Human Immunodeficiency Virus Exposure but Not Early Cytomegalovirus Infection Is Associated With Increased Hospitalization and Decreased Memory T-Cell Responses to Tetanus Vaccine

Abstract

Background

Human immunodeficiency virus (HIV)-exposed, uninfected (HEU) infants experience high rates of infectious morbidity. We hypothesized that early cytomegalovirus (CMV) infection was associated with increased hospitalization rates and decreased vaccine responses in HEU compared with HIV-unexposed (HUU) infants.

Methods

Among infants enrolled in the Tshipidi study in Botswana, we determined CMV infection status by 6 months of age and compared hospitalization rates and responses to tetanus and Bacille Calmette-Guérin vaccines among HEU and HUU vaccinees.

Results

Fifteen of 226 (6.6%) HEU infants and 17 (19.3%) of 88 HUU infants were CMV-infected by 6 months. The HEU infants were approximately 3 times as likely to be hospitalized compared with HUU infants (P = .02). The HEU peripheral blood cells produced less interleukin (IL)-2 (P = .004), but similar amounts of interferon-γ, after stimulation with tetanus toxoid. Antitetanus immunoglobulin G titers were similar between groups. Cellular responses to purified protein derivative stimulation did not differ between groups. Maternal receipt of 3-drug antiretroviral therapy compared with zidovudine was associated with increased IL-2 expression after tetanus toxoid stimulation. The infants’ CMV infection status was not associated with clinical or vaccine response outcomes.

Conclusions

We observed that increased rates of hospitalization and decreased memory T-cell responses to tetanus vaccine were associated with HIV exposure and incomplete treatment of maternal HIV infection, but not early CMV infection.

We found increased rates of infection-related hospitalization and decreased cellular responses to the tetanus vaccine among HIV-exposed, uninfected compared with HIV-unexposed infants. Incomplete treatment of maternal HIV, but not early infant CMV infection, was associated with decreased vaccine responses.

Antiretroviral therapy (ART) has reduced rates of perinatal human immunodeficiency virus (HIV) transmission to <1%, leading to a surge in the global population of HIV-exposed, uninfected (HEU) infants [1, 2]. The HEU infants are more susceptible to infections, leading to elevated rates of morbidity and mortality in this population [1, 3]. In Botswana, HEU infants represent <30% of all newborns but account for approximately 50% of under-2 mortality in the country [4].

Multifaceted immune abnormalities have been identified in HEU infants that may account for their increased susceptibility to infection, including decreased transfer of maternal antibodies, increased T-cell activation and senescence, and innate immune dysfunctions [3]. Vaccine responses provide a good model for evaluating the immune status of the host. Tetanus toxoid is an alum-adjuvanted vaccine that induces both humoral and cellular immune responses. Most studies suggest that African HEU infants mount appropriate antibody responses to tetanus vaccine [5–7]; however, 2 studies from Brazil identified lower antitetanus antibody titers in vaccinated HEU compared with HIV-unexposed (HUU) infants [8, 9]. Cellular responses to tetanus vaccine appear consistently impaired in HEU infants, with decreased cellular expression of interferon (IFN)γ, interleukin (IL)-2, and tumor necrosis factor (TNF)α [10, 11]. Bacille Calmette-Guérin (BCG) is a live-attenuated vaccine administered at birth in countries where tuberculosis is endemic. Studies of HEU infants’ cellular responses to BCG have reported variable results according to country, laboratory assay, and infant age. Some studies showed that HEU infants had decreased T-cell proliferation [12, 13], IFNγ and TNFα production [12, 14, 15], and polyfunctional cytokine expression [16] compared with HUU infants, whereas other studies showed similar responses [10, 11, 17, 18]. These studies, taken together, demonstrate that cellular immune defects may play a greater role than humoral defects in the altered immune response to pathogens in HEU infants.

