A systematic review and meta‐analysis of the prevalence of human cytomegalovirus shedding in seropositive pregnant women

Abstract The detection of human cytomegalovirus (HCMV) in an individual's bodily fluid by culture techniques or through HCMV DNA detection by polymerase chain reaction, is known as HCMV shedding. Human cytomegalovirus shedding has the potential to transmit HCMV infection, where an individual can become infected with HCMV through contact with the bodily fluid of another individual containing HCMV. Human cytomegalovirus shedding can occur in primary infection and in non‐primary infection for individuals with prior infection (HCMV seropositive). Human cytomegalovirus infection causes few or no symptoms in a pregnant woman, but can cause significant harm to her foetus if congenital CMV (cCMV) infection occurs. The association between HCMV shedding in HCMV seropositive pregnant women and the vertical transmission of HCMV to result in cCMV infection is poorly investigated, challenged by a limited understanding of the distribution of HCMV shedding in HCMV seropositive pregnant women. We systematically reviewed the published literature to describe the prevalence of HCMV shedding in HCMV seropositive women during pregnancy up to delivery. This analysis identified nine studies that met our eligibility criteria. In these studies, the prevalence of HCMV shedding in any bodily fluid of HCMV seropositive women during pregnancy and at delivery ranged from 0% to 42.5%. A meta‐analysis, performed on six of the nine studies with suitable sample sizes, estimated a pooled prevalence of 21.5% [95% CI 12.7%,30.3%]. To our knowledge, this is the first review to systematically search the literature to summarise the prevalence of HCMV shedding in HCMV seropositive pregnant women. These estimates can help in the development of disease burden models and therapeutic or preventative strategies against cCMV infection in the context of non‐primary maternal HCMV infection.


| INTRODUCTION
Following primary infection, human cytomegalovirus (HCMV) can establish life-long latency in an infected individual and may reactivate, known as non-primary HCMV infection. 1 Non-primary HCMV infection can also occur from reinfection with a different HCMV strain. 1 The current standard for diagnosing primary HCMV infection is the detection of HCMV immunoglobulin G (IgG) with or without HCMV immunoglobulin M (IgM) in an individual with previously undetectable HCMV IgG and IgM, known as seroconversion. 2 The diagnosis of non-primary HCMV infection in an individual with prior infection, known as being HCMV seropositive, cannot therefore rely on conventional serological findings alone. 2 The detection of HCMV in an individual's bodily fluid by culture techniques or through HCMV DNA detection by polymerase chain reaction (PCR), is known as HCMV shedding. 3  Congenital CMV infection is the most common congenital infection in the UK and worldwide. 4,5 Up to 25% of children with cCMV infection will have long-term neurodevelopmental impairments, the most common being sensorineural hearing loss. 6,7 Learning disability, behavioural problems, visual impairment, cerebral palsy, epilepsy and autism spectrum disorder have also been associated with cCMV infection. [6][7][8] Children with cCMV infection may require significant input from health and social care services, which can have a large impact on families and is associated with a significant cost to society. 8,9 Due to the challenges of diagnosing non-primary HCMV infection serologically, the detection of HCMV shedding by culture or PCR in HCMV seropositive pregnant women may offer a suitable diagnostic approach. This may be a valuable tool for the evaluation of preventative and therapeutic strategies, such as vaccine development and risk-

| Search strategy
The systematic review protocol, conduct and report, were executed in accordance with the criteria set out by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. 10,11 International prospective register of systematic reviews (PROS-PERO) 12

| Statistical analysis
A narrative synthesis of the included studies, structured around the condition, context and population specified and the quality assessment, was produced. Individual study characteristics structured around the data extraction categories were summarised in a descriptive table. Meta-analysis of the prevalence of HCMV shedding in HCMV seropositive pregnant women, overall, and separately by bodily fluid site where at least two studies reported prevalence data by site that could be extracted, was performed to provide a summary estimate on the global burden of HCMV shedding prevalence in HCMV seropositive pregnant women. The meta-analysis was performed using Microsoft Excel, 14 grouped by random-effects model with 95% CI to allow for between-study variation, and to adjust the weight of each study based on the assumption that observed variability is due to both sampling error and true variability in the population. 13,14 Analysis of heterogeneity across the studies was performed through the calculation of Higgins I 2 (I 2 ) to assess the extent of inconsistency of findings across the studies included in the meta-analysis. 13,14 The minimum sample size for studies to include in the meta-analysis was 94, calculated based on the adequate sample size in prevalence study formula 15 The prevalence estimated from the study with the highest prevalence was used in the prevalence study formula as the expected prevalence was not known. A forest plot with a pooled estimate of the effect sizes 14 (prevalence of HCMV shedding in HCMV seropositive pregnant women overall) was created.

