Anatomy and Clinical Implications of OSB

The OSB results from a failure of neurulation in early pregnancy (28 days of gestation). Two forms can be distinguished: Myelomeningocele (MMC; spina bifida cystica aperta) or myeloschisis (the noncystic variant). Along with non-neurulation of a part of the spinal cord, the vertebral arches and the soft tissue fail to close. In MMC, the non-neurulated spinal cord sits on top of a cystic sac formed dorsally by pia and ventrally by dura and the abnormally shaped arachnoidal space in between contains cerebrospinal fluid. In myeloschisis, the arachnoidal space and thus, the cyst are collapsed, and the non-neurulated spinal cord resides on the floor of the vertebral canal (Fig. 1). Importantly, in both variants, the neural tissue remains directly exposed to the amniotic fluid prenatally and to air postnatally.²

Beside the malformation at the back, a constellation of structural defects of the brain, cerebellum, and brain stem (subsumed as Chiari II malformation) are associated with OSB. Typical, but not the sole features of the Chiari II malformation are hydrocephalus, hindbrain herniation (a displacement of part of the brain stem and of the cerebellum through the foramen magnum in to the vertebral canal), and a small, “crowded,” posterior fossa (Fig. 2).

Patients born with this severe congenital malformation suffer from a cluster of lifelong disabilities despite postnatal repair. The most important are paraparesis or paraplegia, neuropathic bladder and bowel dysfunction, and a shunt-dependent hydrocephalus. The former two are virtually always present, and hydrocephalus occurs in over 70%.³ Hindbrain herniation causes in up to 33% of patients operated postnatally, the dysfunction of cranial nerves, cerebellum, and medullary respiratory center.^4-6 Corrective, rehabilitative, or palliative therapeutic measures are adopted to address the wide array of neurosurgical, neurologic, orthopedic, endocrinologic, sexual, and psychosocial issues that accompany the disabilities of OSB. A real cure, however, does not exist.

Pathogenesis of OSB and the Rationale for Prenatal Repair

For many years, the peripheral neurologic deficits seen in patients with OSB were thought to be the direct result of the non-neurulation. With the advent of fetal surgery at the end of the last century, attention was directed to the prenatal history of congenital malformation. It became clear that in some malformations, the evolution during gestation could be affected by negative processes that would eventually become clinically relevant at birth. Following that spirit, the prenatal natural history of spina bifida was studied extensively in the 1990s. Analysis of the spina bifida lesions of aborted fetuses^7-9 demonstrated that although the spinal cord tissue within the lesion was always non-neurulated, the microstructure of the cord as well as the sensorimotor projections were regularly present. However, the more advanced the gestation was, the more damage, such as abrasion, erosion, disruption, hemorrhage, inflammation, and degeneration of the neural tissue was found, in some specimens up to complete disappearance.² These observations led to the “two-hit-hypothesis”: A first hit being the non-neurulation and a second hit an in utero acquired spinal cord destruction by the direct and prolonged exposure to the amniotic fluid. Consequently, the enthralling idea came up that timely in utero coverage could protect the spinal cord and stop the otherwise progressing destruction.

Figs 1A to F

Myelomeningocele: (A) Intraoperative image, preoperative fetal MRI; (B) sagittal view; (C) axial view. The non-neurulated placode (arrow) resides on the cystic sac (arrowhead). Myeloschisis: (D) intraoperative image, preoperative fetal MRI; (E) sagittal view; (F) axial view. There is no cystic sac. The non-neurulated placode sits on the floor of the vertebral canal (arrow)

Figs 2A to C

(A) Preoperative fetal MRI: Hindbrain herniation (cerebellar herniation down to C3) and small “crowded” posterior fossa (arrow). (B) Three weeks after fetal spina bifida repair: Hindbrain herniation has already resolved, cerebrospinal fluid is detectable in the posterior fossa (arrow). (C) Postnatal MRI: No herniation of cerebellum or brain stem. Basal cistern with normal width

Persuasive proof of the two-hit hypothesis was achieved experimentally with different animal models.^9-15 It was possible to mimic the in utero progressive damage to the normal spinal cord by exposing it to amniotic fluid, and it was demonstrated that in utero coverage of the spinal cord could hinder damage and thus spare function. Additionally, the animal experiments showed also that by creating a spina bifida-like lesion and thus, a cerebrospinal fluid leak, a hindbrain herniation could be provoked. Not only sealing the leak by in utero repair reversed this experimentally induced “Chiari malformation”.^16,17 but the evidence obtained by all these experiments paved the way for fetal OSB repair to be commenced in human fetuses.

