Farmer DL, Von Koch CS, Peacock WJ, Danielpour M, Gupta N, Lee H, Harrison MR. In utero repair of myelomeningoceleExperimental pathophysiology, initial clinical experience, and outcomes. Arch Surg. 2003;138(8):872-878. doi:10.1001/archsurg.138.8.872
Experimental work raises the possibility that in utero repair of myelomeningocele (mmc) may improve lower extremity, bladder, and bowel function, ameliorate the arnold-chiari malformation, and decrease the need for postnatal shunting.
We previously developed fetal lamb models to create and reverse lower extremity damage and the arnold-chiari malformation in utero. we then applied our extensive experience with fetal surgery, including fetal endoscopic (fetoscopic) surgical manipulation, to develop techniques for mmc repair.
A tertiary referral center.
All patients treated between 1998 and 2002 for a prenatally diagnosed mmc.
Either fetoscopic mmc repair, fetoscopic patch repair, or limited maternal hysterotomy and microsurgical 3-layered fetal mmc repair was performed.
Main outcome measures
Gestational age at delivery, survival, neurologic outcome, and need for ventricular shunting at 1 year.
Complete fetoscopic repair was accomplished in 1 fetus. two other fetuses underwent partial fetoscopic procedures. the remaining 10 patients underwent limited maternal hysterotomy and microsurgical 3-layered fetal mmc repair. four of 13 patients died, and the mean gestational age at delivery of 11 fetuses born alive was 31 weeks. five of 9 required ventricular shunting by age 1 year. in 2 patients, lower extremity function improved by more than 2 vertebral levels compared with prenatal ultrasonography. five of 10 patients who lived longer than 3 weeks required postnatal wound revision within 7 days after birth.
Fetoscopic repair, although feasible, does not yet yield optimal surgical results. open surgical repair before 22 weeks' gestation is physiologically sound and technically feasible. one third of patients appear to be spared the need for a shunt at age 1 year, but improvement in distal neurologic function is less clear. additionally, fetal mortality is associated with this procedure. our results complement the data published by groups at children's hospital of philadelphia, in pennsylvania, and vanderbilt university, nashville, tenn. a national institutes of health–sponsored prospective randomized trial is now underway at these 3 centers to compare fetal repair with postnatal repair.
MYelomeningocele (mmc) is a devastating birth defect that affects 5 of 10 000 live births worldwide. a nonlethal but highly morbid form of neural tube defect in which the meninges and neural tissue are exposed to the intrauterine environment, mmc is postulated to cause neurologic deficits in the fetus in stages. first, the original defect in neurulation creates the neural tube defect and any associated myelodysplasia, and then exposure to the intrauterine environment causes secondary trauma to the spinal cord. theoretically, fetal surgery for mmc would help avoid this latter injury. prenatal repair of mmc, with the initial intent to preserve distal neurologic function by covering the exposed spinal canal, has now been performed in the united states for approximately 5 years.
More than a decade ago, the fetal treatment center laboratory at the university of california, san francisco (ucsf) began to investigate the feasibility and efficacy of in utero repair of mmc. the rationale for early repair was based on pathological studies of human mmc that suggested that much of the damage to the spinal cord was secondary to exposure to amniotic fluid and trauma. we hypothesized that repairing the defect and protecting the cord before birth could preserve lower extremity neurologic function. using a fetal lamb model, we surgically exposed the spinal cord and demonstrated neurologic damage to lower extremity, bladder, and bowel function comparable to that seen in human fetuses with mmc1; we then showed that covering the spinal defect ameliorated neurologic damage.2
A second series of experiments tested the hypothesis that cerebral spinal fluid (csf) leakage from the open central canal into the amniotic fluid (as happens in human mmc) is related to the development of the arnold-chiari malformation and, perhaps, the development of hydrocephalus. opening the central canal of the spinal cord in fetal lambs produced hindbrain herniation analogous to the arnold-chiari malformation, and repairing this lesion, ie, sealing the csf leak, prevented hindbrain herniation in fetal lambs.3
Although the optimal timing of fetal repair is unknown, indirect evidence suggests that repair early in gestation would most likely yield improvement in neurologic function. ultrasonogram evaluation documenting the natural history of fetal mmc suggests that both the central and peripheral nervous system insults may be progressive.4- 7 The arnold-chiari malformation and ventriculomegaly often progress during fetal development, lower extremity movement seen before 17 to 20 weeks' gestation often diminishes later in gestation,6 and clubbing of the feet from the associated mmc appears to be progressive throughout gestation. second, experimental and clinical work in other areas of nervous system development suggest that plasticity is greatest in the young brain and nervous system.8,9 For example, recovery of function after central nervous system damage from traumatic and hypoxic insults is better in neonatal animals than in adult animals.
