P = .10). The large French Perinatal Cohort study tested 4797 HIV-exposed infants for CMV by urine culture in the first 10 days of life (Table 1) . There was a significantly higher prevalence of congenital CMV infection among HIV-infected neonates, compared with HIV-uninfected neonates (Table 1). Similar results were seen in a smaller study of 51 HIV-exposed infants in Kenya , as well as in a retrospective case-control analysis of HIV-exposed infants enrolled in a clinical trial comparing long and short durations of maternal and infant use of zidovudine for the prevention of MTCT of HIV in Thailand . Most recently, a subanalysis of data from a study in Malawi revealed that congenital CMV infection was more common among infants with in utero HIV infection, compared with HIV-exposed-uninfected infants (Table 1) .
Other studies have not shown significant differences between congenital CMV rates of HIV-infected and HIV-exposed-uninfected neonates, although most had small samples of HIV-infected infants (Table 1) [12, 22, 23, 26].
Of the 6 studies assessing postnatal CMV infection, most found a significantly higher rate of CMV infection in HIV-infected children (Table 1) [12, 18, 22, 26–28]. The study by Chandwani et al was the only one that found no significant difference in postnatal CMV rates by HIV status; however, the P value for the comparison was marginally significant (Table 1), and the cumulative infection rate, including congenital CMV infection, was significantly higher in HIV-infected children, compared with HIV-exposed-uninfected children (30% vs 17%; P = .010) .
While there are several articles describing rates of HIV/CMV coinfection in infants (see the previous subsection), little research has examined whether CMV infection is associated with increased susceptibility to subsequent MTCT of HIV. Two studies met the eligibility criteria for this question; one addressed sequence of perinatal infection events, while the other examined postnatal infection [28, 30].
A study from Thailand analyzed results of HIV and CMV testing performed on HIV-exposed infants over time from birth through age 18 months. The sequence of infection with CMV and HIV was discernable from the longitudinal data. In utero CMV infection, defined as a positive CMV IgM titer in cord blood or detection of CMV DNA in an infant peripheral blood sample obtained within 10 days of birth. Intrapartum/neonatal CMV infection was defined as CMV-negative cord blood and neonatal peripheral blood samples, with a serum sample positive for CMV IgM or DNA at 6 weeks of age. In utero and perinatal HIV infection were defined as a positive HIV-specific PCR test within 7 days after birth and between 8 days and 1 month after birth, respectively. While congenital CMV infection was associated with intrapartum HIV infection (P = .03), perinatal CMV infection was not associated with in utero HIV infection (P = 1.00). While not conclusive, this suggests that the timing of CMV infection may be important in determining increased susceptibility to HIV MTCT .
A more recent study examined the risk of HIV infection through breastfeeding in Malawian infants with early CMV infection, defined as CMV DNA detection at age 6 months. HIV DNA testing was performed at birth and ages 2, 12, 28, and 48 weeks. Detection of CMV in plasma via PCR analysis at age 6 months was not associated with an increased risk of subsequent HIV acquisition through breastfeeding (hazard ratio [HR], 4.52; 95% CI, .58–35.3), while there was a marginally significant increased risk for the combined outcome of HIV acquisition or infant death (HR, 4.27; 95% CI, .99–18.4) . The study was underpowered to detect a difference in HIV transmission events.
Six articles met eligibility criteria for this question (Table 2) [21, 22, 26, 27, 31, 32]. Although our focus was disease progression during infancy, most studies reported disease progression for longer periods.
Two small, retrospective studies conducted in the United States in the early 1990s assessed the association of CMV coinfection with mortality in HIV-infected infants [27,31].While one study reported a significant association (P < .05), the Fisher exact test we conducted yielded a P value of .095, indicating no significant difference (Table 2) .The second study also failed to find a statistically significant association (Table 2) .In a small prospective cohort study, Gabriel et al reported no difference in survival or HIV disease progression among perinatally HIV-infected children by CMV infection status (Table 2) .
Another study of 37 HIV-infected infants found that CMV infection in the first 6 months of life was associated with higher mean p24 antigen concentrations and higher mean CD8+ T-lymphocyte proportion but not with CD4+ T-lymphocyte proportion (Table 2) . A similar study found absolute CD4+ T-lymphocyte counts, CD4+ T-lymphocyte percentage, and ratios of CD4+ to CD8+ T cells were significantly lower in CMV/HIV-coinfected infants, compared with infants without CMV infection (Table 2). Mean survival time for CMV/HIV-coinfected infants was 25 months, compared with 39 months for infants not infected with CMV; this difference was not statistically significant (Table 2) .
