Neonates born with myelomeningoceles have been recognized for decades to be at risk for having life-threatening dysfunction ascribed to brainstem-associated deficits. In 1986, Gilbert et al. reported a high incidence of brainstem dysgenesis (76%) at autopsy in children with spina bifida who were younger than 2 years of age.¹ This sentinel paper provided evidence, although in a select population, that the brainstem and cranial nerve nuclei can be severely dysplastic or poorly developed in some children with spina bifida. Hence, the debate of whether such brainstem dysfunction represented solely compressive factors associated with the Chiari II malformation, or an inherent dysplasia, and therefore surgically irreversible, was given hard data supporting the latter mechanism. Concurrent within this time frame, however, was the clinical realization that brainstem dysfunction as manifested by apnea, neurogenic dysphagia, aspiration, or vocal cord paralysis, at times, was potentially reversible by ensuring adequate CSF diversion and, occasionally, decompressive surgery.²

The introduction of prenatal closure of myelomeningoceles has provided not only improved outcomes regarding hydrocephalus but also insight into the etiology of the Chiari II malformation with reduction in anatomical severity of this entity after prenatal closure.^3–6 While the anatomical changes can be dramatic with rapid reversal of hindbrain hernia after prenatal closure, the functional consequences of this anatomical change remain poorly defined.

The purpose of this study was to compare the incidence of significant brainstem dysfunction (SBD) as defined by the need for airway surgery, gastrostomy tube, or intubation because of apnea, aspiration, or airway control problems in neonates with a history of myelomeningocele who have been treated with either prenatal or postnatal closure at our institution.

Methods

The records and imaging of all children (n = 87) who had undergone closure of myelomeningocele at our institution whether prenatally (n = 27) or postnatally (n = 60) over a period of 78 months, from December 2014 until May 2021, served as the data. Prenatal closure was available at our institution in the last 44 months of this period. Prior to this, all mothers and fetuses who met criteria for prenatal closure were referred elsewhere if the family desired. Offering of prenatal closure at our institution was made using maternal and fetal criteria as outlined in the Management of Myelomeningocele Study (MOMS) publication.³ No fetus was refused intervention because of the size of the defect on prenatal imaging. Hydrocephalus outcomes were discussed with families after the 2015 publication by Tulipan et al., which described no apparent benefit regarding the need for surgical treatment of hydrocephalus when the fetal ventricular size was greater than 15 mm.⁷ All surviving children had at least 4 months of follow-up after birth either through neurosurgical clinic visits in Kansas City or by phone if mothers (n = 2) undergoing prenatal intervention delivered outside of our region and obtained postnatal evaluation and care elsewhere.

The Children’s Mercy Hospital institutional review board deemed this study exempt from review given the nature of data collection. Data prior to 2018 were retrospectively obtained, and subsequent data were collected prospectively. A spreadsheet was maintained with medical record number, fetal ventricular size, gestational age at fetal imaging and delivery, postnatal ventricular size, need for and type of hydrocephalus treatment, spinal neurological level at birth, prenatal or postnatal repair, presence of SBD, Chiari grade by postnatal imaging, and death. The myelomeningocele neurological level was determined by clinical examination by a fellowship-trained pediatric neurosurgeon after delivery in all neonates and before surgical closure in the postnatal group (Table 1). Distribution graphs and median and modal levels were calculated for prenatal, postnatal, and SBD for comparison. Fetal ventricular sizes from either prenatal ultrasound or MRI were recorded, along with gestational age at the time of imaging. Mean and median fetal ventricular sizes were calculated for the prenatal cohort and the postnatal cohort; within the postnatal cohort, the sizes were also calculated for those with SBD and those without SBD. Gestational age at imaging was similarly analyzed. Unpaired t-tests were performed for statistical analysis using GraphPad when comparing fetal ventricular sizes, gestational ages, or age at first shunt placement between cohorts. Fisher’s exact test was used to compare the incidence of SBD between cohorts.

The first cranial imaging study after delivery was used to measure postnatal ventricular size as the greatest width of the atrium of the larger ventricle. MRI performed after delivery up to 1 year of age was used to provide an anatomical grade of the Chiari II malformation as outlined by Tulipan et al. in 1999 (Table 2).⁶ The clinical course was reviewed for evidence of SBD as defined by 1) neurogenic dysphagia requiring gastrostomy; 2) airway or breathing problems requiring tracheostomy, glottic, or vocal cord surgery; or 3) apnea, aspiration, or airway control problems requiring intubation. Each child was labeled as either having SBD or not having SBD. No case of SBD was the result of an etiology not ascribable to the central nervous system.

