Reinfection With Cytomegalovirus During Pregnancy: A Prospective Cohort Study in Canada

ABSTRACT

Congenital cytomegalovirus infections (cCMV) are an important cause of childhood neurodevelopmental deficits. Most cCMV are the result of maternal non‐primary infections during pregnancy, which can be due to reactivation or reinfection. To identify the rate of CMV reinfection during pregnancy and its risk factors. We performed a secondary analysis of CMV seropositive participants from two prospective cohort studies in Quebec, Canada. Antibody responses to four strain‐specific CMV epitopes located in glycoproteins B and H were measured by enzyme‐linked immunosorbent assay. CMV reinfection was defined as the appearance of an antibody response to a new epitope in the third compared to the first trimester. Risk factors for reinfection were assessed. Among 1614 participants, CMV reinfection was identified in 2.7% of participants, representing an incidence of 54.99 per 1000 person‐years at risk (95% confidence interval 39.95–73.82). Age, marital status, household income, continent of birth or ethnicity were not associated with reinfection during pregnancy. The incidence of CMV reinfection during pregnancy is like what has been reported for primary infection in Quebec. A greater understanding of the patterns of reinfection is needed to inform strategies to reduce the burden of disease from cCMV.

1. Introduction

Cytomegalovirus humanbeta5 (CMV), member of the Orthoherpesviridae family and Cytomegalovirus genus [1] is the most common cause of congenital infection [2]. It is a major cause of permanent disability, especially sensorineural hearing loss [3]. The birth prevalence of congenital CMV infections (cCMV) is 0.64%, with 11% of infants with cCMV being symptomatic at birth [4]. The risk of cCMV depends on prior maternal immunity, timing of infection during pregnancy, and CMV population seroprevalence [5]. cCMV can be the result of either primary infection during pregnancy, or “non‐primary” infection, defined by serologic evidence of maternal CMV infection that was acquired before pregnancy [6]. It is estimated that in the United States, 75% of cCMV are the results of the latter [7]. Mechanisms of “non‐primary” infection include reactivation of a latent CMV strain or reinfection with an exogenous strain during pregnancy [8].

The frequencies of reinfection versus reactivation, as well as their relative contributions to cCMV, are poorly characterized [7]. Importantly, severe sequelae from cCMV occur both in the context of maternal primary and non‐primary infections [9]. The main objective of this study was to describe the rate of CMV reinfection during pregnancy in Quebec, Canada. Secondary objectives were to describe CMV strain‐specific serological profiles in the first trimester and to explore risk factors associated with CMV reinfection during pregnancy.

2. Materials and Methods

We conducted a secondary analysis of two prospective cohort studies with banked data and specimens. The Grossesse en santé (GES) cohort was conducted in Québec City (Canada) between 2005 and 2010 at the Centre Hospitalier Universitaire (CHU) de Québec‐Université Laval to study complications of pregnancy [10, 11]. A total of 7855 pregnant women were recruited at their first antenatal visit. Eligibility criteria included age at least 18 years, gestational age at least 10 weeks, and no hepatic or renal disease. Pregnancy terminations, miscarriages or intrauterine fetal demise before 24 weeks were excluded. Among excluded cases with presentations compatible with cCMV and available whole cord blood, there was no cases of cCMV (negative CMV PCR on whole cord blood). The Design, Develop, Discover (3D) cohort study included 2366 women in nine centers in Québec (Canada) between 2010 and 2012 to address the determinants of adverse birth outcomes [12]. Eligibility criteria included ages 18–47 years and ability to communicate in French or English. Exclusion criteria included intravenous drug use, severe illness or life‐threatening conditions, and multiple gestation pregnancies. Biological samples, sociodemographic and clinical information were collected for both cohorts.

For this sub‐study, we selected participants from both cohorts who had available serum specimens from the first trimester (T1), corresponding to the recruitment visit, at delivery (T3), and cord blood samples. Additional inclusion criteria were to be CMV seropositive at T1 without recent primary infection, according to serological testing (CMV IgM, IgG, and IgG avidity). We extracted clinical and sociodemographic characteristics from both cohorts including age, parity, employment status, level of education, marital status, annual household income, country of birth, race/ethnicity, gestational age at birth, infant sex, and birth weight percentiles according to Canadian reference [13].

