- Email: [email protected]
The case for and against intrauterine surgery for myelomeningoceles
European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
The case for and against intrauterine surgery for myelomeningoceles Carys M. Bannister Fetal Management Unit, St. Mary's Hospital, Hathersedge Road, Whitwoth Park, Manchester, UK
Abstract The case for and against intrauterine surgery on a myelomeningocele depends on how and in what order the malformations in the spinal cord and brain are thought to develop. If the brain defects arise in the embryonic period and undergo no further change and if the spinal cord is so deformed that it is functionless from the start, then operative intervention before birth will have no signi®cant effect on the ultimate neurological defect. If, on the other hand, the brain lesions evolve during gestation and the deformed spinal cord has some useful function that can be lost by contact with the amniotic ¯uid or is susceptible to mechanical damage, then intrauterine surgery may have a bene®cial role. Long-term follow-up of children who have already undergone intrauterine surgery should answer whether this novel form of treatment imparts signi®cant bene®t to justify the risks it imposes on the mother and fetus. # 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Spina bi®da aperta; Intrauterine surgery; Arguments for and against
1. Introduction Spina bi®da aperta, of which myelomeningocele is a major part, is at the more serious end of the spectrum of lesions classi®ed as spina bi®da. In virtually all affected individuals the entire neuroaxis is to some degree maldeveloped. A number of theories have been put forwards to explain the abnormal development of the brain and the spinal cord in spina bi®da aperta. They include developmental arrest, tissue overgrowth, the effects of abnormal cerebrospinal ¯uid hydrodynamics, and bursting open of the previously closed neural tube. 2. Embryogenesis 2.1. Developmental arrest The developmental arrest theory proposes that part of the neural plaque destined to become the spinal cord fails to roll up and fuse, leaving a ¯attened plaque of undeveloped neural tissue on the dorsal surface of the fetus, which often involves ®ve or more vertebrae in the thoracic, lumbar or sacral spine [1]. The observation that myelomeningoceles most commonly arise in the lower lumbar and sacral regions suggests that abnormal closure of the posterior neuropore is somehow involved in their etiology. The abnormalities
E-mail address: [email protected] (C.M. Bannister).
affect not only those structures derived from the neural tube, but also those arising from the associated mesoderm with the result that abnormalities of the vertebrae and the meninges are always present. The Chiari II malformation almost inevitably accompanies a myelomeningocele. The arrest theory has also been used to explain the development of this posterior fossa malformation. It is suggested that because the pontine ¯exure fails to form, the pons, medulla and the adjacent part of the fourth ventricle are too long and cannot ®t into the posterior fossa, they to lie in the upper cervical spinal canal. When the related parts of the cerebellar vermis and choroid plexus begin to development at about 12 weeks, they do so in the upper cervical spinal canal [2±4]. 2.2. Overgrowth The overgrowth theory was ®rst proposed by Cleland in 1883 who suggested that excessive growth of tissue at the edges of the neural plate prevents it from inverting normally and it is therefore unable to fuse [5]. Overgrowth of the cerebellum and the brainstem along with a posterior fossa that is smaller than normal, leads to downward displacement of these structures and the development of the Chiari II malformation [6]. 2.3. Hydrodynamic theory The hydrodynamic theory was suggested by a number of authors before Gardner extended it further [7]. Gardner
0301-2115/00/$ ± see front matter # 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 3 0 1 - 2 1 1 5 ( 0 0 ) 0 0 4 3 3 - 4
110
C.M. Bannister / European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
believed that prior to the eighth or ninth week of gestation when the foramina of Luschka and Magendie normally develop in the roof of the fourth ventricle allowing CSF to escape from the ventricular system, a state of physiological hydrocephalus exists with distension of the central canal of the neural tube. If this is not relieved by perforation of the roof of the fourth ventricle at the correct time, further production of CSF causes the Chiari II malformation by pushing the cerebellar tonsils and part of the fourth ventricle into the upper cervical spinal canal and the myelomeningocele by bursting open the lower end of the distended central canal of the neural tube. 2.4. Secondary rupture Padget suggested that the primary de®cit in spina bi®da aperta was the bursting open of the lower part of the closed neural tube [8]. Escape of ¯uid from the central canal raised a bleb in the mesoderm beneath the ectoderm, that later ruptured to form the myelomeningocele. In this theory, the Chiari II malformation is a secondary response to the escape of ¯uid from the neural tube leading to microcephaly and a small posterior fossa. Later development of hindbrain structures in the small posterior fossa was thought to result in herniation of the cerebellar tonsils and part of the fourth ventricle into the upper cervical spinal canal [8]. McLone and associates in laboratory studies observed failure of neural tube closure after exhibition of a teratogen or rupture of a previously closed tube in mouse embryos exposed to large doses of vitamin A [9,10]. From the foregoing it is apparent that the exact embryogenesis of a myelomeningocele and the associated Chiari II malformation is still in dispute. What is undoubted is that irrespective of how the abnormalities arise in the spinal cord and the brain, they are identi®able in the embryo from as early as the eighth week of gestation, and are established in the fetus by the twelfth week. 3. Functional developemnt Throughout fetal life the spinal cord in spina bi®da aperta is represented by the neuroplaque lying on the dome of the myelomeningocele. The neuroplaque is comprised of very abnormal and deformed neural tissue that is extremely unlikely to undergo differentiation and maturation. Whilst it cannot be proved beyond doubt as to whether or not it has any function, it is dif®cult to imagine such a deformed structure ever becoming capable of performing many, if any, of the in®nitely complex functions associated with a normally developed spinal cord. A myelomeningocele is frequently not the only abnormality present in the spinal cord. The central canal remains open in many cases and an associated syrinx may arise at any time during the life of the affected individual. Magnetic resonance scans of spina bi®da aperta subjects frequently show that the spinal cord
