To evaluate the impact of prenatal myelomeningocele repair on fetal head biometry.
To evaluate the impact of prenatal myelomeningocele repair on fetal head biometry.
Fifty fetuses underwent open fetal myelomeningocele repair at our institution between January 1998 and July 2002. All had serial head circumference (HC) and lateral ventricular diameter (VD) measurements taken preoperatively and weekly for 8 weeks after repair. Cortical index (CI) was defined as HC/VD. Measurements were compared with gestational age-matched values from nomograms. One-sample t-test, ANOVA and repeated measures analysis were used to assess HC, VD and CI after fetal repair.
Preoperatively, the HC in fetuses with myelomeningocele was smaller than control values (186.4 vs. 198.8 mm, P = 0.0004). Eight weeks' postoperatively this difference had resolved (293 vs. 301.6 mm, P = 0.76). The mean increase in CI after repair was 20% (P = 0.02) compared with the predicted 51% in normal cases. The average increase in VD was 3.9 mm (38.8%, P < 0.001).
Mid-gestational repair of myelomeningocele alters fetal head growth. Increased CI suggests HC changes are not due to ventriculomegaly alone. Copyright © 2004 ISUOG. Published by John Wiley & Sons, Ltd.
Prenatal myelomeningocele repair remains controversial as a non-lethal indication for maternal–fetal surgery. The rationale for fetal repair is not to prevent fetal or neonatal death, but rather to prevent or reduce the significant lifelong disabilities associated with myelomeningocele, which include various degrees of cognitive impairment, motor dysfunction in the legs, hydrocephalus, and bladder and bowel incontinence1. Previous reports on prenatal head growth in fetuses with unrepaired myelomeningocele showed disproportionately small head circumference (HC) measurements associated with a progressive enlargement of the lateral ventricular diameter (VD)2–4. Several studies have shown a negative correlation between the severity of ventriculomegaly and cognitive outcome in children born with myelomeningocele5, 6. Furthermore, indicators of the brain mass volume such as the cortical index (CI), have been shown to remain nearly constant or progressively decrease over gestation in fetuses with myelomeningocele, in contrast to the linear increase seen in normal fetuses2, 4, 7, 8.
Postnatal treatment for myelomeningocele consists of surgical closure of the spinal canal at birth and lifelong supportive care. However, the severe morbidity and significant mortality of myelomeningocele combined with promising results of research in animal models led to the consideration of prenatal intervention for this disorder. Initial clinical experience suggests that mid-gestational fetal myelomeningocele repair may reduce morbidity from hydrocephalus and Arnold Chiari II malformation by reversal of the hindbrain herniation component, arrest or slow the progressive ventriculomegaly, decrease the need for ventriculoperitoneal shunting and result in better than predicted leg function when compared with non-repaired fetuses with the same skeletal lesion9–13.
However, it is unknown whether maternal–fetal surgery for myelomeningocele changes the abnormal patterns of fetal head biometry and brain growth associated with myelomeningocele in reported historic observational case studies. The purpose of this paper was to assess the impact of in-utero myelomeningocele repair on fetal head growth as reflected by fetal head biometry measurements and CI.
This study was approved by the Committee for Protection of Human Subjects Institutional Review Board at the Children's Hospital of Philadelphia.
From January 1998 to July 2002, 145 fetuses with myelomeningocele were referred to The Center for Fetal Diagnosis and Treatment at the Children's Hospital of Philadelphia. Our institution's inclusion criteria for fetal repair included pregnancy less than 26 + 0 weeks at the time of surgical intervention, a normal fetal karyotype, no other major fetal abnormality which might impact on long-term prognosis, lateral ventricular size of less than 17 mm, the presence of a type II Arnold Chiari malformation, an S1-level lesion or higher, and evidence of intact neurological function of both lower extremities by demonstration of flexion and extension at the level of the hips, knees, ankles and toes as well as the absence of clubfoot deformity. During the study period 50 patients met the selection criteria and underwent fetal myelomeningocele repair. Table 1 summarizes the demographic data of the study population. We have described the preoperative evaluation procedure, surgical approach and postoperative obstetric management for in-utero myelomeningocele closure previously13.
|Variable||Mean ± SD|
|Maternal age at fetal surgery (years)||30.4 ± 4.8|
|Gestational age (weeks) at:|
|initial evaluation||22 ± 0.2|
|fetal myelomeningocele repair||22.5 ± 1.9|
|delivery||34.1 ± 3.6|
|No. of weeks of postoperative in-utero follow-up||11.5 ± 4.1|
|Birth weight (g)||2482 ± 707|
The prenatal ultrasound reports of the 50 cases that underwent fetal surgical myelomeningocele repair were reviewed. Data were collected preoperatively and at 2-, 4-, 6- and 8-week intervals postoperatively. One patient was excluded from the analysis because of premature delivery within 1 week of surgery. In 31/49 (63.3%) cases all five sets of measurements, in 10/49 (20.4%) the first four, in 5/49 (10.2%) the first three and in 3/49 (6.1%) only the first two sets of measurements were available for analysis. The numbers of available measurements reflect the gestational age at delivery or patients who returned home for follow-up care (n = 3).
