Morphometric Analysis of Spina Bifida after Fetal Repair Shows New Subtypes with Associated Outcomes.

ABSTRACT

Importance The binary classification of spina bifida lesions as myelomeningocele (with sac) or myeloschisis (without sac) belies a spectrum of morphologies, which have not been correlated to clinical characteristics and outcomes. Objective: To characterize spina bifida lesion types and correlate them with preoperative presentation and postoperative outcomes. Design: Secondary analysis of images and videos obtained during fetoscopic spina bifida repair surgery from 2020-2023. Setting: Fetal surgery was performed at a quaternary care center. Participants: A prospective cohort of patients referred for fetal spina bifida underwent fetoscopic repair under an FDA-approved protocol. Of 60 lesions repaired, 57 had available images and were included in the analysis. Interventions or Exposures We evaluated lesion morphology on high-resolution intraoperative images and videos to categorize lesions based on placode exposure and nerve root stretching. Main Outcomes and Measures: The reproducibility of the lesion classification was assessed via Kappa interrater agreement. Preoperative characteristics analyzed include ventricle size, tonsillar herniation level, lower extremities movement, and lesion dimensions. Outcomes included surgical time, need for patch for skin closure, gestational age at delivery, preterm premature rupture of membranes (PPROM), and neonatal cerebrospinal fluid (CSF) diversion. Results: We distinguished five lesion types that differ across a range of sac sizes, nerve root stretching, and placode exposure, with 93% agreement between examiners (p<0.001). Fetal characteristics at preoperative evaluation differed significantly by lesion type, including lesion volume (p<0.001), largest ventricle size (p=0.008), tonsillar herniation (p=0.005), and head circumference (p=0.03). Lesion level, talipes, and lower extremities movement did not differ by type. Surgical and perinatal outcomes differed by lesion type, including need for patch skin closure (p<0.001), gestational age at delivery (p=0.01), and NICU length of stay (p<0.001). PPROM, CSF leakage at birth, and CSF diversion in the NICU did not differ between lesion groups. Linear regression associated severity of ventriculomegaly with lesion type, but not with tonsillar herniation level. Conclusions and Relevance There is a distinct phenotypic spectrum in open spina bifida with differential baseline presentation and outcomes. Severity of ventriculomegaly is associated with lesion type, rather than tonsillar herniation level. Our findings expand the classification of spina bifida to reveal a spectrum that warrants further study.