The mechanisms that trigger immune dysregulation in HEU infants are not understood, but early cytomegalovirus (CMV) infection may contribute. Early infancy represents a critical period of immune development, and CMV infection during this time is associated with multiple immune cell perturbations, including an expansion of activated and apoptosis-vulnerable T-cells [19–21] and redistribution of natural killer cell subsets [22]. In addition, CMV has adapted several mechanisms to suppress host immune responses, including downregulation of human leukocyte antigen molecules, and expression of a viral homolog of IL-10, which may alter the immune response to heterologous antigens [23–25]. In adults, CMV infection is associated with decreased cellular and humoral responsiveness to the influenza and pertussis vaccines [26, 27]. In infants, early CMV infection is associated with decreased cellular IFNγ production to measles vaccine [28] but no change in humoral responses to measles, tetanus, Haemophilus influenzae, meningococcus, or oral polio vaccines [28–30]. We hypothesized that HEU infants, and particularly HEU infants with early CMV infection, experience higher hospitalization rates and decreased cellular responses to BCG and tetanus vaccines compared with HUU infants.

METHODS

Study Population

This study used existing samples and data from the Tshipidi study, a prospective observational cohort study that enrolled pregnant women and their infants in Botswana between 2010 and 2012 [31]. Women ≥18 years of age (454 HIV-infected and 458 HIV-uninfected) were enrolled during pregnancy (88%) or within 7 days of delivery (12%). Women living with HIV (WLHIV) with a CD4 count ≤350 cells/mL or World Health Organization stage 3 or 4 disease were eligible for 3-drug ART; all other WHIV received prophylaxis with zidovudine (ZDV) during pregnancy and single-dose nevirapine (NVP) during labor and delivery, consistent with local guidelines at the time. The HEU infants received prophylaxis with a single dose of NVP and 1 month of ZDV. During the study period, the Botswana HIV program guidelines promoted formula feeding for HEU and provided free infant formula to WHIV. Mothers and their infants were observed for 24 months postpartum to evaluate the effects of in utero exposure to HIV and ART on infant outcomes. Illness and hospitalization data were collected through parental report and medical chart review. Human immunodeficiency virus-1 deoxyribonucleic acid (DNA) polymerase chain reaction (PCR) was performed up to 12 months, and HIV-1 enzyme-linked immunosorbent assay (ELISA) was performed at 18 months of age. Infant plasma, buffy coat, and/or viable peripheral blood mononuclear cells (PBMCs) samples were stored at birth, 1, 6, and 18 months. Infants were eligible for inclusion in this substudy if they were HEU or HUU (not HIV-infected) and had at least 2 stored samples from 2 different study visits available for analysis.

All mothers provided written informed consent. Investigation of the role of CMV in child outcomes was prespecified in the protocol. The Tshipidi study was approved by the institutional review boards (IRBs) of the Botswana Health Research Development Committee and the Harvard TH Chan School of Public Health. This substudy of deidentified samples and data was approved by the Colorado Multiple IRB.

Immunization Schedules

According to the Botswana Immunization Schedule, infants receive BCG vaccine at birth and a tetanus-containing vaccine at 2, 3, 4, and 18 months of age. Pregnant women receive tetanus boosters during each pregnancy.

Determination of Early Cytomegalovirus Infection

We defined early CMV infection as detectable CMV DNA in a blood sample and/or evidence of an immune response to CMV in the first 6 months of life. Because not all sample types were available from every visit, infants were considered CMV-infected or -uninfected as per the criteria in Table 1, or CMV-indeterminate. Deoxyribonucleic acid was extracted from buffy coat, PBMCs, and/or plasma samples using the QIAamp DNA Blood Mini Kit (QIAGEN), and CMV DNA was quantified using the LightCycler-FastStart DNA Master Hybridization Probes kit (Roche Diagnostics) as previously described [32]. Qualitative anti-CMV immunoglobulin (Ig)M antibodies and quantitative IgG antibodies were measured in plasma with ELISA (Abcam, Diamedix). Based upon known CMV epidemiology in sub-Saharan Africa [33], we expected that almost all infants would have detectable anti-CMV IgG antibodies at birth due to transplacental transfer, and that the IgG titer would steadily decline over time as maternal antibody wanes [34]; therefore, we defined seroconversion to CMV as an IgG titer that was greater than or equal to any previous titer.

Table 1.

