Does prenatal surgical repair of myelomeningocele lead to better school-age outcomes?

Title: Does prenatal surgical repair of myelomeningocele lead to better school-age outcomes?

Manuscript citation: Houtrow AJ, Thom EA, Fletcher JM, Burrows PK, Adzick NS, Thomas NH, et al. Prenatal Repair of Myelomeningocele and School-age Functional Outcomes. Pediatrics. 2020 Feb;145(2):e20191544.

Reviewed by: Faith Kim MD¹, David A. Bateman MD¹, Donna M. Garey MD MPH¹

Affiliations: ¹NewYork- Presbyterian Children’s Hospital of New York, Columbia University Medical Center, Division of Neonatology and Department of Pediatrics.

Corresponding Author: Faith Kim MD, 3959 Broadway, PH-17, NewYork- Presbyterian Children’s Hospital of New York, NY, 10032. Email: [email protected]. Phone: 757-892-4018.

Keywords: myelomeningocele, fetal surgery, hydrocephalus, functional outcomes

Type of investigation: Prognosis; exploratory secondary analysis of an interventional randomized controlled trial (RCT) entitled “A randomized trial of prenatal versus postnatal repair of myelomeningocele.” (MOMS1 trial). 

Question: Do children who underwent prenatal repair of myelomeningoceles have improved functional outcomes at 6 to 11 years of age compared to those who underwent postnatal repair? 

Methods: Design: Follow-up cohort study of a previous multicenter RCT

Allocation: In this current study (MOMS2 trial), families were offered enrollment when the child was between 6 and 11 years old to assess neurodevelopmental and behavioral outcomes.

Blinding: Study-designated psychologists and physical therapists were blinded to the treatment group, as were the radiologists who reviewed MRI images. Caregivers were not blinded.

Follow-up period: Children were followed up between the ages of 6 to 11 years. 

Setting: Three clinical centers that participated in the MOMS1 trial or caregivers’ homes.

Patients: The study population consisted of 161 families out of the original 183 families. Not assessed were 8 children who died following randomization in the first trial, 10 who declined to participate and 4 who were lost to follow-up. Among those seen at school age, 79 participants were in the prenatal surgery group compared to 82 participants in the postnatal surgery group.

Intervention: In the MOMS1 trial, 183 eligible women with a fetal diagnosis of myelomeningocele were randomly assigned to undergo standard postnatal repair or prenatal repair < 26 weeks gestation at one of three participating centers.

Outcomes: Primary outcome: Composite score from the Vineland Adaptive Behavior Scales.

Secondary outcomes: Neuropsychological and behavioral outcomes included the Kaufman Brief Intelligence Test (KBIT-2) scores, Woodcock-Johnson Tests of Achievement (WJ-III) reading and math subtests, California Verbal Learning Test-Children’s Version, Children’s Memory Scale (CMS) visual memory subtests, Berry-Buktenica Developmental Test on Visual-Motor Integration, Purdue Pegboard Test, Developmental Neuropsychological Assessment (NEPSY-II) word generation subtest, Behavioral Rating Inventory of Executive Function (BRIEF), SNAP-IV assessment of inattention and hyperactivity-impulsivity, and the Child Behavioral Checklist (CBCL). Other outcomes included parental stress assessment and quality of life at home, motor outcome and activities of daily living using the Functional Rehabilitation Evaluation of Sensori-Neurologic Outcomes (FRESNO) validated questionnaire, and the frequency of procedures such as shunt placement or revision and presence of abnormalities on MRI (e.g. hindbrain herniation, syringomyelia, epidermoid cysts).

Analysis and Sample Size: At the time of calculation for the MOMS2 study, there were 177 survivors. The authors estimated a power of 90% to detect a 0.5 SD difference in the composite score from the Vineland assessment with a type 1 error rate of 5%. Intention-to-treat principle was used to analyze outcomes.

Patient follow-up: Out of 175 survivors, 161 families (92%) were enrolled in this follow-up study. Table 1 depicts a breakdown of participants that completed their assessments.

Table 1: Number of children from the original MOMS1 study who completed assessments and imaging in the follow-up cohort at school age.  