is pathologically thin especially in the cervical region. Except for the syrinx, which may be a secondary phenomenon, the other developmental abnormalities are almost certainly primary. In subjects with spina bi®da aperta it is usual for the Chiari II malformation to be documented as the only abnormality present in the brain. Even when the Chiari II malformation is described, commonly the only features mentioned are elongation of the cerebellar tonsils together with the part of the fourth ventricle lying in the upper cervical canal. In reality the abnormalities of the Chiari II malformation are much more complex and include beaking of the tectum; kinking and caudal displacement of the pons, medulla and cervicomedullary junction; absence of the cisterna magna; and small, sometimes very small cerebellar hemispheres. All of these structures lie in an abnormally small posterior fossa. As well as these posterior brain defects, there are also numerous abnormalities in the cerebral hemispheres which can include one or more of the following; polymicrogyria, cortical heterotopia, an enlarged massa intermedia and ventriculomegaly. Hydrocephalus, de®ned as ventriculomegaly with raised intraventricular pressure, seldom develops in the fetus but does so after birth in at least 80% of conventionally managed cases. Until recently, with the exception of hydrocephalus, all the abnormalities of the brain were accepted as representing primary congenital defects. Both Sutton and Bruner and their colleagues have reported ®nding evidence that repair of the myelomeningocele in utero results in return of the cerebellar tonsils to the posterior fossa together with establishment of a cisterna magna. They have also recorded in their treated fetuses a reduced incidence of post-natal shunt dependent hydrocephalus [11,12]. These observations suggest that what was previously thought to be part of a ®xed lesion in the Chiari II malformation may not be so after all. Whether or not this ultimately proves to be the case, what remains doubtless is that there are widespread abnormalities present throughout the brain and spinal cord showing the extent and serious nature of the grossly disturbed development of the whole neuroaxis in spina bi®da aperta. 4. Neurologial dysfunction The maldevelopment of the spinal cord causes the majority of patients to be incontinent of urine and faeces, the former currently managed most often by three-to-four hourly intermittent catheterisation, and the latter by the administration of daily mini-enemas. Sooner or later their majority of patients will have to have major surgery on the bladder and bowel to deal with incontinence problems not solved by more simple measures. The maldevelopment of the spinal cord is almost always associated with some degree of weakness and lack of sensation in the lower limbs. Children under the age of eight years can often walk, sometimes without any aids or with lower limb splints,
C.M. Bannister / European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
and even those with more severe degrees of involvement may ambulate using devices such as reciprocating gait orthoses or parawalkers. As the children get older and heavier their weight/power ratio decreases and ambulation becomes more dif®cult, and in many the use of arm crutches becomes a necessity. Later on the majority of children give up the struggle to walk and become permanent wheelchair users, although with the advantage that they expend much less energy to get about and are able to carry their belongings with them. It should be expected that such widespread abnormalities would produce a recognisable pattern of abnormal neurological function in subjects af¯icted with spina bi®da aperta; and this is indeed broadly the case, although the severity of the symptoms in an individual varies in degree depending on extent of the abnormality in each of the affected parts of the neuroaxis. Whilst there is some correlation between the severity of the abnormalities in the spinal cord and those in the brain (in general the higher the lesion in the spinal cord, the more severe the abnormalities in the brain), this is by no means always so and any combination of abnormalities can and does occur. The consequence of this is that it is dif®cult, if not impossible, to construct meaningful matched groups of patients on whom to carry out controlled studies. Matched groups should contain equal numbers of individuals with the same degree of learning dif®culties due to poor short-term memory problems, poor concentration and poor sequencing abilities. The myelomeningocele should be at the same vertebral level and affect the same number of vertebrae. Finally they should also have the same type of neurogenic bladder and bowel dysfunction. 5. Treatment options The rationale for the treatment of spina bi®da aperta depends to a great extent on how the neuraxis lesions are believed to develop and what functional capabilities the abnormal structures are thought to have. 5.1. Embryomic Intervention All of the abnormalities might be correctable if it were thought that the lesions developed in sequence. It might be possible, for instance, to correct the ®rst lesion and in so doing prevent the subsequent ones from developing. This would be the case if hydrocephalus developed ®rst and caused the pressure in the ¯uid trapped in the neural tube to rise and the lower end of the spinal cord to rupture. If this is indeed what happens then the myelomeningocele could justi®ably be classi®ed as a secondary phenomenon. The expectation would be that correcting the intraventricular pressure would allow the development of the whole neuraxis to normalise, and that would include all of the abnormalities associated with the Chiari II malformation as well as those in the cerebral hemispheres. Even if there
111
