Prenatal predictors of mortality in fetuses with congenital diaphragmatic hernia: a systematic review and meta-analysis

This study aimed to evaluate prenatal predictors of mortality in fetuses with congenital diaphragmatic hernia (CDH). A systematic literature search was performed to identify relevant observational studies that evaluated the ability of lung-to-head ratio (LHR), observed-to-expected LHR (o/e-LHR), observed-to-expected total fetal lung volume (o/e-TFLV), lung-to-thorax transverse area ratio (L/T ratio), intrathoracic herniation of the liver and the stomach, and side of diaphragmatic hernia, using a threshold for the prediction of mortality in fetuses with CDH. Study quality was assessed using the QUADAS-2 tool. Hierarchical summary receiver operating characteristic curves were constructed. A total of 50 articles were included in this meta-analysis. The QUADAS-2 tool identified a high risk of bias in more than one domain scored in all parameters. Among those parameters, the diagnostic odds ratio of mortality with o/e-LHR < 25%, o/e-TFLV < 25%, and L/T ratio < 0.08 were 11.98 [95% confidence interval (CI) 4.65–30.89], 11.14 (95% CI 5.19–23.89), and 10.28 (95% CI 3.38–31.31), respectively. The predictive values for mortality were similar between the presence of liver herniation and retrocardiac fetal stomach position. This systematic review suggests that o/e-LHR, o/e-TFLV, and L/T ratio are equally good predictors of neonatal mortality in fetuses with isolated CDH.


Introduction
Congenital diaphragmatic hernia (CDH) is a disease, that affects 1 in 4000 live births, in which the abdominal organs herniate into the thoracic cavity through a congenital defect in the diaphragm [1]. The severity of CDH varies widely from mild, which are completely asymptomatic at birth, to the most severe cases, in which death occurs immediately after birth. The severity of CDH depends on the presence of pulmonary hypoplasia and pulmonary hypertension [2][3][4][5].
Recently, prenatal imaging such as prenatal ultrasound or magnetic resonance imaging (MRI) has been used to assess the degree of lung hypoplasia in CDH, which is directed at improving its fidelity and prognostic ability. Advancements in prenatal diagnostic imaging and perioperative respiratory and circulatory management have recently improved the prognoses of patients. An accurate prenatal assessment of severity including pulmonary hypoplasia is essential for the standardization of perinatal care, and trials of prenatal therapy such as fetoscopic tracheal occlusion (FETO) [6,7]. Furthermore, the prenatal prognostic classification of CDH would provide more precise information about the estimated prospective course of treatment, and allow the establishment of a standardized protocol based on the prenatal findings.
Several antenatal prognostic parameters of postnatal survival in fetuses with CDH have been reported [8][9][10][11][12][13][14][15][16][17][18][19][20][21]. Most of them were direct or indirect measurements of fetal lung Kazunori Masahata, Masaya Yamoto, Satoshi Umeda contributed equally to this work. size, either by ultrasound or MRI. It has been reported that the antenatal predictors, which include lung area-to-head circumference ratio (LHR) [10,11], observed-to-expected LHR (o/e-LHR) [22,23], observed-to-expected total fetal lung volume (o/e-TFLV) [15,18,24,25], and lung-to-thorax transverse area ratio (L/T ratio) [8,16,26] are reliable predictors for postnatal mortality and morbidity. Other parameters are related to a worse prognosis such as the presence of herniated intrathoracic abdominal organs, among which the commonly evaluated was liver and stomach herniation, either classified as abdominal (entire viscera within the abdomen) or thoracic (any portion of the viscera into the thorax) [9, 13, 14, 27-33], and a right side of the diaphragmatic defect in patients with CDH [34][35][36]. However, the evidence supporting the use of specific antenatal parameters to predict postnatal outcome in prenatally diagnosed patients with CDH, which was the optimal threshold, is limited. Therefore, this study conducted a systematic review and meta-analysis of currently available literature to evaluate the utility of antenatal predictors of postnatal mortality in fetuses with CDH.

Search strategy
An electronic and systematic literature search for prenatal diagnostic studies predicting the mortality in fetuses with CDH was performed on December 6, 2020 and December 7, 2020 using MEDLINE and the Cochrane Library to identify studies using relevant keywords, Medical Subject Headings, and subheadings (Supplement 1).

