Prenatal closure of myelomeningocele has been shown to decrease the rate of ventriculoperitoneal shunt (VPS) placement by 1 year of age, as well as to improve composite mental and motor function scores.1–3 Further analysis of these data found that ventricular size of 15 mm or larger at prenatal screening (19–25 weeks of gestation) negated the benefit of using intrauterine closure to reduce the rate of VPS placement.4 This report appropriately led to changes in counseling and practice recommendations for intrauterine closure at some fetal surgery centers.
In addition to decreased VPS rates, additional benefits such as improvement in Chiari malformation type II characteristics, improved quality of life, and decreased healthcare costs, have also been observed and may be associated with whether VPS placement or other CSF diversion therapy is needed.5–15
Thus, intrauterine ventricular width has become an important measurement. In our experience, ventricular size measurements seemingly vary significantly depending on the imaging modality (ultrasound [US] vs MRI). In addition, interobserver variability may also play a role in measurement disparity. This study sought to determine whether imaging modality and interobserver variability were factors in any ventricular size disparity seen on imaging studies.
Methods
The imaging studies of 15 consecutive fetuses who underwent prenatal myelomeningocele repair between October 2017 and February 2019 at Children’s Mercy Fetal Health Center, Kansas City, Missouri, were reviewed. All fetuses were imaged with prenatal US and fetal MRI. We recorded gestational age at the time of US and MRI. At our institution, prenatal US is interpreted by maternal-fetal medicine specialists (MFMs), and fetal MRI is interpreted by pediatric radiologists (PRs). A single MFM reviewed all the initial US images. Blinded review of each US and MRI study was performed by a PR who was not involved in the initial review of any study. These reviewers were selected for their specialization in fetal diagnostic and neurological imaging. Although MFMs are proficient in US interpretation, they do not participate in formal MRI interpretation, and thus we did not include MRI interpretations by MFMs. Ventricular size was determined by measuring the maximal ventricular width at the level of the atrium in both ventricles. By using these data, we performed three comparisons.
Comparison 1: Retrospective Assessment
We retrospectively compared the dictated ventricular widths obtained with US by MFMs and those obtained with MRI by PRs. These data were recorded in our electronic medical record system. Mean ventricular widths were tabulated for the two modalities and compared.
Comparison 2: Interobserver Assessment
The initial interpretations of US were performed by MFMs, and those of fetal MRI by PRs. US and MR images were then loaded into a de-identified folder so that a PR could perform blinded measurements of both studies. These blinded measurements were then compared with the original dictated measurements to assess interobserver variability.
Comparison 3: Intermodality Assessment
Intermodality variability between the blinded MRI and US measurements by the PR was assessed. Differences between imaging modalities and observers were analyzed with the paired t-test. The mean difference in millimeters and 95% CI was calculated for each comparison. All descriptive statistics and statistical analyses were performed with GraphPad Prism version 8.1.2.
Results
Comparison 1: Retrospective Assessment
On average, fetal US (mean 22.3 weeks of gestation) was performed after MRI (21.9 weeks of gestation). The two imaging modalities were often performed only days apart (average [range] 3.8 [0–20] days; median 2 days). Four fetuses underwent imaging studies on the same day. Five US studies were done prior to MRI, on average (range) 3 (1–6) days earlier. Six MRI studies were performed prior to US, on average (range) 7 (1–20) days earlier.
Ventricular sizes ranged from 3.6 to 16.3 mm on US and from 5.0 to 17.2 mm on MRI. The right ventricle was larger on MRI than US in all but 1 fetus by an average of 2.1 mm. Similarly, the left ventricle was, on average, 2.0 mm larger on MRI than US in all but 2 fetuses. The mean difference between sides was 2.0 mm, and MRI was associated with larger measurements than US (95% CI 1.43–2.69, p < 0.001) (Table 1).