was evidence to support this view of how a myelomeningocele and its associated abnormalities arise, for treatment to be effective it would have to be carried out very early in gestation, when none of the currently used procedures would be feasible. 5.2. Pre-labour delivery by caesarean section Treatment aims might be more realistic if, irrespective of how the lesions of spina bi®da aperta arose, the neuroplaque was believed to have useful function that could be lost at a later period of gestation or during delivery of the infant. It has been suggested that infants born by caesarean section have one or two lower vertebral levels of leg function compared to those born by vaginal delivery [13,14]. To account for these ®ndings it has been postulated that passage of the infant through the birth canal damages the neuroplaque. Cochrane and colleagues further suggested that it was not only the delivery of the infant that was damaging, but also the contractions of the uterus during labour that were harmful, and they recommended that the infants should be delivered by caesarean section before the onset of labour. However, more recent studies have been unable to demonstrate incontrovertibly that all infants with spina bi®da aperta bene®ted from being delivered by caesarean section before the amniotic membranes ruptured [15,16]. 5.3. Intrauterine repair Tulipan and his colleagues suggest that in the later weeks of gestation the amniotic ¯uid is damaging to the neuroplaque and for this reason, he and his team have been repairing myelomeningoceles in utero [17]. He reports that leg function after interuterine surgery is better than in infants treated post-natally. A number of issues are raised in assessing these results. As previously stated, there are considerable dif®culties when trying to match patients to make up control groups for comparison. It would be helpful if prenatal assessment of leg movements could be made. However, attempts at analysing leg movements in utero have been fraught with problems. Movement of limbs ¯oating in a ¯uid medium are not necessarily initiated and maintained by neuromuscular mechanisms under the control of the normal central nervous system, and these are the only important movements for forecasting future leg function. Unfortunately in utero there is still no certain way of distinguishing active movements from passive ones. An unexpected result of repairing the myelomeningocele lesion in utero has been the observation, by both the Sutton and Bruner groups, that cerebellar tonsilar herniation appears to be reduced post-operatively and that this is accompanied by the development of a cisterna magna [11,12]. Sutton and colleagues suggest that these changes are due to altered cerebrospinal ¯uid dynamics. They also claim that the incidence of infants developing hydrocephalus post-natally is reduced; in their series only one of the nine
112
C.M. Bannister / European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
surviving fetues who underwent intrauterine repair of a myelomeningocele had, at the time of reporting, a shunt inserted for hydrocephalus. On the other hand, 59% of the 29 fetuses operated on in the Bruner series have required shunts, these infants having been followed up for a minimum of 6 months. 5.4. Conventional treatment If one believes that the myelomeningocele and its associated abnormalities in the rest of the central nervous system are ®xed from the time they ®rst develop, then the de®cits seen at birth are the same as those which were present in the embryonic period and treatment, whenever it is carried out, cannot signi®cantly effect the ultimate functional outcome. The primary objective of conventional treatment is not to protect the neuroplaque but to prevent infection, particularly ascending infection of the CSF pathways which, if ventriculitis supervenes, can have a devastating effect on cerebral function. Conventional management of an infant born with a myelomeningocele is repair of the spinal cord lesion 24 to 48 h after birth. The procedure consists of separating the neuroplaque from the surrounding membranes, rolling it up into a tube and suturing the edges together and then letting it drop into the open spinal gutter. Membranes mobilised from the deep tissues of the back are sutured over the reconstructed neural tube, and ®nally the skin is freed up from the subcutaneous layers and closed. In at least 80% of cases hydrocephalus develops within 2 weeks of operation because the escape route for the CSF out of the ventricular system down the central canal and around the spinal cord is lost. The indications for inserting a shunt for hydrocephalus are those of raised intraventricular pressure and not the presence of ventriculomegaly. The most usual signs of raised intraventricular pressure in a neonate are a disproportionate increase in the head circumference, dilatation of veins on the scalp and forehead, a tense fontanelle, separation of the sutures of the cranial vault, the appearance of squints and `sunsetting' of the eyes. 6. Arguments for and against intrauterine surgery The case for or against intrauterine repair of the back lesion in fetuses with a myelomeningocele therefore depends on the evidence supporting either (a) whether the neuroplaque has any useful function at the time of its embryogenesis, and if function is present, will it be lost by damage from constituents in the amniotic ¯uid or by uterine contractions and passage of the infant through the birth canal, and (b) whether the cerebellar tonsils are shifted upwards by a change in the hydrodynamics of the CSF after repair of the myelomeningocele, and whether this, together with the appearance of a cisterna magna, decreases the incidence of post-natal shunt dependent hydrocephalus. At the moment, it is not known whether any of these are
valid, and it is still an open question whether it is justi®ed to institute any form of treatment before the child is born. Even if the aforementioned arguments in favour of intrauterine repair for myelomeningoceles turn out to be favourable, they would only be worthwhile if the treatment was risk-free for both the mother and the fetus any risks they would have to be judged against the outcome. There have already been reported complications for the mother and the fetus. In the Sutton series, there are no reported maternal complications; but in the Bruner series one mother had a small bowel obstruction 5 weeks after the hysterotomy; there was dehiscence and infection of the hysterotomy wound to which bowel became adherent. A second mother also had dehiscence of the hysterotomy wound. It should not be forgotten that all the mothers had to undergo at least two operative procedures for the intrauterine treatment of their fetuses. So far there is no data available to tell us whether this will affect the ability of these women to become pregnant again. Fetal complications have been more common. In the Sutton series one infant was delivered very prematurely immediately after the intrauterine repair of a myelomeningocele and died 2 days later. Three other infants were delivered at or before 30 weeks of gestation and all required ventilatory support. Another infant had a CSF leak from the site of the back repair and required an operation to close the ®stula. A further child had an operation at the age of seven months to revise the operation site for what was stated to be tethering of the spinal cord. In the Bruner series, ®ve infants were delivered before 30 weeks of gestation and one of these was extruded along with the placenta at the time of the hysterotomy. Taking these complications into consideration, they represent a signi®cant risk for both the mother and the fetus. As already pointed out, whether they can be justi®ed will depend on demonstrating improvement in the neurological status of the fetus, and it will be argued below any improvement must be dramatic if intrauterine repair of myelomeningoceles is to become standard treatment. 