Neurodevelopmental outcomes in children with isolated congenital diaphragmatic hernia: A systematic review and meta‐analysis

Background: Congenital diaphragmatic hernia (CDH) reportedly has neurologic consequences in childhood however little is known about the impact in isolated CDH.


| INTRODUCTION
Congenital diaphragmatic hernia (CDH) is a serious birth defect that occurs in 3-4/10000 fetuses (ORPHA:2140). 1 The condition is characterized by a defect in the diaphragm allowing abdominal organs to herniate into the thorax thus interfering with normal lung development. This leads to bilateral lung hypoplasia, which at birth causes respiratory insufficiency and pulmonary hypertension. Survival rates in large volume centers using standardized neonatal treatment protocols are over 70% in isolated cases. 2 However, survival may come at the cost of high medium-to long term morbidity rates including chronic lung disease, variable pulmonary hypertension, gastroesophageal reflux, feeding and growth problems, neurocognitive delay, hearing loss, thoracic deformations, and hernia recurrence. [3][4][5][6][7][8][9][10] For parents of children born with CDH, the most relevant long term morbidity may be neurodevelopmental impairment (NDI).
Several studies have indicated that CDH-patients are at risk for cognitive and motoric dysfunction with numbers varying from 16% to 80%. [11][12][13][14][15][16][17][18][19] Extracorporeal membrane oxygenation (ECMO), supplemental oxygen requirement beyond 30 days of life and neuromuscular hypotonicity are known (postnatal) predictors of NDI, next to prenatal predictors such as disease severity or gestational age at delivery. 11,15,20,21 However, the majority of these studies included both isolated and nonisolated CDH cases. It is important to differentiate between these two groups as the rate of NDI is increased in children with nonisolated CDH presenting with various chromosomal and/or structural abnormalities. [22][23][24][25] Second, the ongoing TOTALtrial is investigating the effect of a prenatal intervention for CDH, is restricted to fetuses with isolated CDH. 26,27 When counseling parents prenatally and evaluate if they are suitable candidates for fetal therapy, it is important to be able to present reliable data to estimate the risk of severe comorbidities, including NDI of their child. Therefore, we aimed to systematically review the literature on the risk of NDI in fetuses and infants with CDH, either isolated or with associated structural or genetic abnormalities.

| Protocol and registration
This systematic review was conducted in accordance with the guidance on Preferred Reporting Items for Systematic reviews and Metaanalyses. The protocol was registered with the International Prospective Register of Systematic Reviews (PROSPERO 2018 CRD42018115854). 28

| Eligibility criteria
Randomized controlled trials (RCT), cohort and case-controlled studies and case series reporting the neurologic outcomes of children born with CDH were considered eligible. No date or language restrictions were applied. Systematic reviews, narrative review articles, case reports, incomplete reports, book chapters, conference abstracts, letter to the editors, and comments were excluded. Case series reporting on less than five patients were excluded.

| Search strategy
A systematic search was conducted in MEDLINE, EMBASE, and Cochrane databases using free text and Medical Subject Headings as described in Appendix S1. Subsequently, reference lists of relevant review articles were manually checked to identify relevant cross-references (snowballing). Covidence (Veritas Health Innovation Ltd) was used to eliminate duplicate articles and manage study screening. We excluded duplicates type I (among different databases) and type-II (in different journals/issues).

| Study selection
Two authors (Lennart Van der Veeken and Simen Vergote) screened all titles and abstracts independently, excluded irrelevant studies and then independently assessed the remaining full-text articles for eligibility; disagreements were resolved by consensus or if necessary by a third author (Karl Kristensen). Studies were excluded if the full text was unavailable online and the abstract contained insufficient information. Authors from studies from which data could not be extracted (e.g., composite or combined outcomes given) were contacted by e-mail for further clarification. If authors did not reply, studies were listed as "awaiting classification". Studies containing patient cohorts which appeared to have been published previously by the same authors were excluded. In that case we opted only to include the cohort with the largest number of patients for each outcome measure. If cohorts were tested twice, we included most recent results, that is oldest age of children at testing.

