Maternal Antibody Responses and Nonprimary Congenital Cytomegalovirus Infection of HIV-1–Exposed Infants

Abstract

Risk of congenital cytomegalovirus (cCMV) transmission is highly dependent on the presence of preexisting maternal immunity, with the lowest rates observed in CMV-seroimmune populations. Among infants of CMV-seroimmune women, those who are exposed to human immunodeficiency virus (HIV) have an increased risk of acquiring cCMV infection as compared to HIV-unexposed infants. To better understand the risk factors of nonprimary cCMV transmission in HIV-infected women, we performed a case-control study in which CMV-specific plasma antibody responses from 19 CMV-transmitting and 57 CMV-nontransmitting women with chronic CMV/HIV coinfection were evaluated for the ability to predict the risk of cCMV infection. Primary multivariable conditional logistic regression analysis revealed an association between epithelial-tropic CMV neutralizing titers and a reduced risk of cCMV transmission (odds ratio [OR], 0.18; 95% confidence interval [CI], .03–.93; P = .04), although this effect was not significant following correction for multiple comparisons (false-discovery rate, 0.12). Exploratory analysis of the CMV specificity of plasma antibodies revealed that immunoglobulin G (IgG) responses against the glycoprotein B (gB) neutralizing epitope AD-2 had a borderline association with low risk of transmission (OR, 0.72; 95% CI, .51–1.00; P = .05), although this was not confirmed in a post hoc plasma anti–AD-2 IgG blocking assay. Our data suggest that maternal neutralizing antibody responses may play a role in protection against cCMV in HIV/CMV-coinfected populations.

Cytomegalovirus (CMV), most widely known for its lifelong persistence and association with life-threatening multiorgan disease in immunocompromised individuals, is also the most common congenital infection worldwide [1–4]. Much like the emerging Zika virus, CMV transmitted in utero can result in severe fetal neurological deficits, including microcephaly, intrauterine growth restriction, cognitive impairment, and significant sensorineural hearing loss [4–6]. With an alarming 40 000 newborns infected with congenital CMV (cCMV) in the United States annually, of whom up to 8000 develop long-term disabilities, there is great need for the development of a protective maternal vaccine [7, 8]. In the absence of a protective vaccine, 30%–40% of CMV-seronegative women who acquire primary CMV infection during pregnancy will transmit the virus to their infant in utero [7]. In contrast, only 1%–2% of CMV-seroimmune women, in whom the virus is persistent, will transmit the virus to their fetus. Moreover, infants born to CMV-seropositive women are less likely to exhibit signs of CMV infection at birth and typically develop only mild neurological sequelae, most commonly sensorineural hearing loss [9, 10].

Similar to observations in CMV-naive women, pregnant women with chronic human immunodeficiency virus (HIV) infection have an elevated risk of transmitting CMV in utero, despite widespread CMV seroprevalence among HIV-infected women of childbearing age [11, 12]. While the risk of cCMV transmission among HIV-infected women varies depending on the severity of HIV-induced immunosuppression, rates of transmission in studies conducted prior to widespread availability of antiretroviral therapy (ART) have been reported to be as high as 11% [13–15]. Importantly, maternal ART has been an extremely effective strategy for the reduction of in utero HIV transmission, yet the burden of cCMV infection in HIV-exposed infants remains high in ART-recipient maternal populations (1.5%–3.5% transmission rate) [16–21]. Furthermore, HIV-exposed infants, regardless of their HIV transmission status, more frequently sustain symptomatic cCMV infection, compared with congenitally infected, HIV-unexposed infants [22]. Thus, as the number of HIV-exposed, CMV-infected infants with long-term neurodevelopmental effects rises in the ART era, it is critical to gain a better understanding of the impact of maternal HIV infection on cCMV infection.

Previous studies in women with chronic HIV infection have shown that maternal CD4⁺ T-cell count is inversely correlated with the risk of cCMV transmission [20]. We reported a similar finding in our newly developed rhesus monkey model of cCMV transmission, which demonstrated that pregnant monkeys lacking peripheral CD4⁺ T-cells had high rates of placental rhesus CMV transmission, potentially attributable to the observed delay in the maternal CMV-specific neutralizing antibody response [23]. Thus, it is plausible that the effect of HIV on CD4⁺ T-cells may similarly impair the functional CMV-specific antibody responses, leading to enhanced cCMV transmission. To define the role of maternal CMV-specific antibody responses in protection against cCMV infection of HIV-exposed infants, we assessed the magnitude, function, and CMV glycoprotein specificity of maternal plasma immunoglobulin G (IgG) responses in CMV-transmitting and nontransmitting women with chronic HIV/CMV coinfection. Using multivariable conditional logistic regression modeling, we investigated whether any maternal CMV-specific antibody responses predicted the risk of cCMV infection in a US cohort of ART-naive, HIV-infected women and their infants.