In 1999, Kovacs et al reported the results of a study that examined the association of CMV infection with HIV disease progression among HIV-infected infants. Infants were recruited from multiple high-risk obstetric clinics in the United States and followed until age 18 months. CMV infection was assessed by urine culture at birth and every 6 months thereafter. Two infants had congenital CMV infection. By age 18 months, infants coinfected with CMV had higher rates of HIV disease progression than infants infected with HIV alone (Table 2). Additionally, CMV infection during the first 18 months more than doubled the risk of HIV disease progression (relative risk, 2.59; 95% CI, 1.13–5.95) .
This systematic review synthesizes the evidence on the relationship between HIV and CMV infections in infants but also highlights several limitations and gaps in the existing literature. On the question of whether HIV-exposed infants have higher rates of congenital CMV infection, the limited information available suggests that the main determinant may be the mother’s level of HIV-induced immunosuppression [10, 16–18]. Most studies in which HIV-infected mothers were not immunosuppressed found no difference in congenital CMV infection rates [10, 18], whereas those with immunosuppressed women found increased congenital CMV infection in HIV-exposed infants. This is biologically plausible, as immunosuppression correlates with CMV shedding in the genital tract of HIV-infected women [14, 17, 29, 33]. Additionally, emerging evidence suggests that maternal ART may decrease the rate of congenital CMV infection [12, 20, 23–25], likely by improving maternal health and immunity, resulting in decreased mucosal shedding of CMV during birth and a reduced incidence of reinfection or disease reactivation. It is more difficult to assess any effects of maternal ART on postnatal CMV transmission, particularly in settings where breastfeeding is practiced, given generally very high rates of CMV transmission via breastfeeding [30, 34, 35]. As an-tiretroviral programs roll out in sub-Saharan Africa and as more women initiate cART even prior to their pregnancy, studies will need to reevaluate the frequency of congenital, perinatal, and postnatal CMV infection in HIV-exposed infants.
The majority of studies point to an increased prevalence of congenital CMV infection among HIV-infected neonates, compared with HIV-exposed-uninfected neonates [21, 24, 28, 29]. Similarly, the available evidence indicates a higher frequency of postnatal CMV acquisition among HIV-infected infants, compared with HIV-exposed-uninfected infants [12, 22, 26, 27]. This could be due to a higher degree of immunosuppression in HIV-transmitting mothers, with concomitant increased risk of CMV reactivation, reinfection, and shedding and, thus, an increased risk of CMV transmission to the infant . Additionally, immunosuppression in HIV-infected infants may make them more vulnerable to coinfection.
Only 2 studies were found that could directly assess the question of whether early CMV infection increases the risk of subsequent MTCT of HIV. These 2 studies indicate that congenital CMV infection does increase the risk of subsequent HIV infection for infants during the intrapartum period  and postnatally . Even though there is biological plausibility for these findings, more epidemiological evidence is needed to answer this question definitively.
Few studies have assessed the role of early CMV infection on HIV disease progression in infants, and most have small sample sizes and limited power with methodologic differences, making it difficult to directly compare results. Most but not all evidence suggests that early CMV infection accelerates HIV disease progression in infants. However, most included studies were conducted exclusively in the pre-cART era [21, 22, 27, 31, 32]. Of interest, recent, yet unpublished evidence indicates that level of CMV viremia during early life in HIV-infected infants is a predictor of the size of HIV reservoir after cART-induced virologic suppression , possibly through coinfection of the same long-lived memory lymphocytes.
A limitation of this review was the inconsistency in specimen type and laboratory methods used for detecting CMV. However, while serological analysis may detect more cases of infant CMV infection, several studies have demonstrated that PCR-based methods for CMV detection are highly correlated with serological testing [38–41]. CMV PCR testing in urine and saliva has very high sensitivity and specificity (>90%), compared with serological testing [38, 41]. Additionally, a recent study assessing the validity of PCR testing in blood found that PCR detected most infant CMV infections .
There are remaining gaps in our understanding of the role of CMV in HIV acquisition and health outcomes of HIV-infected infants and HIV-exposed-uninfected infants. Expanded access to maternal cART, starting prior to or early in pregnancy, has many benefits beyond directly decreasing HIV MTCT, including improvements in maternal health and immunity, and decreasing CMV infection risk for the infant. Decreasing CMV transmission prenatally and postnatally might further decrease HIV transmission risk, as well as lead to better health and developmental outcomes among HIV-exposed-uninfected infants. Clinical trials may need to assess methods to further decrease infant CMV exposure while maintaining breastfeeding, which has multiple proven benefits. A trial of valacyclovir starting at 34 weeks of gestation and continuing postnatally for 12 months in mother with HIV and herpes simplex virus coinfection did not affect postnatal CMV transmission via breastfeeding but modestly decreased cervical CMV shedding .Higher doses or novel anti-CMV drugs in the pipeline  may offer opportunities to further investigate the effects of inhibition of CMV replication in HIV-infected mothers as a way to further decrease MTCT of both viruses and improve infant survival and health, particularly in resource-limited countries.
Financial support. This work was supported by the CDC.
Disclaimer. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention (CDC).
Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.