Preclinical stem cell therapy in fetuses with myelomeningocele: A systematic review and meta‐analysis

Abstract Objective We performed a systematic review to summarize the efficacy and safety of in utero stem cells application in preclinical models with myelomeningocele (MMC). Methods The study was registered with PROSPERO (CRD42019160399). We searched MEDLINE, Embase, Web of Science, Scopus and CENTRAL for publications articles on stem cell therapy in animal fetuses with MMC until May 2020. Publication quality was assessed by the SYRCLE's tool. Meta‐analyses were pooled if studies were done in the same animal model providing similar type of stem cell used and outcome measurements. Narrative synthesis was performed for studies that could not be pooled. Results Nineteen and seven studies were included in narrative and quantitative syntheses, respectively. Most used mesenchymal stem cells (MSCs) and primarily involved ovine and rodent models. Both intra‐amniotic injection of allogeneic amniotic fluid (AF)‐MSCs in rat MMC model and the application of human placental (P)‐MSCs to the spinal cord during fetal surgery in MMC ovine model did not compromise fetal survival rates at term (rat model, relative risk [RR] 1.03, 95% CI 0.92–1.16; ovine model, RR 0.94, 95% CI 0.78–1.13). A single intra‐amniotic injection of allogeneic AF‐MSCs into rat MMC model was associated with a higher rate of complete defect coverage compared to saline injection (RR 16.35, 95% CI 3.27–81.79). The incorporation of human P‐MSCs as a therapeutic adjunct to fetal surgery in the ovine MMC model significantly improved sheep locomotor rating scale after birth (mean difference 5.18, 95% CI 3.36–6.99). Conclusions Stem cell application during prenatal period in preclinical animal models is safe and effective.


| INTRODUCTION
Myelomeningocele (MMC) is a severe congenital malformation of the central nervous system resulting from an incomplete closure of the neural tube during the third-fourth week of embryonic development. 1 The prevalence of MMC varies greatly among geographical areas ranging from 0.3 to 59.0 cases per 10,000 births. 2 MMC is characterised by the protrusion of the neural placode and its meninges through a malformed vertebral arch and skin defect. The condition can be detected by prenatal ultrasound scan as early as the first trimester; however, the majority of cases are diagnosed during the second trimester (anomaly) ultrasound scan. 3,4 Apart from preventive therapy using periconceptual vitamins such as folic acid, current management following prenatal diagnosis may include termination of pregnancy, postnatal or more recently fetal surgery. 5 The rationale for fetal repair before birth is that MMC is a 'progressive' condition with cumulative spinal cord functional loss throughout gestation, as demonstrated in clinical and animal studies. [6][7][8] Fetal surgery can arrest this deterioration and improve the patients' ability to walk unaided at 30-month old. [9][10][11] However, the benefit of the surgery to bladder function is still under review. [12][13][14][15][16] Despite these improvements, there are several shortcomings of fetal surgery. Although the number of centres offering fetal surgery for MMC has been increasing, 17 global availability is still limited. Furthermore, fetal surgery is usually performed in the late second trimester, between 23 and 26 weeks' gestation, to reduce the risk of chorioamniotic membrane separation and associated preterm birth. 18,19 Moreover, fetal surgery is not a cure. When considering patient outcomes at 30-month-old age; for example, approximately half of the fetal treated patients have to rely on clean intermittent catheterization to pass urine and more than half cannot walk without the aid of orthosis. 11,12 Additional interventions during fetal life such as the use of stem cells, may improve the shortcomings of fetal surgery. Stem cell transplantation, particularly of mesenchymal stem cells (MSCs), have been reported in both animal and clinical studies for spinal cord injury. [20][21][22] In clinical cases of individuals suffering from spinal cord injury, stem cell therapy improves light-touch and pinprick sensory function, bladder function and also increases the score of the daily living activities when compared to patients who receive only rehabilitation. 22 For treatment of MMC, in utero stem cell therapy has been reported to improve outcome in several animal studies, but as yet no human trials have been conducted.
Several animal models have been used to evaluate pathophysiology and treatment options for MMC. These models can be divided into surgically and non-surgically induced models. All ovine, rabbit and chick models involve surgical manipulation; laminectomy and resection of dura mater, to create an MMC-like lesion. 23,24 In contrast, in the rat model, the lesion is induced by gavaging retinoic acid to pregnant dams early in gestation. Retinoic acid is a well-known teratogen that disrupts the process of neural tube closure leading to the MMC defect in the pups. 25 All of the aforementioned models, both surgical and non-surgical, have been applied to study feasibility, safety and efficacy of in utero stem cell transplantation for MMC.
In this study, we systematically reviewed the application of stem cells in preclinical animal models of MMC with regards to their safety, efficacy and to justify the possibility of translation into a clinical study. from inception until May 2020. The search strategy included both Medical Subject Headings term and free text words (Data S1). Topicrelated reviews were manually searched to retrieve additional relevant articles. Endnote X9 (Thomson Reuters) was used to remove duplicate studies based on names of the authors, titles, and year of publications.

| Inclusion and exclusion criteria
The population was MMC animals receiving an in vivo, in utero application of stem cells. The intervention included any type of stem cells; embryonic stem cells (ESCs), pluripotent stem cells (IPSCs), neuronal stem cells (NSCs), neural crest stem cells (NCSCs) and MSCs. Comparator group was animals receiving only fetal surgery, saline injection or no treatment at all. Studies were excluded if stem cells were administered after birth or was published in non-English language. Outcomes examined were related to safety, survival and efficacy as described below. No date restrictions were applied.
Editorial comments, review studies and publications without full-text accessibility were excluded.

