A new look at myelomeningoceles: functional level, vertebral level, shunting, and the implications for fetal intervention
Natalie E. Rintoul
ABBREVIATION. MMC, myelomeningocele.
On the basis of available data, a patient who is born with a myelomeningocele (MMC) will have an 80% to 85% chance of developing hydrocephalus that requires the placement of a cerebrospinal fluid shunt. (1,2) Hydrocephalus is often not apparent at birth, but shunting is usually required within the first week of life. It is the impression of many pediatric neurosurgeons that the incidence of shunting varies with the level of the spinal defect, which may be defined in terms of the functional neurologic assessment or bony vertebral defect. However, the incidence of shunting as a function of either functional or bony level has not previously been reported.
The relationship between functional level and bony level in these patients is also not well established. Two studies involving 11 and 15 patients, respectively, suggested good correlation between lesion level as determined by prenatal ultrasound and postnatal neuromuscular level. (3,4) However, no large study has compared level of bony defect with functional level.
Preliminary data from 2 centers suggest that fetal MMC closure may favorably influence the outcome of this disease. (5,6) In each study, the incidence of shunting was reported to be lower than expected in MMC patients who underwent surgical repair before birth. This benefit may have resulted from reversal of the Chiari II malformation, which has been demonstrated with magnetic resonance imaging in MMC patients. (6) However, only a subset of fetuses who receive a diagnosis of MMC are considered candidates for fetal intervention. Therefore, selection bias may also have accounted for this apparent benefit. Furthermore, if the level of the spinal defect does influence the requirement for shunting, then it would be important to stratify patients according to this variable when assessing the effects of fetal repair. It has also been suggested that fetal closure may improve motor function in MMC patients. An accurate understanding of the expected level of motor function based on the bony anatomy of the lesion is important in determining whether fetal surgery influences neurologic outcome.
Ultimately, the true value of fetal surgery for MMC is best determined by a multicenter, prospective, randomized, clinical trial. The optimal design of such a trial will require accurate and detailed historical outcome data. In this article, we examine the natural history of spina bifida in patients who were followed after standard postnatal closure at the Children’s Hospital of Philadelphia. We sought to establish definitive rates of shunting for each level of lesion, defined by both functional and vertebral criteria. The relationship between these 2 sets of criteria was also examined. These data will facilitate assessment of the first patients treated with fetal surgery and will contribute to the power calculations needed for a nationwide, randomized, controlled trial comparing fetal with postnatal closure.
A retrospective chart review of all 450 patients evaluated in the spina bifida clinic at the Children’s Hospital of Philadelphia was conducted. Only those patients who were delivered after February 1983 were included, when computed tomography scanning became available for measurement of ventricular size and shunting was not performed on the basis of clinical criteria alone. Patients with occult dysraphic lesions and multiple congenital anomalies were excluded. Several patients were discharged from the hospital without early closure at the parents’ request. Although some of these patients later returned for delayed closure, they were also excluded. Thus, 297 patients with open MMCs were included in the analysis.
The level of each lesion was established on the basis of both functional motor examination and the level of spinal dysraphism as evident on radiograph. Functional level was determined both by physical therapy and orthopedic surgery assessments, and a level was assigned on the basis of the best functional myotome. For patients whose functional level worsened over time as a result of excessive weight gain, injury, hydromyelia, or tethered cord, the patient’s best functional level was recorded. The higher (worse) level was recorded for patients with different neuromotor levels on the left and right.
To determine bony level, we reviewed radiographs obtained for 189 patients as voiding cystourethrogram scout films or shunt radiographs. A single radiologist (A.M.H.), who was blinded to the patients’ functional level, read each film and assigned a bony level on the basis of the superior extent of the spina bifida defect. On an anterior view of the abdomen or spine, the level of abnormality was determined by noting widening of the interpedicular distance, flattening of the pedicles, and, in children >1 year of age, the level at which the posterior spinous process was no longer identified.