6.1. Goals for treatment Given the degree of disability suffered by the majority of patients with myelomeningoceles, new treatments that offer minimal bene®t should be assessed with all the above points in mind. New treatments deserve to be assessed but this must be carried out in a meaningful manner, and it is unlikely that the answers will be provided by a randomised controlled trial. Instead, goals for treatment should be set. From the point of view of the brain, not only should new treatment make it unnecessary to insert a cerebrospinal ¯uid shunt system, brain development should be normalised rendering the subject capable of living independently as an adult and able to earn a living. In addition, and ranking even higher in importance, the function of the spinal cord should be improved to such an extent that the subject will have normal bladder and bowel control and males normal sexual func-
C.M. Bannister / European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
tion. Finally, fully-grown subjects should be able to walk without or with only minimal leg bracing, and free hands. Any new treatment that ful®ls these goals would be worthy of being heralded as a major advance in the treatment of myelomeningoceles. The fetuses who have already been operated on, if they are followed up carefully and for a suf®cient length of time, should be able to provide the answer as to whether intrauterine repair of the back lesion meets these goals. If the goals are not met then for those parents who elect to continue a pregnancy with a fetus diagnosed as having a myelomeningocele, the best and safest management for both the mother and the fetus is to continue with the present policy of delivering the child through the birth passages and carrying out a repair of the back lesion soon after birth, and inserting a CSF diversion system should the patient develop signs of raised intraventricular pressure. 6.2. Parental expectations It must also be taken into account that the parental expectations of major improvements are a powerful argument for them to choose treatment; particularly ®rst-time parents of a fetus with spina bi®da aperta. No matter how much informants think they are giving unbiased information about the positive and negative aspects of the condition, loving parents will almost invariably only hear the good things and will consequently feel obliged to give their fetus the chance of what they hope and expect will be a normal or near normal life. Whilst it may be dif®cult to think of it in this way, until the bene®ts of pre-natal treatment are proven, the offer of pre-natal surgery is depriving parents of the choice of terminating the pregnancy which pre-natal diagnosis made possible in the past. If a good outcome from treatment is not realised, then these unfortunate parents will be adding countless numbers of children to those already born with spina bi®da aperta. These children have abnormally developed brains, whether or not a shunt is inserted, suffer from poor short-term memory, poor concentration, and inability to sequence life management events, all of which result in learning dif®culties and an almost certainly lifetime dependence on others for daily living needs [18]. If a CSF diversion system has been implanted, it will probably require many procedures to be carried out during the rest of the patient's life to maintain its function. It cannot be contested that spina bi®da aperta is anything other than a major disaster affecting the whole of the central nervous system, and minor improvements in the
113
neurological status do not make the outcome signi®cantly better. References [1] Warkany J, Wilson JC, Geiger JF. Myeloschisis and myelomeningocele produced experimentally in the rat. J Comp Neurol 1958;109:34±41. [2] Daniel PM, Strich SJ. Some observations on the congenital deformity of the central nervous system known as the Arnold±Chiari malformation. J Neuropathol 1958;17:255±60. [3] Peach B. The Arnold±Chiari malformation: morphogenesis. Arch Neurol 1965;12:527±38. [4] Emery JL, Levrick RK. The movement of the brain stemand the vessels around the brain stem in children with hydrocephalus and the Arnold±Chiari malformation. Dev Med Child Neurol 1966;11:49±53. [5] Cleleand J. Contributions to the study of spina bifida, encephalocele and anencephalus. J Anat Physiol 1883;17:238±58. [6] Barry A, Patten BM, Stewart BH. Possible factors in the development of the Arnold-Chiari malformation. J Neurosurg 1957;24:285±90. [7] Gardner WJ. The dysraphic states from syringomyelia to anencephaly, Amsterdam, Excerpta Medica, 1973, pp. 1±201. [8] Padget DH. Spina bifida embryonic neuroschisis Ð a causal relationship. Definition of the postnatal conformations involving a bifid spine. Johns Hopkins Med J 1968;128:233±41. [9] McLone DC, Suna J, Collins J. Neuroulation: biochemical and morphological studies on primary and secondary neural tube defects. In: Humphreys RP (Ed.), Conceps in Pediatric Neurourgery Vol. 4. Basel, S. Karger, 1983;15±19. [10] McLone DG, Knepper PA. On the role of complex carbohydrates and neurulation. Pediatr Neurosc 1986;12:2±8. [11] Sutton LN, Adzick NS, Bilaniuk LT, Johnson MP, Crombleholme TM, Flake AW. Improvement in hindbrain herniation demonstrated by serial fetal magnetic resonance imaging following fetal surgery for myelomeningocele. JAMA 1999;282:1826±31. [12] Bruner JP, Tulipan N. Fetal surgery for myelomeningocele and the incidence for shunt-dependent hydrocephalus. JAMA 1999;282: 1819±25. [13] Bensen JT, Dillard RG, Burton BK. Open spina bifida: does cesarean section improve prognosis? Obstet Gynecol 1988;71:532±4. [14] Cochrane D, Aaronyk K, Wilson D, Steinbok P. The effects of labour and delivery on spinal cord function and ambulation in patients with meningomyelocele. Childs Nerv Syst 1991;7:312±5. [15] Hill AE, Beattie F. Does cesarean section delivery improve neurological outcome in open spina bifida? Eur J Pediat Surg 1994;4(Suppl. 1):32±4. [16] Liu SL, Shurtleff DB, Ellenbogen RG, Loeser LD, Kroop R. 19 year follow-up of fetal myelomeningocele brought to term. Eur J Pediat Surg 1999;9(Suppl. I):12±4. [17] Tulipan N, Bruner MHernanz-SchulmanJP. Reduced hindbrain herniation after intrauterine myelomeningocele repair: a report of four cases. Pediatric Neurosurgery 1998;29:274±8. [18] Hunt GM. Non-selective intervention in newborn babies with open spina bifida: the outcome 30 years on for the complete cohort. Eur J Pediat Surg 1999;9(Suppl. 1):5±8.
