Impact of Maternal Immunity on Congenital Cytomegalovirus Birth Prevalence and Infant Outcomes: A Systematic Review

Congenital cytomegalovirus (cCMV) is the leading non-genetic cause of sensorineural hearing loss (SNHL), and efforts are geared towards prevention through vaccine development. Transmission rates following primary maternal infection occur at rates of 30–40%, however reported placental rates upon non-primary maternal infection is reported to be less than <4%. There is significant debate about whether this reduction in transmission rate is due to pre-existing maternal immunity, which could identify possible immunologic targets for vaccines. To address this question, we performed a systemic review of the literature using Preferred Reporting Items for Systematic Review and Analysis (PRISMA) guidelines. We identified cohort studies in high CMV seroprevalent (>80%) areas or in developing regions that examined a cohort of at least 50 infants for congenital CMV acquisition. We identified 19 articles that met criteria and were further categorized based on pre-conception serology, maternal seroprevalence, or previously known seroprevalence. Birth prevalence rates ranged from 0.4% to 6% (median 1.1%), with the studies reporting on clinical outcome (16/19 studies) noting the majority of infected infants as asymptomatic. We also utilized a recent study that differentiated primary maternal infections from chronic infections in a highly seropositive population to calculate a placental transmission rate in women with pre-existing immunity compared to that of no pre-existing immunity. This work confirms a low cCMV birth prevalence in highly seropositive populations, indicates via a calculated placental transmission rate that the CMV placental transmission rate is lower in non-primary infection than that of primary infection, and reveals gaps in data for further research aiming to identify targets for vaccine development.

Approximately 85-90% of infants born with cCMV are asymptomatic, while only 10% of infants are symptomatic at birth [9,10]. Of those infants who demonstrate CMV-associated sequelae at birth, 40-60% will develop permanent sequelae including cerebral palsy, cognitive impairment, microcephaly and sensorineural hearing loss (SNHL) [7,10,11]. Those who are born asymptomatic are also at risk of developing sequelae; as 6-23% of asymptomatic infants eventually develop SNHL. Thus, cCMV is the who independently screened all available randomized clinical trial, cohort, and case-control studies cited within MEDLINE (via PubMed), EMBASE (via Elsevier), CINAHL (via EBSCO), Scopus (via Elsevier), and ClinicalTrials.gov from inception through 9 October 2018.
The search was conducted by a professional medical librarian and included a mix of keywords and subject headings representing cytomegalovirus, seroprevalence, and the pediatric population, respectively. Reproducible search strategies can be found in the Supplemental Materials Table S2. Editorials, letters, and comments were excluded, as were animal-only studies and studies not in English.
Studies were excluded if they did not clearly meet inclusion criteria based on title and/or abstract review. All disagreements regarding eligibility were resolved through discussion between the reviewers or, if unable to be resolved, by adjudication by a third reviewer. Study selection is presented by flowchart as per PRISMA guidelines ( Figure 1). control studies cited within MEDLINE (via PubMed), EMBASE (via Elsevier), CINAHL (via EBSCO), Scopus (via Elsevier), and ClinicalTrials.gov from inception through 9 October 2018. The search was conducted by a professional medical librarian and included a mix of keywords and subject headings representing cytomegalovirus, seroprevalence, and the pediatric population, respectively. Reproducible search strategies can be found in the Supplemental Materials Table S2. Editorials, letters, and comments were excluded, as were animal-only studies and studies not in English.
Studies were excluded if they did not clearly meet inclusion criteria based on title and/or abstract review. All disagreements regarding eligibility were resolved through discussion between the reviewers or, if unable to be resolved, by adjudication by a third reviewer. Study selection is presented by flowchart as per PRISMA guidelines ( Figure 1). The search identified a total of (n = 609) references that were imported into an EndNote X9 (Clarivate Analytics) library and uploaded into Covidence, a systematic review screening tool. Duplicate citations (n = 280) were automatically identified and removed by Covidence. Two reviewers independently screened a total of (n = 329) references by title and abstract. Of those, (n = 270) references were deemed irrelevant and excluded. Tied references were broken by a third independent reviewer.
A total of (n = 59) articles were assessed in the full-text screening phase for eligibility given the following inclusion criteria: (1) employed a randomized control trial, cohort, or case-control study design; (2) studied a region of established high (>80%) maternal CMV seroprevalence or developing region; (3) screened at least (N = 50) neonates for CMV; and 4) utilized CMV DNA Polymerase Chain Reaction (PCR) of neonatal urine, saliva, blood, or dried blood spot (DBS) samples acquired within the first three weeks of life to confirm diagnosis of CMV infection. Studies that only utilized serological testing, such as detection of anti-CMV specific IgM antibodies via enzyme-linked immunosorbent assay (ELISA), to confirm diagnosis of congenital CMV infection were excluded on the basis that only 20-70% of infected infants present with anti-CMV specific IgM antibodies [25]. A further 40 studies were excluded based on study population, screening method for CMV, outcomes, study design and number of infants screened. A total of (n = 19) studies were ultimately included in the review. Data including birth prevalence rates (number of infants infected with CMV at birth/total number of infants studied), placental transmission rates (number of infants infected/number of mothers who had primary or non-primary CMV infection) symptoms, and symptom rates were extracted into an Excel spreadsheet (Table 1). The search identified a total of (n = 609) references that were imported into an EndNote X9 (Clarivate Analytics) library and uploaded into Covidence, a systematic review screening tool. Duplicate citations (n = 280) were automatically identified and removed by Covidence. Two reviewers independently screened a total of (n = 329) references by title and abstract. Of those, (n = 270) references were deemed irrelevant and excluded. Tied references were broken by a third independent reviewer.
A total of (n = 59) articles were assessed in the full-text screening phase for eligibility given the following inclusion criteria: (1) employed a randomized control trial, cohort, or case-control study design; (2) studied a region of established high (>80%) maternal CMV seroprevalence or developing region; (3) screened at least (N = 50) neonates for CMV; and 4) utilized CMV DNA Polymerase Chain Reaction (PCR) of neonatal urine, saliva, blood, or dried blood spot (DBS) samples acquired within the first three weeks of life to confirm diagnosis of CMV infection. Studies that only utilized serological testing, such as detection of anti-CMV specific IgM antibodies via enzyme-linked immunosorbent assay (ELISA), to confirm diagnosis of congenital CMV infection were excluded on the basis that only 20-70% of infected infants present with anti-CMV specific IgM antibodies [25]. A further 40 studies were excluded based on study population, screening method for CMV, outcomes, study design and number of infants screened. A total of (n = 19) studies were ultimately included in the review. Data including birth prevalence rates (number of infants infected with CMV at birth/total number of infants studied), placental transmission rates (number of infants infected/number of mothers who had primary or non-primary CMV infection) symptoms, and symptom rates were extracted into an Excel spreadsheet (Table 1).

