Prenatal ultrasound for the diagnosis of the agenesis of corpus callosum: a meta-analysis

Abstract Background Prenatal ultrasound has been regularly used as the screening tool for agenesis of corpus callosum (ACC) of the fetuses, which were mainly suspected on the basis of indirect signs rather than the visualization of the CC. However, the diagnostic accuracy of prenatal ultrasound for ACC, compared to the gold standard of postmortem diagnosis or postnatal images, is still unknown. This meta-analysis was performed to comprehensively evaluate the efficacy of prenatal ultrasound for the diagnosis of ACC. Methods Studies evaluating the diagnostic accuracy of prenatal ultrasound for ACC compared to postmortem diagnosis or postnatal images were retrieved by searching PubMed, Embase, and Web of Science databases. Pooled sensitivity and specificity were calculated with a random-effects model. The diagnostic accuracy was measured by summarized area under the receiver operating characteristic (AUC) curve. Results Twelve studies involving 544 fetuses with suspected anomalies of central nervous system were included, and 143 of them were with validated diagnosis of ACC. Pooled results showed that prenatal ultrasound has satisfying diagnostic efficacy for ACC, with the pooled sensitivity, specificity, positive and negative likelihood ratios of 0.72 (95% confidence interval [CI]: 0.39–0.91), 0.98 (95% CI: 0.79–1.00), 43.73 (95% CI: 3.42–558.74, and 0.29 (95% CI: 0.11–0.74), respectively. The pooled AUC was 0.94 (95% CI: 0.92–0.96), suggesting good diagnostic performance of prenatal ultrasound. Subgroup analysis according to the prenatal ultrasound procedures showed a better diagnostic efficacy of neurosonography than that of regular ultrasound screening (sensitivity: 0.84 versus 0.57, specificity: 0.98 versus 0.89, and AUC: 0.97 versus 0.78). Conclusions Prenatal ultrasound, particularly for the neurosonography, confers satisfying efficacy for the diagnosis of ACC.


Introduction
Agenesis of corpus callosum (ACC) is one of the most prevalent fetal anomalies of the central nervous system (CNS) [1,2].The incidence of ACC is difficult to estimate because the diagnosis is still challenging and partial of the patients are asymptomatic [1].The pathogenesis of ACC is complicated, which involves the interactions of genetic, infective, vascular and toxic factors during pregnancy [3,4].The prognosis of fetuses with ACC varied according to the concurrent CNS and non-CNS anomalies, which could be from normal to severely impaired neurodevelopmental outcomes [5][6][7].Accordingly, prenatal diagnosis of fetuses with ACC is important in clinical practice [8,9].Currently, prenatal ultrasound has been used as a first-line screening imaging technique for ACC in pregnant women with a mean gestational age (GA) of 20 weeks [10].However, based on the major society guidelines, ultrasound screening in the second trimester is not required the visualization of the CC and a subsequent diagnosis of ACC may be suspected on the basis of indirect signs of ACC, such as the absence of cavum septi pellucidi (CSP) and ventriculomegaly, etc. [1].Accordingly, fetuses with suspected ACC are often referred to fetal magnetic resonance imaging (MRI) for further confirmation and identification of other co-existing anomalies [11,12].Recent International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) Practice Guidelines recommended the use of the multiplanar ultrasonic assessment of the brain for detection of ACC in fetuses, which is termed as neurosonography [10,13,14].Although prenatal screening for ACC using ultrasonic examination has been well applied, its diagnostic accuracy compared to the gold standards of postmortem diagnosis or postnatal image is still unknown.Here, we performed a meta-analysis to comprehensively evaluate the efficacy of prenatal ultrasound for the diagnosis of ACC.

Methods
The recent edition of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement [15,16] were followed through the study.The design, implementation, and reporting of the study was in accordance with the instruction of the Cochrane Handbook [17] guideline.

Literature search
According to the aim of the meta-analysis, we used a combined search term to identify relevant studies by systematic search of the electronic databases, including PubMed (MEDLINE), Embase, and Web of Science.The terms that we used were "sonography" OR "ultrasound" OR "ultrasonic" OR "prenatal diagnosis" OR "antenatal", and paired with "corpus callosum".The search was limited to studies in humans and published from the index to 1 November 2021.As a supplement, a manual analysis of the references of the related articles was also performed to identify possible related studies.