2.1. Serological Testing

Serological tests to establish prior CMV infection were performed in the context of two previously published studies on CMV seroprevalence in these cohorts [14, 15]. For the GES cohort, baseline maternal seropositivity was established by CMV IgG on T1 samples using Abbott Architect Platform at the CHU‐Sainte‐Justine virology laboratory. If IgG was positive (> 6.0 arbitrary units per milliliter (AU/mL)), IgM was subsequently tested. In the event of positive IgM (> 0.85 signal to cutoff (s/co)), IgG avidity was measured to assess chronicity of infection. An avidity cut‐off of 60% was used to exclude infections within the last 4 months [14]. For the 3D cohort, IgG positivity was determined semi‐quantitatively at T1 by ELISA using Captia CMV IgG kits (Trinity Biotech, USA) and analyzed on a Triturus automated system (Grifols) [15]. As IgM status was not established in the previous study, IgG positive and equivocal serum samples at T1 were tested for CMV IgM and IgG avidity following the same schematic on the Abbott Architect Platform. For CMV IgG, Abbott Architect and Captia Trinity methods have been shown to have a concordance of 82.8% (95% confidence interval (CI) 76.7–87.6) and a Kappa score of 0.65 (95% CI 0.55–0.75) [16].

CMV reinfections were identified by antibody responses to CMV strain‐specific epitopes on glycoproteins B and H (gB, gH) from two laboratory strains (AD169, Towne) measured by enzyme‐linked immunosorbent assay (ELISA) using T1 and T3 serum samples [17, 18]. CMV‐seronegative sera were used as controls. The four polymorphic peptides were expressed in Escherichia coli Rosetta, and purified by affinity column, as described previously [17]. Strain‐specific ELISA was performed on Corning® 96 Well EIA/RIA Assay Microplate (Corning, Somerville, Massachusetts, USA). 50 µL of each antigen (or carbonate buffer alone (blank) were coated overnight at 4°C, then the plates were blocked for 2 h, 37°C, in borate buffer containing 3% goat serum. Serum samples were diluted 1:100 in blocking buffer then 100 µL was added to the coated antigens and incubated 1 h, 37°C. Detection of specific IgG was done using goat F(ab′)2 anti‐human IgG‐HRP (horseradish peroxidase) (Southern Biotech, Birmingham, AL 35209, USA). Revelation was done by adding 50 µL TMB substrate (1‐step Turbo TMB ELISA Substrate Solution (Thermo Fisher Scientific, Waltham, Massachusetts, United‐States) to each well then 50 µL of stop solution (2 M H₂SO₄). Optical density (OD) at 450 nm was then read using a plate reader and blank value of each sample was subtracted from the OD value of each antigen for that same sample. A positive response was defined as an OD value of > mean + 3STD result of seronegative samples for each antigen. The use of four antigens allowed the detection of 16 different CMV serological profiles. Reinfection was defined as the appearance of an antibody response targeting a new CMV epitope at T3 compared to T1. Samples from the GES cohort were tested at the University of Alabama at Birmingham and samples from the 3D cohort were tested at the CHU‐Sainte‐Justine‐Research‐Centre. The same methodology was used at both centers, and similar results were obtained for 14 samples tested in both centers.

cCMV cases were identified by CMV PCR on whole cord blood using the AltoStar CMV PCR Kit 1.5 (Altona, Germany) platform [19, 20].

2.2. Statistical Analyses

Incidence of CMV reinfection during pregnancy was defined as the number of reinfections between T1 and T3 and expressed per 1000 person‐years at risk, with a CI calculated using the Poisson exact method. Responses to each of the four epitopes of the strain‐specific serological assay were evaluated, yielding 16 potential profiles. Descriptive statistics were performed to illustrate these serological profiles at T1, and chi‐square test for independence to contrast them by region of birth and age.