Main Results: Out of 159 families that completed Vineland-3 assessments, the composite score was not significantly different between the two groups (89  9.6 in the prenatal group versus 87.5  12 in the postnatal group; p=0.35) even after adjustment for sex (number of females were significantly less in the prenatal group, 43% versus 65%; p=0.006). The average age of children at their visit was similar in both groups (7.8 years) Though, children in the prenatal group were born at a younger gestational age (34.4  2.6 weeks versus 37.4  1 weeks; p<0.001). As shown in Table 2, children in the prenatal group demonstrated significantly improved scores on multiple assessments compared to the postnatal group; however, there were 5 children in the postnatal group who were low-functioning and unable to complete the tasks and were assigned the lowest score. When they were removed in a sensitivity analysis, only the WJ-III reading composite and Purdue Pegboards scores remained significantly different between the two cohorts. Children in the prenatal group demonstrated improved sensorimotor outcomes compared to the postnatal group (92%  9% versus 85%  18%; p=0.001). They required fewer shunts for hydrocephalus (49% versus 85%; p<0.001) and fewer shunt revisions (47% versus 70%; p=0.02). Families in the prenatal group reported overall improved quality of life and less financial and personal strain.

Table 2: Outcomes of Vineland-3 composite scores (primary outcome) and other neuropsychological assessments in the prenatal and postnatal surgical groups

Study Conclusion: Although overall adaptive behavior, as measured by the Vineland composite scale, was not significantly different between the two groups, children who underwent prenatal repair continued to demonstrate improved gross and fine motor outcomes, reduced neuroimaging abnormalities, and overall improved quality of life at school age for both children and families.

Commentary: Myelomeningocele (MMC) is the most common central nervous system malformation associated with morbidities including hydrocephalus, motor disabilities and mental impairment [1,2]. According to a study in California, 20% of mothers chose to terminate following this diagnosis and despite improved survival rates up to 80% into adulthood, up to one-third of patients require some form of daily assistance and multidisciplinary care coupled with lifelong medical expenses [2]. In utero closure was first proposed in the 1980s and performed successfully in various animal models demonstrating improved neurologic outcomes, laying the foundation for the MOMS1 trial [3].

MOMS1 was a multicenter, RCT including 158 healthy mothers with a singleton fetus diagnosed with a MMC randomized to prenatal surgery before 26 weeks of gestation versus postnatal surgery. The study was terminated early on the basis of efficacy of prenatal surgery [3]. The prenatal surgery group had fewer females, more premature birth, and fetuses with a more severe lesion level. By 12 months, fewer infants in the prenatal surgery group died or met indication for a shunt compared to the postnatal group (68% versus 98%, RR 0.70 [98% CI 0.58, 0.84]; p<0.001) [3]. In a subsequent study involving the entire anticipated cohort (183 mothers), the findings were similar in terms of death and need for shunt, but there was also a reduction in the number of children who required a shunt revision by 12 months in the prenatal surgery group compared to the postnatal group (RR 0.38 [95% CI 0.22,0.66]; p<0.0001) [4]. At 30 months, the prenatal group demonstrated higher rates of independent ambulation, functional level atleast two better than the anatomic level, and higher psychomotor scores [5].  

In the MOMS2 trial, investigators hypothesized that the benefits of prenatal surgery would result in improved adaptive behavior using the Vineland Adaptive Behavior Scales and other neurocognitive and behavioral outcomes [6]. The Vineland-3 measures the level of adaptive ability by assessing three core domains including communication, daily living skills and socialization, and has been well-validated in previous studies [6]. Although children in the prenatal group continued to have improved motor and neurologic outcomes, prenatal surgery was not associated with improved adaptive behaviors. There were small but significantly higher scores on the KBIT-2 Composite IQ score, WJ-III Reading composite, Child Memory Scale: Dot learning, and Purdue pegboard test in the prenatal repair group. When the five low-functioning children in the postnatal group were removed in a sensitivity analysis, only reading skills and Pegboard test scores were significantly higher in the prenatal group.

The pattern of higher reading skills and verbal scores in the prenatal group, albeit still lower than the general population, along with limited executive function and visual processing deficits are comparable to previous studies that have defined the “cognitive phenotype” of individuals with spina bifida [7]. Limitations of this study included a lack of adjustment for socioeconomic status when assessing cognitive outcomes, inability for caregivers to be blinded, and a study population that may not reflect the general demographics of patients with MMCs, limiting the generalizability of these findings. However, the MOMS1 trial along with its follow-up studies have continued to demonstrate the safety and efficacy of fetal repair. Long-term follow-up of these patients into adulthood will be the true testament to the risks and benefits of this approach.