| Data extraction
Two authors independently extracted data (Simen Vergote and Lennart Van der Veeken) and entered this into a standardized Excel form (Microsoft; Appendix S2) with study and condition characteristics.
Disagreements were resolved by consensus. Neurologic outcomes were categorized in eight categories: general neurologic outcome, development, cognition, motor, language, vision, hearing, and behavior.
Outcomes were the number of children scoring abnormal and borderline (abnormal was defined as scoring more than two standard deviations below average on standardized tests and borderline as a score between two and one standard deviation below average). The primary outcome was neurodevelopmental delay. Secondary outcomes included, motor skills, intelligence, vision, hearing, language, and behavior abnormalities. Data were separated between isolated and nonisolated studies. The definition for isolated CDH was: (1) absence of major structural abnormalities at birth and (2) normal genetic findings.
If possible, data from isolated cases was extracted from studies including both isolated and nonisolated cases. If it was not possible to distinguish the two groups, all data were considered as nonisolated.

| Quality assessment of studies
Study quality and risk of bias were analyzed using a standardized form. Randomized trials were analyzed using the Cochrane Collaboration's tool for assessing risk of bias. 29 Case-control studies were analyzed using the Newcastle-Ottawa Scale for assessing the quality of nonrandomized studies. 30 Case series were analyzed using the National Institutes of Health study quality assessment tool. 31

| Assessment of heterogeneity
Methodological and clinical heterogeneity of data per study were evaluated. Variables were tested for statistical heterogeneity by applying the I 2 test to determine whether data could be pooled. An I 2 value less than 40% was taken to indicate heterogeneity not to be important; 30%-60% may represent moderate heterogeneity; 50%-90%: may represent substantial heterogeneity, 75% to 100% considerable heterogeneity. 32

| Meta-analysis
Meta-analysis for all outcomes was carried out using MedCalc statistical software version 15.4 (MedCalc Software). Results were expressed as proportions with 95% confidence intervals (CI) as all outcomes were categorical variables. Pooled proportions were calculated using the random effects model as we anticipated the presence of heterogeneity among the included studies. Results from meta-analyses are presented as percentage with confidence interval, 95% (CI). Comparisons between groups are presented as risk ratios.

| Search results
The electronic literature search (April 2020) identified 2292 study records. After deduplication, 1854 articles were screened by title and abstract, and a further 1672 were excluded as irrelevant. Snowballing identified an additional 10 records and the full texts of the remaining 192 articles were reviewed. After assessment of the full text, 102 articles were excluded: 14 did not report neurologic outcomes, 17 were not randomized trials, case-control series or case series, four studies did not report on CDH, 36 were excluded for overlapping data and for 31 studies we are at the moment of writing, still waiting for response to clarify the data ( Figure 1).

| Risk of bias
Quality assessment of the studies is displayed in Table S2. No studies were found to have a high risk of bias or to be low quality.

| Statistical heterogeneity
Neurologic outcomes data were pooled in 40 separate meta-analyses. The results and levels of heterogeneity per outcome measure are listed in Appendix S3.

| Vision and hearing
Seven studies reported on visual problems using a variety of tests.

| Neurologic morbidity nonstandardized tests
Using nonstandardized tests, neurologic development was assessed in 23 studies using a wide variety of tests ranging from physical examination to nonstandardized questionnaires. The majority of these patients were included in large series including ECMO treated patients. Due to the heterogeneity of testing methods used it was not appropriate to perform a meta-analysis on this data.

| Isolated versus nonisolated cases
Nine studies compared isolated versus nonisolated CDH or reported on the relative risk (RR). 20

| Severity
Six studies investigated the relationship between neurologic outcomes and severity. Church et al. described smaller lung volumes on prenatal MRI to be associated with lower fine motor scores. 45 Bojanic et al. did not see a correlation between liver status and motor scores. 39 The correlation between cognition and lung size was examined in five studies but none found this to be correlated. 21

| Fetoscopic endoluminal tracheal occlusion (FETO)
One study compared neurologic outcome in patients that were treated with FETO versus patients that were not treated prenatally. 7 They did not detect any differences in cognition, motoric score,