Ventricular sizes and differences between MRI and US measurements
| Patient No. | US | MRI | Difference* | |||
|---|---|---|---|---|---|---|
| Rt Side | Lt Side | Rt Side | Lt Side | Rt Side | Lt Side | |
| 1 | 4.4 | 4.5 | 7.5 | 8.4 | 3.1 | 3.9 |
| 2 | 10.1 | 8.9 | 11 | 13 | 0.9 | 4.1 |
| 3 | 9.6 | 7.6 | 9 | 11.6 | −0.6 | 4 |
| 4 | 3.6 | 4.7 | 7 | 7 | 3.4 | 2.3 |
| 5 | 5.4 | 6.3 | 7.9 | 5.3 | 2.5 | −1 |
| 6 | 5.3 | 8 | 6.7 | 7.5 | 1.4 | −0.5 |
| 7 | 4.6 | 5.9 | 9.4 | 8.2 | 4.8 | 2.3 |
| 8 | 14.3 | 13.4 | 17 | 16 | 2.7 | 2.6 |
| 9 | 7.7 | 7.8 | 12.5 | 12.4 | 4.8 | 4.6 |
| 10 | 15.5 | 14.3 | 16 | 16 | 0.5 | 1.7 |
| 11 | 15.7 | 16.3 | 16.6 | 17.2 | 0.9 | 0.9 |
| 12 | 11.5 | 12.3 | 13 | 12 | 1.5 | −0.3 |
| 13 | 6.6 | 7 | 9.6 | 7.9 | 3 | 0.9 |
| 14 | 4 | 9.7 | 5 | 14 | 1 | 4.3 |
| 15 | 9.4 | 9.2 | 11 | 10 | 1.6 | 0.8 |
Ventricles were measured in millimeters at the point of maximal ventricular width.
Width measured with MRI minus that measured with US is shown.
The right and left lateral ventricles were 14.3 mm and 13.4 mm, respectively, as measured with the initial US study in 1 fetus (patient 8). MRI was performed 6 days later. The widths of the right and left ventricles were 17 mm and 16 mm. This is an important example of how different modality measurements could affect family counseling (Fig. 1). When either modality shows a ventricular width greater than 15 mm, the family receives counseling about the increased risk of CSF diversion; however, intrauterine closure may still be offered.
Fetal US (A) and MR (B) images obtained 6 days apart. The lateral ventricles were 14.3 mm and 13.4 mm on US, and 17 mm and 16 mm on MRI. Figure is available in color online only.
Comparison 2: Interobserver Assessment
For the comparison of dictated US results with blinded measurements, the imaging study of 1 fetus was deemed to have too much motion to provide reliable measurement and was excluded. Twenty-six of the 28 remaining ventricles were larger when measured with US by PRs in a blinded fashion than when measured by MFMs (Fig. 2).
Differences in lateral ventricular sizes (in millimeters) obtained with prenatal US between MFMs in clinical practice and blinded PRs. The 95% CI (dashed lines) is shown.
On average, the right ventricle was only 0.5 mm larger when measured with US by PRs when compared with measurements by MFMs (95% CI 0.33–0.74, p > 0.001). Similarly, the left ventricle was only 0.6 mm larger when measured by PRs (95% CI 0.08–1.14, p = 0.03). The average difference in ventricular width between ventricles measured with US by PRs in a blinded fashion and those dictated by MFMs was 0.6 mm, and this difference was statistically significant (95% CI 0.31–0.84, p < 0.001). Although statistically significant, the clinical significance of this submillimeter disparity is less certain.
For MRI, the average absolute (either larger or smaller) differences between dictated results and blinded measurements were 0.4 mm for the right ventricle, 1.0 mm for the left ventricle, and 0.7 mm for all ventricles. Nine of 30 ventricles were measured as smaller with blinded interpretation. One ventricle had identical measurements with both modalities. The remaining 20 ventricles were measured as larger with blinded interpretation. The difference between the clinical measurements in the patient charts and the blinded measurements obtained with MRI was statistically significant (95% CI 0.03–1.42, p = 0.042); blinded measurements tended to be larger (Fig. 3).
Differences in lateral ventricular sizes (in millimeters) obtained with fetal MRI between PRs in clinical practice and blinded PRs. The 95% CI (dashed lines) is shown.