Both functional and bony levels were designated by the use of a 7-point scale (7 = any thoracic, 6 = L1, 5 = L2, 4 = L3, 3 = L4, 2 = L5, and 1 = any sacral). The 2 methods of assessment were compared by subtracting the radiologic level from the functional level. A negative value indicated that the functional level was worse (more cephalad) than the vertebral level and a positive value indicated that the functional level was better than the vertebral level.
An evaluation of the correspondence between the 2 methods of assigning the lesion level was performed with a Cohen’s [kappa]. A continuous measure of difference between the 2 ratings was also determined. The shunt rates for each level of MMC were compared with a [chi square] test. The number of patients shunted within each level (thoracic, lumbar, and sacral) according to each method of assessment was compared using a log linear analysis. The median number of shunt revisions at each level of lesion was assessed using a median test. The effect of the mode of delivery on the difference between functional and vertebral level was assessed using a Kruskal-Wallis test.
Level of Lesion
All 297 patients were assigned a lower extremity functional level (Fig 1). Lumbar function was found in the majority (184 patients [62%]), with a preponderance of L3 function within the group. Sixty-seven patients (23%) functioned at a thoracic level, and 46 patients (15%) functioned at a sacral level. Radiology records were available for 218 (73%) of the 297 patients. Of these, 29 patients had no plain film. Therefore, a total of 189 patients were assigned a vertebral anatomic level from radiographs (Fig 1). Thirty-five patients (19%) were thoracic, 114 (60%) were lumbar, and 40 (21%) were sacral.
[FIGURE 1 OMITTED]
The correlation between functional and anatomic levels was generally good if patients were classified only as thoracic, lumbar, or sacral (Cohen’s [kappa] = 0.523), with concordance in 74% of patients. When a more detailed analysis was performed with assignment of specific lumbar levels, the correlation was weaker (Cohen’s [kappa] = 0.273) and only 39% of patients were concordant. In discordant patients, there was a tendency for the functional level to be higher (worse) than the anatomic level by an average of 2 spinal segments (89 patients [47%]). However, in 27 patients (14%), the functional level was better than the anatomic level (Fig 2). A log linear analysis was performed comparing the group with functional level better than the anatomic level to the group with the functional level worse than the anatomic level, which showed a significant difference between the groups (P < .001).
[FIGURE 2 OMITTED]
Requirement for Shunting
Of the 297 patients reviewed, 242 (81%) were shunted. Of these, the date of the initial shunt procedure was available for 211 (87%) of 242. Some of the patients initially had been cared for at another hospital and the records were incomplete. Most closures (94%) were performed within the first week of life, and all were within the first 30 days. The majority of the initial shunts were also performed within the first week of life (62%; Fig 3), and 26% were performed at the same time as the MMC closure. The median age at shunt placement was 5 days of age. In 1 case, the shunt was placed 1 year after the MMC was closed. Timing of shunt placement was independent of the level of lesion. The median number of shunt revisions was 2 per patient with a range of 0 to 21. Of these, 55% occurred in the first year of life. The number of shunt revisions was independent of the level of the lesion (P = .36).
[FIGURE 3 OMITTED]
The rate of shunt placement varied with the functional level of the lesion. As shown in Table 1, 97% of the thoracic-level patients, 87% of the lumbar-level patients, and 37% of the sacral-level patients were shunted (P < .001). A similar pattern of shunt placement was obtained for vertebral-level of lesion with 100% thoracic, 88% lumbar, and 68% sacral shunted (P = .006; Table 2). A significant difference in shunt incidence was found for patients with sacral MMCs when the lesions were defined by functional level rather than bony level (P < .001; Table 3). There was no significant difference in age at shunt placement among the different spinal levels regardless of how the level was determined.
A posterior fossa decompression was performed in 9.4% (28/297) patients. All patients who underwent posterior fossa decompression were shunted. The overall mortality was 3% (8/297). Of these, 6 died of Chiari complications, 1 of cardiac arrest with pneumococcal meningitis, and 1 secondary to cardiac arrest related to congenital heart disease. It was not possible to assess neonatal mortality in this study as we reviewed all of the charts of the patients who were alive to discharge and seen in the spina bifida clinic.
