Cochlear implant performance in children deafened by congenital cytomegalovirus—A systematic review

Congenital cytomegalovirus (cCMV) infection is a major cause of sensorineural hearing loss in children.


Congenital infection with cytomegalovirus (cCMV) is common in Wes-
tern countries such as the Netherlands, with a birth prevalence of 0.53%. 1,2 A meta-analysis of the worldwide literature revealed an overall birth prevalence of 0.64%. 3 An infant may acquire cCMV when its mother encounters or reencounters the virus during pregnancy, which is transmitted through bodily fluids. Approximately 10%-15% of infants infected congenitally have clinical evidence of the disease (are symptomatic) at birth. 4 Symptomatic infants present with symptoms such as intrauterine growth retardation, low birth weight, prematurity and hepatosplenomegaly. 5 Contrarily, the majority of infected infants is asymptomatic at birth. In approximately 7%-25%, asymptomatic infants develop symptoms of the infection later in life. 5,6 The most common postnatal symptom of cCMV infection is sensorineural hearing loss (SNHL), presenting in 15%-65% of infected children. [7][8][9][10] Additional symptoms or comorbidities associated with cCMV are visual impairment, cognitive and motor deficits which can result in neurodevelopmental delay and balance problems. In addition, these children are prone to neurodevelopmental dysfunction such as intellectual disability, autism spectrum disorder (ASD), attention deficit hyperactivity disorder (ADHD) and pervasive developmental disorder (PDD). [11][12][13] The SNHL is often progressive in the first 4 years of life for both asymptomatic and symptomatic CMV infected children. 14,15 When it advances to severe to profound SNHL, cochlear implantation may be considered. Considering the comorbidities associated with cCMV, the question arises whether cCMV children are able to achieve the same performance level with a cochlear implant (CI) compared to non-cCMV children with a CI. A growing number of studies have investigated the performance of cCMV children after cochlear implantation on various speech perception, speech production, receptive language and auditory performance outcomes. A systematic review by Shin et al. 9 reviewed the literature up to 2011 mainly focusing on the effect of antiviral medication on cCMV related SNHL and secondarily the effect of cochlear implantation on cCMV-related SNHL. As for the results on cochlear implantation, they found contradictory results in a combination of case-control studies and case series: some studies reported equal performance in cCMV children compared to non-cCMV children, often 1 year after implantation. Other studies found lower levels of CI performance in cCMV children compared to non-cCMV children. The effect of cCMV-related comorbidities on CI performance was not investigated specifically. This can be a major interest to analyse additionally. The primary aim of this systematic review was to evaluate CI performance in children deafened by cCMV compared to non-cCMV children. Our secondary aim was to investigate the effect of cCMV-related comorbidities on CI performance in cCMV children.

| Ethical considerations
No ethical considerations were made as this is a review of existing literature.

| Search and selection
A systematic literature search in PubMed, Embase and the Cochrane library was conducted from database inception up to the 15th of May 2017. Search terms were limited to the terms: "children," "cochlear implant" and "performance" with all relevant synonyms (see Table S1). To avoid a too narrow search, we did not search on terms corresponding to "CMV." Duplicates were were included when they examined speech perception, speech production, receptive language and auditory performance in children with SNHL due to cCMV infection after cochlear implantation. Exclusion criteria were adult patients and/or single-sided deafness. Commentaries, systematic reviews, non-English/Dutch studies and non-human studies were excluded. Unpublished studies were not excluded.

| Study assessment
Eligible articles were independently assessed by two authors for directness of evidence (DoE), data extractability and risk of bias (RoB) using predefined criteria (see Table 1). DoE was scored by evaluating the population, intervention and outcome. In addition,