Criteria for Early CMV Infection^a

Infected	Uninfected
CMV DNA in any sample type ≤6 months of age	Undetectable CMV DNA in samples of any type from birth AND/OR 1 month, AND 6 months
CMV IgM in plasma ≤6 months of age (in the presence of any level of CMV IgG)	CMV DNA in samples of any type at 18 months, AND undetectable CMV DNA at 6 months
CMV IgG titer at 1 month or 6 months ≥IgG titer at birth	Undetectable CMV IgG in plasma at 6 months and/or 18 months
CMV IgG titer at 6 months ≥IgG titer at 1 month	CMV IgM in plasma at 18 months
	CMV IgG titer at 18 months ≥CMV IgG titer at 6 months

Abbreviations: CMV, cytomegalovirus; DNA, deoxyribonucleic acid; Ig, immunoglobulin.

^aEarly CMV infection is defined as acquisition before 6 months of age. See rationale for criteria in Supplementary Table 1.

Vaccine Responses

Antitetanus IgG antibodies were measured in plasma with quantitative ELISA (IBL America). T-cell responses to tetanus and purified protein derivative (PPD) were measured using dual-color IFNγ/IL-2 FluoroSpot (Mabtech). The PBMCs for the FluoroSpot assay were thawed and washed twice with Roswell Park Memorial Institute (RPMI) 1640 medium containing L-glutamine (Gibco) with 10% fetal bovine serum (SAFC Biosciences) and 2 μL/mL Benzonase (Novagen), then resuspended in RPMI 1640 containing L-glutamine supplemented with 10% human AB serum (GemCell) and 1% penicillin-streptomycin (Gibco). Cell counts and viability were obtained using a Guava easyCyte instrument (MilliporeSigma). Samples with a minimum of 70% viability and 600 000 viable cells were rested in complete media at 37°C in a humidified 5% CO₂ incubator overnight before stimulation. Peripheral blood mononuclear cells were added to precoated FluoroSpot 96-well plates at a concentration of 250 000 cells per well in 100 µL complete media with PPD of Mycobacterium tuberculosis (1 µg/mL) (Statens Serum Institut), tetanus toxoid (2.5 µg/mL) (MilliporeSigma), phytohemagglutinin ([PHA] 5 µg/mL) (MilliporeSigma), or media control. Cells were incubated for 48 hours at 37°C in a humidified 5% CO₂ incubator, then IFNγ- and IL-2-producing cells were detected using monoclonal antibodies, and spot-forming units (SFU) were counted using a Cellular Technology Limited ImmunoSpot analyzer. Results are expressed as SFU per 1 million PBMCs, after subtraction of the negative control count from the count for each stimulated condition.

Data Analysis

The goal of this substudy was to determine the association between the primary exposures of interest (perinatal HIV exposure and early CMV infection) and the outcomes of infant hospitalization and vaccine responses. Demographic characteristics were compared between HEU and HUU infants using Fisher’s exact or χ ² tests for categorical variables and Mann-Whitney test for continuous variables. Fisher’s exact test and/or logistic regression was used to determine the association between HIV-exposure status, early CMV infection, and hospitalization rates. For logistic regression models, potential covariates included sex, gestational age at birth, and site of enrollment (urban vs rural). Mann-Whitney tests, t tests, and/or linear regression models were used to determine the association between HIV exposure status, early CMV infection, and infant vaccine responses. For linear regression models, potential covariates included sex, tetanus IgG titer at birth, and maternal receipt of a tetanus-containing vaccine during pregnancy. Among HEU infants, a separate linear regression model was used to determine the association between maternal HIV characteristics (viral load, CD4 count, and receipt of ART versus ZDV) and infant vaccine responses. All analyses were 2-sided, and P < .05 defined statistically significant differences. Statistical analyses were completed using R version 3.4.1 and GraphPad Prism version 8.0.1.

RESULTS

Cohort Characteristics

A total of 226 HEU and 88 HUU infants met criteria for inclusion in this substudy, including 3 sets of twins. Demographic characteristics of the mothers and infants are reported in Table 2 and Supplementary Table 2. Mothers of HEU infants were more likely to be enrolled at an urban site, be older at enrollment, have fewer years of education, and be employed. Among WHIV, the median viral load and CD4 at entry were 1950 copies/mL and 429 cells/mm³, respectively. The majority of WHIV (84%) received ZDV with single-dose NVP as perinatal prophylaxis. Rates of prematurity and low birth weight did not differ between groups. The HEU infants and their mothers were less likely to receive vaccines according to the recommended schedule. As per study-contemporaneous Botswana HIV program guidelines, 96% of HEU infants were fed formula, whereas 100% of HUU infants were breastfed.