Results
The 19 studies that met our established criteria for studying cCMV transmission in highly seropositive populations were further broken down into three tiers. The Tier 1 studies (n = 3) addressed the chronicity of maternal CMV infection in study participants via assessment of IgG avidity and/or preconception CMV IgG immunity to differentiate primary from chronic maternal CMV infection. All Tier 1 studies completed IgG avidity testing on at least a portion of the pregnant subjects. The IgG avidity testing indicates whether an infection was primary or chronic based on the strength of IgG binding ratio with and without a dissociating reagent [46]. The Tier 2 studies (n = 12) assessed seroprevalence in mothers but did not specifically differentiate between primary or chronic CMV infections. Finally, Tier 3 studies (n = 4) were performed in highly CMV seroprevalent populations, however, did not formally test the seroprevalence of the mothers involved in the study. To correct for this, we utilized seroprevalence data from previous studies in similar regions [44,45]. Across the 19 studies in all three tiers, 386 infants were diagnosed with cCMV infections. The birth prevalence across all three tiers was 0.4 to 6%.

Tier 1 Studies
All Tier 1 studies utilized infant saliva or urine PCR with or without urine culture for confirmation of cCMV infection and diagnosed a majority (>89%) of pregnant subjects with chronic CMV infections based on high avidity IgG binding. In Tier 1 studies, cCMV birth prevalence rates in highly seropositive populations ranged from 0.5% to 1.1% with a median rate of 1% [19,27,47] (Table 1, Figure 2). All studies included IgG avidity testing to confirm chronic maternal infection on at least a subset of the pregnant subjects. Only one study, Mussi-Pinhata et al. in 2018, specifically examined cCMV rates in maternal primary versus chronic CMV infection within a highly seroprevalent maternal population. CMV serostatus was assessed in the first trimester, and the cCMV rates for both seronegative women and seropositive women, as well as the seronegative women with primary CMV acquisition during pregnancy were reported. Notably, there was a reduced cCMV birth prevalence among seropositive women (0.5%) versus preconception seronegative women (birth prevalence 2.8%) [19]. Moreover, there was a high rate of primary CMV acquisition during pregnancy in this population (14%, 5 of 36), demonstrating the frequent maternal CMV exposure in this population, and the observed placental transmission rate of 20% (1 in 5) is similar to what is expected in the setting of primary CMV acquisition. If the maternal CMV re-infection rate for the chronically infected mothers were assumed to be the same as that of the seronegative population (14% of seroimmune mothers potentially re-infected with CMV during pregnancy) producing a birth prevalence of 0.5%, then the calculated placental transmission rate for the chronically infected mothers would be 3.5%.