The mode of delivery for each patient was assessed to determine whether the functional level was better than the vertebral level in infants who were delivered by cesarean section versus vaginal birth. The mode of delivery was recorded for 80 of the 189 children for whom data were available for both functional and vertebral levels. Of the total number, 40 were delivered vaginally and 40 were delivered via cesarean section. Table 4 illustrates the distribution of functional and vertebral level differences for both modes of delivery and for children for whom no method of delivery was recorded. The radiologic level tended to be better than the functional level for both modes of delivery, with no significant difference between vaginal birth and cesarean section. These 2 groups (cesarean and vaginal delivery) did not differ from patients for whom no mode of delivery was recorded (P = .530; [chi square] = 1.27).
This study established outcome data for the conventional postnatal treatment of patients with MMC, recognizing that the epidemiology of this disease is in evolution. In 1973, 80% of lesions were reported to be lumbar and <15% were thoracic or sacral. (7) In our population, 23% of lesions were functionally thoracic, 62% were lumbar, and 15% were sacral. This difference may reflect prenatal diagnosis and termination or the effects of maternal folic acid supplementation. Since 1992, the Public Health Service has recommended that folic acid be taken by all women during the periconceptional period. (8,9) Despite the evolving epidemiology of this disease, neural tube defects remain the most common birth defect. MMCs will continue to occur in both unplanned pregnancies in which folic acid is not taken in the periconceptional period and planned pregnancies of which 30% are thought to be folic acid resistant.
We found our overall shunt rate of 81% to be consistent with the generally accepted range of 80% to 85% for an unselected spina bifida population. Furthermore, we found that the rate of ventricular shunting for hydrocephalus in infants born with MMC was related to the level of the lesion. There was a significantly lower shunt rate for patients with better functional levels. There was also a higher shunt rate for sacral-level lesions when level was determined radiographically compared with functionally. This is the first study to examine the incidence of ventricular shunting at specific spinal levels.
In the 189 patients for whom both functional and radiologic data were available, we found that there was not complete correspondence between these modes of assessment. These data differ from data reported by both Kollias et al (3) and Coniglio et al, (4) who found that the functional level was accurately predicted by the anatomic level in 10 of 11 and 12 of 15 patients studied, respectively. Our data also differ from the larger study by Luthy et al, (10) who reported data on 160 infants and concluded that motor function is equal to or better than the last intact lamina.
We found that the functional level was equal to the vertebral level in 39% of patients. In nearly half, the functional level was 2 levels above (worse than) the vertebral level. This finding might be explained by inferior displacement of the spinal cord segments in relation to the vertebral segments as a result of congenital tethering. Furthermore, we found no difference between the patients delivered by cesarean section and vaginal birth. This varies from the findings of Luthy et al, (10) who reported a motor level caudad to the vertebral level by 3.3 [+ or -] 3.0 segments for 47 patients delivered by cesarean section before labor, 1.1 [+ or -] 2.3 for 78 infants delivered vaginally, and 0.9 [+ or -] 4.1 for 35 infants delivered by cesarean section after the start of labor. (10) The differences in findings may be explained in part by differences in the timing of back closure postnatally. As well, there is evidence that rupture of the amniotic fluid membranes in patients with large MMCs may be a significant factor in nerve damage. (11) Peripartum information was not available in this retrospective review.
Because of the retrospective design of this study, many different health professionals performed the physical examinations. However, by using each patient’s best neuromotor level, we attempted to minimize the effects of interobserver variability with regard to the assessment of functional levels. The radiologic levels all were assigned by a single radiologist, who was blinded to each patient’s functional level, thus providing a more standardized assessment of vertebral level. Nonetheless, a lack of prospective standardized physical examinations may have influenced the results of this investigation. In addition, it is important to note that our population derives from a span of nearly 2 decades. In that time, folic acid supplementation was introduced. It is unclear whether folic acid has changed the severity of presentation in MMC patients or whether its effects are purely an all-or-none phenomenon.