The case for and against intrauterine surgery for myelomeningoceles Carys M. Bannister Fetal Management Unit, St. Mary's Hospital, Hathersedge Road, Whitwoth Park, Manchester, UK
Abstract The case for and against intrauterine surgery on a myelomeningocele depends on how and in what order the malformations in the spinal cord and brain are thought to develop. If the brain defects arise in the embryonic period and undergo no further change and if the spinal cord is so deformed that it is functionless from the start, then operative intervention before birth will have no signi®cant effect on the ultimate neurological defect. If, on the other hand, the brain lesions evolve during gestation and the deformed spinal cord has some useful function that can be lost by contact with the amniotic ¯uid or is susceptible to mechanical damage, then intrauterine surgery may have a bene®cial role. Long-term follow-up of children who have already undergone intrauterine surgery should answer whether this novel form of treatment imparts signi®cant bene®t to justify the risks it imposes on the mother and fetus. # 2000 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Spina bi®da aperta; Intrauterine surgery; Arguments for and against
1. Introduction Spina bi®da aperta, of which myelomeningocele is a major part, is at the more serious end of the spectrum of lesions classi®ed as spina bi®da. In virtually all affected individuals the entire neuroaxis is to some degree maldeveloped. A number of theories have been put forwards to explain the abnormal development of the brain and the spinal cord in spina bi®da aperta. They include developmental arrest, tissue overgrowth, the effects of abnormal cerebrospinal ¯uid hydrodynamics, and bursting open of the previously closed neural tube. 2. Embryogenesis 2.1. Developmental arrest The developmental arrest theory proposes that part of the neural plaque destined to become the spinal cord fails to roll up and fuse, leaving a ¯attened plaque of undeveloped neural tissue on the dorsal surface of the fetus, which often involves ®ve or more vertebrae in the thoracic, lumbar or sacral spine [1]. The observation that myelomeningoceles most commonly arise in the lower lumbar and sacral regions suggests that abnormal closure of the posterior neuropore is somehow involved in their etiology. The abnormalities
E-mail address: [email protected] (C.M. Bannister).
affect not only those structures derived from the neural tube, but also those arising from the associated mesoderm with the result that abnormalities of the vertebrae and the meninges are always present. The Chiari II malformation almost inevitably accompanies a myelomeningocele. The arrest theory has also been used to explain the development of this posterior fossa malformation. It is suggested that because the pontine ¯exure fails to form, the pons, medulla and the adjacent part of the fourth ventricle are too long and cannot ®t into the posterior fossa, they to lie in the upper cervical spinal canal. When the related parts of the cerebellar vermis and choroid plexus begin to development at about 12 weeks, they do so in the upper cervical spinal canal [2±4]. 2.2. Overgrowth The overgrowth theory was ®rst proposed by Cleland in 1883 who suggested that excessive growth of tissue at the edges of the neural plate prevents it from inverting normally and it is therefore unable to fuse [5]. Overgrowth of the cerebellum and the brainstem along with a posterior fossa that is smaller than normal, leads to downward displacement of these structures and the development of the Chiari II malformation [6]. 2.3. Hydrodynamic theory The hydrodynamic theory was suggested by a number of authors before Gardner extended it further [7]. Gardner
0301-2115/00/$ ± see front matter # 2000 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 3 0 1 - 2 1 1 5 ( 0 0 ) 0 0 4 3 3 - 4
110
C.M. Bannister / European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
believed that prior to the eighth or ninth week of gestation when the foramina of Luschka and Magendie normally develop in the roof of the fourth ventricle allowing CSF to escape from the ventricular system, a state of physiological hydrocephalus exists with distension of the central canal of the neural tube. If this is not relieved by perforation of the roof of the fourth ventricle at the correct time, further production of CSF causes the Chiari II malformation by pushing the cerebellar tonsils and part of the fourth ventricle into the upper cervical spinal canal and the myelomeningocele by bursting open the lower end of the distended central canal of the neural tube. 2.4. Secondary rupture Padget suggested that the primary de®cit in spina bi®da aperta was the bursting open of the lower part of the closed neural tube [8]. Escape of ¯uid from the central canal raised a bleb in the mesoderm beneath the ectoderm, that later ruptured to form the myelomeningocele. In this theory, the Chiari II malformation is a secondary response to the escape of ¯uid from the neural tube leading to microcephaly and a small posterior fossa. Later development of hindbrain structures in the small posterior fossa was thought to result in herniation of the cerebellar tonsils and part of the fourth ventricle into the upper cervical spinal canal [8]. McLone and associates in laboratory studies observed failure of neural tube closure after exhibition of a teratogen or rupture of a previously closed tube in mouse embryos exposed to large doses of vitamin A [9,10]. From the foregoing it is apparent that the exact embryogenesis of a myelomeningocele and the associated Chiari II malformation is still in dispute. What is undoubted is that irrespective of how the abnormalities arise in the spinal cord and the brain, they are identi®able in the embryo from as early as the eighth week of gestation, and are established in the fetus by the twelfth week. 3. Functional developemnt Throughout fetal life the spinal cord in spina bi®da aperta is represented by the neuroplaque lying on the dome of the myelomeningocele. The neuroplaque is comprised of very abnormal and deformed neural tissue that is extremely unlikely to undergo differentiation and maturation. Whilst it cannot be proved beyond doubt as to whether or not it has any function, it is dif®cult to imagine such a deformed structure ever becoming capable of performing many, if any, of the in®nitely complex functions associated with a normally developed spinal cord. A myelomeningocele is frequently not the only abnormality present in the spinal cord. The central canal remains open in many cases and an associated syrinx may arise at any time during the life of the affected individual. Magnetic resonance scans of spina bi®da aperta subjects frequently show that the spinal cord