Results
The 19 studies that met our established criteria for studying cCMV transmission in highly seropositive populations were further broken down into three tiers. The Tier 1 studies (n = 3) addressed the chronicity of maternal CMV infection in study participants via assessment of IgG avidity and/or preconception CMV IgG immunity to differentiate primary from chronic maternal CMV infection. All Tier 1 studies completed IgG avidity testing on at least a portion of the pregnant subjects. The IgG avidity testing indicates whether an infection was primary or chronic based on the strength of IgG binding ratio with and without a dissociating reagent [46]. The Tier 2 studies (n = 12) assessed seroprevalence in mothers but did not specifically differentiate between primary or chronic CMV infections. Finally, Tier 3 studies (n = 4) were performed in highly CMV seroprevalent populations, however, did not formally test the seroprevalence of the mothers involved in the study. To correct for this, we utilized seroprevalence data from previous studies in similar regions [44,45]. Across the 19 studies in all three tiers, 386 infants were diagnosed with cCMV infections. The birth prevalence across all three tiers was 0.4 to 6%.

Tier 1 Studies
All Tier 1 studies utilized infant saliva or urine PCR with or without urine culture for confirmation of cCMV infection and diagnosed a majority (>89%) of pregnant subjects with chronic CMV infections based on high avidity IgG binding. In Tier 1 studies, cCMV birth prevalence rates in highly seropositive populations ranged from 0.5% to 1.1% with a median rate of 1% [19,27,47] (Table 1, Figure 2). All studies included IgG avidity testing to confirm chronic maternal infection on at least a subset of the pregnant subjects. Only one study, Mussi-Pinhata et al. in 2018, specifically examined cCMV rates in maternal primary versus chronic CMV infection within a highly seroprevalent maternal population. CMV serostatus was assessed in the first trimester, and the cCMV rates for both seronegative women and seropositive women, as well as the seronegative women with primary CMV acquisition during pregnancy were reported. Notably, there was a reduced cCMV birth prevalence among seropositive women (0.5%) versus preconception seronegative women (birth prevalence 2.8%) [19]. Moreover, there was a high rate of primary CMV acquisition during pregnancy in this population (14%, 5 of 36), demonstrating the frequent maternal CMV exposure in this population, and the observed placental transmission rate of 20% (1 in 5) is similar to what is expected in the setting of primary CMV acquisition. If the maternal CMV re-infection rate for the chronically infected mothers were assumed to be the same as that of the seronegative population (14% of seroimmune mothers potentially re-infected with CMV during pregnancy) producing a birth prevalence of 0.5%, then the calculated placental transmission rate for the chronically infected mothers would be 3.5%.

Tier 2 Studies
Among the Tier 2 studies, 5 out of 12 utilized dried blood PCR for diagnosis of cCMV, whereas the remaining 7 studies utilized saliva or urine PCR and/or urine culture for diagnosis of cCMV. In these Tier 2 studies, the birth prevalence rate ranged from 0.4% to 3%, with a median of 1.85% [28][29][30][31][32][33][34][35][36][37][38][39]. These studies did not utilize avidity testing to indicate specifically if the maternal infection was recent. The study by Stagno et al. in 1982 had a limitation in its analysis of cCMV, as 79 samples were either contaminated or spilled, reducing the total number of samples (119 of 197, 60%), and possibly resulting in an over reported prevalence of 1.7% [36].

Tier 3 Studies
All 4 studies in Tier 3 utilized urine and saliva PCR for diagnosis of cCMV, and 2 of 4 studies also utilized blood PCR as an alternative method of analyzing cCMV prevalence rates. In Mwaanza et al. four different methods were utilized to detect transmission, including PCR saliva, PCR urine, PCR sera, and IgM sera [42]. There was not significant overlap between the assays with only 3 of 15 samples (20%) testing positive in more than one domain. Among these Tier 3 studies, the birth prevalence rate ranged from 0.4% to 6% with a median rate of 2.15% [40][41][42][43].