Results of Prenatal Repair

In 1997, Bruner et al¹⁸ from Vanderbilt University reported on the very first case of a endoscopic human fetal MMC repair. In 1998, two groups [Tulipan from Vanderbilt University¹⁹ and Adzick et al²⁰ from the Children's Hospital of Philadelphia (CHOP)] reported independently the first cases of successful open in utero repair. Both groups published a year later sister publications in the same JAMA issue: They reported that fetal OSB repair reversed hindbrain herniation in a majority of fetuses and reduced the need for ventriculoperitoneal shunting by about 50%.^21,22 Several follow-up studies by both groups confirmed the reversibility of hindbrain herniation (Fig. 2) and the dropping of shunt rate.^23-26 Moreover, favorable evolution with regard to head size,²⁷ brain stem function,²⁸ motor function of the legs,²⁹ and different neurodevelopment parameters could be observed.^30,31 However, bladder dysfunction, one of the major problems of OSB, seemed not to improve after prenatal repair.^32-34 Another downside was that some patients had worsening neurology due to tethering of the spinal cord at the repair site.³⁵ Further, prenatal OSB repair was associated with increased maternal risks (including preterm labor and uterine dehiscence), risk of premature birth, and increased risk of fetal or neonatal death.^22,25 Despite the maternal risks at the time of prenatal OSB repair, the CHOP group could show that reproductive capacity was not lost after prenatal OSB repair and that the hysterotomy risks ware comparable to patients with classical cesarean section.³⁶

These clinical data with all the above-described positive and negative aspects prompted the conception of a prospective, multicenter, randomized controlled trial to compare safety and efficacy of prenatal vs postnatal OSB repair (Management of Myelomeningocele Study, MOMS trial).¹

The MOMS trial began in 2003 and took 7 years. It was stopped prematurely after inclusion of 183 of the planned 200 cases based on the efficacy of the prenatal group. The prenatal group did significantly better with regard to hindbrain herniation (no herniation: Prenatal/postnatal = 36%/4%, p < 0.001, severe herniation = prenatal/postnatal: 6%/22%, p < 0.001), shunt placement at the age of 12 months (prenatal/postnatal = 40%/82%, p < 0.001), a composite score derived from the Bayley Mental Development Index and the difference between the functional and the anatomical levels of the lesion at 30 months (p = 0.007), and, finally, regarding independent walking at 30 months (prenatal/postnatal: 42%/21%, p < 0.01). Yet, prenatal surgery was associated with an increased risk of preterm delivery and uterine dehiscence at delivery. In detail, oligohydramnios (prenatal/postnatal 21%/4%), spontaneous rupture of membranes (46%/8%), spontaneous labor (38%/14%), and preterm delivery (79%/15%, average ages 34.1 vs 37.3 weeks) were significantly more frequent in the prenatal than in the postnatal group (p < 0.001). Although maternal safety was preserved, the hysterotomy site was found thinned in 25% of the mothers and dehiscent in 10%.

In summary, the trial demonstrated unequivocally that prenatal OSB repair is not completely curative and that risks are associated with it, but also that it definitely yields the best overall results achievable today.

After the MOMS trial, centers offering prenatal OSB repair (among them our Zurich Center for Fetal Diagnosis and Therapy with over 60 cases, data to be published soon) showed that the MOMS-trial results can also be reproduced outside the setting of a rigorous trial.^37-40 and additional studies looked at different aspects of prenatal OSB repair. The data from the MOMS trial were further analyzed in substudies. One study focused on the need for shunt placement⁴¹ after prenatal OSB repair. It demonstrated that the ventricle size at initial screening is an important predictor. In the prenatal group, 20% of those with ventricle <10 mm, 45.2% with ventricle size of 10 to 15 mm, and 79% with ventricle size ≥15 mm received a shunt. In the postnatal group, percentages were 79.4, 86.0, and 87.5% respectively (p = 0.02). Thus, patients that have a ventricle size at screening over 15 mm do not have a benefit from fetal repair regarding hydrocephalus. Another study analyzed the effect of prenatal repair on bladder function.⁴² Although prenatal OSB repair did not significantly reduce the need for clean intermittent catheterization at 30 months; it was shown that patients had less bladder wall trabeculation and less open bladder neck. As the author stated, the implication of these findings are unclear at this point. In the same year, Carr⁴³ published a 5-year follow-up of 54 patients operated at CHOP before the MOMS trial. This group demonstrated a greater likelihood to successfully toilet-train than historical controls. In this respect, our own data published a year later by Horst et al⁴⁴ suggest a positive effect of prenatal MMC closure on lower urinary tract function. However, several other studies, some of them with a weak study design, were not able to demonstrate improved urological outcome after OSB repair.^45-47 Hence, further (prospective) data are needed and hopefully, the follow-up study of the MOMS trial patients at school age will provide a clearer view of the urological outcome after prenatal OSB.