Other medical centers pursuing the strategy of in utero repair of mmc began to apply this work clinically.10- 15 Initially, the open approach was applied relatively late in gestation (after 28 weeks) on the principle that if preterm labor and delivery ensued the fetus would still be viable. although the results were equivocal, showing no demonstrable effect on lower extremity function and bladder or bowel control, a serendipitous finding was the decreased need for csf shunting after birth and the improvement of the arnold-chiari malformation on imaging studies.13,14 A recent publication of early experience with mmc at other centers reports improved ventricular shunt rates if fetal repair is performed before 25 weeks' gestation.15,16 This experimental work and the serendipitous clinical findings suggest that there may be benefit to in utero repair of mmc, particularly if performed early in gestation.
The technical feasibility of repair in the first half of gestation and the optimal method for intervention had not yet been determined. myelomeningocele is a nonlethal defect and does not justify significant maternal morbidity or fetal mortality. therefore, we initially chose to pursue a less invasive fetal endoscopic, or fetoscopic, approach to repair.
We report the development of our surgical approach to fetal mmc repair and describe the characteristics and outcomes of our initial clinical experience in 13 patients.
After obtaining committee on human research approval (no. h5881-22141), we searched the fetal treatment center database to identify which patients were referred to ucsf between 1998 and 2002 for further evaluation of fetal mmc. of the 86 patients referred with mmc, we identified 13 who underwent prenatal intervention for mmc at ucsf. of these 13 patients, the first 3 operations were performed under the auspices of our institutional fetal oversight committee, whereas later patients were treated as part of an institutional study protocol (committee on human research no. h603-16259). inclusion criteria for fetal mmc repair at ucsf included normal karyotyping, no associated developmental anomalies on fetal magnetic resonance imaging (mri), presence of arnold-chiari malformation, and presence of thoracic and/or lumbar mmc.
The 3 modes of repair used, fetoscopic mmc repair, fetoscopic patch repair, and limited maternal hysterotomy with microsurgical 3-layered fetal mmc repair, are described in the "results" section. the outcome measures evaluated were gestational age at delivery, survival, neurologic outcome, the need for ventricular shunting at age 1 year, and the need for postnatal wound revision.
A woman carrying a fetus with mmc at 25 weeks' gestation underwent prenatal evaluation using ultrasonography and mri, which revealed a severe arnold-chiari malformation, effacement of the cisterna magna, unilateral ventriculomegaly, and a spinal defect at the level of the l2 through s1 vertebrae. she underwent attempted fetoscopic placement of an alloderm patch (lifecell corporation, branchburg, nj). placement of a transuterine stay suture in the fetal buttock stabilized the fetus in the amniotic fluid. trocars were inserted, and dissection of the fetal skin surrounding the mmc defect was completed, both without difficulty. the alloderm patch was inserted into the uterus and tacked in 4 quadrants to the fetal skin after a circumferential edge of fetal skin was undermined. just prior to completion of the fetal patch closure, bleeding from the placenta necessitated opening the uterus to complete the procedure and control the bleeding. the remainder of the procedure was without incident, and the uterus was closed. no transfusions were required, and the patient had an uncomplicated postoperative recovery. she received magnesium sulfate tocolytics for 2 days, followed by indomethacin. the patient required oxygen by nasal cannula for 4 days and had mild interstitial pulmonary edema on a chest radiograph. the fetus showed no evidence of complications and was delivered at 35 weeks' gestational age by elective cesarean section. an ultrasonogram taken 2 days after delivery compared with an ultrasonogram taken at the time of fetal repair revealed an improved arnold-chiari malformation but increased ventriculomegaly. during definitive neurosurgical repair after birth, the patch was noted to have partially pulled away from the fetal skin, and its removal made definitive repair more difficult. nonetheless, standard postnatal repair was accomplished. the infant did well; however, she developed progressive hydrocephalus and required ventricular peritoneal shunting 6 weeks postnatally. at age 1 year, this child met all of her neurodevelopmental milestones, and her lower extremity motor and sensory function extended to the l4 level.