Table 2.

Demographic characteristics of infants and their mothers^a.

Characteristic	HEU n=226	HUU n=88	p-value
Enrollment site
Rural	95 (42%)	53 (60%)	0.006
Urban	131 (58%)	35 (40%)
Maternal receipt of tetanus-containing vaccine during pregnancy	136 (60%)	72 (82%)	0.0003
Maternal antiretrovirals in pregnancy
ZDV alone	189 (84%)	NA	NA
cART	37 (16%)	NA
Maternal baseline viral load at entry in copies/mL (median, range)	1950 (<40 – 160,000)	NA	NA
Maternal baseline CD4 count at entry in cells/mm3 (median, range)	429 (48 – 1349)	NA	NA
Infant sex, number of female	113 (50%)	49 (56%)	0.38
Gestational age at delivery in weeks (median, range)	40 (28–45)	40 (31–45)	0.52
Premature delivery	33 (15%)	11 (13%)	0.71
Birthweight in kg (median, range)	3.1 (1.6–5.7)	3.3 (2.0–4.8)	0.003
Birthweight categorical
>2.5 kg (normal BW)	190 (84%)	79 (91%)	0.15
1.5 – 2.5 kg (LBW)	36 (16%)	8 (9%)
Missing	0	1 (1%)
Feeding method
Breastfed	10 (4%)	88 (100%)	<0.0001
Formula fed	216 (96%)	0
Infant zidovudine prophylaxis	225 (99%)	NA	NA
Infant single dose nevirapine prophylaxis	213 (94%)	NA	NA
Infant receipt of BCG vaccine	170 (75%)	76 (86%)	0.03
Infant receipt of ≥3 doses tetanus vaccine	158 (70%)	81 (92%)	<0.0001

^aDemographic characteristics were compared using Fisher’s exact test or Chi-Square test for categorical variables and Mann-Whitney test for continuous variables.

BCG, Bacille Calmette-Guérin; BW, birthweight; cART, combination antiretroviral therapy; HEU, HIV-exposed, uninfected; HUU, HIV-unexposed, uninfected; LBW, low birthweight; ZDV, zidovudine

Rates of Early Cytomegalovirus Infection

Using the criteria established in Table 1, we determined that among 226 HEU infants, 15 were CMV-infected in the first 6 months of life, 118 were CMV-uninfected, and 93 were CMV-indeterminate (Supplementary Table 3). Among 88 HUU infants, 17 were CMV-infected in the first 6 months of life, 8 were CMV-uninfected, and 63 were CMV-indeterminate. Three infants had detectable CMV DNA at birth representing congenital infection; all 3 were HEU infants.

Association of Hospitalization With Human Immunodeficiency Virus Exposure and Cytomegalovirus Infection

Forty-five infants in this substudy were hospitalized at least once by age 24 months. Forty-two hospitalizations occurred among 39 HEU infants, and 6 hospitalizations occurred among 6 HUU infants (Figure 1). The odds ratio of hospitalization for HEU compared with HUU infants was 2.85 (95% confidence interval [CI], 1.17–6.57). Among HEU infants, 41% of hospitalizations occurred by 6 months of age, 28% occurred between 6 and 12 months, and 31% occurred between 12 and 24 months. Among HUU infants, 16% of hospitalizations occurred by 6 months of age, 33% occurred between 6 and 12 months, and 50% occurred between 12 and 24 months. The indications for hospitalization are presented in Table 3. Most hospitalizations were for an infectious indication (83% and 66% among HEU and HUU infants, respectively). Unadjusted logistic regression models did not find any association between hospitalization and any of the following covariates: early CMV infection, sex, gestational age at birth, site of enrollment.

Figure 1.

Hospitalization rates by 24 months among human immunodeficiency virus (HIV)-exposed, uninfected (HEU) and HIV-unexposed (HUU) infants. Proportions of children experiencing at least 1 hospitalization by age 24 months; rates are compared between HEU (n = 226) and HUU (n = 88) infants using Fisher’s exact test.

