Stem Cell Therapies for Gastrointestinal Anastomotic Healing: A Systematic Review and Meta-Analysis on Results from Animal Studies

Despite advances in surgery, anastomotic leaks (ALs) after gastrointestinal surgery are common, occurring in up to 20% of colorectal [1] and 11.4–21.2% of esophageal anastomoses [2, 3]. Conventional views implicate tension at the anastomotic site, poor surgical technique, and local ischemia as culprits, but experimental data have not verified causal links [4‒6]. Specific factors have been described to be associated with ALs, including preoperative radiation, male sex, and obesity [1], but specific pathophysiologic mechanisms remain elusive. Recent studies have shed some light into this surgical conundrum as certain bacterial strains were shown to cause ALs in experimental models, through the production of collagen-degrading enzymes and activation of matrix metalloproteinases [7‒9].

Multiple interventions have been proposed to assist anastomotic healing and prevent leaks, including stem cell therapies, hyperbaric oxygen, matrix metalloproteinase inhibitors, and growth factors [10]. Stem cell therapies are particularly appealing since mesenchymal stem cells can differentiate into multiple different cell types and assist healing through multiple independent mechanisms. Stem cells promote wound healing through epithelial regeneration [11], differentiation into different cell types, and promotion of angiogenesis [12]. Several experimental studies have assessed the efficacy of stem cell therapy in gastrointestinal anastomotic healing, but results have been conflicting, with some demonstrating a favorable effect [13] and others reporting no difference in outcomes [14]. Due to these discrepancies, there is a need to assess the literature in a systematic fashion. Caziuc and colleagues [15] previously performed a systematic review of 3 studies evaluating the role of stem cells in colorectal anastomotic healing. The purpose of the present study is to perform a systematic review and meta-analysis to assess the efficacy of stem cell therapy on gastrointestinal anastomotic healing. We assessed surgical anastomoses of the esophagus, stomach, and small and large intestines.

A systematic review of the literature was conducted based on the PRISMA guidelines [16]. The study protocol was established and agreed by all authors. The protocol was not registered. Institutional Review Board approval was waived.

All studies evaluating the effects of stem cell therapies for gastrointestinal anastomotic healing were considered. Gastrointestinal anastomoses were defined as the surgical connection of the gastrointestinal tract from the esophagus to the rectum. Biliary anastomoses were outside the scope of the study and were not considered in the analysis. We only focused on peer-reviewed studies, and conference abstracts were excluded from the analysis. Studies where a full-text was not available in English were also excluded.

The literature search was performed in Medline (PubMed), Web of Science, and the Cochrane Library by two authors (A.G., L.K.). The last day of the search was February 14, 2021. The queries used were (anastomosis stem cell) and (anastomotic stem cell). The identified studies were screened for additional records that may have been missed during the search. Review articles were also identified and screened for records.

Data were extracted from studies independently by two authors (A.G., L.K.), and discrepancies were addressed by resolved by a third reviewer (E.F.). The main outcomes of interest were AL rates on postoperative day (POD) 7 and at the latest time point reported. Authors of specific studies were contacted to confirm specific results. Subanalyses were performed among studies focusing on high-risk anastomoses (i.e., ischemia, radiation, or colitis) and among studies with similar design. A secondary outcome was anastomotic bursting pressures on POD 7. Data regarding anastomotic bursting pressures were meta-analyzed only among studies focusing on bowel anastomoses in rats as swine bursting pressures are expected to differ considerably. Meta-analysis of other outcomes reported by some studies, such as histologic scores and immunohistochemical assessments, were not performed due to discrepancies in individual methodologies.

Statistical analysis was performed using RStudio (v.1.2.5019) and the packages “meta” and “metafor.” Outcomes were reported as odds ratios (ORs) for categorical outcomes and standardized mean differences for continuous variables. All results were reported with 95% confidence interval (CI), and all p values were two-tailed. We used random-effects meta-analysis due to considerable heterogeneity in experimental methods among different studies. Heterogeneity was measured using the I² statistic and Cochran’s Q statistic.

The PRISMA flowchart is displayed in online supplementary Figure 1 and checklist in online supplementary Table 1 (for all online suppl. material, see www.karger.com/doi/10.1159/000526603). Out of 731 records identified, 611 remained after removal of duplicate records. After screening records by title and abstract, 25 were assessed as potentially eligible. Ten records were further excluded for various reasons. Three studies were excluded due to focusing on biliary anastomoses (Hara et al. [17], Zhang et al. [18] and Ismail et al. [19]), which were outside the scope of the present study. The study by Sukho et al. [20] was excluded as it had duplicate data as a previous study by the same authors [21]. The study by Xue et al. [22] was excluded as the full-text was only available in Chinese. The studies by Reischl et al. [10] and Caziuc et al. [15] were review articles that were excluded from our qualitative synthesis but were individually screened for additional records. Finally, the studies by Karakaya et al. [23], Chen et al. [24], Wang et al. [25], and Karakaya et al. [26] were excluded as they were conference abstracts and full-text was not available.