Definitions of prenatal predictors and thresholds
The following parameters were included: (1) LHR was the ratio of the contralateral lung area, which was the product of the longest diameter of the lung and lung circumference via manual tracing, to the head circumference measured by ultrasound [10, 11]; (2) o/e-LHR was the ratio of the observed LHR measured by ultrasound to the expected LHR obtained from data of normal fetuses [22,23,37]; (3) o/e-TFLV was the ratio of total lung volume (both right and left lungs) measured by MRI, to the expected total lung volume for gestational age from the reference tables [15]; (4) L/T ratio was the ratio of the area of the contralateral lung via manual tracing, to the area of the thorax defined as the space surrounded by the inner border of the bilateral ribs, the sternum, and the vertebra [38] measured using ultrasonography of the transverse section containing the four-chamber view of the heart [8], (5) intrathoracic herniation of the liver (liver-up) was defined as any part of the liver observed in the thorax space using ultrasound or MRI; (6) stomach in chest was the presence of any portion of the stomach above the level of the diaphragm by prenatal imaging or stomach-down [30, [39][40][41][42]. Retrocardiac stomach herniation was defined as herniation of more than half of the stomach into the contralateral thoracic cavity or others; and (7) a side of diaphragmatic hernia (right or left). The continuous parameters related to fetal lung size, such as LHR, o/e-LHR, o/e-TFLV, and L/T ratio, were converted to binary variables using the threshold values. The threshold of LHR, o/e-LHR, o/e-TFLV, and L/T ratio were determined according to previous reports [22,38,[43][44][45]. Accordingly, we defined the thresholds of < 1 for LHR, < 25% for o/e-LHR, < 25% for o/e-TFLV, and < 0.08 for L/T ratio.

Inclusion and exclusion criteria
Inclusion criteria were peer reviewed full papers published from 2000 onward (to ensure that the findings were retrieved in accordance with, and relevant to, current care protocols for patients with CDH). Each study had to fulfill the following criteria: (1) studies with prenatally diagnosed CDH patients, (2) observational studies (prospective or retrospective cohort studies, case-control studies, and case series) evaluating the ability of LHR, o/e-LHR, o/e-TFLV, and L/T ratio or prenatal imaging parameters, such as intrathoracic herniation of liver and stomach, right-sided CDH, using an ultrasound or MRI for the purpose of prognostication to predict survival, (3) studies which reported adequately on study population and definition of prenatal predictors, (4) studies with sufficient information on the outcome (survive or not) to construct 2 × 2 tables, (5) studies with the main outcome measures were mortality of prenatally diagnosed patients with CDH, and (6) studies with fetuses that underwent fetal therapy were included.
The studies with an article type of review, case report, or letter with no full text available, studies that included postnatally diagnosed as CDH but did not focus on the imaging parameters using the defined threshold, or studies that had insufficient reporting of data for analysis were excluded.

Data extraction from included studies
A predesigned form was used for data extraction, where titles and abstracts were independently screened by two individual authors. The following characteristics were collected for each eligible study: study design, authors' names, year of the publication, study duration, country of the study, testing sample size, characteristics of the population (isolated or non-isolated CDH, side of the diaphragmatic hernia), gestational age of measurement, prenatal imaging (ultrasound or MRI), prenatal predictors with a threshold, mortality of each parameter, and the follow-up period. Two individual authors independently screened full text papers. After completing the data abstraction, a second reviewer resolved the disagreements between prior reviewers at full text stage.

Assessment of risk of bias
Quality assessment was done using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool consisting of four domains to assess the risk of bias [46], including patient selection, index test (i.e., the antenatal predictors), reference standard (the test considered as the gold standard against which index test results are compared), and flow and timing of patient. All domains are rated with regard to the risk of bias, and the first three items are also rated in terms of concerns regarding applicability to the research question. Studies were categorized as having low, high, or unclear risk, according to the judgment of the reviewers about each domain. Quality assessment was performed by two reviewers and checked by the second reviewer. Any disagreements were resolved through discussion and consensus.

Statistical analysis
For each study included in the meta-analysis, we examined the information to produce a 2 × 2 table of the number of true positives (TP), true negatives (TN), false positives (FP), and false negatives (FN), for the calculation of sensitivity and specificity. Using Review Manager (RevMan) version 5.4 software, forest plots were constructed to show sensitivity and specificity for all included studies graphically. The bivariate random-effect model and the hierarchical summary receiver operating characteristic (HSROC) model were used to estimate summary points at 95% confidence intervals for sensitivity, specificity, positive likelihood ratio (PLR), negative likelihood ratio (NLR), and diagnostic odds ratio (DOR). Furthermore, the summary curve, including the 95% credibility regions, 95% prediction region, and the area under the HSROC curve (AUC) were estimated. Statistical analyses were performed using STATA SE software, version 16 (Stata Corp, College Station, TX, USA).