Comparison 3: Intermodality Assessment
Similar to the findings of our retrospective review, the average size of the lateral ventricles measured with US by PRs in a blinded fashion was significantly different from that measured with MRI (95% CI 1.26–2.67, p < 0.001). The average size of the lateral ventricles was significantly larger when assessed with MRI versus US, with a mean difference of 2.0 mm (Fig. 4).
Differences between lateral ventricular sizes (in millimeters) obtained with prenatal US and fetal MRI. The mean difference (dashed line) and 95% CI (dotted lines) are shown. Fetal MRI was consistently biased toward larger ventricular size (+ 2 mm) compared with prenatal US.
Discussion
Intrauterine myelomeningocele closure has multiple beneficial outcomes, and decreased need for CSF diversion is considered the major benefit.1–3,5–15 Since Tulipan et al. reported the findings of a post hoc analysis showing that prenatal ventricular size greater than or equal to 15 mm conferred no reduction in need for CSF diversion with prenatal surgery, many have used this metric to counsel families and not offer prenatal closure.4 Bruner et al. also found that ventricular size at prenatal myelomeningocele closure affected shunt rates.16 However, possible disparities in ventricular width measurements as a function of specific imaging modality are rarely discussed.
Our analysis found that imaging modality coupled with interobserver variability can result in an average variation in ventricular measurements as great as 3 mm. MRI-based measurements were routinely larger than US-based measurements, even when the two imaging modalities were used only days apart. A significant difference in ventricular size on US and MRI was still observed when measurements were obtained by PRs in a blinded fashion. Although prior studies utilized US for ventricular measurements, this may not be the practice at all centers that now offer intrauterine myelomeningocele closure.1,4 The imaging modality needs to be accounted for when offering prognostic information regarding future likelihood of CSF diversion. Each center should consider performing its own comparison; as in our experience, this could be worthwhile.
Although we found a statistically significant difference in US measurements between observers, clinical significance is unclear because the mean difference was only 0.6 mm. Of note, these measurements were performed by physicians with different specialties (MFMs and PRs). In 26 of 28 ventricles measured (one imaging study was deemed to have too much motion for measurement), the PR measured the ventricles as larger than the initial interpretation by the MFM. Thus, although an interobserver difference was detected, it does not appear to be clinically significant.
The modality used to measure ventricular size appears to be more clinically significant. In our experience, MRI overestimates ventricular size by an average of 2 mm in comparison with US. Absolute differences in increased ventricular size due to progressive hydrocephalus between studies cannot be the explanation because MRI was often performed before US to obtain measurements. Because Tulipan et al.4 reported US-based data, counseling regarding the benefits of prenatal closure and CSF diversion should be based on a careful review of both US and fetal MRI data. If MRI shows a ventricular size of 15 mm or larger, then one should strongly consider reviewing US data before prognosticating about the need for CSF diversion.
Limitations and Future Thoughts
Our data were obtained at a single site. Comparison with data from other fetal centers, to determine variations in ventricular measurements obtained with MRI and US, would be important to analyze reproducibility. Analysis of combined experiences from other centers could further delineate measurement points for prognostication with MRI measurements.
We did not include clinical data for our cohort, as that information would not impact the purpose of this study regarding analysis of variability in ventricular size measurement.
Conclusions
It is critical that clinicians have the best understanding of the outcomes, risks, and benefits of intrauterine myelomeningocele closure when discussing options with mothers who are pregnant with a fetus with myelomeningocele. We hope to bring to light the issue of the modality used to measure intrauterine ventricular size, to ensure accurate measurements before excluding fetuses from or recommending fetuses for intrauterine surgery with all the significant risks and benefits. We found that choice of imaging modality resulted in significantly different fetal ventricular measurements, and measurement with MRI was larger than that with US. Every center involved in counseling families about the risks and benefits of fetal intervention for spina bifida needs to be aware of these possible imaging-based disparities.
Disclosures
The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.
Author Contributions
Conception and design: Lundy, Domino, Mitchell, Fickenscher, Grabb. Acquisition of data: Lundy, Vlastos, Mitchell, Fickenscher, Grabb. Analysis and interpretation of data: all authors. Drafting the article: Lundy, Domino, Grabb. Critically revising the article: Lundy, Domino, Grabb. Reviewed submitted version of manuscript: Lundy, Vlastos, Mitchell, Fickenscher, Grabb. Approved the final version of the manuscript on behalf of all authors: Lundy. Statistical analysis: Domino, Grabb.