A 24-year-old woman presented to the ucsf fetal treatment center carrying a 22-week-gestational-age fetus with mmc. preoperative evaluation revealed an anterior placenta, a severe arnold-chiari malformation, no evidence of hydrocephalus, and partial agenesis of the corpus callosum. a sacral mmc defect at the l3 level as well as bilateral clubfeet were seen. at 24 weeks' gestation, she underwent maternal laparotomy with complete 2-layer fetoscopic repair of the mmc. we exposed the maternal uterus as previously described,17- 19 then placed three 5-mm radially expanding trocars and, again, a stay stitch through the fetal buttock. the placode was dissected free, and the neighboring dura was identified and closed over the neural elements using fine sutures. as a second layer, the skin was mobilized laterally using fetoscopy and brought together under some tension with 2-0 sutures tied with intracorporeal knots. the sutures tore, and additional sutures had to be placed. small gaps remained in the midline, and it was hoped that these would close as part of fetal wound healing. the fetus and mother tolerated the procedure well without complication. the mother had an unremarkable postoperative course with minimal need for tocolytics or supplemental oxygen. six weeks postoperatively, she developed spontaneous rupture of membranes, and she delivered 1 week later at 31 weeks' gestation. prenatal ultrasonography prior to delivery revealed no evidence of intracranial hemorrhage, but there was some increase in the size of the ventricles consistent with mild ventriculomegaly, which had increased since the first fetal preoperative scan. at birth, csf was leaking from the site of the intrauterine fetoscopic repair, and thus postnatal re-repair was required to close the csf defect. ventriculoperitoneal shunting was performed at the same time. lower extremity motor function was at the l3-4 level. the infant re-presented a month later with evidence of urosepsis and ultimately died of a septic illness.
A 29-year-old woman was admitted with a fetus at 19 weeks' gestation with a diagnosis of lumbosacral mmc. preoperative sonographic evaluation revealed a spinal mmc defect extending from the level l3 through s1, bilateral clubfeet, and a significant arnold-chiari malformation; however, the ventricles were normal in size. she underwent in utero repair of the mmc defect using maternal laparotomy and attempted fetoscopy, followed by a hysterotomy for open 3-layer microneurosurgical repair. because of the very large anterior placenta, laparotomy was required, with delivery of the gravid uterus. the placenta was mapped, and 3 radially expanding trocars were placed in the posterior portion of the uterus to allow visualization of the fetus. again, positioning of the fetus was challenging, and we ultimately chose to proceed with an open repair. the trocar sites were connected using the intrauterine stapler, and hysterotomy was performed. the neurosurgeons proceeded with a 3-layer repair including mobilization and untethering of the neural placode, retubularization of the placode, and dural closure. skin flaps were raised, and soft tissue coverage was attempted. because we were unable to completely approximate the skin edges without significant tension, a chorioamnion patch was harvested from within the uterus and used to complete soft tissue coverage. the neurosurgical procedure was completed without incident. however, the fetus had a dislocated ankle with laceration, most likely caused by initial attempts to position the fetus. this was also repaired in utero. the intrauterine procedure was further complicated by a period of fetal bradycardia. after epinephrine was administered and the fetus and cord were repositioned well within the fluid-filled amniotic cavity, the fetal heart rate returned to normal. at the completion of the procedure the fetus was stable. the uterus was closed in the standard 2-layer fashion. mother and fetus had an initially stable postoperative course. she required minimal tocolytics and no supplemental oxygen and was discharged on postoperative day 6 while taking nifedipine. postoperative ultrasonograms revealed no evidence of intracranial hemorrhage and no significant change in ventricular size. at 21 weeks' gestation, she developed premature rupture of membranes and underwent spontaneous abortion of a nonviable fetus.