Key points
• Congenital cytomegalovirus (cCMV) infection is a major cause of sensorineural hearing loss in children.
• In addition to sensorineural hearing loss, congenital cytomegalovirus can cause severe neurodevelopmental comorbidities.
• Cochlear implant performance in children deafened by congenital cytomegalovirus is inferior compared to children deafened by other aetiologies Therapy: •, Cochlear implantation before the age of 18 years; ○, other therapy. Outcome: •, Speech perception, speech production, receptive language or auditory performance; ○, other outcome. | 1285 studies were scored satisfactory when data from the cCMV population was extractable and unsatisfactory when it was not.
RoB was scored on selection bias, standardisation of outcome, blinding, missing data and follow-up. Selection of population was scored satisfactory when there was no selection bias, moderate when authors selectively included non-missing data and unsatisfactory when sample selection was deemed susceptible to bias. We scored whether a test was externally validated (standardisation of outcome). Missing data were subdivided into less than 10%, between 10% and 20%, more than 20% or not reported. Finally, follow-up was scored satisfactory when measurements were performed at set times and unsatisfactory when follow-up was unequal between patients. Adhering to the Grade system for assessing RoB, no studies were excluded based on RoB. A sensitivity analysis was performed removing studies with a moderate or high RoB. Therefore, studies were excluded based on DoE and data extractability only.

| Data extraction
Study characteristics, such as study population, non-cCMV group, groups.
To answer our second research question, scores of asymptomatic cCMV children versus symptomatic cCMV children and asymptomatic cCMV children versus non-cCMV children were extracted and compared by the reviewers. Between-group analyses were performed using the Mann-Whitney U test for non-parametric numeric data and Fisher's exact test for non-parametric ordinal data in SPSS, and a P-value of <.05 was deemed significant.

| Search strategy and study selection
As shown in Figure 1, our search identified 5280 unique articles.
After screening titles and abstracts in inclusion and exclusion criteria, 288 articles were left for full-text screening. Cross-reference screening did not yield additional articles. Corresponding authors were contacted when full texts were not available, which resulted in one additional full text. Consequently, 30 articles were eligible for critical appraisal.

| Data extraction
Primarily, we provided a descriptive table of study results from the study population and comparison group. Meta-analyses of repeatedly reported outcomes (SIR, CAP) were attempted but not reported due to insufficient reporting within the original studies. Computing a forest plot was attempted but failed due to the lack of reporting of measures of uncertainty in the cCMV group and/or control group and great variation in follow-up. Contacting corresponding authors for additional data yielded no response, which made it impossible to gain a sensible result. Secondarily, when available, scores of asymptomatic cCMV children versus symptomatic cCMV children and asymptomatic cCMV children versus asymptomatic non-cCMV children were extracted, displayed in tables and compared by the reviewers.

| Study characteristics
The characteristics of the 12 studies selected after critical appraisal are presented in Table 2. Sample size in these studies varied from 2 cCMV vs 5 non-cCMV 19 to 16 cCMV vs 131 non-cCMV. 18 Children were divided into two groups based on SNHL with or without comorbidities, respectively: symptomatic and asymptomatic cCMV children. The majority of studies included a combination of symptomatic and asymptomatic cCMV children versus a non-cCMV group. Four studies did not report the presence of comorbidities in the non-cCMV group. 18,21,22,25 As can be seen in Table 2, one study included only asymptomatic cCMV children and a non-cCMV group without comorbidities. 19 The following outcome measurements were described: auditory performance was tested in six studies [21][22][23][24][25]27 , speech perception in four studies 5,13,18,20 , speech production in nine studies 5,18-21,24-27 , receptive language in three studies, 23,24,26 and language-social developmental quotients in one study. 13 A list of abbreviations and explanation of outcome measures can be found in Table S2.

| Prognostic value of cCMV on CI performance
and role of cCMV-related comorbidities 3.5.1 | Primary comparison: cCMV vs non-cCMV As shown in Table 3, seven of twelve studies 5,13,18,20,22,23,27 , showed worse outcome in cCMV children (symptomatic and/or asymptomatic) compared to non-cCMV groups on various speech and language outcomes. Six studies 5,13,18,20,22,27     In sum, the majority of studies found worse outcomes in the cCMV children compared to non-cCMV children, which was attributed by most authors to be related to cCMV comorbidities.