| DISCUSSION
In this systematic review, we report a 16% occurrence of neurodevelopmental delay in children with isolated CDH. There was a wide variety in domains tested, as well as the methods to test these domains, leading to heterogeneity in the prevalence of NDI.
Nonetheless, we came to a motoric deficit rate of 13%, abnormal cognitive function in 5% and aberrant hearing in three %. Although earlier studies reported that associated anomalies increase the risk for NDI, we only found a slightly lower rate of developmental delay in isolated CDH. However, hearing impairment and cognitive delay were found to be less frequent in isolated CDH compared to combined isolated and nonisolated cases. The relative risk on the other hand clearly demonstrated an increased risk in case of nonisolated CDH. The findings in the nonisolated group may be an underestimation because these studies include both isolated and nonisolated cases with a variable percentage (1%-37%) of nonisolated cases.
Many studies still rely on nonstandardized assessment tools, only four out of 10 studies on NDI in isolated CDH used standardized tests, the other six varied widely in methodology and definitions. Based on these four studies, the rate of NDI was 16%. This is much higher than in the general population, which is around 4%-8%. [119][120][121] If mild NDI is included this number is reported around 15%, and also mental health problems have been reported to be as high as 20%, 122 however we restricted in this study to severe NDI. NDI was mainly attributed by an increased rate of motor problems, present in 13% of cases. Cognitive problems were only seen in 5%. We can only speculate on why the motor function is more often affected than the cognitive function.
Possibly, the decreased exercise capacity or oxygen dependency limits the motor function and development in children with CDH. Second, these children suffer from nutritional and growth problems caused by gastrointestinal disorders which can impact motor development. 123 Inadequate development of brain regions that are responsible for motor function, might also play a part. Radhakrishnan et al. found a correlation between the lung volumes and cerebellar dimensions. 124 On the other hand, this might not be a consequence of the disease, but of the required postnatal surgery, anesthesia, complications, and other neonatal interventions as a motor deficit occurs also in other surgical correctable congenital anomalies. 125 It is unclear whether this impairment persists later in life due to the lack of studies reporting on the motor or cognitive outcome in adults. Four studies reporting on outcomes in adolescents found cognitive impairment to be similar to our meta-analysis which suggests that the cognitive impairment lasts at least until adolescence. 63 Severe cases these days are offered to opt for prenatal surgery as FETO has been shown to increase survival. 26,27

| Strengths and limitations
This study has limitations. First the quality of a systematic review highly depends on the published data and our effort to provide a tool for counseling fetal surgery patients is therefore compromised by the poor quality of the primary data. In this review, only three RCTs and four cohort studies were included, all other studies did not include a control group. A second limitation is that we pooled data to report on a single number for each outcome in the absence of clinical and statistical homogeneity. Another limitation is that although we tried to identify studies only including isolated cases, some studies included additional anomalies, though minor, that were of unknown nature. Lastly, although we checked for overlapping data, it is possible the same children were included in multiple studies, reporting on different neurologic domains which increases the risk for over estimation. Therefore, to avoid double counting, we did not report on an overall rate of neurologic complications.
This study also has some strengths. We performed a systematic search according to current guidelines for systematic reviews and included all articles reporting on neurologic outcome in CDH.
Because we included case series, we maximized the number of cases included. Secondly, by analyzing neurologic domains separately, we were able to report impairment rates different outcomes. Thirdly, we are, to our knowledge the first to focus on outcome in isolated CDH which is important for counseling of fetal surgery candidates.

| CONCLUSION
Children with CDH seem to be at increased risk for neurodevelopmental delay, mainly attributed to abnormal motor function whereas cognition and hearing are less frequently affected. Only a minority of published data deals with isolated CDH fetuses. The outcomes of these were better. In view of the potential of antenatal management and its complications, it would be interesting to determine the relationship of the severity of pulmonary hypoplasia, the gestational age at delivery, and conduct studies on the difference between isolated and nonisolated cases. This systematic review highlights the current need for standardized core outcomes for CDH patients (https://ern-ernica.eu/). Fund.

CONFLICT OF INTEREST
The authors report no conflicts of interest.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.