After our initial technical experience and based on accumulated clinical experience at other centers, we temporarily abandoned the fetoscopic approach for mmc repair and proceeded with open fetal repair in the next cohort of 10 patients. briefly, a maternal laparotomy followed by a hysterotomy was performed in preparation for an open 3-layer microsurgical mmc repair. the neural placode was dissected free, allowed to drop into the spinal canal, and retubularized. the dura was mobilized laterally and closed over the placode in a watertight fashion. the skin was undermined extensively in attempts to achieve a complete closure. an amnion patch was used in 4 patients to improve skin closure. the fetus was returned to the uterus, and the uterus and maternal abdomen were closed in a standard fashion, as described above.
The clinical characteristics and outcomes of the first 13 patients who underwent fetal mmc repair are outlined in Table 1. our limited initial experience parallels that reported at other centers.10- 14 Follow-up times for the 5 male and 5 female fetuses surviving for longer than 1 week postnatally ranged from 1 to 45 months (mean, 17 months). the mean gestational age at mmc repair was relatively early at 22 weeks (range, 19-25 weeks). no maternal mortality and no significant maternal morbidity were seen. finite fetal or neonatal mortality did exist—4 (31%) of 13 patients—primarily due to preterm labor and delivery. the mean gestational age at delivery of viable fetuses was 31 weeks (range, 25-35 weeks). despite the use of fetoscopic and open mmc repair techniques, watertight dural and especially skin closure posed a challenge. we used amnion patches in 4 of 12 multilayer repairs to enhance skin closure. however, 5 of 10 patients surviving longer than 3 weeks required wound revisions within 7 days postnatally.
No fetus can be said to be cured of the problems associated with mmc. however, some improvements have been seen. the shunt rate in this series at 1 year was 5 (56%) of 9 patients, compared with the historic control shunt rate of nearly 90% for all fetuses with mmc, regardless of the level of lesion. three patients died before 1 year and did not have a shunt, and 1 patient is alive but has not yet reached the 1-year follow-up.
Most infants demonstrated minimal or no improvement in distal neurologic function over 1 to 2 levels, which is what can be expected with postnatal care. two patients surviving longer than 1 week had improved lower extremity function of 3 and 4 levels, respectively (patients 6 and 10; Table 1). however, postnatal mri or a radiograph was not available for confirmation of the postnatal level for 1 of the 2 patients (patient 10). for patient 10, the level of function at the postnatal examination was compared with the prenatal ultrasonogram. patient 9 died before we could confirm unexpectedly good results on the examination performed immediately postnatally. one patient underwent further untethering of the spinal cord at age 15 months after progressive worsening of lower extremity function. urologic function also does not appear to be improved, and all patients had variable amounts of bowel and bladder dysfunction.20
Myelomeningocele is a devastating birth defect that results in lifelong lower extremity neurologic deficiency, fecal and urinary incontinence, the arnold-chiari malformation, and hydrocephalus, most often requiring ventriculoperitoneal shunting. experimental work in animal models as well as early clinical work suggested that ongoing damage to the spinal cord might be alleviated by in utero repair and that the alteration in spinal fluid dynamics, ultimately leading to the need for ventriculoperitoneal shunting, might be decreased by in utero repair. the fetal natural history and the biological development of the nervous system suggest that early repair is more likely to improve altered spinal fluid dynamics and prevent ongoing injury to the exposed neural placode; therefore, we believe that early fetal repair is necessary, ideally prior to 20 weeks' gestational age. to facilitate early fetal repair and minimize maternal morbidity, we initially chose a fetoscopic repair. however, experience with our first 3 cases suggested that, although fetoscopic repair is feasible, it does not yet yield the optimal surgical result. we further demonstrated that repair at an early gestational age is physiologically sound and technically feasible using a limited open surgical approach with a microscopic 3-layer neurosurgical repair.
We found a reduction in the need for postnatal csf shunting similar to what has been reported in the literature.12,13 A recently published study combined patients with mmc from children's hospital of philadelphia, in pennsylvania, and vanderbilt university, nashville, tenn, to compare shunt rates of patients treated in utero with postnatal repair using historic controls.16 Patients had to reach at least 1 year of age to be included in the study. overall, only 55% of patients with fetal mmc repair required shunting compared with 86% of patients with postnatal repair. further, in patients at less than 25 weeks' gestation at the time of fetal mmc repair, the shunt rate was only 44%, suggesting a rationale for earlier fetal mmc repair. the results reached statistical significance. our patient group, although much smaller, showed a similar trend. all our fetal patients were at less than 25 weeks' gestation at the time of repair, and a shunt rate of 56% at age 1 year or older was seen. in contrast, no proven benefit to lower extremity neurologic function or bowel and bladder control with in utero mmc repair compared with postnatal repair has been demonstrated to date.