Supplemental Information
Previous Presentations
Portions of this material were previously presented as a poster at the American Association of Neurological Surgeons/Congress of Neurological Surgeons Section of Pediatric Neurological Surgery Meeting, Scottsdale, Arizona, December 7, 2019.
References
- 1↑
Adzick NS, Thom EA, Spong CY, et al.. A randomized trial of prenatal versus postnatal repair of myelomeningocele. N Engl J Med. 2011;364(11):993–1004.
- 2
Danzer E, Gerdes M, Bebbington MW, et al.. Fetal myelomeningocele surgery: preschool functional status using the Functional Independence Measure for children (WeeFIM). Childs Nerv Syst. 2011;27(7):1083–1088.
- 3
Hisaba WJ, Cavalheiro S, Almodim CG, et al.. Intrauterine myelomeningocele repair postnatal results and follow-up at 3.5 years of age—initial experience from a single reference service in Brazil. Childs Nerv Syst. 2012;28(3):461–467.
- 4↑
Tulipan N, Wellons JC III, Thom EA, et al.. Prenatal surgery for myelomeningocele and the need for cerebrospinal fluid shunt placement. J Neurosurg Pediatr. 2015;16(6):613–620.
- 5
Adzick NS. Fetal myelomeningocele: natural history, pathophysiology, and in-utero intervention. Semin Fetal Neonatal Med. 2010;15(1):9–14.
- 6
Adzick NS. Fetal surgery for spina bifida: past, present, future. Semin Pediatr Surg. 2013;22(1):10–17.
- 7
Elbabaa SK, Gildehaus AM, Pierson MJ, et al.. First 60 fetal in-utero myelomeningocele repairs at Saint Louis Fetal Care Institute in the post-MOMS trial era: hydrocephalus treatment outcomes (endoscopic third ventriculostomy versus ventriculo-peritoneal shunt). Childs Nerv Syst. 2017;33(7):1157–1168.
- 8
Meuli M, Moehrlen U. Fetal surgery for myelomeningocele is effective: a critical look at the whys. Pediatr Surg Int. 2014;30(7):689–697.
- 9
Moldenhauer JS, Soni S, Rintoul NE, et al.. Fetal myelomeningocele repair: the post-MOMS experience at the Children’s Hospital of Philadelphia. Fetal Diagn Ther. 2015;37(3):235–240.
- 10
Mummareddy N, Dewan MC, Huang A, et al.. Intrauterine closure of myelomeningocele is associated with superior long-term quality of life than postnatal closure: a single-center study. J Neurosurg Pediatr. 2019;24(2):115–119.
- 11
Lundy P, Vlastos E, Grabb PA. Prenatal closure of myelomeningocele reduces Chiari associated brainstem dysfunction. Poster presented at: 48th Annual AANS/CNS Section on Pediatric Neurological Surgery Meeting; December 7, 2019;Scottsdale, AZ.
- 12
Rintoul NE, Sutton LN, Hubbard AM, et al.. A new look at myelomeningoceles: functional level, vertebral level, shunting, and the implications for fetal intervention. Pediatrics. 2002;109(3):409–413.
- 13
Silberberg A, Robetto J, Grillo M. Ethical issues in intrauterine myelomeningocele surgery. New Bioeth. 2018;24(3):249–257.
- 14
da Costa MDS, Nicacio JM, Dastoli PA, et al.. Alterations in skull base anatomy in intrauterine and postnatal repaired myelomeningoceles. Childs Nerv Syst. 2020;36(11):2757–2763.
- 15
Werner EF, Han CS, Burd I, et al.. Evaluating the cost-effectiveness of prenatal surgery for myelomeningocele: a decision analysis. Ultrasound Obstet Gynecol. 2012;40(2):158–164.
- 16↑
Bruner JP, Tulipan N, Reed G, et al.. Intrauterine repair of spina bifida: preoperative predictors of shunt-dependent hydrocephalus. Am J Obstet Gynecol. 2004;190(5):1305–1312.