The decision to treat hydrocephalus in neonates without SBD regardless of prenatal or postnatal closure was guided by any history of new irritability or excess vomiting, head circumference changes, ventricular size changes on serial imaging, fontanel status, and extraocular eye motion findings.

Results

There were 60 neonates born at a mean gestational age of 37.3 weeks (SD 2.98 weeks) who underwent postnatal closure of their myelomeningoceles and 27 fetuses born at a mean gestational age of 36.1 weeks (SD 1.83 weeks) who underwent prenatal closure. There were no discernible trends with the institution of our fetal program regarding annualized number of myelomeningocele closures. Fifteen neonates (25%) met the criteria for SBD in the first 6 weeks of life. All 15 of these neonates underwent postnatal closure of their defects. Two of them, however, were born extremely prematurely. Excluding these 2 neonates would give an incidence of 22%. Neonates who underwent prenatal myelomeningocele closure did not display SBD (p = 0.008).

Hydrocephalus in neonates with SBD was managed with ventriculoperitoneal shunt insertion (n = 13) in all but the 2 children who had their defects closed postnatally and were born extremely prematurely. If a shunt had been inserted prior to recognition of SBD (n = 4), then shunt function assessment at time of SBD diagnosis was performed. Physical examinations focused on fontanel status, findings concerning for malfunction such as subgaleal fluid around the shunt CSF leakage from any incision, head size, neck arching, eye motion and position, and stridor. Repeat cranial imaging was done in all cases. Shunt taps were performed in all to exclude infection even if function was deemed adequate. Shunt exploration was performed if any concern of shunt malfunction persisted after the above assessment (n = 3 of 4 with shunts inserted prior to diagnosis of SBD). No child’s SBD resolved with either shunt insertion or shunt revision.

Fetal Ventricular Size

The prenatal diagnosis of spina bifida was made in all but 5 fetuses. Prenatal imaging was available for all patients undergoing prenatal closure and 55 of the 60 patients who underwent postnatal closure, leaving 82 prenatal imaging studies available for review (18 ultrasound and 64 MRI). The cohort with prenatal closure underwent imaging on average almost 4 weeks earlier versus the postnatal cohort. Comparison of the fetal ventricular sizes of those who underwent postnatal closure versus those with prenatal closure showed no significant difference (Fig. 1 and Table 3). Comparisons of fetal ventricular sizes in those with postnatal closure with SBD versus those with prenatal closure without SBD showed no significant difference. No significant difference existed when comparing fetal ventricular sizes in the cohort with prenatal closure versus the cohort experiencing SBD. The median fetal ventricular sizes across all cohorts were identical at 13 mm. In addition, the prenatal cohort underwent imaging significantly earlier in gestation (mean 22.6 weeks’ gestation, SD 1.4 weeks’ gestation) compared with the postnatal cohort (mean 26.1 weeks’ gestation, SD 4.7 weeks’ gestation) (p = 0.0003). Similarly, fetuses with SBD (mean 26.4 weeks’ gestation, SD 4.4 weeks’ gestation) underwent imaging significantly later (p = 0.0003) than the prenatal cohort. Hence, if one could correct for gestational age, the average fetal ventricular size likely would have been larger in the prenatally closed cohort if measured around 26 weeks’ gestation versus 22.6 weeks’ gestation. We excluded 2 neonates who underwent postnatal closure and were delivered extremely prematurely as their need for prolonged intubation could be at least partially attributed to their prematurity. They had fetal ventricular sizes of 9 and 11 mm and would have driven the average postnatal cohort ventricular size lower if included. No difference in fetal ventricular sizes among the various cohorts could be detected that would be a factor in predicting better or worse outcomes regarding SBD.

Boxplots of fetal ventricular widths (in mm) of the cohort with prenatal closure (left) and the cohort with postnatal closure (right). The x indicates the mean.

TABLE 3.

Statistical comparisons of fetal ventricular size cohorts

	Mean Ventricular Size ± SD (mm)	95% CI (mm)	p Value*
All postnatal closure (n = 55) vs prenatal closure (n = 27)†	13.3 ± 6.2 vs 12.8 ± 3.4	−3.08 to 2.01	0.68
Postnatal closure w/ SBD (n = 11) vs w/o SBD (n = 42)‡	13.8 ± 7.6 vs 13.5 ± 6.0	−3.99 to 4.70	0.87
Postnatal closure w/ SBD (n = 11) vs prenatal closure (n = 27)	13.8 ± 7.6 vs 12.8 ± 3.4	−2.56 to 4.57	0.57

^*Unpaired t-test.