To assess the risk factors of maternal CMV reinfection, the demographic characteristics of participants with and without CMV reinfections were compared using either Fisher exact test or Student t‐test depending on the type of variable. CMV IgM status at T1 was also compared between groups, as a dichotomous variable for the whole cohort (positive, negative) and a continuous variable among participants with positive IgM (levels in s/co). The association between the presence of a specific response at T1 with reinfection was determined using chi‐square test. The variables used in the multivariate logistic regression models were selected from the literature and included age, parity, socioeconomic status, marital status, region of birth and race/ethnicity. Region of births were grouped by income as reported by The World Bank [21]. Variables such as employment status, level of education and household income were tested as proxies of socioeconomic status and models were compared using the Akaike information criterion (AIC). The model with the lowest AIC was presented. Statistical significance was defined as p‐value < 0.05. All data analyses were performed using RStudio version 2023.06.1 + 524.

3. Results

3.1. Study Cohort

The breakdown of eligible participants from the GES and 3D cohorts is illustrated in Figure 1. The demographic characteristics of selected participants are presented in Table 1. Most women were born in North America or Europe and identified as Caucasian, less than 35 years old and multiparous. Most women were employed, had completed collegiate or university‐level education and an annual household income of more than $60 000 CAD. The mean gestational age was 13.3 weeks at T1 and 39.4 weeks at T3.

Figure 1.

Study cohort based on serologies from the first trimester. CMV, cytomegalovirus; GES. Grossesse en santé.

Table 1.

Association between CMV reinfections and sociodemographic factors.

	No reinfection n = 1570 (%)	Reinfection n = 44 (%)	Unadjusted OR 95% CI	Adjusted a OR 95% CI
Age (years)
18–29	614 (39.1)	18 (40.9)	Reference	Reference
30–34	614 (39.1)	16 (36.4)	0.88 [0.42–1.87]	0.56 [0.20–1.45]
35–47	342 (21.8)	9 (20.5)	0.90 [0.35–2.14]	0.81 [0.26–2.22]
Parity
Nulliparous	595 (37.9)	17 (38.6)	Reference	—
Multiparous	974 (62.0)	27 (61.4)	0.97 [0.50–1.91]	—
Employment status
Unemployed	219 (13.9)	4 (7.1)	Reference	—
Employed	1225 (78.0)	40 (90.9)	1.79 [0.64–6.96]	—
Level of education
Secondary or less (including professional)	341 (21.7)	7 (15.9)	Reference	—
Collegial	434 (27.6)	12 (27.3)	1.36 [0.49–4.12]	—
University level	711 (45.3)	25 (56.8)	1.72 [0.71–4.76]	—
Marital status
Single, separated, or divorced	108 (6.9)	3 (6.8)	Reference	Reference
Common‐law partner or married	1384 (88.2)	41 (93.2)	1.01 [0.33–5.45]	1.68 [0.31–31.09]
Annual household income
Less than 39 999$	365 (23.2)	10 (22.7)	Reference	Reference
40 000$–59 999$	263 (16.8)	7 (15.9)	0.97 [0.31–2.85]	1.03 [0.30–3.36]
More than 60 000$	755 (48.1)	24 (54.5)	1.16 [0.53–2.75]	1.16 [0.44–3.34]
Continent of birth
North America and Europe	1135 (72.3)	31 (70.5)	Reference	Reference
South America	119 (7.6)	5 (11.4)	1.53 [0.46–4.08]	0.34 [0.04–2.81]
Asia	80 (5.1)	3 (6.8)	1.37 [0.26–4.53]	0.78 [0.14–4.27]
Africa	163 (10.4)	5 (11.4)	1.14 [0.34–3.01]	0.55 [0.11–2.81]
Race/Ethnicity
Caucasian	760 (48.4)	14 (31.8)	Reference	Reference
Other races/ethnicities	390 (24.8)	15 (34.1)	2.10 [0.93–4.74]	3.35 [0.79–10.73]

Note: Numbers do not add up to 1614 (i.e., number of pregnant women included in this study) because of missing values for some variables.

Abbreviations: CI, confidence interval; CMV, cytomegalovirus; n, number of women in category; OR, odds ratio.

^aAdjusted for: age, marital status, annual household income, continent of birth, and ethnicity.