Inclusion and exclusion criteria
We applied the following inclusion criteria to identify possible related studies [1]: studies published as fulllength article in English [2]; included pregnant women with fetuses of suspected CNS anomalies who underwent prenatal ultrasound examination [3]; aimed to evaluate the diagnostic performance of prenatal ultrasound for the diagnosis of ACC [4]; the reference standard for the validation of ACC was postmortem diagnosis or postnatal images; and [5] essential data of true-and false-positive and true-and false-negative values were reported or could be estimated so as to create the standard 2 Â 2 table.The protocol of prenatal ultrasonic examination and the referenced validation methods were consistent with which were applied in the original studies.Reviews, postnatal studies, and studies that did not evaluate ACC were excluded.

Data collection and quality assessment
Two independent authors performed literature analysis, data collection, and study quality assessment based on the inclusion criteria, separately.If discrepancies occurred, discussion with the corresponding author was indicated for the final judgment.Study information, characteristics of the fetuses, GA at diagnosis, and details of reference standard, as well as the data of true-and false-positive and true-and falsenegative values were extracted for each study.The quality of the included studies were evaluated with the QUADAS-2 (quality assessment tool for diagnostic accuracy studies) scale [18].Based on the risk of bias of the study and the applicability of the protocol, each study was rated as of low, high, or unclear risk of bias for each domain.

Statistical methods
The sensitivity, specificity, and positive and negative diagnostic likelihood ratios (DLRs) were pooled from the 2 Â 2 tables and summarized with the corresponding 95% confidence interval (CI).Besides, diagnostic odds ratio [OR], which represents the ratio for the odds of a correct diagnosis to the odds of a misdiagnosis [19], was also calculated to indicate the efficacy of diagnosis.The diagnostic accuracy of the prenatal ultrasound for ACC was evaluated by summarizing the area under the receiver operating characteristic (AUC) curve by pooling the data from each included studies.The extent of between-study heterogeneity was evaluated by the Cochrane's Q test, and the estimation of I2 statistic [17].An I 2 > 50% reflects significant heterogeneity [20].A random-effects model was used in this study because this is a conservative way to incorporate the influence of possible between-study heterogeneity.Deeks' funnel plots were constructed and the asymmetry test was applied to evaluate the risk of publication bias.STATA 12.0 was used for statistical analyses, and a P value < 0.05 indicates statistical significance.

Results of literature search
As shown in Figure 1, the initial database search retrieved 1473 studies, and 1247 remained after removing of the duplications.Then, by analyzing the contents of titles and abstracts, 1206 studies were further removed mainly because they were not relevant to the objective of the meta-analysis.In the remaining 41 studies, 29 were subsequently excluded after reading the full texts for the reasons listed in Figure 1.Finally, 12 studies [21][22][23][24][25][26][27][28][29][30][31][32] were available for the metaanalysis.

Study characteristics and quality assessment
The summarized features of the included studies are displayed in Table 1.Overall, 12 studies [21][22][23][24][25][26][27][28][29][30][31][32] including 544 fetuses of suspected CNS anomalies were included into the meta-analysis.Of them, 143 were with validated diagnosis of ACC.These studies were published between 2003 and 2021 and performed in Austria, Israel, China, United Kingdom, the United States, Korea, Poland, Sweden, Turkey, and Spain, respectively.The median GA for the prenatal ultrasonic diagnosis varied between 18 and 20 weeks.The protocol of prenatal ultrasonic examination was described as neurosonographic examination in seven studies [21,22,25,27,28,31,32], and as ultrasonic screening in five studies [23,24,26,29,30].Postmortem diagnosis or postnatal images were used as the reference examinations for the validation of ACC.Details of quality evaluation of the included studies via the QUADAS-2 criteria are shown in Table 2.The risk of bias and applicability concerns of the index test was deemed to be of unclear risk since the details of ultrasonic examination was not described in five studies [23,24,26,29,30].For other domains, all the included studies were judged as low risk of bias.

Meta-analysis results
Pooled results of the 12 studies showed that prenatal ultrasound has a satisfying diagnostic efficacy for ACC, with pooled sensitivity and specificity of 0.72   3).Subgroup analysis according to the prenatal ultrasound procedures showed a better diagnostic efficacy of neurosonography than that of regular ultrasound screening (sensitivity: 0.84 versus 0.57, specificity: 0.98 versus 0.89, and AUC: 0.97 versus 0.78; Table 3 and Figure 4) for ACC in fetuses with suspected CNS anomalies.The average GAs for the diagnosis of ACC cases in studies using neurosonography and ultrasound screening were not statistically significant (25.1 ± 3.6 weeks versus 25.5 ± 5.4 weeks, p ¼ .61).