is pathologically thin especially in the cervical region. Except for the syrinx, which may be a secondary phenomenon, the other developmental abnormalities are almost certainly primary. In subjects with spina bi®da aperta it is usual for the Chiari II malformation to be documented as the only abnormality present in the brain. Even when the Chiari II malformation is described, commonly the only features mentioned are elongation of the cerebellar tonsils together with the part of the fourth ventricle lying in the upper cervical canal. In reality the abnormalities of the Chiari II malformation are much more complex and include beaking of the tectum; kinking and caudal displacement of the pons, medulla and cervicomedullary junction; absence of the cisterna magna; and small, sometimes very small cerebellar hemispheres. All of these structures lie in an abnormally small posterior fossa. As well as these posterior brain defects, there are also numerous abnormalities in the cerebral hemispheres which can include one or more of the following; polymicrogyria, cortical heterotopia, an enlarged massa intermedia and ventriculomegaly. Hydrocephalus, de®ned as ventriculomegaly with raised intraventricular pressure, seldom develops in the fetus but does so after birth in at least 80% of conventionally managed cases. Until recently, with the exception of hydrocephalus, all the abnormalities of the brain were accepted as representing primary congenital defects. Both Sutton and Bruner and their colleagues have reported ®nding evidence that repair of the myelomeningocele in utero results in return of the cerebellar tonsils to the posterior fossa together with establishment of a cisterna magna. They have also recorded in their treated fetuses a reduced incidence of post-natal shunt dependent hydrocephalus [11,12]. These observations suggest that what was previously thought to be part of a ®xed lesion in the Chiari II malformation may not be so after all. Whether or not this ultimately proves to be the case, what remains doubtless is that there are widespread abnormalities present throughout the brain and spinal cord showing the extent and serious nature of the grossly disturbed development of the whole neuroaxis in spina bi®da aperta. 4. Neurologial dysfunction The maldevelopment of the spinal cord causes the majority of patients to be incontinent of urine and faeces, the former currently managed most often by three-to-four hourly intermittent catheterisation, and the latter by the administration of daily mini-enemas. Sooner or later their majority of patients will have to have major surgery on the bladder and bowel to deal with incontinence problems not solved by more simple measures. The maldevelopment of the spinal cord is almost always associated with some degree of weakness and lack of sensation in the lower limbs. Children under the age of eight years can often walk, sometimes without any aids or with lower limb splints,
C.M. Bannister / European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
and even those with more severe degrees of involvement may ambulate using devices such as reciprocating gait orthoses or parawalkers. As the children get older and heavier their weight/power ratio decreases and ambulation becomes more dif®cult, and in many the use of arm crutches becomes a necessity. Later on the majority of children give up the struggle to walk and become permanent wheelchair users, although with the advantage that they expend much less energy to get about and are able to carry their belongings with them. It should be expected that such widespread abnormalities would produce a recognisable pattern of abnormal neurological function in subjects af¯icted with spina bi®da aperta; and this is indeed broadly the case, although the severity of the symptoms in an individual varies in degree depending on extent of the abnormality in each of the affected parts of the neuroaxis. Whilst there is some correlation between the severity of the abnormalities in the spinal cord and those in the brain (in general the higher the lesion in the spinal cord, the more severe the abnormalities in the brain), this is by no means always so and any combination of abnormalities can and does occur. The consequence of this is that it is dif®cult, if not impossible, to construct meaningful matched groups of patients on whom to carry out controlled studies. Matched groups should contain equal numbers of individuals with the same degree of learning dif®culties due to poor short-term memory problems, poor concentration and poor sequencing abilities. The myelomeningocele should be at the same vertebral level and affect the same number of vertebrae. Finally they should also have the same type of neurogenic bladder and bowel dysfunction. 5. Treatment options The rationale for the treatment of spina bi®da aperta depends to a great extent on how the neuraxis lesions are believed to develop and what functional capabilities the abnormal structures are thought to have. 5.1. Embryomic Intervention All of the abnormalities might be correctable if it were thought that the lesions developed in sequence. It might be possible, for instance, to correct the ®rst lesion and in so doing prevent the subsequent ones from developing. This would be the case if hydrocephalus developed ®rst and caused the pressure in the ¯uid trapped in the neural tube to rise and the lower end of the spinal cord to rupture. If this is indeed what happens then the myelomeningocele could justi®ably be classi®ed as a secondary phenomenon. The expectation would be that correcting the intraventricular pressure would allow the development of the whole neuraxis to normalise, and that would include all of the abnormalities associated with the Chiari II malformation as well as those in the cerebral hemispheres. Even if there
111