Obstetrical outcomes and risk factors for obstetrical complications were analyzed in a third substudy.⁴⁸ The article updated and expanded the information presented in the original MOMS report. Finally, the long-term impact and parental stress on the families of the women who participated in the MOMS trial were assessed.⁴⁹ Families of women who had prenatal repair had a significantly lower overall negative impact of caring for a child with spina bifida up to 30 months of age when compared with those that had postnatal repair.

Diagnostic Workup and Eligibility for Prenatal OSB Repair

Details on detection of spina bifida by ultrasound is a topic of its own and beyond this review article. However, ultrasound is not only crucial to diagnose the presence of a spina bifida, it also should provide information on the kind of spinal dysraphism (open or closed), on the functional level of the lower extremity, and on whether additional anomalies concerning the fetus or the placenta/uterus might be present that would preclude fetal repair. The sonographic appearance of the lesion itself may not always allow a distinction of the type of spinal dysraphism; however, presence or absence of hindbrain herniation can be used as an indirect indicator. In fact, in closed spinal dysraphism, such as meningocele, lipomyelocele, and myelocystocele, the Chiari II malformation is not typically present. Although presence or absence of Chiari II can be assessed with ultrasound, fetal magnetic resonance imaging (MRI) allows to best visualize it and is thus an absolutely necessary tool in the workup for prenatal OSB repair.^50,51 Figure 3 depicts a case with a closed dysraphism (meningocele) and absence of Chiari II.

Fig. 3

A fetus with a meningocele (closed dysraphism) and, therefore, not eligible for fetal spina bifida repair. There is no hindbrain herniation (asterisk). The lesion (arrowhead) could be misinterpreted as myelomeningocele. The thin skin coverage is not always clearly distinguishable. The thickened filum terminale (arrow) could be misjudged as protrusion of a placode

Both dynamic MRI sequences and ultrasound are useful to assess the functional level of the lower extremities and are thus central for counseling. With ultrasound, Carreras et al⁵² demonstrated in patients with prenatal OSB an agreement of over 88% between prenatal and postnatal segmental levels. Hence, the functional level at the time of screening is helpful to give an individualized prognosis in regard of future lower extremity function.

For the MOMS trial, several exclusion criteria were defined. Some of them represented clear contraindications to fetal surgery; others were specifically formulated for the trial to minimize confounding variables. Although most centers performing prenatal repair employ the majority of the criteria set in the MOMS trial (Table 1), some criteria have been discarded and others have been revised.⁵³ Generally speaking, for a patient to qualify for fetal surgery, the mother must be healthy and the fetus must not suffer from other pathologies than the spina bifida complex. Figure 4 shows an example of a fetus not eligible for fetal surgery due to a severe kyphosis and a suspected caudal regression syndrome. Further, prenatal OSB should be performed between 23.0 and 25.9 weeks of gestation. In the MOMS trial, the window for prenatal OSB repair was set between 19.0 and 25.9 gestational weeks. Study results from before the MOMS trial had shown that repair after 26 weeks would no longer yield a substantial benefit.²⁵ A study from CHOP that was published after the MOMS trial demonstrated that repair before 23 weeks is associated with higher rates of preterm premature rupture of membranes and chorioamniotic membrane separation;⁵⁴ it is, therefore, not any longer recommended to offer prenatal OSB repair before 23 weeks.