Study selection
A total of 1260 studies were identified and screened in the systematic database search. The selection process is presented in the flow diagram ( Fig. 1). Of the 1260 search results, 1070 studies were excluded on first screening based on titles and abstracts. The full texts of all the potentially relevant citations were obtained. After the second screening, the remaining 190 studies, based on their full texts, were narrowed down to 75 records from 50 articles published from 2000 to 2020. The characteristics of the studies are shown in Fig. 2 [76][77][78][79][80][81][82]. The parameters in relation to the fetal lung volume and intrathoracic fetal stomach position included all patients with isolated CDH, while other parameters of both intrathoracic herniation of the liver and the right side of diaphragmatic hernia included studies with non-isolated CDH patients. The total sample size of each parameter was 344 for LHR, 1018 for o/e-LHR, 321 for o/e-TFLV, 355 for L/T ratio, 2756 for liver position, 713 for stomach in chest, 817 for retrocardiac stomach position, and 2649 for the side of diaphragmatic hernia (Table 1).

Quality assessment of studies
The summary of the analyzed quality assessment for each included study, according to the QUADAS-2 tool, is presented in Fig. 3. The detailed analysis of each study was captured in Supplement 2. In all studies included in this review, the high risk of bias in each domain was 46.0% (23/50) for patient selection, 22.0% (11/50) for index test, 46.0% (23/50) for reference standard, and 6.0% (3/50) for flow and timing. The high risk of bias in more than one domain was observed in all parameters. Hence, there was a high risk of quality concerns in the domains of patient selection, index test, and reference standard. The risk of bias in patient selection was considered high in all parameters, mainly because majority of the studies were case-control or case series which did not use random sampling. Additionally, some studies did not include all cases. The risk of bias in performance of the index test and the reference standard was also considered high in five and six parameters, respectively. Meanwhile, the risk of bias arising from patient flow and timing of procedures was considered low or unclear in the majority of the studies, owing to missing information in almost all studies, and it was not clear if a reference standard was performed in every patient. With regards to applicability, there was a high risk identified for patient selection and reference standard, but a low risk for index test application as most included studies reported detailed information with validated tests.

Quantitative analysis of antenatal predictors
The summary results of sensitivity, specificity, DOR, PLR, NLR, AUC, and the 95% CIs of each antenatal parameter are listed in Table 1. The constructed HSROC curves from the meta-analysis of the antenatal predictors are shown in Fig. 4. Among those parameters, the high predicting values of mortality included PLR and DOR observed in o/e-LHR, o/e-TFLV, and L/T ratio, respectively. On the contrary, the right side of the hernia had the lowest predicting values in sensitivity and DOR. With regards to the viscera herniated into the chest, the predicting values for mortality were similar between liver-up and retrocardiac stomach position. Among them, the stomach in chest had the lowest NLR which was 0.04 [95% confidence interval (CI) 0.0-2.42], and the highest DOR which was 31.82 (95% CI 0. 49-2061.73), with an AUC of 0.52 (95% CI 0.47-0.56). The prediction contours of LHR, o/e-LHR, o/e-TFLV, L/T ratio, liver-up, stomach in chest, and retrocardiac stomach position are large, which indicates a high possibility of the presence of heterogeneity between the studies in the HSROC curve.