| Systematic review study selection
We conducted a systematic literature search to identify studies detailing the prevalence of HCMV shedding in HCMV seropositive pregnant women. Out of 709 articles generated from our literature search, 395 articles were selected for abstract evaluation following deduplication ( Figure 1). Of the 395 articles, 43 studies were screened for full-text evaluation, and nine studies met the inclusion criteria ( Figure 1). Three studies appeared to have met the inclusion criteria but were excluded: the seropositive status of the pregnant women with HCMV shedding was unclear in two studies, 16,17 and the source of shedding was of a foetal bodily fluid for one study. 18 A description of the nine studies included in the systematic review derived from the data extraction is shown in Table 1. Only data that met the inclusion criteria from included studies are stated, data not meeting the inclusion criteria from these same studies are not stated.

| Design and methods of selected studies for analysis
All studies included in the systematic review employed a cohort study design, with a sample size ranging between 8 and 564 women (Table 1). Across the studies, the bodily fluid sites for HCMV shedding detection were saliva, urine, vaginal secretions, cervical secretions, and blood. Six studies (

| Demographics of individuals in selected studies for analysis
The nine studies in the systematic review represented four continents: three from North America, three from Asia, two from Europe and one from South America. Reporting of population age varied across studies, with mean age reported in four studies (

| Prevalence of human cytomegalovirus shedding in selected studies for analysis
The prevalence of HCMV shedding in pregnant women, at least once in any bodily fluids and at any point during pregnancy up to delivery, ranged from 0% to 42.5%. The highest prevalence was reported by

| Pooled prevalence and meta-analyses of data in selected studies
The pooled prevalence estimate of HCMV shedding in any bodily fluids in HCMV seropositive pregnant women from the metaanalysis performed was 21.5% [95% CI 12.7%,30.3%], as shown in Figure 2. Six studies met the sample size criteria (

| DISCUSSION
To our knowledge, our review is the first to systematically summarise the prevalence of HCMV shedding in HCMV seropositive pregnant women, in different populations around the world.
We have assembled evidence that HCMV shedding is indeed detected in HCMV seropositive women during pregnancy, up to delivery. Furthermore, we found a pooled prevalence estimate that 21.5% of HCMVseropositive pregnant women will shed HCMV on at least one time-point during pregnancy. As HCMV shedding may suggest that an individual is acutely infected with HCMV, 1,2 HCMV shedding in a HCMV seropositive pregnant woman may suggest that she has developed a non-primary maternal HCMV infection. It is known that there is an association between cCMV infection and nonprimary maternal HCMV infection. 28 Note: Table detailing

F I G U R E 2
Forest plot displaying a random effects meta-analysis of the prevalence of human cytomegalovirus (HCMV) shedding in any bodily fluid of HCMV seropositive pregnant women on selected studies. Forest plot displaying a random effects meta-analysis of the prevalence of CMV shedding in any bodily fluid of CMV seropositive pregnant women on selected studies with suitable sample size. Each square represents the prevalence for each study included in the meta-analysis, placed in position to the corresponding study on the y-axis and the corresponding prevalence estimate on the y-axis. The pooled estimate prevalence is marked by a diamond. The length of the line on either side of each square and diamond represents the extent of the prevalence 95% CI. The prevalence estimate and 95% CI for each study and for the pooled estimate are also displayed on the far right, including Higgins I 2 score for the pooled estimate displayed as I 2 to analyse heterogeneity across the studies. The association between HCMV shedding in pregnant women and the vertical transmission of cCMV has been poorly investigated.
We have gathered evidence here to suggest that this important gap in our knowledge needs to be addressed, and it may contribute towards the development of disease burden models and therapeutic or preventative strategies against cCMV infection.
The potential risk factors for HCMV shedding in HCMV seropositive pregnant women also needs to be investigated further.
Exposure to young children was identified as a risk factor for HCMV shedding in one of the three included studies in the systematic review, that reported varying data on this. Young children commonly shed HCMV for prolonged periods of time and have been shown to be a common source of HCMV transmission to pregnant women. 1,3 Exposure to young children is likely to be a risk factor for HCMV shedding in HCMV seropositive pregnant women too, however this was not shown conclusively in all studies. Although the studies represented a considerable range of countries and continents, only studies published in English were included, which may limit the applicability of these results internationally. Indeed, a third of the studies were from the USA, and no studies represented Africa or Australia continents. Therefore, it may be difficult to generalise our findings for populations that were not studied.
The high level of statistical heterogeneity 32 identified across the studies included in the meta-analyses means that the pooled prevalence estimates from the meta-analyses should be interpreted with caution. The population demographics such as ethnicity, socio-economic status, age and status of exposure to children also varied amongst all studies included in this systematic review. Due to this variability, it was not possible to determine the association of these factors with HCMV shedding, but it may have had an impact on the prevalence of HCMV shedding found.

| CONCLUSION
To our knowledge, our review is the first to systematically search the literature to summarise the available evidence on the prevalence of HCMV shedding in HCMV seropositive pregnant women.