Fetal HC was measured at the level of the biparietal diameter (BPD). In-utero repair HC data were compared with HC measurements at the 50th centile for age-matched values from nomograms constructed by Lessoway et al.7. The maximum diameters of the left and right lateral ventricles were measured individually at the level of the atria. VD was calculated as the mean of the lateral ventricular diameters. For comparison purposes normal VD was defined as 7.5 mm on the basis of the findings of Pilu et al.8. The degree of ventriculomegaly was classified as: none (≤10 mm), mild (10.1–15 mm), moderate (15.1–25 mm) or severe (>25 mm)3. CI was defined as HC divided by VD. CI was chosen as an indirect reflection of brain size4. For comparison purposes, normal CI was calculated from predicted HC measurement of age-matched normal values divided by 7.5 mm7, 8.
To evaluate whether the gestational age of fetal myelomeningocele repair influenced changes in head biometry after fetal surgery, patients were divided into three groups: fetal closure before 22 weeks' gestation (early repair), closure between 22 and 24 weeks (intermediate repair) and 24 to 26 weeks (late repair). In-utero repair patients were also classified according to prenatal sonographically determined spinal lesion levels as: thoracic (≥T12), high lumbar (L1–L2), low lumbar (L3–L5) or sacral level defects (S1) and HC, VD and CI were compared with predicted age-matched normal values.
HC, VD and CI were examined for normality using Shapiro and Wilks's W statistic. When significantly non-normal distributions were found, data were transformed using the ‘Box-Cox’ method to normalize the distributions before applying parametric tests such as ANOVA. In cases in which no transformation normalized the data, nonparametric tests were used. After transformation, data were examined for outliers, and none were found. Repeated measures analysis was performed to assess HC, VD and CI after fetal repair. A secondary repeated measures model was used to examine any correlation between postoperative HC, VD and CI changes and gestational age at fetal surgery and spinal lesion level of the myelomeningocele. All values are reported as mean ± SD. Statistical significance was set at P < 0.05. Analyses were performed using SAS statistical software (Cary, NC, USA, Version 8.02).
Preoperatively, the HC of fetuses with myelomeningocele was significantly smaller than that of the age-matched normal values (186.4 mm, 10th centile vs. 198.8 mm, 50th centile, P = 0.0004). Eight weeks after fetal myelomeningocele closure, however, no significant difference was found (292.9 mm, 25th centile vs. 301.6 mm, 50th centile, P = 0.76). Before fetal surgery, 34% (17/49) of fetuses were below the 10th centile whereas 8 weeks postoperatively only 12.2% (4/31) fetuses were below the 10th centile. The mean HC increase of in-utero repaired fetuses was 57.1% (106.5 ± 14.9 mm), while the mean HC increase was 51.7% (102.8 ± 15.6 mm) for the age-matched normal values.
For fetuses with repaired myelomeningocele, the mean VD preoperatively was 10.3 ± 2.2 (range, 5.6–15) mm and 8 weeks' postoperatively it was 14.3 ± 3 (range, 9.0–21.6) mm, with an average increase of 3.9 ± 2.6 mm (38.8%, P < 0.001) during the 8-week study period. Using an upper limit of 10 mm, preoperatively normal lateral ventricles were seen in 32.6% of cases, mild ventriculomegaly in 63.3%, moderate ventriculomegaly in 4.1% and none had severe ventriculomegaly. Eight weeks postoperatively, normal lateral ventricles were seen in 9.1% of cases, 57.6% had mild ventriculomegaly, and 33.3% had moderate ventriculomegaly. None of the fetuses had severe ventriculomegaly.
Preoperatively, the mean CI for the in-utero repair groups was 18.5 ± 5.0 (range, 10.9–32.1); postoperatively, the CI increased and 8 weeks after fetal repair the mean CI was 22 ± 5.1 (range, 16.1–33.2). The average increase in CI after fetal myelomeningocele closure was 20% (P = 0.02, Figure 1). However, the mean CI of age-matched normal fetuses increased from 26.6 ± 2.4 (range, 21.7–32.0) to 40.2 ± 1.7 (range, 36.5–43.7), representing an increase of 51.1%. Pre- and postoperatively, the mean CI was significantly higher in age-matched normal fetuses compared with that in the myelomeningocele group (P < 0.0001).
