This study protocol was developed through discussion between our research team which includes clinical and preclinical research experts (CMB, EH, BT, EF), an information specialist (BS), and research methodologists with expertise in knowledge synthesis (NA, KDC). The protocol was prepared using the PRISMA-P reporting guidelines (39) and the protocol format for the preparation, registration and publication of systematic reviews of animal intervention studies published by de Vries et al (40). This protocol was registered using the Open Science Framework (osf.io/5rc6t). A completed PRISMA-P checklist for this protocol is provided as an additional file (see additional file 1). Post-hoc modifications to the current protocol incurred during the review process will be documented in the publication of the final report to ensure transparency.
Eligibility criteria
We will include all study designs of research describing preclinical NEC models that evaluate the efficacy of MSCs or other cell therapy treatments that are published in English. Potentially relevant non-English materials will be provided in the appendix section but will not be included in the analysis. We will not limit our eligibility criteria by year of publication or publication status. Details on the study PICOS are provided below.
Population
The population will include all preclinical in vivo animal models of experimentally induced NEC. To enhance the generalizability of our findings, NEC will be defined as any histological intestinal tissue injury. Various mechanisms of NEC have been reported in animal models in the literature (41, 42). These include gavage formula or hyperosmolar formula gavage feeds, exposure to hypoxia and/or hypothermia and administration of bacteria.
Intervention
The intervention of interest is all MSCs cell types, independent of the source, in animal models. MSCs will be defined by using the International Society for Cellular Therapy consensus statement. The minimal criteria to meet to define human MSC are: (1) adhesion to plastic; (2) expression of the cell surface markers CD105, CD73 and CD90 and lack of CD45, CD34, CD14 or CD11b, CD79ɑ or CD19 and HLA-DR surface molecules expression and (3) differentiation to osteoblasts, adipocytes and chondroblasts in vitro (43).
All types of MSCs will be considered in the review to enhance the generalizability of our study findings. Any therapeutic MSCs regimen (prophylactic – before or during induction of NEC, or rescue after induction of NEC) will be considered.
Comparison
In our population study of experimental animal models of NEC, we will consider all types of reported comparator. We will also consider groups with no comparator.
Outcome
The primary outcome is histological signs of NEC measured at any time after the administration of MSCs. The secondary outcomes measured are: survival, bowel perforation, gut permeability (e.g. disruption of thigh junction proteins, liver fatty acid-binding protein level and intestinal fatty acid binding protein level), gut motility, levels of inflammatory markers (e.g. white blood cells count, C-reactive protein level and fecal calprotectine) cytokine levels (e.g., interleukine-8 and toll-like receptor 4) and any describe adverse events.
Study design
We will include any study design in the search including both comparative and non-comparative studies as well as empirical research.
Information sources
Search strategy
The search strategies will be developed and tested through an iterative process by an experienced medical information specialist in collaboration with the study team members. The MEDLINE strategy will be peer-reviewed by another senior information specialist prior to execution using the Peer Review of Electronic Search Strategies (PRESS) guidelines (44). Proposed databases include MEDLINE, Embase, and Web of Science. Searches will utilize a combination of controlled vocabulary (e.g., “Stem cells”, “Mesenchymal Stem Cells”, ”Stromal Cells” and “Enterocolitis, Necrotizing”) and keywords (e.g., “MSCs”, “stromal cell”, “neonatal enterocolitis”). Vocabulary and syntax will be adjusted across the databases. To enhance search efficacy, we will use animal filters validated for Pubmed/MEDLINE and Embase (40, 45, 46), amended as appropriate to accommodate new developments in vocabulary. There will be no language or date restrictions on any of the searches. We will perform a grey literature search of conference websites and animal research organizations’ websites to identify literature not included in the aforementioned databases, and also perform general searches using Google Scholar.
Study records
Data management
DistillerSR (Evidence Partners Inc, Ottawa, Canada) software will be used for records and data management throughout the review. The resulting search citations will be deduplicated and then imported into DistillerSR. Study screening and data extraction forms will be created on the platform to facilitate review of citations obtained from the search.
Selection process
Two reviewers, independently and in duplicate, will carry out citation screening in two stages, reviewing records against the a priori eligibility criteria using an online systematic review software program (Distiller Systematic Review (DSR) Software; Evidence Partners Inc, Ottawa, Canada). At Stage 1, the titles and abstracts of all records obtained from the search will be screened in duplicate independently by CMB and EH while Stage 2 screening will encompass full-text review of the articles deemed potentially relevant at Stage 1. If there are disagreements between the reviewers, it will be resolved by a consensus or by a third member of the study team (EF or BT). Both stages of screening will begin with a calibration exercise to ensure consistent application of eligibility criteria between reviewers. This will include pilot screening of titles and 25 citations at Stage 1 and 25 full-text articles at Stage 2. In accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (39), the exclusion reasons of potentially eligible studies will be recorded.