METHODS

Study Population

A total of 215 HIV-infected US women enrolled into the Women and Infants Transmission (WITS) study between 1988 and 1994 were retrospectively screened for chronic CMV infection. The 215 women selected for our study had not received ART, had CD4⁺ T-cell counts of >200 cells/mL, and had maternal plasma collected between the beginning of the second trimester and 2 months after delivery, as well as a paired infant plasma sample collected within the first month of life. The serologic tests used in our study to identify women with chronic CMV infection (193 [90%]) are described in the Supplementary Materials and Methods.

Identification of CMV-Transmitting and Nontransmitting Mothers

Nineteen CMV-transmitting women (cases) were defined by either the detection of CMV DNA via quantitative polymerase chain reaction (PCR) analysis or by the presence of CMV-specific immunoglobulin M (IgM) by commercial enzyme-linked immunosorbent assay (ELISA; Creative Diagnostics) in paired infant plasma samples that had been collected at or before 1 month postpartum. Cases were matched (1:3) to CMV nontransmitting controls (n = 57) by age (<25 or ≥25), race/ethnicity (white or other), and parity (nulliparous or nonnulliparous). Details of each retrospective diagnostic assay are explained in the Supplementary Materials and Methods.

Cell Growth and Virus Culture

Human retinal pigmented epithelial (ARPE-19) cells (ATCC) and human lung (MRC-5) fibroblasts (ATCC) were cultured for no more than 35 and 20 passages, respectively. Details of cell culture maintenance are described in Supplementary Materials and Methods. AD169 (ATCC), AD169 revertant virus (AD169r; a gift from Merck) as described previously [24], and TB40/E-mCherry (a gift from N. Moorman [25]) virus stocks were prepared in T75 culture flasks. Supernatant containing cell-free virus was collected when 90% of cells showed cytopathic effects and was then cleared of cell debris by low-speed centrifugation before passage through a 0.45-µm filter.

CMV IgG Binding and Avidity Enzyme-Linked Immunosorbent Assays (ELISAs)

The magnitude of maternal CMV-specific plasma IgG was determined by whole-virion ELISAs, using CMV strains AD169 and TB40/E, whereas avidity was tested against CMV TB40/E only. The details of these experiments are described in the Supplementary Materials and Methods.

High-Throughput Immunofluorescence Neutralization Assays

ARPE-19 and MRC-5 cells were seeded into 96-well plates and incubated for 2 days at 37°C and 5% CO₂ to achieve 100% confluency. After 2–3 days, 3-fold dilutions (1:10–1:30 000) of heat-inactivated maternal plasma samples were incubated with 3 × 10⁴ plaque-forming units (PFU) of AD169r or 8.9 × 10² PFU of AD169 virus stock in a total volume of 50 µL for 45 minutes at 37°C and then added in duplicate to wells containing ARPE-19 or MRC-5 cells, respectively. Virus was removed after 2 hours, and cells were washed once before they were incubated for an additional 17 hours at 37°C. Infected cells were then fixed for 10 minutes with 3.7% paraformaldehyde and permeabilized for 10 minutes with Triton × 100, and subsequently processed for immunofluorescence with mouse anti-HCMV IE1 monoclonal antibody (MAB810, Millipore) followed by goat anti-mouse IgG-AlexaFluor488 (Millipore). To extract the fluorescent conjugate into solution, a 50-µL volume of alkaline carbonate buffer (0.1 M NaOH and 0.5% Na₂CO₃) was added to each well for 1 hour, transferred to a 384 black-well plate, and read at a wavelength of 450 nm with a Victor 3X. Calculation of the 50% infectious dose (ID₅₀) and validation of our assay are described in the Supplementary Materials and Methods.