| Secondary comparison 2: Asymptomatic vs symptomatic cCMV
Data of asymptomatic cCMV children versus symptomatic cCMV children were extracted of three studies to elucidate the effect of comorbidities in the performance of cCMV implantees (Table 5)

| Prognostic value of cCMV on CI performance and role of time post-implant
Follow-up ranged from 3 to 77 months. The six studies 5,13,18,20,22,27 that showed a significant difference between the cCMV and non-cCMV group reported this difference at a single test 13,18,20,27 and most of them at a short follow-up (less than 2 years 5,13,20,22 ). Two studies 18,20 reported a significantly worse performance in the cCMV children at multiple test moments: in Yoshida et al., 20 a significant difference between a cCMV group and a non-cCMV group before 12-month follow-up was noted, while after more than 12 months, no significant differences were noted. In the analysis of the study by Ramirez and Nikolopoulos 18 , follow-up varied between 1 and 5 year in the cCMV group (4 cases at 1 year, 7 cases at 3 years, 2 cases at 4 years and 3 cases at 5 years). A worse performance than the non-cCMV group at the final follow-up year was seen in 38% of cases (P = .04).

| Summary of main findings
The aim of this systematic review was to evaluate the prognostic value of cCMV infection as a cause of SNHL on CI performance in cCMV children. In addition, we evaluated the effect of cCMV-related comorbidities on CI performance in cCMV children. Six of seven T A B L E 4 Speech and language outcome after cochlear implantation in children with asymptomatic cCMV infection compared to a non-cCMV group were not able to recognise a pattern between the severity of comorbidities and speech and language development.
The majority of the studies that found lower performance in the cCMV group did so within 2 years after implantation. Lower performance mostly disappeared after a longer follow-up. This indicates that CI performance in cCMV children may be delayed but can progress to the level of other children after 2 years. This finding emphasises the benefit of counselling patients and their parents during the first years of speech and language development and explaining that cCMV children may take longer to achieve similar results as non-cCMV children but that they are able to achieve a similar level of speech and language development in approximately 2 years.

| Comparison with literature
The finding of a lower CI performance score in cCMV implantees compared to other aetiologies is in agreement with the systematic review performed in 2011. 9 In their subanalysis on outcome of surgical therapy for CMV-related SNHL, the authors state that children with congenital CMV will advance more slowly than those with other causes of SNHL. However, no analysis was performed to explore the role of associated comorbidities. The current study adds knowledge to the latter question.
T A B L E 5 Speech and language outcome after cochlear implantation in children with symptomatic cCMV infection compared to asymptomatic cCMV infection Methodologically, the current review differs from the previous. 9 The emphasis on the effect of cCMV and related comorbidities on speech and language development after cochlear implantation resulted in a broad search up to 2017 without the use of automated implosions through the databases described earlier. Our search strategy resulted in twelve comparative studies that were deemed valid for data extraction, of which only four were included in the earlier review. All seven articles in that review were included in the original search results of this review, indicating no articles were missed with this extensive search. All but one of the articles included in the previous review were included for critical appraisal in ours except for one case report. 46 Certainly, this systematic review also has its limitations. A limitation of this review is that a meta-analysis could not be performed, due to previously mentioned reasons.

| CONCLUSION
This systematic review shows that children deafened by cCMV are associated with impaired CI performance compared to non-cCMV implantees. In the majority of studies, the cCMV group reached lower levels of CI performance, especially in the first years post-implant.
Inferior CI performance in cCMV children was attributed to comorbidities in the majority of studies and confirmed by additional statistical comparisons by the reviewers. Therefore, we urge clinicians to take into account the negative effects of comorbidities associated with cCMV-related deafness during the counselling of cCMV implantees.
Regardless of the above mentioned, all studies revealed that children with CMV related SNHL benefit from cochlear implantation.

ACKNOWLEDGEMENT
We would like to thank Hanneke Bruijnzeel, MD for her efforts in the earlier stages of setting up this research.

CONF LICT OF I NTEREST
None.