Of the 49 fetuses included in our study, 12 (24.5%) fetuses underwent fetal myelomeningocele repair before 22 weeks' gestation, 21 (42.9%) between 22 and 24 weeks, and 16 (32.6%) between 24 and 26 weeks. Pre- and postoperative values for HC, VD and CI for groups repaired early, intermediate and late are summarized in Table 2.
|Measurement||Timing of maternal–fetal surgery|
|Early (< 22 weeks)||Intermediate (22–24weeks)||Late (> 24 weeks)|
|Mean preoperative (n) [centile*]||169.5 mm (n = 12)||183.7 mm (n = 21)||203.9 mm (n = 16)|
|[5–10th centile]||[10–25th centile]||[10–25th centile]|
|Mean 8 weeks postoperative (n) [centile*]||279.9 mm (n = 8)||293.2 mm (n = 14)||301.1 mm (n = 9)|
|[25th centile]||[25th centile]||[10–25th centile]|
|P†||< 0.0001||< 0.0001||< 0.0001|
|8 weeks postoperative (mm)||14.0||13.8||13.8|
|P†||< 0.03||< 0.01||< 0.01|
|8 weeks postoperative||23||22.3||21.8|
Of the 49 fetuses included in the study, five (10.2%) had thoracic lesions, six (12.2%) had high lumbar lesions, 34 (69.4%) had low lumbar lesions and four (8.2%) had sacral lesions. Changes after surgery in HC, VD, and CI according to spinal lesion level are shown in Table 3.
|Measurement||Level of defect|
|Thoracic (> T12)||High lumbar (L1–L2)||Low lumbar (L3–L5)||Sacral (S1)|
|Mean preoperative (n) [centile*]||192.7 mm (n = 5)||186.8 mm (n = 6)||185.3 mm (n = 34)||194.7 mm (n = 4)|
|[10–25th centile]||[10–25th centile]||[10th centile]||[10–25th centile]|
|Mean 8 weeks postoperative (n) [centile*]||305 mm (n = 4)||286 mm (n = 3)||290.9 mm (n = 22)||301.5 mm (n = 2)|
|[40–45th centile]||[10–25th centile]||[40–45th centile]||[50th centile]|
|P†||< 0.0001||< 0.0001||< 0.0001||< 0.0001|
|8 weeks postoperative||16.9||14.8||13.8||12.8|
|P†||< 0.001||< 0.01||0.0006||< 0.001|
|8 weeks postoperative||20.4||19.4||22.6||23.4|
Several investigators have described the natural history of head growth based on biometry in fetuses with myelomeningocele. In a retrospective study of 70 fetuses diagnosed with myelomeningocele before 23 weeks, Nicolaides et al.14 found that 26% and 61%, respectively, had HC and BPD below the 5th centile, although 86% had already developed ventriculomegaly. Roberts and Campbell15 documented the reduction in BPD in fetuses with myelomeningocele and suggested that this was due to ‘abnormal brain development’. Van der Hof et al.2 reported the sonographic findings of 130 fetuses with open spina bifida and noted that HC measurements were significantly smaller in fetuses with myelomeningocele compared with normal fetuses. They found that 20 of 107 (19%) at 24 weeks or less and 10 of 23 (43%) at greater than 24 weeks had HC measurements below the 2.5th centile, despite enlargement of the lateral ventricles. Using the anterior and posterior ventricle-to-hemisphere ratio as an indirect indicator of brain size, they suggested progressive ‘loss of brain mass’ in fetuses with myelomeningocele and proposed that this may be due to further herniation of the cerebellum with subsequent ventriculomegaly2.
Bannister et al.4 evaluated 21 fetuses with myelomeningocele with the diagnosis of spina bifida between the 16th and 23rd weeks of gestation. On the initial scan 84.2% had a HC measurement below the 3rd centile and 86% had enlarged ventricular measurements. In five fetuses, serial sonographic biometric measurements were made. Whereas CI in the normal age-matched fetuses progressively increased throughout gestation, in fetuses with myelomeningocele, the CI remained constant or decreased as pregnancy progressed.
More recently, Babcook et al.3 studied 51 fetuses with myelomeningocele, looking at the correlation between VD and gestational age, and showed that the prevalence as well as the severity of ventriculomegaly in fetuses with myelomeningocele significantly increased with advancing gestational age. Only 44% of fetuses with myelomeningocele demonstrated ventriculomegaly before 24 weeks' gestation, while 94% had ventriculomegaly diagnosed after 24 weeks. Before 24 weeks, only 20% (3/15) had moderate ventriculomegaly and none had severe ventriculomegaly, while after 24 weeks 69% (11/16) developed moderate ventriculomegaly and 31% (5/16) progressed to severe ventriculomegaly.