CMV Binding Antibody Multiplex Assay

Preparation of the CMV gB domain I and domain I + II, described by Potzsch et al [26], is explained fully in the Supplementary Materials and Methods. For the multiplex assay, carboxylated fluorescent beads (Luminex) were covalently coupled to purified CMV antigens and subsequently incubated with maternal plasma in assay diluent (phosphate-buffered saline, 5% normal goat serum, 0.05% Tween 20, and 1% Blotto milk) at various dilutions. The antigen panel included gB (Sanofi-Pasteur), gH/gL (Novartis), gH/gL/gO (Novartis), gH/gL/UL128-UL130-131 pentameric complex (Novartis), gB domain I, gB domain I+II (see above), gB AD-1 (myBiosource), biotinylated linear gB AD-2 (biotin-NETIYNTTLKYGD), and biotinylated linear gH neutralizing peptide from 2 different strain sequences (biotin-LDKAFHLLL and Biotin-LDPHAFHLLL) [27]. CMV glycoprotein–specific antibodies were detected with phycoerythrin-conjugated goat anti-human IgG (2 µg/mL, Southern Biotech). Beads were washed and acquired on a Bio-Plex 200 instrument (Bio-Rad), and results were expressed as mean fluorescence intensity. Assay controls and criteria are further described in the Supplementary Materials and Methods.

Statistical Analysis

Primary immune predictors were analyzed for their association with the outcome of cCMV transmission among a cohort of women with chronic HIV/CMV coinfection, using a multivariate conditional logistic regression model that controlled for both CD4⁺ T-cell count and plasma HIV load. For each of the 3 predefined primary predictors (IgG avidity, epithelial cell neutralizing titers, and CMV pentamer–specific IgG), a false-discovery rate (FDR) was computed to account for multiple comparisons. Additionally, conditional logistic regression was used to assess the association between a set of hypothesis-generating secondary immune predictors and the risk of cCMV transmission. A post hoc analysis explored the association of maternal antibody binding to the CMV gB AD-2 domain with cCMV transmission, using data acquired from a CMV gB-2 (TRL-345; Trellis Bioscience) blocking assay. As part of this analysis, CMV gB AD-2 blocking data were evaluated as both a continuous exposure and as a dichotomous variable. The Pearson correlation coefficient was used to determine the correlation between findings of the CMV gB AD-2 binding antibody multiplex assay and those of the gB AD-2 monoclonal antibody blocking assay.

RESULTS

Screening for Maternal CMV Serostatus and Identification of Study Cases

Banked maternal plasma samples from 215 HIV-infected pregnant women from the WITS cohort collected at or before delivery were retrospectively analyzed to determine maternal CMV serostatus. As outlined in Figure 1, the majority of women (193 of 215 [90%]) were chronically infected with CMV, as defined by the presence of CMV-specific plasma IgG with no detectable IgM reactivity or by CMV-specific IgM and IgG plasma responses with high IgG avidity. The remaining women tested either seronegative for CMV (10 [5%]) or IgM seropositive but IgG seronegative for CMV (12 [6%]), suggestive of an acute infection. To identify cases of nonprimary cCMV transmission, plasma samples obtained from infants aged ≤1 month who were born to each of the 193 women with chronic HIV/CMV coinfection were assessed for cCMV infection by undergoing tests for CMV-reactive IgM or DNAemia (Table 1), as urine or saliva samples were not available. Nineteen infants (10%) met our definition of cCMV infection, with either a positive CMV-specific IgM response (11 infants) or detectable CMV DNAemia by quantitative PCR (12 infants). Four infants had positive results of both tests. Although samples collected from infants during the first 3 weeks of life are required to distinguish cCMV from postnatal CMV infection, women in the WITS cohort abstained from breastfeeding, to avoid postnatal HIV transmission, thereby also reducing the likelihood of postpartum CMV transmission [28]. We cannot, however, rule out the possibility that a proportion of the CMV-infected infants in our study contracted the infection during delivery via contact with virus in the genital tract.

Figure 1.

Maternal cytomegalovirus (CMV) seroprevalence among human immunodeficiency virus (HIV)–infected women from the Women and Infants Transmission (WITS) cohort. A total of 215 HIV-infected, antiretroviral therapy (ART)–naive women enrolled in the WITS study were retrospectively screened for CMV-specific immunoglobulin G (IgG) and immunoglobulin M (IgM) plasma antibodies by commercial enzyme-linked immunosorbent assay. Further assessment of CMV-specific IgG antibody avidity was performed on maternal plasma with detectable IgG and IgM, to distinguish between primary and nonprimary maternal CMV infections. Avidity data are reported as relative avidity indexes (RAIs).