Publication bias
The Deeks' funnel plots for the meta-analysis were shown in Figure 5, and the asymmetry test showed a low risk of publication bias (p ¼ .30).

Discussion
This meta-analysis combined the results of 12 available diagnostic studies, and the results showed that compared to the validation test of postmortem diagnosis and/or postnatal images, prenatal ultrasound has a good diagnostic performance for ACC, with pooled sensitivity, specificity, and AUC of 0.72, 0.98, and 0.94, respectively.Further subgroup analysis by protocols of prenatal ultrasound showed that the prenatal neurosonography seemed to be associated with a better diagnostic performance for ACC than ultrasonic screening (sensitivity: 0.84 versus 0.57, specificity: 0.98 versus 0.89, and AUC: 0.97 versus 0.78).Collectively, these results suggested that prenatal ultrasound, particularly for the neurosonography, confers satisfying efficacy for the diagnosis of ACC in fetuses with suspected CNS anomalies.
To the best of our knowledge, our study may be the first meta-analysis comparing the diagnostic efficacy of prenatal ultrasound for ACC in comparison with postmortem diagnosis and/or postnatal images.Although prenatal ultrasound examination at approximately 20 weeks of GA has been well-applied as the first-line screening methods for ACC, the diagnostic performance of the examination for ACC remains not fully determined, particularly when it is compared with postmortem diagnosis and/or postnatal images [8].The uncertainty for the efficacy of the prenatal ultrasound for the diagnosis of ACC may be because that the diagnosis mainly depends on indirect signs rather than direct signs of ACC.Indeed, although non-visualization of corpus callosum (CC) could be detected in prenatal ultrasonic examination for some cases of fetuses (direct signs), suspected ACC were mostly noticed by the detection of indirect signs, including loss of CSP, widening of the interhemispheric fissure, cerebral ventricular abnormalities, changes of the pericallosal artery, and radial arrangement of cerebral sulci near the third ventricle, etc. [33,34].Revealing the indirect signs of ACC on prenatal ultrasonic examination relies on the experiences of the examiners and standard protocols of the examination.Interestingly, results of our subgroup analysis showed that using the neurosonography with multiplanar ultrasonic assessment of the brain as recommended by ISUOG seemed to be associated with an even better diagnostic performance than those ultrasonic screening [13].Of note, the sensitivity of diagnosis with neurosonography was significantly improved, which is important for the prenatal ultrasound examination at GA of 20 weeks as a general screening tool for the anomalies of the fetuses.The advantages of prenatal ultrasonic examination are also obvious, which is noninvasive, cheap, less dependent on the environment of examination, and repeatable [35].Taken together, results of our finding support the use of prenatal ultrasonic examination, particularly of the neurosonography as a screening tool for the diagnosis of ACC in fetuses with suspected CNS anomalies.Currently, fetal MRI is also available and as the confirmative tool to validate the diagnosis of ACC suspected by prenatal ultrasonic examination.Accumulating studies have been performed to compare the diagnostic efficacy between prenatal ultrasound and fetal MRI for ACC [36][37][38][39].However, the results were not consistent.In a recent important systematic review and meta-analysis, Sileo et al. summarized the results of 14 relevant studies, and showed that associated anomalies detected exclusively at fetal MRI in isolated ACC undergoing neurosonography is lower than previously reported [11].It has been demonstrated that fetal MRI is crucial for determining the prognosis of fetuses with cortical and posterior fossa anomalies, which exist exclusively on MRI [11].A previous multicenter, retrospective cohort study involving 14 fetal medicine centers in Italy, UK, Portugal, Canada, Austria and Spain showed that in fetuses with isolated anomaly of the CC diagnosed on antenatal neurosonography, MRI can identify a small proportion of additional anomalies, mainly malformations of cortical development, which are not detected on ultrasound [40].An early meta-analysis also showed that in fetuses undergoing dedicated neurosonography, the rate of a CNS anomaly detected exclusively on MRI is lower than that reported previously [41].Moreover, MRI performed in the third trimester may be associated with a better detection rate for some types of anomaly, such as cortical, white matter and intracranial hemorrhagic anomalies [41].In addition, for fetuses with isolated mild or moderate ventriculomegaly on neurosonography, although the incidence of an associated fetal anomaly missed on ultrasound and detected only on fetal MRI is lower