Data are now available from 2 groups reporting the early experience with fetal MMC closure. (5,6) The initial rationale for this radical new approach was animal work and some human data, which suggested that at least some of the lower extremity dysfunction seen in patients with MMC was acquired in utero and could potentially be prevented by early coverage of the neural placode. (12-14) An unexpected finding from this experience was that the hindbrain herniation component of the Chiari II malformation was much less prevalent in patients who underwent closure before birth and could even be seen to reverse on serial fetal magnetic resonance imaging studies performed after the procedure. (6) Perhaps related to this, the incidence of hydrocephalus requiring early shunting was decreased among patients who underwent fetal closure early in gestation.
Sutton et al (6) reported only 1 of 9 surviving patients who underwent fetal MMC closure at or before 25 weeks of gestation were shunted at a mean of 182 days of follow-up. Bruner et al (5) reported a 59% incidence of shunting in their series of 29 patients who underwent fetal closure at an average gestational age of 27 weeks compared with a 91% shunt rate in their historical control group of 23 patients. In these 2 series, the patients who required shunts developed symptomatic hydrocephalus at an older age than the control subjects. The fetal patients were shunted at a mean of 182 days of age compared with our historical series in which the median age to shunt placement was 5 days. It was speculated that reversal of the hindbrain hernia might open the cerebrospinal fluid drainage pathways (15) and reduce the incidence of obstructive hydrocephalus. Alternatively, the delay in requirement for shunt placement may reflect different indications for shunting.
It is difficult to compare the incidence of shunting in patients with MMC repaired prenatally with those patients repaired postnatally because, historically, shunts were often placed based on ventriculomegaly without other signs or symptoms of increased intracranial pressure. This emphasizes the need for a prospective randomized trial with commonly agreed-on indications for shunt placement for both fetal surgery patients and conventional postnatal closure patients.
Our review of a large single-center population of MMC patients has demonstrated some disparity between the radiologic and functional assessment of these lesions. In addition, the incidence of shunting for hydrocephalus was found to vary according to the level of MMC and the specific definition (functional versus radiologic) used to assign level. These findings are important not only for counseling individual patients but also for the design of fetal intervention trials.
TABLE 1. Incidence of Shunting by Functional Level
Functional Total Patients Shunted Patients Shunt
Level (n [%]) (n [%]) Required
T 67 (22) 65 (27) 97%
L1 11 (4) 9 (4) 82%
L2 4 (1) 3 (1) 75%
L3 106 (36) 92 (38) 87%
L4 47 (16) 45 (19) 96%
L5 16 (5) 11 (5) 69%
S 46 (15) 17 (7) 37%
All levels 297 242 81%
TABLE 2. Incidence of Shunting by Vertebral Level
Vertebral Total Patients Shunted Patients Shunt
Level (n [%]) (n [%]) Required
T 35 (19) 35 (22) 100%
L1 6 (3) 5 (3) 83%
L2 15 (8) 13 (8) 87%
L3 23 (12) 22 (14) 96%
L4 33 (17) 30 (19) 91%
L5 37 (20) 30 (19) 81%
S 40 (21) 27 (17) 68%
All levels 189 162 86%
TABLE 3. Summarized Ventricular Shunt Rates
According to the Method of Assessment
Level of Functional Radiological P Value *
Lesion Assessment Assessment
Thoracic 97% 100% NS
Lumbar 87% 88% NS
Sacral 37% 68% <.001
P value * <.001 .006
* [chi square] testing.
TABLE 4. Difference Between Functional and Radiologic Assessment
When Patients Were Categorized by Mode of Delivery
Diffe- Cesarean Vaginal Not Total
rence * Section Delivery Reported (n [%])
(n [%]) (n [%]) (n [%])
+5 1 (3) 0 (0) 0 (0) 1 (0.5)
+4 0 (0) 1 (3) 0 (0) 1 (0.5)
+3 0 (0) 0 (0) 1 (1) 1 (0.5)
+2 1 (3) 0 (0) 5 (5) 6 (3)
+1 5 (13) 4 (10) 9 (8) 18 (10)
0 9 (22) 16 (40) 48 (44) 73 (39)
-1 11 (28) 5 (13) 18 (17) 34 (18)
-2 10 (25) 9 (23) 18 (17) 37 (20)
-3 3 (8) 5 (13) 8 (7) 16 (8)
-4 0 (0) 0 (0) 2 (2) 2 (1)
-5 0 (0) 0 (0) 0 (0) 0 (0)
Total 40 40 109 189
Median -1 0 0 0
* The scaled radiologic level minus the scaled functional level.