| Study selection
Titles and abstracts were independently screened and selected for relevance by two reviewers (Yada Kunpalin and Sindhu Subramaniam). A full-text review was performed for all the selected studies based on the aforementioned criteria. Any disagreement was resolved through discussion with a third reviewer (Silvia Perin). In case of overlapping studies, only the most recent publication was included.

| Data extraction
A predefined pro forma was created by the reviewers before data extraction. Extracted information included year of publication, types of animal model, number of animals, sample randomization and gestation age (GA) when the defect was created. Treatment information included source and types of stem cells, dosage, type of vehicles, controls and GA when stem cells were administered, and GA at euthanasia. Extracted outcomes were animal survival rate, defect coverage, spinal cord histopathology and neurological function.
Corresponding authors were contacted for further/missing data.

Risk of bias was independently assessed by Yada Kunpalin and
Sindhu Subramaniam by the Systematic Review Centre for Laboratory Animal Experimentation's (SYRCLE's) tool for animal interventional studies. 27 Discrepancies between the reviewers were resolved through consensus by the third reviewer (Silvia Perin).

| Data synthesis and statistical methods
Meta-analyses were performed only if studies were consistent with regards to the type of animal model, stem cells and outcome measurements. For studies that could not be pooled, we present a narrative data synthesis with descriptive statistics.
Meta-analyses were carried out using the software provided by the Cochrane Collaboration, Review Manager (RevMan) version 5.3.
Quantification of the heterogeneity across the included studies was assessed by chi-squared value test and inconsistency index (I 2 ). I 2 of >50% and <0.1 of α value of chi-square were deemed to have significant heterogeneity. 28 Consequently, a random-effect model was used to analyse the data; otherwise, the fixed-effect model was applied. In terms of animal survival rate and MMC defect coverage rate, the results were represented by relative risk (RR). For Sheep Locomotor Rating (SLR) scale, the improvement was displayed with mean difference.

| Risk of bias assessment
Risk of bias of the included studies is shown in Figure 2

| Study characteristics
The characteristics of the included studies, such as type and source of stem cells, animal models and available outcomes, are shown in

| Animal survival
Twenty-one studies reported data on animal survival after in utero stem cell application, 30 Figure 3B).
Outcomes of defect coverage are summarized in Table 3. In terms of human xenogeneic transplantation, one study found that intra-amniotic injection of human AF-MSCs in the retinoic acidinduced fetal rat MMC model at E17 significantly reduced the area of the MMC defect compared to saline injection (Table 3). 44 Another study demonstrated that in utero transplantation of 3-dimensional (3D) skin generated from human AF-derived iPSCs resulted in more rats having some degree of defect coverage compared to no transplantation (Table 3). 45

| Spinal cord histopathology and function
There were 11 studies reporting the effect of stem cells on spinal cord histopathology and/or function with almost all using MSCs (90.1%, 10/11); 63.6% (7/11) of studies applied human MSCs. 34,35,38,44,47,[50][51][52][53][54][55] Fetal surgical ovine and retinoic acid-induced fetal rat models of MMC were used in 54.5% (6/11) and 45.5% (5/11) of the studies, respectively. Improvement of spinal cord outcomes are shown in Table 4. Meta-analysis to study the spinal cord function was possible in five studies in the surgically created ovine model of MMC ( Figure 3D). Incorporation of P-MSCs at the time of MMC fetal surgical closure improved motor function of the lower limbs compared to fetal surgery alone, as determined by SLR scale (mean difference 5.18, 95% CI 3.36-6.99; Figure 3D). The density of large neurons was also found to be increased with the intervention (Table 4).
In small animal models, injection of adult rat BM-MSCs at E16 into the spinal cord of retinoic acid generated fetal rats with MMC, was associated with a reduction in spinal cord cell death assessed by TUNEL analysis at E20 (death cells; 4.8 � 0.3% vs. 8.9 � 0.6%, p < 0.05), 34  Although the cells did not differentiate, xenogeneic cells were able to engraft and produce the neurotrophic factors glial cell line-derived neurotrophic factor and brain-derived neurotrophic factor. 47 Another study demonstrated that human xenogeneic NCSCs deliv- -291 In light of this, the efficacy of intra-amniotic AF-MSCs to induce defect coverage and eventually to improve neurological functions remains to be evaluated in both small and large animal models. This is important if we consider that, in rodents and rabbits, the volume of intra-amniotic cavity and the gestation are respectively smaller and relatively shorter than in the ovine and/or eventually the human. The use of large animal models will provide further information that can be translated in future clinical trials; for example, the technique for stem cell delivery, the determination of appropriate stem cell dosage and the number of injections required to achieve a complete defect coverage. 58 As the intra-amniotic volume of humans is much larger than that of the rat, improvements in a technique or vehicle to deliver stems needs further development in order to promote cell survival, migration and attachment. The longer gestational period in large animal models would also allow information on medium-to-long term effects of MSCs such as cell engraftment and characterisation of regenerated skin layers.
The rationale for incorporating stem cells as an adjunct to fetal surgery is to regenerate the 'already damaged' spinal cord as even after fetal surgery, more than half of children with MMC were unable to walk without orthoses. 11 In this systematic review, we found a