^†Five neonates in the postnatal closure group did not undergo diagnostic prenatal imaging.

^‡Two patients in the postnatal closure group were excluded because of extreme prematurity, and 2 patients did not undergo diagnostic prenatal imaging.

Spinal Neurological Level

Myelomeningocele levels varied from T12 to S1 for the postnatal group and from L1 to S1 for the prenatal group. The median level for all cohorts was L4 (Fig. 2). The modes were L4, L5, and L4 for the prenatally closed, postnatally closed, and SBD groups, respectively (Fig. 3). Hence, there is no evidence of the group that had prenatal closure having been preselected as having more favorable (more caudal) spinal lesions compared with the group that had postnatal closure. In addition, the group with SBD did not appear to have more rostral (less favorable) lesions compared with either the entire group with postnatal closure or the group with prenatal closure.

Bar graph showing the spinal neurological level at time of birth stratified by prenatal closure, postnatal closure, or SBD.

Bar graph showing the neurological level at time of birth stratified by prenatal closure, postnatal closure, or SBD.

Discussion

Our study was not randomized but used the spinal neurological levels and fetal ventricular sizes as indicators of severity of the myelomeningocele/Chiari/hydrocephalus complex and found no differences in these indicators between prenatal and postnatal cohorts. While selection bias for prenatal closure could be a factor that contributes to different outcomes as this was not a randomized process, we are yet to be able to identify a specific factor other than prenatal closure as an influencer of SBD occurrence.

Compared with MOMS, there was less severe prematurity (4% vs 13%) in our experience as defined by gestational age less than 30 weeks.³ Our 2 most severely premature infants both underwent postnatal closure (both born by spontaneous vaginal delivery at 27 weeks). Our prenatal cohort had a mean gestational age of 36 weeks versus a mean of 34 weeks in the MOMS report. In our study, no patients died in the prenatal cohort. All 3 deaths occurred in the postnatal group, 2 in the neonatal period and 1 at 5 years. All 3 of these deaths were attributed to poor respiratory drive and/or airway protection and, hence, lower brainstem pathophysiology. There were 4 deaths within 1 year in the MOMS trial, with 2 deaths in each cohort. The 2 patients who died in the postnatal cohort were described as having died due to their Chiari malformation. The MOMS report did not describe neurogenic dysphagia, need for airway surgery, tracheostomy, or intubation otherwise. We had a single neonate who had undergone a prenatal closure that manifested "respiratory distress." This compares with 21% of the prenatal cohort in the MOMS report described as having "evidence of the respiratory distress syndrome" after 24 hours of life. This was ascribed to the higher incidence of prematurity compared with their postnatal cohort. Given our longer gestational periods, our incidence of respiratory distress was subsequently less. Notable differences in our experience compared with MOMS were likely from our reduced incidence of prematurity and associated longer mean and median gestational ages, as surgical techniques and postoperative medical management of these pregnant women have improved with collective experience.⁸

Another large analysis of neonates with myelomeningocele who underwent either prenatal or postnatal disclosure described a high incidence of respiratory distress in all groups regardless of gestational age as defined by requiring respiratory support (> 2 L/min of oxygen by nasal cannula or surfactant or caffeine administration) in the first 24 hours of life.⁹ The authors of this report concluded, "Fetal repair may decrease this risk slightly."

Regarding SBD, however, we found a 25% incidence in the postnatal group. This incidence is in line with reports by others of an approximately 20% incidence of "symptomatic" Chiari malformations in infants with myelomeningoceles.^10,11 Mortality, however, has been reported as high as 50% in children with symptomatic Chiari II malformations,^10–12 whereas our mortality rate was 20% (n = 3/15) in children with SBD. Our reduced mortality by comparison may be from more aggressive hydrocephalus management and improved neonatal intensive care capabilities.

