The first cases of in-utero spina bifida repair were performed in 1994 using an endoscopic technique.29
This technique proved unsatisfactory and was abandoned. Percutaneous fetoscopic patch coverage has been tried more recently in a small series of patients, and has also proved problematic.30
In 1997, in-utero closure of spina bifida defects was performed by hysterotomy at Vanderbilt University31
and at CHOP.32
The selection criteria were different at the two institutions. At Vanderbilt, patients were not excluded based on prenatal ventricular size, late gestational age, spinal level, or presence or absence of fetal leg motion by in-utero sonogram. At CHOP, a fetus was only considered for surgery if the gestational age at the time of the proposed surgery was 26 weeks or less, if the transatrial ventricular diameter was less than 16 mm (normal being less than 10 mm), if the estimated level of the lesion was S1 or above, and if there was convincing leg and foot motion on ultrasound and in the absence of foot or leg deformity.
The early experience at these institutions suggested that compared with babies treated postnatally, those treated in utero
had a decreased incidence of hindbrain herniation,33,34
and that ascent of the hindbrain structures could be demonstrated within 3 weeks of the fetal closure using serial MRI. It is clear that the radiographic appearance of hindbrain herniation (the Chiari II malformation) is improved by the procedure but the other manifestations of the Chiari complex, such as thinning of the corpus callosum and polymicrogyria, are not. It is not yet clear whether this translates into improved survival or functional outcome. Although the posterior fossa volume of the normal developing fetus has been measured using MRI,35
the volume of the posterior fossa in fetal myelomeningocele patients has not. It is hypothesized that the volume is small and might be expanded by fetal surgery, but this remains unproven and is the subject of ongoing research.
The overall fetal head size has been demonstrated to be small in myelomeningocele patients, and to increase to normal after fetal surgery; the significance of this is uncertain. It appears that the head enlargement is largely due to restoration of the cerebrospinal volume, which is indicative of reversal of the hindbrain herniation.36
The fetal ventricles typically enlarge throughout gestation following fetal surgery. At present, however, no consideration is being given to placement of fetal shunts. With very short follow-up, it also appeared that this might have resulted in a decreased need for shunting. With somewhat longer follow-up this effect has been maintained to some extent, although some infants who did not require shunts in the newborn period have required shunts later on, usually within the first year. In the combined series of fetal surgery patients from CHOP and Vanderbilt, 104 patients followed for at least 1 year had an overall incidence of shunting of 54%, compared with 86% for a historical control group from CHOP.37
The effect was most evident for those with lumbar lesions, perhaps because of the larger number of these resulted in increased statistical power. The incidence of shunting in those patients who underwent fetal closure prior to 26 weeks of gestation was 42.7%, but was 75% in those who had fetal surgery after 25 weeks of gestation. It was hypothesized that early fetal closure of the spinal lesion eliminated the leakage of spinal fluid from the back, which put back-pressure on the hindbrain. This allowed reduction of the hindbrain hernia and relieved the obstruction of the outflow from the fourth ventricle. This apparent benefit might be due to selection bias or change in the indications for placing a shunt over time.
Most infants and children who have undergone fetal myelomeningocele closure have persistent ventriculomegaly, but often do not have overt signs or symptoms of increased intracranial pressure. The prevailing opinion is that these patients do not require shunts, but it is not yet known if the developmental and cognitive level of function of these children would be improved by more aggressive treatment of the ventriculomegaly.
Benefit in lower extremity function or sphincter continence has been difficult to demonstrate. Children with spina bifida treated with conventional postnatal closure have a level of neuologic function that correlates very well with the bony level of the defect as determined radiographically.6
The Vanderbilt series of early and late gestation fetal closures showed no improvement in leg function compared with historical controls for comparable spinal level, but no attempt was made to ascertain the degree of leg function prenatally.38
The CHOP criteria demanded intact leg and foot motion to be present prior to fetal surgery, and only included early-gestation repairs. In our series, 57% had better-than-predicted leg function at birth in the thoracic and lumbar patients, but follow-up was short.39
There is concern that some of the early benefit in terms of leg function might be at risk. Virtually all of the postnatal lumbosacral MRI studies of these patients suggest tethering, and recently some of the patients have developed symptomatic epidermoid inclusion cysts, which have required repeat surgery.40
It is unclear at this point whether this is a problem unique to fetal closure, or simply that it is being found because of the careful surveillance that these patients are required to undergo. Clinically symptomatic tethering and epidermoid inclusions are also seen in conventionally treated infants with myelomeningocele, particularly in those who undergo intensive neurourological surveillance.41
It is presumed that if neurological functioning of the lower extremities is preserved by fetal closure, symptomatic cord tethering is likely to be even more of a problem, as there is more function to be lost. Interestingly, the six CHOP patients who have required re-exploration for tethering have all had intraoperative electrophysiological monitoring, and all have shown intact motor nerve conduction even to the lower sacral levels (unpublished data).
The effect of fetal surgery on cognitive functioning has also been difficult to assess. It is known that the average IQ of children with myelomeningocele is significantly lower than that of control children, and that this appears to be due to the disease process itself, rather than associated complications such as shunt infection.42
Fetal surgery could theoretically improve outcome by reducing the incidence of hydrocephalus, or adversely impact outcome by increasing the incidence of prematurity. Preliminary data from CHOP showed a mean Mental Developmental Index of 90.8 in patients who underwent fetal surgery, which is probably not significantly different from postnatally treated individuals, and suggest no major effect of fetal surgery.43
An unexpected finding of fetal surgery is improved wound healing and decreased scar formation, resulting in a cosmetically more favorable back wound. This phenomenon has been extensively studied and has been attributed to down regulation of a transforming growth factor-beta modulator44
and increased endothelial growth factor.45
Fetal surgery is not without risk. Perinatal mortality at CHOP has been 6% (3/50), due to extreme prematurity associated with intrauterine infection in one case.39
The mean gestational age at delivery was 34 weeks 4 days.
There have been no maternal deaths in any fetal surgery series. No patient experienced hysterotomy dehiscence or rupture. As fetal surgery requires a classic cesarean hysterotomy high in the uterus, all future pregnancies require cesarean delivery. No data have been presented to suggest diminished fertility in any of the women undergoing fetal surgery for this or any other condition.