Table 1.

Detection of Congenital Cytomegalovirus (CMV) Infection Among Human Immunodeficiency Virus–Exposed Infants Aged ≤1 Month

IgM Reactivity	qPCR Result
	Positive	Negative	Total
Present	4 (2.1)	7 (3.6)	11a
Absent	8 (4.2)	174 (90.2)	182
Total	12a	181	193

Data are no. or no. (%) of infants.

Abbreviations: IgM, immunoglobulin M; qPCR, quantitative polymerase chain reaction.

^a A total of 19 infants exhibited plasma CMV-specific IgM reactivity (n = 12) and/or had positive qPCR results (n = 11).

Characteristics of the Study Population and Selection of Case Controls

As shown in Table 2, demographic characteristics of 19 CMV-transmitting and 174 nontransmitting HIV/CMV-coinfected women from the WITS cohort revealed that similar proportions were of a race/ethnicity other than white (79% and 91%, respectively), at least 25 years of age (63% and 75%, respectively), and had previously given birth to at least 1 other child (79% and 76%, respectively). The rate of cCMV transmission in this cohort of HIV-exposed infants (10%; Table 3) did not appear to be affected by infant sex or the mode of delivery (74% and 71% infants born vaginally to CMV-transmitting and nontransmitting women, respectively). Infant clinical characteristics including birth weight, head circumference, and gestational age, which can be influenced by cCMV infection and HIV coinfection, were similar across the CMV-infected and uninfected groups in this study, demonstrating limited overt sequelae of cCMV present at birth in this population (Table 3) [4, 21]. Similar to previous studies [17], a high percentage (53%) of infants with a diagnosis of cCMV infection were coinfected with HIV, compared with only 18% of infants who did not acquire CMV. Among the coinfected infants, 20% of HIV infections were established to have occurred in utero, by positive HIV nucleic acid testing at birth, whereas 50% occurred during the peripartum period, and the remainder had an unknown HIV transmission mode [29].

Table 2.

Characteristics of Mothers Coinfected With Human Immunodeficiency Virus and Cytomegalovirus (CMV), by CMV Transmission Status

Characteristic	CMV Transmitters, No. (%)	CMV Nontransmitters, No. (%)
Overall	19 (10)	174 (90)
Race/ethnicity
White	4 (21)	15 (9)
Other	15 (79)	159 (91)
Age at delivery, y
<25	7 (37)	44 (25)
≥25	12 (63)	130 (75)
Parity
0	1 (5)	19 (11)
1	8 (42)	43 (25)
2	3 (16)	36 (21)
>2	4 (21)	53 (30)
Unknown	3 (16)	21 (13)
AIDS status
Stage 0 or missing	7 (41)	65 (59)
Stage 1	1 (6)	10 (9)
Stage 2	4 (24)	23 (21)
Stage 3	5 (29)	13 (12)

Table 3.

Characteristics of Human Immunodeficiency Virus (HIV)–Exposed Infants With or Without Cytomegalovirus (CMV) Infection

Characteristic	CMV Infected	CMV Uninfected
All infants	19 (10)	174 (90)
Gestational age, wk
<38	3 (16)	60 (34)
≥38	16 (84)	114 (66)
Mode of delivery
Cesarean section	4 (21)	39 (22)
Vaginal	14 (74)	128 (74)
Unknown	1 (5)	7 (4)
Sex
Male	9 (47)	90 (52)
Female	10 (53)	84 (48)
Head circumference, z score percentile	13.3 (1.4–39.9)	13.3 (4.4–32.7)
Birth weight percentile	22.3 (5.2–48.2)	14.6 (4.0–37.2)
HIV status
Positive	10 (53)	32 (18)
Negative	9 (47)	142 (82)
Timing of HIV infection
In utero	2 (20)	3 (9)
Peripartum	5 (50)	23 (72)
Unknown	3 (30)	6 (19)
HIV-positive infants aged 1 mo
CD4+ T-cell count, cells/mL	2284 (1298–2864)	3100 (2514–4186)
CD4+ T-cell percentage	40 (34–44)	45 (39–52)
Lymphocyte count, cells/mL	5816 (4238–9293)	6989 (5961–8215)
Log10 HIV RNA load, copies/mLa	5.3 (4.9–5.9)	5.5 (5.1–5.9)

Data are no. (%) or median (interquartile range).