While our preoperative results for HC, CI and VD were similar to those seen in other published studies, postoperative serial sonography was indicative of an increase in head growth after prenatal myelomeningocele closure. After myelomeningocele repair, HC growth patterns changed to become similar to those seen in normal fetuses. Ventriculomegaly was still present, but none of the fetuses developed severe ventriculomegaly as is commonly seen in postnatally treated fetuses with myelomeningocele; the majority of prenatally repaired fetuses had only mild to moderate ventriculomegaly. Other studies have shown that this is associated with a lower risk of developmental and neurocognitive delay16 compared with severe ventriculomegaly. When divided into early, intermediate and late myelomeningocele repair groups, HC increased in all three groups after fetal myelomeningocele repair, but CI only increased significantly when repair was performed before 22 weeks' gestation. VD increase in the intermediate and late groups was higher compared with the early group suggesting an advantage to early repair.
Several explanations of our results are possible. The increase of HC and CI suggests that head growth after fetal myelomeningocele repair may be due, at least in part, to an increase in cortical mass, and not an increase in ventriculomegaly alone. Alternatively, head growth may be due to reversal of the hindbrain herniation alone or in combination with an increase in cortical mass. In normal fetuses, the growth velocity of the infratentorial portion of the brain accelerates significantly during the second trimester17. We and others have shown experimentally, and later clinically, reversal of the hindbrain herniation following in-utero myelomeningocele closure10, 18, 19. The combination of increased infratentorial brain growth and increased intracranial tissue due to hindbrain ascent could have an impact on HC and CI. Additionally, pathological studies of the brains of fetses with myelomeningocele revealed abnormal cellular migration patterns in brain stem, cerebellum and cerebral cortex20. In normal fetuses, neuronal numbers and normal migration of neurons peak during the second trimester21, 22. It is possible that reversal of the hindbrain herniation and brainstem compression, decreased progression in ventriculomegaly and re-establishment of extra-axial fluid space after mid-gestational myelomeningocele repair may facilitate both an increase in neuronal number and normal migration patterns during this gestational period, contributing to the increased HC and CI.
Several studies have suggested that the level of the defect influences ventricular size in prenatally diagnosed myelomeningocele. Fetuses with thoracic lesions more often develop severe ventriculomegaly when compared with fetuses with lumbosacral lesions23, 24. Our data indicate that there is a correlation between prenatal anatomical level and postoperative head and brain growth. Fetuses with lower-lumbar lesions (L3–L5) had the highest increase in CI and the lowest degree of increase in ventriculomegaly.
This study suggests that fetal head biometry is altered following in-utero myelomeningocele repair. It is likely that a component of this increase in HC is due to progressive enlargement of the lateral ventricles and reappearance of the subarachnoid fluid space that occurs during reversal of the hindbrain herniation and re-establishment of more normal cerebrospinal fluid hydrodynamics. Although CI is not a direct measurement, it is reflective of cortical tissue mass in the absence of other brain malformations4. Our study found an increase in CI following prenatal repair (Figure 1) that was not seen in previously reported case series. While this may simply be due to less severe ventriculomegaly, it may also represent increased cortical mass. The finding that CI did not achieve the same rate of linear increase as seen in age-matched normal fetuses is most likely due to the increase in VD that occurred despite in-utero repair.
Limitations of this study include the small number of myelomeningocele patients, the unequal distribution in each of the subgroups studied, and the lack of comparable prenatal biometric parameters in the form of a large series of prenatally diagnosed but unrepaired myelomeningocele fetuses for direct comparison. Therefore, comparisons were made with data from previously published case series. Although we and others have tried to gather such information, it is difficult to study a cohort of patients who must return to referral centers to have frequent serial ultrasound evaluations, particularly when such examinations will not change their obstetric care and are not likely to be covered by their medical insurance.
We recognize that CI is an imprecise reflection of cortical mass. Cortical mantle measurements were available in many of the scans we reviewed, but these measurements were not obtained in a consistent manner at a predetermined anatomical landmark. However, present HC, VD, and CI findings are intriguing, and have led to further studies using more precise measurements at specific anatomical landmarks within the fetal brain obtained from ultrafast fetal magnetic resonance imaging studies. Such studies may lead to a better understanding of the observed biometric changes following prenatal myelomeningocele repair. A multicenter, prospective, randomized treatment trial called the Management of Myelomeningocele Study (MOMS) is presently underway to determine whether fetal neurosurgical closure may reduce infant morbidity and mortality and improve long-term neurocognitive and developmental outcome when compared with children who undergo postnatal neurosurgical repair.