Pollack et al. described neurogenic dysphagia in 9 children with myelomeningoceles that required intervention over a span of 10 years.² Many of these children manifested other lower brainstem problems, with 4 requiring tracheostomies. They noted a lack of improvement after craniocervical decompression in the more severe cases. Six of the 9 children were diagnosed in early infancy. However, the authors did not provide the total number of infants born at their institution with myelomeningoceles. Park et al. described 45 infants with myelomeningoceles that underwent posterior fossa decompression, with the major indications being dysphagia, apnea, stridor, and aspiration.¹¹ There was a 38% mortality rate in this group. The authors described a 19% incidence of a symptomatic Chiari II malformation in 262 children who had their myelomeningoceles closed at their institution over 7 years. The majority of these children presented with these brainstem symptoms and signs in the first 3 months of life. This information aligns with our observation of early onset of symptoms and a similar incidence.

Danzer et al. did try to specifically address the impact on brainstem function with prenatal closure by sending a questionnaire to families with children who had undergone prenatal repair.¹³ They had an 89% response rate, with 48 of 54 families responding. There was no control group. One of the children received a gastrostomy tube for neurogenic dysphagia, and no child received a tracheostomy or had significant apnea or required chronic ventilatory support (supplemental oxygen or continuous positive airway pressure). One child underwent a posterior fossa decompression for syringomyelia. The authors described numerous other lesser symptoms such as gastroesophageal reflux and strabismus but overall had outcomes not dissimilar from ours regarding severe brainstem dysfunction. Tulipan et al. commented in their discussion on observing symptomatic Chiari malformations (dysphagia, stridor, or apnea) in only 2 (< 5%) of 42 neonates who underwent prenatal closure.⁶ Our experience would align with these reports regarding both anatomical improvement or normalization of the posterior fossa and reduced functional concerns related to the brainstem in the prenatal cohort.

There exists definitive evidence of brainstem dysplasia in young children with myelomeningoceles who have died. These brainstem findings ranged from aplasia of the tegmentum in 4% to defective myelination in 44%.¹ Other findings included hypoplastic or absent brainstem nuclei. One-fourth of these children in whom autopsy was performed, however, were free of any intrinsic brainstem abnormalities. Prenatal closure seemingly would not affect this dysplasia, as the embryology of brainstem nuclear development and myelination is well underway by 20 weeks. Whether early loss of cells or myelin in the fetus can be differentiated from lack of formation is not discussed.¹ Curnes et al. correlated anatomical severity with symptomatic brainstem and long tract signs. They found no asymptomatic patients with a fourth ventricle or medullary kink below C3, whereas two-thirds of symptomatic patients had medullary kinks at C4 or below.¹⁴ This implies that anatomical severity does correlate to a degree with physiological dysfunction. Our experience would agree with this observation. Wolpert et al., however, concluded that anatomical severity did not correlate with clinical dysfunction.¹⁵ Studies looking at normal brainstem development described the fluctuating presence of molecular factors such as N-myc downstream-regulated gene 2 (NDRG2) in the later gestational period.¹⁶ Whether specific gene expressions involved in brainstem development are affected by changing the hydrodynamics with prenatal closure could be a factor and a topic of exploration. If our observation coupled with those of others that prenatal closure of myelomeningoceles is associated with reduced SBD, then factors besides purely brainstem dysplastic causes defined by traditional histopathology or radiological imaging need to be considered. The clinical syndrome of Chiari type II–associated brainstem dysfunction, however, while likely multifactorial, appears to have a component that can be mitigated by prenatal closure of the myelomeningocele.

Conclusions

Significant brainstem dysfunction occurred in the 1st month of life in approximately one-fourth of neonates born with myelomeningocele that was closed postnatally. One-fifth of these children with SBD died. No neonate with prenatal closure displayed SBD. SBD is a greater risk factor for early death than shunt-dependent hydrocephalus in this population. Our prenatal cohort was not different from the entire postnatal cohort, or the cohort with SBD regarding fetal ventricular size or spinal level of the myelomeningocele. When discussing whether to treat spina bifida prenatally or postnatally, in addition to impact on hydrocephalus and maternal and fetal risks and benefits, the favorable impact on SBD occurrence with prenatal closure should be an important factor in the decision-making process.

Disclosures

The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.

Author Contributions

Conception and design: Grabb. Acquisition of data: Grabb, Vlastos. Analysis and interpretation of data: all authors. Drafting the article: Grabb. Critically revising the article: Grabb, Lundy, Partington. Reviewed submitted version of manuscript: Grabb, Lundy, Partington. Approved the final version of the manuscript on behalf of all authors: Grabb. Statistical analysis: Grabb. Administrative/technical/material support: Grabb, Vlastos. Study supervision: Grabb.

Supplemental Information