^a Data are for 6 infants with and 17 without CMV infection.

Selection of CMV-nontransmitting control mothers was performed using maternal demographic data to match each CMV-transmitting woman (n = 19) to 3 CMV-nontransmitting women (n = 57) on the basis of race/ethnicity, age, and parity, all of which are recognized independent risk factors of cCMV transmission [30, 31]. Case and control mothers proved to be well matched on these characteristics, as determined by a 2-sided Fisher's exact test (Supplementary Table 1). Other variables that may independently influence the risk of cCMV transmission within this population of HIV-infected mothers, including plasma HIV RNA load and peripheral CD4⁺ T-cell count [17–19], were adjusted for in the multivariable logistic regression analysis model and evaluated by the Wilcoxon 2-sample test (Supplementary Table 1). Maternal plasma CMV load was not controlled for in our study, as all of the CMV-transmitting women had undetectable CMV viremia by qPCR.

Primary Analysis of Maternal CMV-Specific IgG Avidity, Epithelial Cell Neutralization, and Pentameric Complex IgG Binding as Covariates of cCMV Transmission

The goal of this case-control study was to determine whether specific humoral immune responses in mothers with chronic HIV/CMV coinfection were predictive of the risk of cCMV transmission. In previous work, both maternal CMV-specific IgG antibody avidity and epithelial cell neutralization potency have been correlated with the risk of cCMV transmission among women with primary CMV infections [22, 32, 33]. Thus, these responses were considered primary immune predictors of cCMV transmission in a multivariable conditional logistic regression model (Figure 2). Additionally, IgG binding to the CMV gH/gL/UL128/UL130/UL131A pentameric glycoprotein complex was also included as a primary covariate, as this complex is required for virus entry into host epithelial cells [34]. Our primary analysis revealed that neither the CMV-specific IgG avidity index, measured in a whole-virus ELISA against epithelial-tropic strain TB40/E (odds ratio [OR], 0.72; P = .35; FDR = 0.53), nor IgG binding to the pentameric complex (OR, 1.14; P = .69; FDR = 0.68) predicted transmission outcome in the conditional logistic regression model (Figure 2A). However, the magnitude of the epithelial cell neutralization titers (measured as the ID₅₀) against CMV strain AD169r had an association with a low risk of cCMV transmission (OR, 0.18; P = .04) before correction for multiple comparisons (Figure 2A and 2B) yet was nonsignificant after adjustment (FDR = 0.12). Because the diagnosis of congenital CMV infection by IgM ELISA is less reliable than detection of viral DNA by qPCR, an additional post hoc analysis was conducted to determine the association between epithelial cell neutralization titers and transmission risk, after removing cases in which infant plasma specimens tested positive by an IgM ELISA only. Reassuringly, this analysis confirmed a statistically significant association between high maternal epithelial neutralizing antibody titers and a reduced risk of congenital CMV transmission (OR, 0.07; 95% CI, .01–.83; P = .0358), despite the smaller cohort size.

Figure 2.

Association of primary maternal immune predictors with the risk of congenital cytomegalovirus (CMV) transmission among women with chronic human immunodeficiency virus (HIV)/CMV coinfection. A, Multivariable logistic regression was used to determine the odds ratios and 95% confidence intervals of primary immune predictors of congenital CMV transmission. Data are plotted on a natural log (ln) scale. P values and the false-discovery rates (FDRs) calculated after a Bonferroni-type correction are included for each variable. The dotted black line indicates an odds ratio of 1. B, Neutralizing antibody titers against CMV strain AD169 encoding a repaired Ulb́ region were measured using a high-throughput immunofluorescence-based assay in retinal pigmented epithelial cells. Fifty percent infectious dose (ID₅₀; dilution) titers were calculated from the total fluorescence value read at 450 nm following extraction of CMV-IE1 AF488–tagged antibodies from fixed monolayers with an alkaline carbonate buffer. Log₁₀ ID₅₀ values indicate the plasma dilution that resulted in 50% reduction in fluorescence signal, compared with virus-only infected controls. Abbreviation: IgG, immunoglobulin G.