was evidence to support this view of how a myelomeningocele and its associated abnormalities arise, for treatment to be effective it would have to be carried out very early in gestation, when none of the currently used procedures would be feasible. 5.2. Pre-labour delivery by caesarean section Treatment aims might be more realistic if, irrespective of how the lesions of spina bi®da aperta arose, the neuroplaque was believed to have useful function that could be lost at a later period of gestation or during delivery of the infant. It has been suggested that infants born by caesarean section have one or two lower vertebral levels of leg function compared to those born by vaginal delivery [13,14]. To account for these ®ndings it has been postulated that passage of the infant through the birth canal damages the neuroplaque. Cochrane and colleagues further suggested that it was not only the delivery of the infant that was damaging, but also the contractions of the uterus during labour that were harmful, and they recommended that the infants should be delivered by caesarean section before the onset of labour. However, more recent studies have been unable to demonstrate incontrovertibly that all infants with spina bi®da aperta bene®ted from being delivered by caesarean section before the amniotic membranes ruptured [15,16]. 5.3. Intrauterine repair Tulipan and his colleagues suggest that in the later weeks of gestation the amniotic ¯uid is damaging to the neuroplaque and for this reason, he and his team have been repairing myelomeningoceles in utero [17]. He reports that leg function after interuterine surgery is better than in infants treated post-natally. A number of issues are raised in assessing these results. As previously stated, there are considerable dif®culties when trying to match patients to make up control groups for comparison. It would be helpful if prenatal assessment of leg movements could be made. However, attempts at analysing leg movements in utero have been fraught with problems. Movement of limbs ¯oating in a ¯uid medium are not necessarily initiated and maintained by neuromuscular mechanisms under the control of the normal central nervous system, and these are the only important movements for forecasting future leg function. Unfortunately in utero there is still no certain way of distinguishing active movements from passive ones. An unexpected result of repairing the myelomeningocele lesion in utero has been the observation, by both the Sutton and Bruner groups, that cerebellar tonsilar herniation appears to be reduced post-operatively and that this is accompanied by the development of a cisterna magna [11,12]. Sutton and colleagues suggest that these changes are due to altered cerebrospinal ¯uid dynamics. They also claim that the incidence of infants developing hydrocephalus post-natally is reduced; in their series only one of the nine
112
C.M. Bannister / European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
surviving fetues who underwent intrauterine repair of a myelomeningocele had, at the time of reporting, a shunt inserted for hydrocephalus. On the other hand, 59% of the 29 fetuses operated on in the Bruner series have required shunts, these infants having been followed up for a minimum of 6 months. 5.4. Conventional treatment If one believes that the myelomeningocele and its associated abnormalities in the rest of the central nervous system are ®xed from the time they ®rst develop, then the de®cits seen at birth are the same as those which were present in the embryonic period and treatment, whenever it is carried out, cannot signi®cantly effect the ultimate functional outcome. The primary objective of conventional treatment is not to protect the neuroplaque but to prevent infection, particularly ascending infection of the CSF pathways which, if ventriculitis supervenes, can have a devastating effect on cerebral function. Conventional management of an infant born with a myelomeningocele is repair of the spinal cord lesion 24 to 48 h after birth. The procedure consists of separating the neuroplaque from the surrounding membranes, rolling it up into a tube and suturing the edges together and then letting it drop into the open spinal gutter. Membranes mobilised from the deep tissues of the back are sutured over the reconstructed neural tube, and ®nally the skin is freed up from the subcutaneous layers and closed. In at least 80% of cases hydrocephalus develops within 2 weeks of operation because the escape route for the CSF out of the ventricular system down the central canal and around the spinal cord is lost. The indications for inserting a shunt for hydrocephalus are those of raised intraventricular pressure and not the presence of ventriculomegaly. The most usual signs of raised intraventricular pressure in a neonate are a disproportionate increase in the head circumference, dilatation of veins on the scalp and forehead, a tense fontanelle, separation of the sutures of the cranial vault, the appearance of squints and `sunsetting' of the eyes. 6. Arguments for and against intrauterine surgery The case for or against intrauterine repair of the back lesion in fetuses with a myelomeningocele therefore depends on the evidence supporting either (a) whether the neuroplaque has any useful function at the time of its embryogenesis, and if function is present, will it be lost by damage from constituents in the amniotic ¯uid or by uterine contractions and passage of the infant through the birth canal, and (b) whether the cerebellar tonsils are shifted upwards by a change in the hydrodynamics of the CSF after repair of the myelomeningocele, and whether this, together with the appearance of a cisterna magna, decreases the incidence of post-natal shunt dependent hydrocephalus. At the moment, it is not known whether any of these are
valid, and it is still an open question whether it is justi®ed to institute any form of treatment before the child is born. Even if the aforementioned arguments in favour of intrauterine repair for myelomeningoceles turn out to be favourable, they would only be worthwhile if the treatment was risk-free for both the mother and the fetus any risks they would have to be judged against the outcome. There have already been reported complications for the mother and the fetus. In the Sutton series, there are no reported maternal complications; but in the Bruner series one mother had a small bowel obstruction 5 weeks after the hysterotomy; there was dehiscence and infection of the hysterotomy wound to which bowel became adherent. A second mother also had dehiscence of the hysterotomy wound. It should not be forgotten that all the mothers had to undergo at least two operative procedures for the intrauterine treatment of their fetuses. So far there is no data available to tell us whether this will affect the ability of these women to become pregnant again. Fetal complications have been more common. In the Sutton series one infant was delivered very prematurely immediately after the intrauterine repair of a myelomeningocele and died 2 days later. Three other infants were delivered at or before 30 weeks of gestation and all required ventilatory support. Another infant had a CSF leak from the site of the back repair and required an operation to close the ®stula. A further child had an operation at the age of seven months to revise the operation site for what was stated to be tethering of the spinal cord. In the Bruner series, ®ve infants were delivered before 30 weeks of gestation and one of these was extruded along with the placenta at the time of the hysterotomy. Taking these complications into consideration, they represent a signi®cant risk for both the mother and the fetus. As already pointed out, whether they can be justi®ed will depend on demonstrating improvement in the neurological status of the fetus, and it will be argued below any improvement must be dramatic if intrauterine repair of myelomeningoceles is to become standard treatment. 