3.2. Reinfection

Participants with CMV‐specific antibody responses to all four epitopes at T1 (n = 9) were excluded from the reinfection analysis since identification of a new response between T1 and T3 would be impossible. Among the remaining study participants, 2.7% (44/1605) acquired new antibody responses indicating a reinfection event. The incidence rate of reinfection was 54.99 per 1000 person‐years at risk (95%CI 39.95–73.82). Risk factors for CMV reinfection during pregnancy are shown in Table 1. No statistically significant risk factors were identified, though there was a trend towards an association between reinfection and self‐reported race/ethnicity other than Caucasian (aOR 3.35; 95%CI 0.79–10.73) adjusting for age, marital status, annual household income and continents of birth. On the contrary, there was a trend towards a negative association between ages 30‐34 years old and the risk of reinfection compared to ages 18–29 years old (aOR 0.56; 95%CI 0.20–1.45).

Among participants without reinfection, 8.0% (126/1570) had positive IgM at T1, compared to 15.9% (7/44) among participants with reinfection (p = 0.06). Among the IgM positive participants, median IgM index was 1.32 s/co among participants with no reinfection identified, and 1.22 s/co among participants with reinfection (p = 0.46). Based on cord blood PCR, there were two cCMV identified (0.1%); both neonates were born to participants without identified reinfection.

3.3. Strain‐Specific Serologic Profiles

Strain‐specific antibody responses grouped by serological profiles are presented in Table 2. The most frequent profiles were profile 1 (26.1%, no response to all four epitopes) and profile 11 (23.2%, responses to both gB‐Towne and gB‐AD169). Looking at the frequency of antibody response to each epitope of the strain‐specific assay at T1, for AD169‐gH, 473 responses were identified (29%), for Towne‐gH 143 (8%), for AD169‐gB 798 (49%), and for Towne‐gB 784 (49%).

Table 2.

Serological profiles.

Profile	gH AD169	gH Towne	gB AD169	gB Towne	Number of participants n = 1614 (%)
1.					421	(26.1)	421 (26.1)
2.	X				187	(11.6)	437 (27.0)
3.		X			31	(1.9)
4.			X		114	(7.1)
5.				X	105	(6.5)
6.	X	X			3	(0.2)	516 (32.0)
7.	X		X		62	(3.8)
8.	X			X	50	(3.1)
9.		X	X		12	(0.7)
10.		X		X	15	(0.9)
11.			X	X	374	(23.2)
12.	X	X	X		0	(0)	231 (14.3)
13.	X	X		X	4	(0.2)
14.	X		X	X	158	(9.8)
15.		X	X	X	69	(4.3)
16.	X	X	X	X	9	(0.6)	9 (0.6)

The frequency of CMV reinfection according to the number of specific antibody responses at T1 is presented in Figure 2. Women with one or more strain‐specific CMV antibody response at T1 had 62% fewer chances of being identified with a reinfection compared to participants with no strain‐specific antibody response (OR 0.38; 95% CI 0.20–0.73).

Figure 2.

Frequency of CMV reinfection based on number of specific antibody responses at baseline. CMV, cytomegalovirus.

When comparing the response to each of the four specific antibodies at T1 with reinfection, only the identification of AD169‐gB was significantly associated with reinfection (p = 0.009).

The relative frequencies of serological profiles at T1 were compared by region of birth, as depicted in Figure 3. We noted a higher proportion of Profile 1 among women born in Europe and North America (30%) compared to South America, Asia and Africa, which were otherwise comparable (15%) (p < 0.001) (data now shown). We also noted a higher proportion of Profile 1 among participants between 18 and 24 years of age (30%) compared to participants more than 24 years (23%) and the number of antibody responses at baseline increased with increasing age categories (p = 0.01) (data not shown).

Figure 3.

Relative frequency of serologic profiles by region of birth.

4. Discussion

We demonstrated an incidence rate of 54.99 CMV reinfections per 1000 person‐years at risk during pregnancy, evidenced by the identification of new antibody response against at least one of four strain‐specific CMV epitopes at T3 compared to T1 in CMV seropositive women. The most observed serological profile in the cohort was the absence of specific responses to all four epitopes (26.1%). Having no strain‐specific antibody response identified at T1 or having an antibody response to AD169‐gB at T1 were significantly associated with reinfection.