Secondary Analysis of Maternal CMV-Specific Humoral Immune Responses as Predictors of cCMV Transmission

In a secondary analysis, we assessed whether maternal plasma CMV-specific IgG levels, CMV neutralization potency in fibroblasts, or IgG specificity against other CMV surface glycoprotein complexes (gH/gL, gH/gL/gO, and gB) were associated with the risk of cCMV transmission in this population. Neither the magnitude of the total CMV-specific IgG binding responses against fibroblasts (OR, 0.67; P = .68) or epithelial-tropic strains of CMV (OR, 0.83; P = .84) predicted the risk of cCMV transmission (Figure 3). Interestingly, although high maternal epithelial cell neutralizing titers trended toward an association with a reduced risk of cCMV transmission in our primary analysis, neutralizing titers against a fibroblast-adapted CMV strain, AD169, did not predict cCMV transmission (OR, 1.44; P = .34; Figure 3A). We next assessed whether IgG specificity against 3 CMV surface glycoprotein complexes, gH/gL, gH/gL/gO, and gB (Figure 3A), predicted the risk of cCMV transmission. Similar to the lack of association between the maternal pentamer-specific IgG response and transmission risk, binding IgG responses against other glycoprotein complexes showed no association with the risk of cCMV transmission (OR, 0.95–1.14; P > .5 for all). To explore whether maternal antibody responses against recently defined glycoprotein neutralizing epitopes predict cCMV transmission, we assessed plasma IgG binding to antigenic domains of gB (domain I, domain II, AD-1, and AD-2) and gH (peptide I and peptide II) proteins (Figure 3A) [26, 27]. Of those tested, only maternal IgG binding to gB AD-2 had a significant, although borderline, association with a low risk of cCMV transmission (OR, 0.72; P = .05; Figure 3A and 3B). To investigate this potential association, we performed a post hoc maternal plasma anti-gB AD-2 IgG blocking ELISA as an alternative method to measure plasma IgG specificity against this gB epitope (Figure 4). While the plasma IgG binding to AD-2 and AD-2 IgG blocking responses were correlated (R = 0.69’ P≤.0001), conditional logistic regression modeling revealed no association with transmission risk when the anti-AD2 blocking response was evaluated as either a continuous exposure (OR, 1) or as a dichotomous variable (OR, 0.99).

Figure 3.

Association of secondary maternal immune predictors with the risk of congenital cytomegalovirus (CMV) transmission among women with chronic human immunodeficiency virus (HIV)/CMV coinfection. A, Multivariable logistic regression was used to determine the odds ratio and 95% confidence intervals of secondary immune predictors of congenital CMV transmission. Data are plotted on a natural log (ln) scale. P values are included for each variable. The dotted black line indicates an odds ratio of 1. B, The specificity of plasma antibody binding responses against neutralizing epitope AD-2 of CMV gB was measured in a multiplex assay for CMV-transmitting and nontransmitting women. Binding strength is shown as mean fluorescence intensity (MFI). Abbreviation: IgG, immunoglobulin G.

Figure 4.

Assessment of plasma antibody specificity to cytomegalovirus (CMV) gB AD-2 in CMV–transmitting and nontransmitting women. A, The magnitude of the plasma binding response to AD-2 (measured as the mean fluorescence intensity [MFI]) for all women was plotted against the corresponding AD-2 blocking percentages. The R value was calculated using a Spearman correlation test. B, The ability of plasma antibodies to block binding of a CMV AD-2 monoclonal antibody to CMV gB was measured by competition enzyme-linked immunosorbent assay. Blocking percentages were obtained using a 1:10 dilution of maternal plasma from CMV-transmitting and nontransmitting women. Abbreviation: OR, odds ratio.

DISCUSSION

In this historical cohort of HIV-infected US women and their infants, we observed a 10% cCMV transmission rate (19 of 193), falling within the previously reported 2.7%–11.4% range among other ART naive, HIV-infected populations [13–17]. A major limitation of this study is the lack of available infant urine or saliva for diagnostic CMV testing. Thus, we relied on a combination of detection of infant plasma CMV-specific IgM and DNAemia as markers of congenital infection. Diagnostic PCR has been shown to have a sensitivity of 95%–100% with the use of DNA from dried blood spots as the PCR template [35]. Yet, plasma CMV DNAemia can be cleared in congenitally infected infants even before birth, limiting the ability of this assay to be used as a diagnostic test [28]. Interestingly, of the 19 infants with a retrospective diagnosis of cCMV infection, only 4 tested positive for both IgM and CMV DNAemia. Although we set strict criteria for the inclusion of IgM-positive infants to those who had positive results of 2 commercial tests, some of these infants may have had a false-positive result.