Hearing Loss among cCMV-Infected Infants Born to CMV Seropositive Mothers in Highly Seroprevalent Populations
In 37% (7/19) of studies, infants were assessed for hearing loss, a known common sequela of cCMV infection. In Tier 1 studies, hearing loss rates measured after birth ranged from 8.6-9.8%. Mussi-Pinhata et al. (2009) found that 8% (2/27) of infants born to mothers with chronic CMV infections had bilateral hearing loss at follow-up, although they were originally asymptomatic ( Table 2) [27]. In the same study, one infant born to a mother with primary infection had late bilateral hearing loss at 24 months [27]. In Tier 2 studies, hearing loss rates ranged from 0-40% (Table 2). In one study by Barbi et al. (2006), hearing loss was transient and improved by future visits [31]. In another study by Dar et al. (2008), hearing loss was not present until four months in 11% (1/9) of patients and was unilateral in nature [38]. Only one Tier 3 study by Karimian

Prematurity Among cCMV-Infected Infants Born to CMV Seropositive Mothers in Highly Seroprevalent Populations
Two studies identified whether neonates infected with cCMV were born premature and differentiated cCMV prevalence rates for pre-term versus term infants [29,37]. The studies differed on cCMV prevalence rates for the pre-term versus term neonates. In Yamamoto et al. (2001), cCMV was found in 2.1% (6/289) of preterm infants and 1.8% (3/163) of term infants, but the difference in birth prevalence between preterm and term infants was not statistically significant [29]. In this study, all cCMV infants that were clinically symptomatic were born prematurely. In Alkhawaja et al. (2012), cCMV was found in 1.9% (2/84) of preterm infants vs. 6.3% (1/16) of term infants [37], but symptoms were not tracked [29,37].