Rates of primary infection during pregnancy for the cohorts used have been previously described. Rates of primary infection were 1.4 and 0.5 per 10 000 person‐days for the 3D and GES cohorts respectively [14, 15]. The incidence rate of reinfection is thus at least as high as the overall rate of primary infection in the same population. Of note, a cohort study in the United States described CMV reinfection in women during the postpartum period. The annualized rate of reinfection was about 10%, similar to the frequency of primary infection in that population [18]. A Brazilian case‐control study reported an annualized rate of reinfection of 9% during pregnancy among controls with no congenitally infected infants [22]. The lower reinfection rate in the present study, along with the relatively low rate of primary infection, likely reflect less overall CMV transmission in the population of Quebec.

Using a population‐based prediction model, the estimated number of neonates with cCMV attributable to non‐primary maternal infections in a population with a seroprevalence of 30% is estimated to be 13 per 10 000 births [23]. We observed two cases of cCMV in our cohort of 1614 participants, consistent with what would be expected. We did not identify an association between reinfection and cCMV as was described in the Brazilian study [22]. This could be explained by the small number of cCMV in our cohort, the lower sensitivity of cord blood PCR, combined with the fact that we were unable to identify reinfection during the first trimester before the recruitment visit [24]. In other words, the two cases of cCMV in our cohort could be the results of reactivation or missed reinfection.

While risk factors for CMV primary infection have been well described [25], few data exist regarding risk factors of CMV reinfection. In an article outlining risk factors of cCMV after nonprimary infection, maternal young age and unemployment were the only two significant risk factors identified [5]. The association with age is consistent with the trend we observed. We did not identify an association between CMV reinfection and known risk factors of CMV primary infection, such as marital status or parity as a proxy of contact with young children [4, 26]. This is perhaps due to low statistical power as we only identified 44 reinfections in our cohort, or because of the secondary usage of sociodemographic data not initially designed to identify such risk factors.

Among mothers of uninfected infants from the Brazilian study, reactivity to at least one CMV epitope was observed in 77% of women [22], which is concordant with our data (74%). In the cohort of seropositive postpartum women, women with one or more specific antibody responses at baseline had 63% decreased risk of CMV reinfection, like the 62% decreased risk observed in our cohort [18]. Serological screening of CMV IgM is known to be a poor predictor of cCMV in the context of primary infection [27]. In our cohort, IgM status and index at T1 did not differ significantly whether reinfection was identified or not and thus does not seem to be a good marker of reinfection between T1 and T3.

Understanding the epidemiology of CMV reinfections during pregnancy has important public health implications. Worldwide, the prevalence of CMV seropositivity is estimated to be of 86% in women of childbearing age and the occurrence of cCMV in children born from CMV seropositive women is a major public health concern [28], For instance, in the United States, 75% of cCMV cases have been attributed to nonprimary infections [7], while in China, the proportion is as high as 98% [29]. Better understanding of the attributable fraction of non‐primary infections caused by reinfection and its risk factors will inform prevention strategy, including vaccine development [8]. Specifically, it is likely that the most relevant immune responses for blocking an incident reinfection with a new viral strain differ from suppressing reactivation of a latent infection. While waiting for the development of CMV vaccines, our results stress the need to provide counseling regarding CMV infection prevention to all women planning a pregnancy, even if they are CMV IgG seropositive.

The differences in strain‐specific antibody responses according to the region of birth and age likely reflect the disproportionately high levels of exposure to CMV in some populations [30]. Individuals who have been exposed to more CMV strains are more likely to have antibody responses that cross‐react with one or more of the epitopes in the strain‐specific ELISA. Furthermore, women with more CMV infections in the past may be expected to encounter more infectious exposures and a greater risk of reinfection compared to those with few or no past infections. The extent to which the acquisition of infections with specific strains provides cross‐protection against other strains is unknown. Only a minority of CMV variable regions show geographic population structure with evidence for African or European clustering, while the majority have similar allele distributions across different continental populations [31]. This may have clinical implication for new immigrants, who may or may not be at higher risk of CMV reinfection at their arrival, as the circulating CMV strain could be different from the ones they have been exposed to in their country of birth.