The Open Repair

The open approach, with the MOMS trial proven efficacy, is nowadays the gold standard for prenatal OSB repair. Here, we provide a short description of the procedure, although it simplifies enormously an operation, i.e., similar to a symphony, highly complex and only successful if orchestrated precisely and performed by adequately trained specialists. The procedure is done under general anesthesia and full uterine relaxation, typically obtained by a volative anesthetic agent. Access to the uterus is obtained by a transverse laparotomy. The border of the placenta is mapped with sterile ultrasound on the uterus and the hysterotomy is planned in a safe distance from it. Before entering the uterus, it is essential that the fetus is positioned in order to have the back in the designated area of the hysterotomy. The hysterotomy is usually performed with a stapler that places absorbable copolymer staples. The stapler has the advantage of compressing the uterine wall, thus impeding bleeding, and to avoid separation of the membranes by fixating them at hysterotomy border. During the whole procedure, the fetal heart rate and myocardial contractility are monitored with ultrasound. The uterine cavity is irrigated with warmed Ringer's solution throughout the procedure. The back of the fetus is centered in the hysterotomy wound and the lesion is repaired in the same technique as used postnatally: The zona epithelioserosa is resected, the dura is closed watertight and reinforced by paraspinal (myo)fascial flaps, and finally the skin is closed. In some cases, and, especially in myeloschisis, large skin defects may not allow a primary skin closure and avital dermal skin substitutes⁵⁵ can be used to close the skin gap. Rotation flaps, as used by our group, offer in these situations the advantage to have normal skin over the repair site.⁵⁶ After completion of the fetal repair, the uterus is closed in two layers and the amniotic fluid is replaced by Ringer's solution. Finally, the laparotomy is closed in layers.

Fig. 4

A fetus with a myelomeningocele (arrow) and a suspected caudal regression syndrome with massive kyphosis (arrowhead). The patient is not eligible for fetal repair

The learning curve of the procedure and the subsequent refinements of operative technique and perioperative management have led to better results and less complications than originally reported in the MOMS trial.^37,38,57,58 For instance, in our own current experience of 65 cases, we had only one patient with mild pulmonary edema (1.5%, MOMS 6%) and no patient requiring transfusion at delivery (MOMS 9%). Further, in our cohort, a higher rate of complete resolution of hindbrain herniation (over 90%)⁵⁹ (MOMS 36%), and a higher median gestational age at birth have been noted.⁶⁰ In addition, a different tocolysis management has led to markedly less side effects without compromising uterine efficacy.⁶⁰

Despite all these improvements, the procedure remains a great challenge and, especially, the hysterotomy remains a critical risk factor. A cesarean section before onset of labor is obligatory to avoid intralabor uterine rupture.

The Fetoscopic Repair

The maternal morbidity of open fetal repair prompted several groups in the world to endeavor to a minimal invasive approach.^52,61-66 While the fetoscopic approach might look at a first sight like an attractive alternative to open surgery (and will hopefully be the method of choice in the future), several unresolved problems do still hold back the procedure to become the gold standard. Although maternal morbidity is clearly decreased by the fetoscopic approach, the fetus is exposed to a significant higher risk of prematurity and to a questionable efficacy of the repair.

Different endoscopic repair techniques have been proposed. The access to the uterus might be achieved by a percutaneous approach or by exposing directly the uterus through a laparotomy as done in open repair. The truly minimal invasive percutaneous approach has important limitations: Port site (membrane) control, positioning, and holding the fetus are impossible or at best extremely challenging, and an anterior placenta may not allow access to the fetus at all. Heavy problems, such as fetal demise, strong trocar site bleeding that required termination of an uncompleted operation, incomplete or even failed patch coverage, oligohydramnios due to port site leaks, premature rupture of membranes, chorioamnionitis, and prematurity as low as 28 weeks have been encountered by groups using the percutaneous approach.^67-69 On the contrary, the technique where the uterus is exposed by a laparotomy allows port site control, a noninvasive positioning and fixation of the fetus, and a safe access also in patients with anterior placenta.⁶³ While in open fetal OSB surgery, the repair is done as postnatally with a three-layer closure (see earlier), the fetoscopic approach adopts simply patches or a direct skin closure. The patch coverage of the lesion might in the best case scenario exert a protective effect on the exposed spinal cord tissue and hinder leakage of cerebrospinal fluid; however, a nonwatertight closure with the patch technique is frequent and several patients have required postnatal neurosurgical repair. How well a patch or a simple skin closure over the exposed non-neurulated spinal cord is protective is not known. In addition, there is a debate whether uterine carbon dioxide insufflation might be harmful to the fetus.⁷⁰ In summary, although the endoscopic approach seems to reduce the bystander risk—i.e., the maternal risk, it also seems to do it on the cost of the actual patient—i.e., the fetus with spina bifida. Until it cannot be demonstrated (ideally by a randomized trial) that outcomes after fetoscopic repair are at least as good or even better than after open repair, the standard technique should remain the open approach.