Intrauterine repair of spina bifida: preoperative predictors of shunt-dependent hydrocephalus.

Objective: The objective of this study was to determine which factors that are present at the time of intrauterine repair of spina bifida could predict the need for ventriculoperitoneal shunt for hydrocephalus during the first year of life.

Study Design: One hundred seventy-eight fetuses have undergone intrauterine repair of spina bifida at Vanderbilt University Medical Center since 1997. Among these, 116 fetuses had a postnatal follow-up period of at least 12 months. The primary outcome of the study was the need for a ventriculoperitoneal shunt for hydrocephalus during the first year of life. The following variables were analyzed: maternal demographics (age, race, gravidity, and parity), gestational age at the time of surgery, ventricular size, degree of hindbrain herniation (determined by magnetic resonance imaging in 33 cases), type of defect (myelomeningocele vs myeloschisis), upper level of the lesion, presence of talipes, and intraoperative use of a lumbar drain. Statistical analysis was performed with logistic regression (to test the association of fetal and maternal factors and the need for ventriculoperitoneal shunting), 2-sample t-tests for comparison of means, and receiver operating curves with the use of the probabilities that were generated by the logistic regression for both continuous and categoric versions of the factors.

Results: Sixty-one of 116 of the fetuses (54%) who underwent operation in utero required the placement of a ventriculoperitoneal shunt before the age of 1 year. The upper level of the lesion was the strongest predictor of shunt requirement (adjusted odds ratio per 1 level increase with the use of continuous variables [S1 through T10], 1.73 [95% CI, 1.22- 2.44]; adjusted odds ratio with the use of upper lesion level >or=L3 vs 25 weeks as a categorized variable, 3.3 [95% CI, 1.28-8.24]), and preoperative ventricular size (adjusted odds ratio per 1 unit increase with the use of continuous variables, 1.17 [95% CI, 1.01-1.36]; adjusted odds ratio with the use of ventricular size >or=14 mm vs <14 mm as a categorized variable, 3.5 [95% CI, 1.08-11.16]). Receiver operating curves with the use of the probabilities that were generated by the logistic regression analyses for both the continuous and categoric versions of the factors were compared. The area under the curve was approximately 0.81 for both methods. Thirty-eight of 48 of the fetuses (79%) with an upper level of the lesion >or=L3 required placement of a ventriculoperitoneal shunt, although 25 of 68 of the fetuses (37%) with lesions or=14 mm (27/32 fetuses) needed a shunt compared with 41% of the fetuses (34/81 fetuses) with smaller ventricles (P=.03). Seventy-one percent of the fetuses who underwent operation at >25 weeks of gestation also required shunt placement (37/52 fetuses); 39% of the fetuses (24/61 fetuses) who were treated
Conclusion: This study suggests that, among fetuses who underwent operation in utero for spina bifida, fetuses with a ventricular size of <14 mm at the time of surgery, fetuses who had surgery at