Table 3.

Indication for Hospitalization

Indication	HEU	HUU
Gastroenteritis with dehydration	20 (48%)	3 (50%)
Lower respiratory tract infection	13 (31%)	0
Neonatal sepsis	2 (5%)	1 (17%)
Chemical ingestion	3 (7%)	0
Meconium aspiration	2 (5%)	0
Eye disorder	1 (2%)	0
Fever	1 (2%)	0
Orthopedic fracture	0	1 (17%)
Failure to thrive	0	1 (17%)
Total	42	6

Tetanus Antibody Titers

At birth, the geometric mean titer (GMT) of antitetanus IgG antibody was significantly lower among HEU compared with HUU infants (2.29 vs 3.79 IU/mL, P = .009) (Figure 2A). In a multivariable analysis controlling for maternal vaccination with a tetanus-containing vaccine, HIV exposure remained significantly associated with low infant tetanus IgG titers at birth (P = .006).

Figure 2.

Child tetanus immunoglobulin (Ig)G antibody titers at birth and 18 months, by human immunodeficiency virus (HIV) exposure status. Antitetanus IgG antibodies were measured in plasma. (A) The geometric mean titer (GMT) is compared between HIV-exposed, uninfected (HEU) (n = 30) and HIV-unexposed (HUU) (n = 54) infants at birth. (B) The GMT is compared between HEU and HUU infants who received 3 (n = 19 and n = 26, respectively) or 4 (n = 10, n = 14, respectively) doses of tetanus vaccine at 18 months. t tests were used to compare groups, and P values are shown where P < .05. Error bars indicate 95% confidence interval (CI).

Analyses of tetanus IgG titers at 6 and 18 months of age were restricted to infants who had received 3 or 4 doses of tetanus vaccine. Among 6-month-old HEU infants who received 3 doses of vaccine, the GMT was 2.05 IU/mL (95% CI, 1.35–3.11). Too few HUU infant plasma samples were available from the 6-month visit to allow a comparison with HEU tetanus titers. Among 18-month-old infants who received 3 doses of tetanus vaccine, the GMT was 0.54 IU/mL (95% CI, 0.26–1.12) among HEU and 0.65 (95% CI, 0.42–1.00) among HUU infants. Among 18-month-old infants who received 4 doses of vaccine, the GMT was 1.88 (95% CI, 0.58–6.16) among HEU and 2.21 (95% CI, 0.83–5.85) among HUU infants. There were no significant differences in the tetanus GMT between HEU and HUU infants at 18 months of age (Figure 2B). All 6- and 18-month-old HEU and HUU infants who received at least 3 doses of tetanus vaccine had a titer above the threshold of protection (0.1 IU/mL). Unadjusted linear regression models did not find any association between tetanus IgG titer at 18 months and any of the following covariates: early CMV infection, sex, tetanus IgG titer at birth, maternal immunization status.

Cellular Vaccine Responses

Cellular responses to tetanus toxoid and PPD were measured in PBMCs from 6-month-old infants who had received 3 doses of tetanus vaccine and/or 1 dose of BCG, respectively. After stimulation with tetanus toxoid, IL-2 production was significantly lower in HEU compared with HUU PBMCs (median 6 [95% CI, 2–10] vs 17 [95% CI, 12–26] SFU/10⁶ PBMC; P = .004) (Figure 3). Interferon-γ production after PBMC stimulation with tetanus toxoid was generally low in HEU and HUU PBMCs and did not appreciably differ between groups (median 0 [95% CI, 0–2] vs 0 [95% CI, 0–6] SFU/10⁶ PBMC). Interleukin-2 and IFNγ responses to PPD ex vivo stimulation were similar between groups (IL-2 median 10 [95% CI, 6–16] vs 14 [95% CI, 6–32] SFU/10⁶ PBMC; IFNγ median 8 [95% CI, 4–14] vs 12 [95% CI, 4–20] SFU/10⁶ PBMC). All HEU and HUU PBMCs demonstrated robust IL-2 and IFNγ production in response to PHA (positive control) stimulation.

Figure 3.