than that reported previously, the large majority of these anomalies are difficult to detect on ultrasound but evident in MRI [42].All of the above findings support the practice of MRI assessment in fetus with ultrasound suggested CNS anomalies.There are, however, some anomalies that remain undetected during prenatal development [11].In fact, using fetal MRI as a reference method for the validation of ACC is also less optimal as compared to validating by postmortem diagnosis and/or postnatal images [43,44].Although fetal MRI is helpful to provide direct brain images for visualizing of CC, the value of fetal MRI in the era of fetal neurosonography is to detection of additional anomalies for prognostic consideration [11,12].Besides, with the integration of other prenatal tests, such as genomic testing [45], may further improve the screening efficacy of prenatal ultrasonic examination for ACC, which should be observed in the future.
Our study has limitations.Firstly, only fetuses with suspected CNS anomalies were included.The screening efficacy of prenatal ultrasound for ACC as a regular procedure for overall pregnant women remains to be determined.However, this is difficult because the substantial of normal fetuses detected by prenatal ultrasound are not likely to receive postnatal imaging examinations of the brain.Secondly, it may be better to include all types of corpus callosum (CC) pathologies and also compared the efficacy of ultrasound, neurosonography and also MRI for different types of CC pathologies.However, a key consideration during the designing of the meta-analysis is to make efforts to reduce the possible source of heterogeneity.Because including multiple corpus callosum pathologies and using multiple diagnostic tools may introduce additional heterogeneity, in this study, according to the aim of the study, we focused on the evaluation of the efficacy of prenatal ultrasound for the diagnosis of ACC, as compared to the gold standard of postmortem diagnosis or postnatal images.These findings are important for the validating the diagnostic efficacy of prenatal ultrasound for ACC.Subsequently, we may consider another meta-analysis to evaluate the diagnostic efficacy of prenatal ultrasound for other pathological types of CC, such as hypoplasia and dysgenesis of CC.Moreover, a differentiation between complete  and partial ACC is necessary in clinical practice when ACC is diagnosed.However, we were unable to perform subgroup analysis according to the diagnosis for the complete and partial ACC because generally, the included studies did not report the diagnostic efficacy of prenatal ultrasound for complete and partial ACC, separately.Accordingly, studies are warranted in the future to validate if the diagnostic efficacies of prenatal ultrasound for complete and partial ACC are similar.In addition, as mentioned previously, the diagnostic efficacy of prenatal ultrasound for ACC may be affected by many factors, such as the position of the fetus, experience of the examiner, timing and times of examination etc. Influences of these factors on the outcome of the meta-analysis could not be determined in this study.Finally, the details of postmortem diagnosis and/or postnatal images for postnatal validation of ACC were rarely reported among the included studies, which may affect the validity of these referenced tests.
In conclusion, results of this meta-analysis suggested that prenatal ultrasound, particularly for the neurosonography, confers satisfying efficacy for the diagnosis of ACC in fetuses with suspected CNS anomalies.Development of standard ultrasonic protocol and integration of other prenatal tests may be considered to further improve the sensitivity of the prenatal ultrasonic examination as a screening tool for ACC.

Figure 1 .
Figure 1.Flowchart representing the process of literature search and study retrieving.

Figure 2 .
Figure 2. Forest plots for the summarized diagnostic efficacy of prenatal ultrasound for the diagnosis of ACC; (A) summarized sensitivity; and (B) summarized specificity.

Figure 3 .
Figure 3. Summarized ROC curves for the performance of prenatal ultrasound for the diagnosis of ACC.

Figure 4 .
Figure 4. Subgroup analyses for the summarized ROC curves according to the different ultrasonic protocol applied; (A) summarized ROC curves for the performance of prenatal neurosonography for the diagnosis of ACC; and (B) summarized ROC curves for the performance of regular ultrasonic screening for the diagnosis of ACC.

Figure 5 .
Figure 5. Deek's funnel plots for the evaluation of possible publication bias underlying the performance of prenatal ultrasound for the diagnosis of ACC.

Table 1 .
Characteristics of the included studies.

Table 2 .
Quality evaluation for the included studies with QUADAS-2 scores.

Table 3 .
Subgroup analyses in studies with prenatal neurosonography or US screening.