Myelomeningocele is the first nonfatal birth defect to be treated in utero. open fetal surgery requiring a maternal hysterotomy is not only a major surgical intervention but also carries a lifelong risk of uterine rupture during future pregnancies. further, most infants were delivered prematurely, which confers the added risks of prematurity to an infant already handicapped with spina bifida.
Because the relative benefits of mmc repair remain unproven for this nonfatal lesion and because there is significant maternal morbidity associated with uterine hysterotomy, we believe that fetal surgery for mmc should be limited to investigation only in the context of a properly controlled randomized trial. we initially submitted a national institutes of health (nih) proposal for a randomized controlled trial in 1998, before any human fetal mmc procedures had been performed. we resubmitted our proposal in 2000, received a favorable score of 152, and began the study. the national institute of child health and human development subsequently asked us to halt the study and join with our colleagues at children's hospital of philadelphia and vanderbilt university to form a multicenter trial. although it has taken 3 more years to accomplish, an nih-sponsored multicenter randomized controlled trial is about to begin. we remain committed to the performance of a proper trial of fetal surgery to treat this nonfatal disorder before the worldwide proliferation of an unproven therapy.
Corresponding author: diana l. farmer, md, fetal treatment center, university of california, san francisco, 513 parnassus ave, hsw-1601, san francisco, ca 94143-0570 (e-mail: firstname.lastname@example.org).
Accepted for publication april 18, 2003.
This study was presented at the 74th annual meeting of the pacific coast surgical association; february 16, 2003; monterey, calif; and is published after peer review and revision. the discussions that follow are based on the originally submitted manuscript and not the revised manuscript.
Robert s. sawin, md, seattle, wash: As dr farmer has shown, the promise of fixing fetal abnormalities before birth has been a sort of holy grail for those of us forced to face newborns with life-threatening problems such as congenital diaphragmatic hernia. however, the work she presented here today has taken a further step across that frontier. this procedure, fetal repair of a meningomyelocele, is being performed for a nonlethal, albeit very morbid, birth defect. as some institutions have been performing this fetal operation with alarming frequency, the san francisco group has been the voice of reason. indeed, through their efforts, the multi-institutional randomized trial dr farmer just described is about to start.
This paper today is another honest assessment and another voice of reason in this controversial topic. the paper highlights some very interesting concerns and raises several others. my questions will focus on 4 major issues. first, you mentioned maternal problems, and among those mentioned in the manuscript was pulmonary edema. has your center or any of the other centers had any life-threatening maternal morbidity with any of the fetal operations? does the hysterotomy that you now have evolved to mandate a cesarean section for all subsequent pregnancies and, if so, should the morbidity of these subsequent c-sections be factored into the risks of fetal surgery?
Second, your mean gestational age at birth in this series was approximately 30 to 31 weeks, and this obviously introduces other risk factors such as pulmonary immaturity, necrotizing enterocolitis, and other problems of premature infants. have you seen any such problems in any of your fetal surgery patients? and how about the potential impact of this prematurity on neurologic function, since intraventricular hemorrhage, cerebral palsy, and other central nervous system insults are much more common in premature infants?
Third, one of the difficulties that some of us have in the assessment of this treatment is identifying the best outcome variable to measure. in particular, as you mention, the indications for ventricular peritoneal shunting may not be clearly defined, and they may be subject to judgment. if so, might investigator bias account for the apparent decreased rate of vp shunting? your mean follow-up is only 17 months. is it possible that the need for shunting is merely deferred or delayed and that with longer follow-up the incidence of shunting will be similar in both groups?
Aside from shunting, why do you think that there has not been any demonstrable improvement in the motor, sensory, bowel, or bladder function, especially since these are the most important morbidities of meningomyelocele?