Child cellular responses to vaccines at 6 months of age, by human immunodeficiency virus (HIV) exposure status. The proportion of peripheral blood mononuclear cells (PBMCs) expressing (A) interleukin (IL)-2 and (B) interferon (IFN)γ after tetanus toxoid stimulation are compared between HIV-exposed, uninfected (HEU) (n = 71) and HIV-unexposed (HUU) (n = 34) infants who received 3 doses of tetanus vaccine at 6 months. The proportion of PBMCs expressing (C) IL-2 and (D) IFNγ after purified protein derivative (PPD) stimulation are compared between HEU (n = 69) and HUU (n = 27) infants who received 1 dose of Bacille Calmette-Guérin vaccine at 6 months. Mann-Whitney tests were used to compare groups, and P values are shown where P < .05. Error bars indicate median and interquartile range.

Unadjusted linear regression models did not find any association between cellular responses to tetanus toxoid or PPD at 6 months and any of the following covariates: early CMV infection, sex, tetanus IgG titer at birth, and maternal immunization status. In a separate model of HEU infants, maternal receipt of 3-drug ART rather than ZDV was associated with an increase in IL-2 production after tetanus toxoid stimulation (P = .049) but no change in IFNγ responses to tetanus toxoid or IL-2 or IFNγ responses to PPD. There were no associations between maternal viral load or CD4 count at study entry and infant cellular responses to tetanus toxoid or PPD.

DISCUSSION

In this cohort from Botswana, we identified an approximately 3-fold increased rate of hospitalization among HEU compared with HUU infants. Most hospitalizations among HEU infants occurred in the first 6 months of life and were attributed to gastroenteritis or lower respiratory tract infection. Our observations are consistent with other studies in sub-Saharan Africa that described increased infection-related morbidity among HEU infants [1, 3]. Notably, the majority of HEU infants in this study did not breastfeed. However, a previous study in Botswana demonstrated similar rates of pneumonia and diarrheal disease among HEU infants randomized to breastfeed versus formula feed [35], indicating that factors other than feeding method likely contribute to increased morbidity and mortality in this population. We did not identify any association between early CMV infection and hospitalization in HEU infants, which is consistent with previous studies [36], although our study may have been powered to detect only large differences.

We found that PBMCs from HEU compared with HUU infants produced less IL-2, but similar amounts of IFNγ, in response to tetanus toxoid stimulation. The FluoroSpot assay detects almost exclusively CD4 T-cell responses after PBMCs are stimulated with inactivated antigens [37]. The assay has been optimized to detect both memory responses, represented by IL-2, and T-helper 1 (Th1) effector responses, represented by IFNγ. Lower production of tetanus-specific IL-2 may indicate that HEU infants have a defect in generating memory T-cell responses, which may be a result of impaired antigen presentation and/or increased regulatory T-cell activity [11, 38]. Alternate techniques including flow cytometry may better discriminate memory from effector responses and identify the mechanism of decreased IL-2 production by HEU T-cells.

We did not identify differences between HEU and HUU infants’ PBMCs in IL-2 or IFNγ expression after PPD stimulation. Bacille Calmette-Guérin is a live vaccine that induces robust immune responses through prolonged antigen exposure and the simultaneous activation of multiple pathogen recognition receptors [39]; thus, BCG is one of the few vaccines that produces adequate Th1 responses in neonates [40, 41]. It is possible that BCG’s ability to induce strong immune responses overcomes the cell-mediated immune defects present in HEU infants. Our findings confirm the results of a previous study that identified similar T-cell responses to BCG but decreased IL-2 production to tetanus toxoid among a smaller cohort of HEU versus HUU infants [11].

We found that although immunized HEU infants’ cellular responses to tetanus toxoid were decreased compared with HUU infants, antibody titers were similar. This finding is consistent with previous studies [5–7, 10, 11] and supports the hypothesis that cell-mediated immune defects play a greater role than humoral immune defects in the increased susceptibility to infection among HEU infants. Although some studies have concluded that breastfeeding enhances serologic vaccine responses [42], the HEU and HUU infants in this study produced similar antitetanus antibody titers despite the fact that fewer than 5% of HEU and 100% of HUU infants breastfed. An explanation previously presented for the robust humoral responses observed in HEU infants is the reduced interference of transplacentally acquired maternal antibody with infant B-cell responses to vaccine antigens [6, 7]. Our study confirmed that antibody titers to tetanus toxoid were lower among HEU infants at birth, but we did not identify any association between tetanus titers at birth and subsequent humoral or cellular responses to tetanus vaccine.