Discussion
Our systematic review identified 19 studies which evaluated CMV-seropositive mothers and the associated cCMV birth prevalence. Of those, 16 studies reported symptoms of affected infants. Overall, cCMV birth prevalence ranged from 0.4% to 6% [19,26,27,[29][30][31][32][33][34][35][36][37][38][39][40][41][42][43]. Tier 1 studies, which differentiated primary versus chronic CMV maternal infections, had a median birth prevalence rate of 1%. Studies collectively showed that the overwhelming majority of cCMV-infected infants were asymptomatic. Of those infants with symptoms, hearing loss (5.7% of patients, 7/16 studies) followed by abnormal brain imaging (5.4% of patients, 8/16 studies) and jaundice (4.4% of patients, 5/16 studies) were the most common symptoms. Cognitive and motor delays were only reported in 3 cCMV-infected infants within 3 studies. There was insufficient information in this review to make conclusions regarding differential rates of cCMV infections amongst preterm and full term infants. Lanzieri et al. (2014) conducted a systematic review evaluating birth prevalence of CMV in developing countries. In their work, eleven studies reported CMV maternal seroprevalence rates of 84-100% and CMV birth prevalence of 0.6% to 6.1% [23]. Our systematic review found similar birth prevalence, however, we were additionally able to report placental transmission rates for a study that distinguished chronic CMV infections from primary infections. The birth prevalence for this study demonstrated a lower cCMV birth prevalence in maternal chronic CMV infections compared with primary infections (0.5% vs. 2.8% respectively) [19]. The median birth prevalence for all three tiers of studies, including primary and chronic CMV infections, ranged from 1% to 2.1%. The highest median birth prevalence rate was 2.15% in Tier 3, among studies that did not specifically test maternal seroprevalence.
Mothers residing in areas of low maternal CMV seroprevalence acquire CMV during pregnancy at a low rate of 1-4% [21]. The placental transmission rate for primary CMV infection is very high at 30-50% [48]. Other studies show that transmission rates for primary infections can be as high as 78%, with the rate increasing with gestational age, as it can range from 30.1% transmission in the first trimester to 72.2% in the third trimester [49,50]. These high transmission rates in women who are acutely infected are in stark contrast to the low birth prevalence in seropositive populations found from our systematic review, which ranged from 0.4% to 6% [19,26,27,[29][30][31][32][33][34][35][36][37][38][39][40][41][42][43], despite expected high CMV exposure rate in these populations. This range mirrors work by Fowler et al., (2003) in whichupon comparing primary versus non-primary maternal CMV infections they found that 3% of infants were infected by primary maternal infection versus 1% by non-primary maternal infection [34]. The only study within our systematic review that directly compared birth prevalence among primary and chronic maternal CMV infections, performed by Mussi-Pinhata et al., (2018) had similar results [12,19]. Of five pregnant women who seroconverted (14%, 5/36), only one gave birth to an infant with cCMV, resulting in a placental transmission rate of 20% [19]. Our calculated 3.5% placental transmission rate for reinfection of chronically infected mothers, which applies the acquisition rate from the seronegative mothers within the study cohort, is considerably lower than the 20% placental transmission rate for primary infected mothers in this study [19]. The cCMV birth prevalence and calculated transmission rates among seronegative women in a highly CMV seroprevalent population compared to that of their seropositive neighbors is revealing and may demonstrate the protective role of pre-existing maternal immunity.
The risk of early symptomatic cCMV due to primary infection is around 10-15%, but cCMV-infected infants who are initially asymptomatic have a 25% risk of developing sequelae by 24 months. In comparison, the risk of symptomatic cCMV due to non-primary infection is as low as 1%, and the risk of developing late sequelae is approximately 8% [21]. Findings from the studies included in our systematic review were not consistent with these prior findings, as some studies showed upwards of 40% of cCMV-infected infants being symptomatic for those infected following chronic CMV infections [32]. When we focused solely on Tier 1 studies that differentiated primary from chronic maternal infections, symptomatic cCMV-infected infants were as high as 10%. The symptomatic infants were born to mothers with chronic CMV infections. In the Mussi Pinhata et al. study in 2018 comparing primary vs. chronic CMV infections, none of the cCMV-infected infants were symptomatic in either primary or chronic group. Despite the high rates of symptoms for cCMV born to chronically infected mothers compared to previous studies, all studies showed that the majority (55-100%) of cCMV infected infants from chronic CMV infections were asymptomatic. It is unclear why the symptom rate is higher than expected from the prior literature, but may be due to small sample size, variable screening for sequelae of infection, and how long the infants were monitored for symptoms.
Overall, this review demonstrated that maternal immunity to CMV conferred protection as evidenced by median birth prevalence rates of 1.0-2.1% amongincluded studies. This low birth prevalence in a setting of high expected CMV exposure in a seropositive population suggests partial maternal immune protection against cCMV transmission. Current efforts for CMV prevention focus on vaccine development. Future studies must look at maternal T cells and antibody responses in transmitting and non-transmitting seropositive mothers in order to identify potential targets for vaccines to prevent congenital CMV. However, there are several limitations to our systematic review. First, the gold standard for identifying cCMV is viral culture of saliva or urine, which is not always collected in newborns. Instead, dried blood spots are more commonly collected. The utilization of PCR has decreased the length of time for detection. Urine PCR detection has sensitivities ranging from 93-100%, while saliva PCR has sensitivity of approximately 97% [47,51]. In contrast, dried blood PCR has lower sensitivity than that of saliva or urine PCR, making the likelihood of missing cases very high [52]. In our systematic review, 37% (7/19 studies) utilized dried blood PCR, however only 11% (2 of 19 studies) relied solely on that test for diagnosis [30,39]. The studies utilizing dried blood spots as an additional method also used urine, saliva or combination of both. Another limitation of our study is that 21% (4/19) of studies did not assess seroprevalence in the study population [40][41][42][43]. Instead, overall seroprevalence was inferred from similar populations within the same country for the purposes of this review. Importantly, none of the studies assessed for maternal reinfection, which does not provide a full assessment of non-primary infections and indicates the need for studies of placental transmission rates following established maternal re-infection. Another limitation considered is the potential underreporting of hearing loss due to limited infant follow up, as many infants may be diagnosed with hearing loss later in development despite being asymptomatic at birth [9]. Many of the studies in our review assessed hearing loss within the first few weeks of life with minimal long-term follow-up for SNHL, likely decreasing the identified rates of symptomatic SNHL in this patient population versus what was expected.

Conclusions
This study is one of the most comprehensive, structured systematic reviews demonstrating the overall low birth prevalence of cCMV and symptoms in highly seroprevalent populations. Importantly, we identified studies that differentiated primary from chronic maternal infections. Moreover, we highlight one study that reported lower birth prevalence for chronically-infected women than that of seronegative women within the same population, and when applying the infection rate in the seronegative women to the chronically infected women, indicates a lower transmission rate in chronically infected women. Our review reaffirms findings from previous studies and provides further evidence for a protective role of pre-existing maternal immunity against cCMV transmission. Further research should focus on studying placental transmission rates following maternal reinfection, as well as the immune responses in seropositive mothers that influence transmission rates to their infants.
Author Contributions: T.C. helped with conceptualization of the paper, searched the literature, screened references, performed formal analysis and served as primary author of the manuscript. J.M. searched the literature, screened references, performed formal analysis and helped original draft preparation as well as reviewing and editing. S.C. developed the search strategies, contributed to writing the methods section, and reviewed the final manuscript. S.P. helped with conceptualization of the paper, served as independent reference screener for tiebreakers, and helped with writing, reviewing and editing of the manuscript.
Funding: This research received no external funding.