Lastly, you described in your manuscript one of your patients who developed a tethered spinal cord after birth. a New england journal article this july described 3 other fetal meningomyelocele patients who developed the same lesion at a surprisingly precocious age of less than a year. do you have any ideas about the apparent reasons for this risk factor? and, is it something that warrants serial mr scans for all infants with fetal meningomyelocele and, if so, for how long?
Again, i want to compliment dr farmer on her very honest assessment of this pioneering work and congratulate her on her membership in the pacific coast surgical association. i have to warn the association however that with 2 new pediatric surgeon members this year and 3 more proposed pediatric surgeon members for next year, we may be taking over the association.
Richard j. mullins, md, portland, ore: You have presented a spectacular novel surgical technique. i predict 5 years from now you are going to be better at this operation than you are now. are you prematurely starting your study? is there a risk you show no benefit because in point of fact you really haven't perfected the operation enough to give it a good trial?
Dr farmer: I will address the last question first. i personally agree 100%. we are perhaps premature in bringing this to the clinical arena, and i personally believe that this will ultimately be done laparoscopically, that we will be using some kind of glue to seal the csf defect, and we may not even be trying to address the distal neurologic function. one of the reasons this trial needed to be done was that there was tremendous worldwide interest in fetal surgery for this particular defect, interestingly. fetal surgery has now been done for 20 years. the first fetal case was completed in 1983. the interest, however, has just skyrocketed in the last 2 or 3 years. all over the world people want to start fetal surgery centers, and they want to do it for this defect, and they want to do open fetal surgery. this is a very vulnerable patient population. if you approached a pregnant woman and said, "if we cut off your right arm, your baby will do better," most women would do that. that is one of the reasons that the nih forced this trial, not only to answer the question but also to try to stop the proliferation of the procedure.
Thank you, dr sawin, for your very insightful questions. with respect to maternal complications, i am happy to report that there have been no maternal deaths associated with fetal surgery in any of the 3 centers in this trial. however, i am sad to report that there has been 1 death in the united states and 3 that have been rumored worldwide associated with fetal surgical interventions. the international fetal medicine and surgical society has now formed a cooperative to track this information and deal with it in some oversight fashion. even more important perhaps are the near misses. there are potential life-threatening complications. there were reports of uterine rupture from some of the other centers in the early reports of fetal myelomeningocele repair, and it is true that having a nonstandard cesarean section necessitates future cesarean deliveries with all pregnancies. if a woman were to go into labor after having had a nonstandard hysterotomy, she in fact would risk her life if she were unable to get to medical care. so it is not trivial, this issue of need for maternal hysterotomy.
With respect to your question regarding prematurity, i think it is very relevant to ask the question, whether adding the additional risk of prematurity to an already handicapped baby warrants the potential benefit that might accrue to some patients and what may be as few as 30% of the patients. sixty percent of the patients may have no improvement, yet risk prematurity.
The issue of how to design the trial and what outcome variables to look at was challenging. the truth of the matter is the only outcome variable that would allow the trial to be performed from a statistical point of view was ventricular peritoneal shunting. what everyone wants to know is do these kids do better? do they walk, and are they smarter? unfortunately, in order to do that, you would have to compare at each level, and you would need so many patients to stratify at each specific level given the small amount of improvement that one would not be able to get enough patients in order to answer that question. so ventricular peritoneal shunting is really the only outcome variable that has a chance of being statistically significant with a reasonable number of patients, not putting too many people at risk.
You could ask the question, since it appears that there has been minimal to no improvement in distal neurologic function, should we be doing the trial at all? or should this entire field be abandoned? i don't know the answer to that question. no child has been cured of spina bifida. no fetus will be able to walk that was not going to be able to walk before. is it good enough to not need a ventricular peritoneal shunt? i am actually concerned that even at the completion of the trial, we won't know the answer to that question. is the benefit of a third of the kids not needing a shunt (shunts are better than they used to be, and that's not such a bad operation to have after birth), does that benefit for a small number of the patients outweigh the risk to the fetuses and to the mother? the only other comparable field is that of living-related liver transplantation. would we do living-related liver transplants to make a third of the patients a little bit better? it's hard to say.
With respect to your final question regarding tethered cord and other late complications that we have seen as a result of fetal surgery, like all new surgical procedures, particularly those in the fetus, only time will tell what some of the secondary and late complications may be, and i think we will have to watch these patients closely.