We did not find any association between early CMV infection and vaccine responses. Previous studies have identified associations between early CMV infection in HEU infants and elevated levels of soluble inflammatory markers or increased CD8 T-cell activation [21, 43]. With respect to vaccine responses, one study identified an association between early CMV infection and decreased cellular responses to measles vaccine [28], but the effect of CMV on cellular responses to tetanus and BCG vaccines has not previously been described. The proportion of HEU infants that we identified with early CMV infection was small (<7%); this is likely because of low rates of CMV transmission in this nonbreastfed population, as well as variable sample availability limiting the ability to determine CMV status for all infants. It is possible that our small sample size precluded the ability to identify an association between early CMV infection and adverse outcomes. Alternatively, early CMV infection may not be an important driver of immune dysregulation in HEU infants. The role that CMV plays in the immune development of HEU infants requires further exploration in larger cohorts.

Another hypothesis regarding the trigger for immune dysregulation in this population is that exposure to the inflammatory in utero environment associated with maternal HIV imprints the developing fetal immune system [44]. We found that maternal receipt of ART compared with ZDV was associated with increased IL-2 production by HEU infants’ PBMCs after tetanus toxoid stimulation. This observation underscores the importance of attaining virologic suppression during pregnancy. It is possible that the immune reconstitution and decreased inflammation associated with virologic suppression provides a healthier environment for fetal immune development. Our findings are in alignment with previous studies that showed lower risk of severe infections in HEU infants born to mothers with higher CD4 cell counts or lower viral load at delivery and who initiated ART before pregnancy [45–47]. We did not identify an association between maternal CD4 count nor viral load and infant vaccine responses, but these maternal variables were measured only at enrollment in our study. Of note, breast milk plays an important role in promoting early immune development and establishing a healthy neonatal gut microbiome [48, 49]; therefore, it is possible that replacement feeding may have contributed to dysfunctional immune development in the HEU infants in this study. Because breastfeeding is discouraged among WHIV in high-income countries, replacement feeding will need to be considered in the context of the many factors that may contribute to altered immune responses in this population.

Ours is one of the few studies to examine the role of early CMV infection in clinical outcomes and vaccine responses in HEU infants. Our study was limited by sample availability, which resulted in a relatively large number of infants with indeterminate CMV status and the inability to compare tetanus antibody responses at 6 months between HEU and HUU infants. We used multiple methods to determine CMV status, because previous studies have shown that diagnostic sensitivity in infants is improved by the use of both PCR and serologic testing [50]. This testing strategy was able to partially, but not completely, overcome our limited sample availability. Strengths of our study include that we were able to determine clinical outcomes and both cellular and humoral vaccine responses within the same cohort of HEU infants. We used an HUU infant control group that was relatively well matched for location, socioeconomic status, and birth outcomes, minimizing differences attributable to genetic or environmental factors. However, it is important to note that HUU infants were breastfed, whereas a majority of HEU infants were not.

CONCLUSIONS

In conclusion, we found increased rates of infection-related hospitalization and decreased cellular, but not antibody, responses to the tetanus vaccine, among HEU compared with HUU infants, suggesting that defects in cellular immune responses may underlie the increased susceptibility to infections in HEU. In addition, we did not identify a relationship between early CMV infection and any adverse clinical outcomes or vaccine responses, but we did find an association between maternal receipt of ART and improved HEU infants’ cellular responses to tetanus vaccine. Further investigation into other potential drivers of HEU infant immune dysregulation is needed to develop interventions that can decrease the rates of infectious morbidity in this population.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Financial support. This work was funded by the National Institutes of Health (Grant Numbers R01 MH087344 and K24 AI131928 to S. L.); Thrasher Research Fund (Grant Number AWD 164515; to C. S.); and the Children’s Hospital Colorado Research Institute.

Potential conflicts of interest. A. W. receives research grant support from Merck and GlaxoSmithKline. 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.

Presented in part: ID Week, October 2017, San Diego, CA; ID Week, October 2018, San Francisco, CA.