To investigate the role of maternal IgG in protection against cCMV transmission in women with chronic HIV infection, we conducted a primary multivariable conditional logistic regression analysis to assess associations between risk of fetal infection and maternal CMV-specific IgG avidity, epithelial cell neutralizing response, and IgG binding strength to the pentamer. In a study conducted by Boppana et al, 76% of nontransmitting pregnant women after primary CMV infection had CMV-specific antibodies with high avidity indexes, compared with only 17% of CMV-transmitting women [32]. We report no association between maternal IgG avidity and risk of CMV transmission among HIV-infected women, yet all had high IgG avidity (RAI range, 0.63–1.00), as expected in a cohort of women with chronic CMV infection. More recently, work by Lilleri et al showed that CMV-nontransmitting women with primary infection had a more rapid appearance of IgG against gH/gL and the pentamer than CMV-transmitting women [33]. In our study, both CMV-transmitting and nontransmitting women had preexisting CMV-specific IgG responses, yet we were able to determine whether the magnitude of pentamer-specific IgG was associated with transmission risk. We observed no association between pentamer-specific IgG levels and the risk of cCMV. Interestingly, high plasma epithelial cell neutralizing titers were predictive of a reduced risk of cCMV transmission (P = .04), yet this finding was not significant following correction for multiple comparisons, possibly owing to our limited sample size. Alternatively, it is possible that some of the infants who tested positive for CMV-specific IgM only had a false-positive result, as our post hoc analysis following removal of these cases confirmed a statistically significant association between high maternal epithelial cell neutralizing antibody titers and a reduced risk of cCMV despite the smaller cohort size. Together, these findings suggest that the neutralization function of CMV-specific IgG responses in HIV/CMV-coinfected women may best predict nonprimary cCMV transmission.

We further explored the role of maternal CMV-specific antibody responses in protection against nonprimary cCMV infection, including fibroblast cell neutralization titers and other CMV glycoprotein–specific IgG responses. Total IgG binding responses to either a fibroblast or epithelial-tropic CMV strain did not predict transmission risk. Similar to previous findings in primary maternal CMV infection, the maternal fibroblast neutralizing response also did not predict cCMV transmission [33]. Last, we assessed the binding specificity of maternal IgG antibodies to the CMV glycoproteins gH/gL, gH/gL/gO, and gB and defined neutralizing epitopes of CMV gB and gH. Our initial results indicated a borderline association between the plasma IgG response to gB AD-2 and a reduced risk of cCMV transmission but no association with any of the other glycoprotein complexes. Interestingly, the AD-2 epitope is a dominant neutralizing epitope of CMV gB, and the presence of antibodies against this region correlated with reduced risk of CMV disease following solid organ transplantation [36]. However, a follow-up plasma anti-gB-AD-2 IgG blocking assay revealed no association between maternal IgG AD-2–specific antibodies and reduced risk of transmission.

Despite growing evidence for a role of maternal antibodies in the protection against cCMV, passive infusion of CMV hyperimmune globulin following primary maternal infection has only moderately reduced the rate of cCMV transmission in case-controlled clinical trials [37]. The failure of these trials may be due to suboptimal dosing or timing of administration, which have not been thoroughly investigated in humans. However, this study suggests that preexposure provision of passive antibodies that mediate potent epithelial cell could provide enhanced protection. Alternatively, as we did not study the role of T-cell immunity in this population, we cannot rule out the possibility that protection against cCMV transmission is dependent on a robust T-cell–mediated memory response, particularly the CD4⁺ T-cell response [20, 23], which is impaired in women with chronic HIV/CMV coinfection. This work suggests that strategies to enhance CMV-specific immunity in seropositive mothers may help to impede secondary cCMV transmission.

Supplementary Data

Supplementary materials are available at http://jid.oxfordjournals.org. Consisting of data provided by the author to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the author, so questions or comments should be addressed to the author.