6.1. Goals for treatment Given the degree of disability suffered by the majority of patients with myelomeningoceles, new treatments that offer minimal bene®t should be assessed with all the above points in mind. New treatments deserve to be assessed but this must be carried out in a meaningful manner, and it is unlikely that the answers will be provided by a randomised controlled trial. Instead, goals for treatment should be set. From the point of view of the brain, not only should new treatment make it unnecessary to insert a cerebrospinal ¯uid shunt system, brain development should be normalised rendering the subject capable of living independently as an adult and able to earn a living. In addition, and ranking even higher in importance, the function of the spinal cord should be improved to such an extent that the subject will have normal bladder and bowel control and males normal sexual func-
C.M. Bannister / European Journal of Obstetrics & Gynecology and Reproductive Biology 92 (2000) 109±113
tion. Finally, fully-grown subjects should be able to walk without or with only minimal leg bracing, and free hands. Any new treatment that ful®ls these goals would be worthy of being heralded as a major advance in the treatment of myelomeningoceles. The fetuses who have already been operated on, if they are followed up carefully and for a suf®cient length of time, should be able to provide the answer as to whether intrauterine repair of the back lesion meets these goals. If the goals are not met then for those parents who elect to continue a pregnancy with a fetus diagnosed as having a myelomeningocele, the best and safest management for both the mother and the fetus is to continue with the present policy of delivering the child through the birth passages and carrying out a repair of the back lesion soon after birth, and inserting a CSF diversion system should the patient develop signs of raised intraventricular pressure. 6.2. Parental expectations It must also be taken into account that the parental expectations of major improvements are a powerful argument for them to choose treatment; particularly ®rst-time parents of a fetus with spina bi®da aperta. No matter how much informants think they are giving unbiased information about the positive and negative aspects of the condition, loving parents will almost invariably only hear the good things and will consequently feel obliged to give their fetus the chance of what they hope and expect will be a normal or near normal life. Whilst it may be dif®cult to think of it in this way, until the bene®ts of pre-natal treatment are proven, the offer of pre-natal surgery is depriving parents of the choice of terminating the pregnancy which pre-natal diagnosis made possible in the past. If a good outcome from treatment is not realised, then these unfortunate parents will be adding countless numbers of children to those already born with spina bi®da aperta. These children have abnormally developed brains, whether or not a shunt is inserted, suffer from poor short-term memory, poor concentration, and inability to sequence life management events, all of which result in learning dif®culties and an almost certainly lifetime dependence on others for daily living needs [18]. If a CSF diversion system has been implanted, it will probably require many procedures to be carried out during the rest of the patient's life to maintain its function. It cannot be contested that spina bi®da aperta is anything other than a major disaster affecting the whole of the central nervous system, and minor improvements in the
113
neurological status do not make the outcome signi®cantly better. References [1] Warkany J, Wilson JC, Geiger JF. Myeloschisis and myelomeningocele produced experimentally in the rat. J Comp Neurol 1958;109:34±41. [2] Daniel PM, Strich SJ. Some observations on the congenital deformity of the central nervous system known as the Arnold±Chiari malformation. J Neuropathol 1958;17:255±60. [3] Peach B. The Arnold±Chiari malformation: morphogenesis. Arch Neurol 1965;12:527±38. [4] Emery JL, Levrick RK. The movement of the brain stemand the vessels around the brain stem in children with hydrocephalus and the Arnold±Chiari malformation. Dev Med Child Neurol 1966;11:49±53. [5] Cleleand J. Contributions to the study of spina bifida, encephalocele and anencephalus. J Anat Physiol 1883;17:238±58. [6] Barry A, Patten BM, Stewart BH. Possible factors in the development of the Arnold-Chiari malformation. J Neurosurg 1957;24:285±90. [7] Gardner WJ. The dysraphic states from syringomyelia to anencephaly, Amsterdam, Excerpta Medica, 1973, pp. 1±201. [8] Padget DH. Spina bifida embryonic neuroschisis Ð a causal relationship. Definition of the postnatal conformations involving a bifid spine. Johns Hopkins Med J 1968;128:233±41. [9] McLone DC, Suna J, Collins J. Neuroulation: biochemical and morphological studies on primary and secondary neural tube defects. In: Humphreys RP (Ed.), Conceps in Pediatric Neurourgery Vol. 4. Basel, S. Karger, 1983;15±19. [10] McLone DG, Knepper PA. On the role of complex carbohydrates and neurulation. Pediatr Neurosc 1986;12:2±8. [11] Sutton LN, Adzick NS, Bilaniuk LT, Johnson MP, Crombleholme TM, Flake AW. Improvement in hindbrain herniation demonstrated by serial fetal magnetic resonance imaging following fetal surgery for myelomeningocele. JAMA 1999;282:1826±31. [12] Bruner JP, Tulipan N. Fetal surgery for myelomeningocele and the incidence for shunt-dependent hydrocephalus. JAMA 1999;282: 1819±25. [13] Bensen JT, Dillard RG, Burton BK. Open spina bifida: does cesarean section improve prognosis? Obstet Gynecol 1988;71:532±4. [14] Cochrane D, Aaronyk K, Wilson D, Steinbok P. The effects of labour and delivery on spinal cord function and ambulation in patients with meningomyelocele. Childs Nerv Syst 1991;7:312±5. [15] Hill AE, Beattie F. Does cesarean section delivery improve neurological outcome in open spina bifida? Eur J Pediat Surg 1994;4(Suppl. 1):32±4. [16] Liu SL, Shurtleff DB, Ellenbogen RG, Loeser LD, Kroop R. 19 year follow-up of fetal myelomeningocele brought to term. Eur J Pediat Surg 1999;9(Suppl. I):12±4. [17] Tulipan N, Bruner MHernanz-SchulmanJP. Reduced hindbrain herniation after intrauterine myelomeningocele repair: a report of four cases. Pediatric Neurosurgery 1998;29:274±8. [18] Hunt GM. Non-selective intervention in newborn babies with open spina bifida: the outcome 30 years on for the complete cohort. Eur J Pediat Surg 1999;9(Suppl. 1):5±8.