Discussion
The current systematic review evaluated the contribution of selected antenatal predictors on mortality in fetuses with prenatally diagnosed CDH. The main finding of the study was that the antenatal parameters related to fetal lung volume were found to be equally good predictors of mortality, including the o/e-LHR < 25%, o/e-TFLV < 25%, and L/T ratio < 0.08, in isolated CDH patients.
The predictive value of LHR in fetuses with CDH was first described in 1996 by Metkus et al. [10]. However, the use of LHR has been controversial, as it was shown to increase according to the gestational age [22,37]. Thereafter, a ratio of the o/e-LHR by Jani et al. [22,23], which is not influenced by gestational age, was more informative in fetuses with isolated CDH. In our review, we found that o/e-LHR had both the highest summary DOR and PLR among those parameters related to fetal lung size, and this was consistent with the previous review [21]. Furthermore, both o/e-LHR on ultrasound and o/e-TFLV on MRI were superior in predicting mortality compared to LHR, due to their ability to control chronological change in gestational age. It has been demonstrated that the lung ipsilateral to the side of the hernia is more hypoplastic than the contralateral lung, which suggests that evaluation of both lungs may better estimate the degree of pulmonary hypoplasia in isolated CDH patients [45,83]. LHR is a simple measurement of the contralateral lung by ultrasound, but MRI is more advantageous in reliably visualizing and measuring both the ipsilateral and contralateral lungs compared to ultrasonography [25, 44,83]. However, our review demonstrated that the predictive values for mortality were similar between o/e-TFLV and o/e-LHR. In severe CDH with o/e-TFLV < 25%, the contribution of the affected ipsilateral lung may be reasonably small, and it may have contributed to discordance.
Among the antenatal parameters related to the lung volume, LHR, o/e-LHR, and o/e-TFLV are the most commonly used worldwide. L/T ratio is an ultrasonographic indicator of the severity of fetal CDH which is widely used in Japan. It was first described in 1990 for the assessment of pulmonary hypoplasia in CDH [8] and has been applied in the evaluation of pulmonary hypoplasia in CDH neonates since then [16,38]. Usui et al. [51] reported that L/T ratio was linearly correlated with o/e-LHR; thus the two parameters can be interconverted. Specifically, an L/T ratio of 0.08 is equivalent to an o/e-LHR of 25%. Calculating o/e-LHR is complex and requires a healthy reference value for the LHR, which may vary depending on nationality and ethnicity. In contrast, the L/T ratio appears to be a reliable predictive parameter as it is reportedly not influenced by gestational age in fetuses with CDH [51]. The combination of L/T ratio and liver-up is useful in prenatal risk stratification of patients with CDH [16,84]. In our review, L/T ratio was also a good predictor for neonatal mortality in fetuses with isolated left-sided CDH.
The intrathoracic herniation of the liver has been associated with a poorer outcome [4,13,19,85], and the amount of liver herniation into the thorax was also related to poor outcomes in fetuses with CDH [28,86]. The presence of liver herniation led to a direct reduction in the sizes of the right and left lungs [87]. However, previous studies reported that the position of the liver did not provide a significant contribution to the prediction of survival [12,49]. This systematic review demonstrated that liver-up has relatively low levels of summary in terms of specificity, DOR, and PLR in predicting mortality, compared to the other parameters related to fetal lung volume. One of the reasons was the heterogeneity of the definition of the liver position. Some of the included studies have prenatal assessment by ultrasound as well as fetal MRI, and revealed wide variations in the definition of liver herniation from authors in different studies. Another reason was the modality, in which some studies described the definition of liver herniation evaluated by the ultrasound, while others applied MRI, or both. The other reason was the degree of herniation, which is consistent with the size of the defect in the diaphragm. The percentage of herniated liver with respect to the total liver volume calculated by MRI was found to have a strong correlation with survival, in comparison to the presence or absence of liver herniation [18,44]. Further evaluation of the degree of intrathoracic herniated liver is warranted as it was not evaluated in this study.
Although the intrathoracic fetal stomach position was an accurate predictor of neonatal prognosis and has been associated with poorer outcome in isolated left-sided CDH [19, 30, 32, 33], it has not been broadly applied as a predictor of postnatal outcome. In patients with isolated CDH, the intra-abdominal fetal stomach position has been associated with a favorable prognosis with a survival rate of more than 90% [9, 32, 61]. Our review also found that the intrathoracic fetal stomach herniation demonstrated the lowest summary of NLR for mortality, with a mortality rate of fetuses with an intra-abdominal stomach of 1.4% (2/140) [19,33,44,47,54,61,75]. Meanwhile, it has already been established that for isolated CDH, more abnormal fetal stomach herniation was associated with adverse outcomes [30, 33]. As per previous reports, our review demonstrated that retrocardiac fetal stomach position was predictive of postnatal mortality in isolated CDH, compared to fetuses with other positions. It provides evidence that fetal stomach position alone was associated with a postnatal outcome in isolated CDH.
In our review, we have demonstrated that the predictive values for mortality in patients with right-sided CDH were low. It has been reported that the patients with right-sided CDH have been found to have a greater mortality rate [34][35][36], larger lesions of diaphragmatic defect, and more type C and D defects as described by the CDHSG classification system [82,88]. Meanwhile, although the high incidence of the predictors for adverse outcomes was observed in right-sided CDH compared with left-sided cases, postnatal mortality remained similar with regard to laterality [53,81]. These findings suggest that right-and left-sided CDH may have a different pathophysiology. Currently, Terui et al. have reported a simple scoring system based on widely available prenatal ultrasound findings that included the right side of the hernia [89], and it is thought to be useful for predicting the prognosis of fetuses with CDH.
We acknowledge that the present study has some limitations. Most studies selected for our analysis were retrospective observation studies, including case-control and caseseries studies, which had very different size populations. Also, selection bias could have occurred because only data from patients who underwent both prenatal MRI and ultrasound were used. The other limitations of the included studies are the high risk of bias in many studies, due to mainly Fig. 2 Accuracy of individual studies of antenatal parameters for mortality prediction. TP true positive, FP false positive, FN false negative, TN true negative, 95% CI 95% confidence interval, LHR lung-to-head ratio, o/e-LHR observed-to-expected LHR, o/e-TFLV observed-to-expected total fetal lung volume, L/T ratio lung-to-thorax transverse area ratio ◂ Table 1 The results of the summary of sensitivity, specificity, DOR, PLR, NLR, and AUC of antenatal parameter CDH congenital diaphragmatic hernia, LHR lung-to-head ratio, o/e-LHR observed-to-expected LHR, o/e-TFLV observed-to-expected total fetal lung volume, L/T ratio lung-to-thorax transverse area ratio, CI confidence interval, DOR diagnostic odds ratio, PLR positive likelihood ratio, NLR negative likelihood ratio, AUC area under the curve patient selection and reference standard. Some studies were excluded due to the lack of outcome data or impossibility of calculating 2 × 2 tables using the difference thresholds of antenatal parameters. The reported difference between the sample size for the true positives and true negatives in the article may have affected the results. Furthermore, the review revealed wide variations in the definition of liver and stomach herniation by various investigators in different countries, and the lack of consistency in the timing of the performance of prenatal assessment. Hence, we recognized the need to standardize the definition. The postnatal mortality with different timings of follow-up in the included studies may have contributed to the postnatal outcome as well. This review included some studies with neonates with CDH who were treated with ECMO. The use of ECMO in CDH patients has been reported as predictive to lower hospital survival [90], and the variability in the use of ECMO might have affected the outcome. Regarding CDH laterality, right-sided CDH was not associated with increased mortality, but was associated with an increased requirement for pulmonary vasodilator therapy, supplemental oxygen at discharge, and a need for tracheostomy [80]. Lastly, it is important to realize that many factors contribute to mortality, such as perinatal management and therapeutic strategies, where practices at each institution have changed over the study period.