EBM Lesson: Worst-Rank Score Analysis; In this study, five low-functioning children in the postnatal surgery group could not perform assessments and were assigned the lowest score. The authors found that by excluding them in a sensitivity analysis, several of their previous findings were no longer significant. If a participant cannot undergo evaluation because of morbidity from the underlying disease being studied or death prior to evaluation, investigators may assign them a “worst-rank score” [8]. Some have criticized this approach as introducing bias, while others have advocated using this method when missing data results from manifestations of the underlying disease [9]. Before including or excluding these patients, investigators should assess each case to determine if there is a relationship between their inability to perform an evaluation and the disease studied. Sensitivity analyses may be performed when missing data exists by assigning them the highest or lowest values and assessing whether altering these values leads to a different conclusion [10]. In this paper, the authors excluded 5 subjects in their sensitivity analysis, which is often done the other way by excluding them in the primary analysis and including them in a sensitivity analysis. However, they were all from the postnatal surgical group and were unable to complete the assessments potentially due to their underlying condition, type of repair or another unobserved variable. By excluding their data altogether, we are assuming their data is missing completely at random, which is likely untrue [10]. This may lead to a type II error since it is plausible that there is a biological relationship between postnatal repair and worse outcomes. Although the principle behind conducting “worst-rank score analysis” is appropriate in this study, excluding their results implies that their findings are not clinically relevant. These children are the most significantly delayed and likely to have lifelong severe disabilities. Excluding them could limit the ability to detect the benefits of effective therapies, particularly if there were no other identifiable variables that could explain their status. Further elaboration about the difference in outcomes with a focus on those who are more severely affected could provide more information about the spectrum of outcomes in this population. This study adds to evidence that there is continued efficacy and safety behind prenatal repair that may extend beyond motor and neurologic outcomes but continues to warrant continued long-term evaluation.

Acknowledgment:

The Journal club is a collaboration between the American Academy of Pediatrics- Section of Neonatal Perinatal medicine and the International Society of Evidence- based neonatology (EBNEO.org).

REFERENCES

1. Moldenhauer JS, Adzick NS. Fetal surgery for myelomeningocele: after the management of myelomeningocele study (MOMS). Semin Fetal Neonatal Med 2017;22: 360–6.
2. Danzer E, Flake AW. In utero repair of myelomeningocele: rationale, initial clinical experience and a randomized controlled prospective clinical trial. Neuroembryology Aging 2006;4: 165–74.
3. Adzick NS, Thom EA, Spong CY, Brock JW, Burrows PK, Johnson MP, et al. A randomized trial of prenatal versus postnatal repair of myelomeningocele. N Engl J Med 201;364: 993–1004.
4. Tulipan N, Wellons JC, Thom EA, Gupta N, Sutton LN, Burrows PK, et al. Prenatal surgery for myelomeningocele and the need for cerebrospinal fluid shunt placement. J Neurosurg Pediatr 2015;16: 613–20.
5. Farmer DL, Thom EA, Brock JW, Burrows PK, Johnson MP, Howell LJ, et al. The management of myelomeningocele study: full cohort 30-month pediatric outcomes. Am J Obstet Gynecol 2018;218: 256.e1-e13.
6. Houtrow AJ, Thom EA, Fletcher JM, Burrows PK, Adzick NS, Thomas NH, et al. Prenatal repair of myelomeningocele and school-age functional outcomes. Pediatrics 2020;145: e20191544.
7. Dennis M, Barnes MA. The cognitive phenotype of spina bifida meningomyelocele. Dev Disabil Res Rev 2010;16: 31–9.
8. Wang C, Scharfstein DO, Colantuoni E, Girard TD, Yan Y. Inference in randomized trials with death and missingness. Biometrics. 2017;73: 431–40.
9. Lachin JM. Worst-rank score analysis with informatively missing observations in clinical trials. Control Clin Trials 1999;20: 408–22.
10. Thabane L, Mbuagbaw L, Zhang S, Samaan Z, Marcucci M, Ye C, et al. A tutorial on sensitivity analyses in clinical trials: the what, why, when and how. BMC Med Res Methodol 2013;13: 92.