Data collection process
The two independent investigators (CMB and EH) conducting screening and extraction will use standardized forms (See drafts in Appendix 1). These forms will be piloted on a set of 10 documents and then discussed. If necessary, changes to the forms will be made and they will be re-piloted on 10 further documents. This process will be repeated until 80% agreement or higher is reached between the reviewers. We will contact the first author of the included studies once, if necessary, for any missing data.
Data items
Once all studies are identified, data extraction will be performed using a standardized data extraction form in Microsoft Excel (Microsoft Corporation, Seattle, Washington, USA). We will collect data related to key study items that include but not limited to variables in Table 1 below. If means and measures of dispersion are reported via figures in the primary studies, they will be approximated using online tools such as Webplotdigitizer (A. Rohatgi, 2020, Pacifica, California, USA).
Outcomes and priorization
The primary outcome is histological signs of NEC measured at any time after the administration of MSCs. The histological signs can be tissue inflammation, bacterial invasion, tissue necrosis and signs of micro-perforation. If applicable, we will use the standard histologic scoring previously described in the literature to assess the severity of histological changes in NEC (41, 47). Histopathology of the lesions will be categorized in four classes: grade 1 focal, mild injury confined to villous types, grade II, partial loss of vili, grade III, necrosis extending to submucosa and grade IV, transmural necrosis.
The secondary outcomes measured will be: survival, bowel perforation, gut permeability, gut motility, levels of inflammatory markers, cytokine levels and any reported adverse events. Survival will be measured in number of days between the diagnosis of NEC and unintentional death. Bowel perforation will be defined as any free air in the abdominal cavity on imaging or visualization of a hole in the wall of the intestinal barrier by direct visualization or by histology. Any possible parameters to assess gut permeability (including but not restricted to disruption of thigh junction proteins, liver fatty acid-binding protein level and intestinal fatty acid binding protein level) and motility will be considered as a valid measure for these outcomes. Any kind of inflammatory markers (including but not restricted to white blood cells count (cells x 109/L), C-reactive protein (mg/L) and fecal calprotectine (mcg/g)) and cytokine levels (including but not restricted to interleukine-8 (pg/mL) and present of toll-like receptor 4) will be recorded. All reported adverse events will be considered.
Risk of bias in individual studies
Two independent investigators (CMB and EH) will evaluate the risk of bias for each study included in the review. We will use the SYRCLE’s risk of bias tool for animal studies (48). This tool was created to establish consistency and avoid discrepancies in the assessment of risk of bias in systematic reviews of animal intervention studies. Each item will be assigned a value of low, high or unclear. Elements of the SYRCLE’s tool include assessment of method used for the sequence generation, description of baseline animal characteristics, allocation concealment, random animal housing, blinding of outcome assessment, random outcome assessment, completeness of outcome data and statement of selective outcome reporting.
Data synthesis
Descriptive statistics and tables of evidence summary will be reported. Study characteristics will be summarized using frequencies and percentages. As a rule, heterogeneity that may be explained by clinical or methodological differences between studies will preclude any planned meta-analyses. If so, study results will be synthesized narratively. Outcomes with dichotomous endpoints (e.g., survival and presence of intestinal perforation) from each included study will be pooled and described using risk ratio and/ or odds ratios and 95% confidence intervals that incorporate a random effect modelling approach. We will use forest plots to visualize the data. Statistical heterogeneity between studies will be quantified with I-squared statistics. Sparse data will not be meta-analyzed but described narratively. Outcomes with continuous endpoints (E.g., gut permeability, gut motility, levels of inflammatory markers and cytokine levels) will be pooled using mean difference or standardized mean difference. Meta-regression with multiple study level covariates will be attempted when there are at least 10 studies in the body of evidence. Otherwise, clinical and methodological diversity in studies will be explored in pre-specified subgroup and sensitivity analyses. We will analyze subgroups for the following: animal models, sex, experimental model of NEC (e.g., hypercaloric stress, exposure to hypoxia or exposure to hypothermia), administration route of MSC (e.g., intravenous or intraperitoneal), dose of MSC, timing of MSC administration after induction of NEC (less or equal to one hour, greater than one hour to less or equal to four hours, more than four hours or multiple doses), other treatment used and control group. To assess the potential meta-bias and publication bias, we will use the funnel plot as a graphical tool. The quality of reporting of individual preclinical studies will be assessed in accordance with the elements of the ARRIVE (Animal Research: Reporting of In Vivo Experiments) guidelines (49).