Conclusion
In summary, our systematic review and meta-analysis suggest that an o/e-LHR of < 25%, o/e-TFLV of < 25%, and L/T ratio of < 0.08 are equally good predictors of the mortality in fetuses with isolated CDH. Furthermore, liver-up and retrocardiac fetal stomach position were also good predictors of mortality, and an intra-abdominal fetal stomach position is a favorable antenatal prognostic parameter. For the antenatal parameters (a-h), in green was highlighted a low risk of bias or low applicability concern and in red the studies with a high risk of bias or high applicability concerns. The bar with blue indicates the studies were where these risks of bias or applicability concerns could not be assessed properly (unclear). QUADAS-2 Quality Assessment of Diagnostic Accuracy Studies, LHR lung-to-head ratio, o/e-LHR observed-to-expected LHR, o/e-TFLV observed-toexpected total fetal lung volume, L/T ratio lung-to-thorax transverse area ratio

Fig. 4
Constructed hierarchical summary receiver operating characteristics (HSROC) curves of antenatal parameters for the prediction of mortality in patients with prenatally diagnosed CDH: a LHR < 1; b o/e-LHR < 25%; c o/e-TFLV < 25%; d L/T ratio < 0.08; e liver-up; f stomach in chest; g retrocardiac stomach position; h right side of diaphragmatic hernia. Each circle on the plot represents the pair of sensitivity and specificity from a study, and the size of the circle is scaled according to the sample size. The solid red block represents the summary of sensitivity and specificity, and this summary point is surrounded by a 95% confidence region (yellow dashed line) and 95% prediction region (blue dotted line). CDH congenital diaphragmatic hernia, LHR lung-to-head ratio, o/e-LHR observed-to-expected LHR, o/e-TFLV observed-to-expected total fetal lung volume, L/T ratio lung-to-thorax transverse area ratio Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Springer Nature or its licensor holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.