For next steps, we aim to improve the methodology to identify reinfections by including a broader range of CMV variant antigens. With the ELISA technique used in this study, we were only able to identify antibody responses to four distinct strain‐specific epitopes in gH and gB of CMV laboratory strains AD169 and Towne. Given the variability of circulating CMV strains [32], we almost certainly underestimated the true incidence of CMV reinfection during pregnancy, and in participants with many strains detected in the first trimester, the chances to detect a new one with the actual assay are limited. For future research, combining the use of viral genotyping with a serology‐based approach, for example, could provide internal validity and additional insights into the response to reinfection [33, 34].

4.1. Strengths and Limitations

The strengths of our study are the use of two large prospective cohort studies and biobanks, allowing us to identify CMV reinfection between the first and third trimesters as well as congenital infections, making it the largest cohort study to date on CMV reinfection during pregnancy. The use of strain‐specific ELISA to measure responses to CMV has been described previously as a reliable immunology‐based approach to identify reinfections [18, 22].

This study has some limitations. Importantly, the strain‐specific assay has a lack of sensitivity due to the small number of epitopes it contains, as demonstrated by the fact that the most common profile among CMV seropositive women is the absence of any strain‐specific response. This also means that in participants with more specific antibody responses at T1, the strain‐specific assay is less likely to detect a reinfection. This suggests that we most likely underestimate the true incidence of reinfection as the test is unable to identify at least 25% of all infections. Moreover, because CMV reinfection may occur with a huge diversity of CMV strains, and the strain‐specific ELISA that was used has limited power to detect these strains, our findings may not be generalizable. Furthermore, the sociodemographic questionnaires were not designed to specifically assess risk factors with CMV reinfection. More details on child rearing, hand‐washing and sexual practices could have allowed us to better characterize risk factors. Saliva, urine and vaginal swabs were not available in the cohorts, limiting our ability to confirm the appearance of a new CMV strain by sequencing if shedding was detected. Finally, the small number of participants with CMV reinfections and cCMV in the cohort limits the power of statistical analyses.

5. Conclusions

In conclusion, this study is the largest cohort to this date focusing on CMV reinfection during pregnancy in a population with a relatively low seroprevalence. We described serological profiles at the first trimester, confirming that the cumulative exposure to CMV varies by continent of birth and age. We identified a relatively low incidence of CMV reinfection during pregnancy, though it was at least as high as the rate of primary infections in the same population, contributing to the understanding of the epidemiology of non‐primary CMV infections. For the next steps, a greater sensitivity to measure strain‐specific responses will allow us to better characterize CMV reinfection during pregnancy and the immune responses required to broadly protect against naturally circulating CMV variants.

Author Contributions

Juliette St‐Georges: data curation, formal analysis, investigation, writing–original draft preparation. Safari Joseph Balegamire: data curation, investigation, methodology. Ariane Larouche: formal analysis, investigation, methodology, validation, writing–reviewing and editing. Suresh Boppana: conceptualization, methodology, validation, resources, writing–review and editing. Christian Renaud: conceptualization, methodology, writing–review and editing. Benoît Mâsse: conceptualization, methodology, writing–review and editing. Yves Giguere: resources, writing–review and editing. Jean‐Claude Forest: resources, writing–review and editing. Valerie Lamarre: resources, writing–review and editing. François Audibert: conceptualization, writing–review and editing. Soren Gantt: conceptualization, funding acquisition, methodology, resources, supervision, writing–review and editing. Isabelle Boucoiran: conceptualization, funding acquisition, methodology, project administration, resources, supervision, writing–review and editing.

Ethics Statement

All aspects of the study were approved by the ethics committee of the CHU‐Sainte‐Justine‐Research‐Centre.

Conflicts of Interest

PCR kits were provided by ALTONA diagnostics; Consulting fees and research funding from Moderna, Merck, GSK and Curevo Vaccine (SG); research funding form Moderna and Ferring (IB).

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.