See text for an explanation of scaled values. A negative value
indicates the functional level was above (worse than) the
We thank Gordon Morewood, MD, for technical assistance during data abstraction, analysis, and the preparation of this manuscript.
(1.) Stein SC, Schut L. Hydrocephalus in myelomeningocele. Childs Brain. 1979;5:413-419
(2.) Pater JD, Lapras C, Guilburd JN. Spina bifida apertamyelomeningocele: hydrocephalus. Neurochirurgie. 1988;34:47-52
(3.) Kollias SS, Goldstein RB, Cogen PH, Filly RA. Prenatally detected myelomeningoceles: sonographic accuracy in estimation of the spinal level Radiology. 1992;185:109-112
(4.) Coniglio SJ, Anderson SM, Ferguson JE. Functional motor outcome in children with myelomeningocele: correlation with anatomic level on prenatal ultrasound. Dev Med Child Neurol. 1996;38:675-680
(5.) Bruner JP, Tulipan N, Paschall RL, et al. Fetal surgery for myelomeningocele and the incidence of shunt-dependent hydrocephalus. JAMA. 1999;282:1819-1825
(6.) 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;282:1826-1831
(7.) Emery JL, Lendon RG. The local cord lesion in neurospinal dysraphism (meningomyelocele). J Pathol. 1973;110:83-96
(8.) MRC Vitamin Study Research Group. Prevention of neural tube defects: results of the Medical Research Council Vitamin Study. Lancet. 1991; 338:131-137
(9.) Recommendations for the use of folic acid to reduce the number of cases of spina bifida and other neural tube defects. MMWR Morb Mortal Wkly Rep. 1992;41(RR-14):1-7
(10.) Luthy DA, Wardinsky T, Shurtleff DB, et al. Cesarean section before the onset of labor and subsequent motor function in infants with meningo-myelocele diagnosed antenatally. N Engl J Med. 1991;324:662-666
(11.) Liu SL, Shurtleff DB, Ellenbogen RG, Loeser JD, Kropp R. 19-year follow-up of fetal myelomeningocele brought to term. Eur J Pediatr Surg. 1999;9:12-14
(12.) Meuli M, Meuli-Simmen C, Yingling CD, et al. Creation of myelomeningocele in utero: a model of functional damage from spinal cord exposure in fetal sheep. J Pediatr Surg. 1995;30:1028-1033
(13.) Meuli M, Meuli-Simmen C, Hutchins G, et al. In utero surgery rescues neurological function at birth in sheep with spina bifida. Nat Med. 1995;1:342-347
(14.) Meuli M, Meuli-Simmen C, Yingling CD, et al. In utero repair of experimental myelomeningocele saves neurological function at birth. J Pediatr Surg. 1996;31:397-402
(15.) McLone DG, Knepper PA. The cause of Chiari II malformation: a unified theory. Pediatr Neurosci. 1989;15:1-12
Natalie E. Rintoul, MD *; Leslie N. Sutton, MD ([section]); Anne M. Hubbard, MD ([paragraph]); Brian Cohen, PhD (#); Jeanne Melchionni, RN ([double dagger]); Patrick S. Pasquariello, MD ([double dagger]); and N. Scott Adzick, MD ([parallel])
From the Department of Pediatrics, * Divisions of Neonatology and ([double dagger]) General Pediatrics; Department of Surgery, ([section]) Divisions of Neurosurgery and ([parallel]) Pediatric General and Thoracic Surgery; ([paragraph]) Department of Radiology; and (#) Division of Biostatistics and Epidemiology, Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania.
Received for publication Apr 23, 2001; accepted Oct 1, 2001.
Reprint requests to (L.N.S.) Division of Neurosurgery, Children’s Hospital of Philadelphia, 34th St and Civic Center Blvd, Philadelphia, PA 10194. E-mail: firstname.lastname@example.org
COPYRIGHT 2002 American Academy of Pediatrics
COPYRIGHT 2002 Gale Group