Table 1 summarizes the characteristics of included studies. Fourteen studies were included, of which 1 was on esophageal anastomoses, 2 on gastric, 2 on small intestinal, and 9 on colorectal anastomoses. Ten studies (64.3%) used adipose-derived stem cells, 3 (21.4%) used bone marrow-derived stem cells, and 1 (7.1%) used myoblasts. In terms of administration, 10 studies instilled stem cells locally (64.3%), 1 injected intravenously (7.1%), and 2 studies used both approaches (14.3%). Eleven studies used rats (78.6%), 2 used swine (14.3%), and 1 used rabbits (7.1%). Eleven studies used high-risk experimental conditions to increase the likelihood of ALs, including radiation (2 studies), insufficiently sutured anastomosis (2 studies), ischemia (4 studies), colitis (1 study), and perforation (2 studies). The most common study design was adipose-derived stem cells for high-risk colorectal anastomoses in rats.

One study by Xue et al. [27] evaluated bone marrow-derived stem cells in fibrin scaffolds on esophageal anastomotic healing. They reported a higher closure rate (83.3% vs. 11.1%, p = 0.02) and lower infection rate (33.3% vs. 88.9%, p = 0.02) among treated animals.

No studies evaluated the effects of stem cells specifically on gastric anastomosis, but we included 2 studies evaluated stem cells in the healing of sutured gastric perforation. Liu et al. [28] investigated the effects of adipose-derived stem cells (sprayed and injected) on the healing of sutured gastric perforation. Tanaka et al. [29] investigated the effects of myoblast cell sheets on the healing of sutured gastric perforation. Since these studies did not directly evaluate a surgical anastomosis, they were not included in the meta-analysis.

Eleven studies reported the effects of mesenchymal stem cells on intestinal (colorectal and small intestinal) anastomoses. Seven studies (Pascual et al. [14], Van de Putte et al. [30], Yoo et al. [31], Sukho et al. [21], Morgan et al. [32], Alvarenga et al. [13], Moussa et al. [33]) evaluated the effects of adipose-derived stem cells on colorectal anastomotic healing in rats. Two studies by Adas et al. [34] and Adas et al. [35] evaluated local and systemic bone marrow stem cell administration on colorectal anastomotic healing in rats. Two studies by Maruya et al. [36] and Pan et al. [37] evaluated the effects of adipose-derived stem cell therapy on small intestinal anastomotic healing in swine. In terms of additional interventions, Yoo et al. [31], Morgan et al. [32], Adas et al. [34], and Adas et al. [35] caused ischemia by selectively devascularizing blood vessels supplying the anastomosis; Van de Putte et al. [30] and Moussa et al. [33] used radiation; Alvarenga et al. [13] used TNBS-induced colitis; and Sukho et al. [21] used a model of insufficiently sutured anastomosis.

In terms of administration techniques, 4 studies used local injection (Yoo et al. [31], Adas et al. [34], Moussa et al. [33], Pan et al. [37]). Four studies used scaffolds placed around the anastomosis (sheets: Sukho et al. [20, 21], Maruya et al. [36], gelatin sponge: Morgan et al. [32]). One study (Pascual et al. [14]) used biosutures. One study (Alvarenga et al. [13]) used local instillation. One study (Adas et al. [35]) evaluated intravenous injection. Finally, 2 studies (Van de Putte et al. [30], Moussa et al. [33]) evaluated a combination of intravenous and local injections.

All studies reported postoperative outcomes, macroscopic appearance of tissues after euthanasia, and histologic findings. The in vitro differentiation potential of stem cells was investigated in 7 studies (50%). Delivery of stem cells to perianastomotic tissues was examined in 7 studies (50%), with immunofluorescence being used in all studies. Specific types of inflammatory cells in perianastomotic tissues were examined in 3 studies. Finally, levels of inflammatory proteins or genes were determined in 7 studies (online suppl. Table 2).

A meta-analysis was performed among the 12 studies evaluating stem cells in gastrointestinal anastomoses (Table 2). Eight out of 12 studies reported AL rates on POD 7. Random-effects meta-analysis was performed due to experimental differences between these studies (e.g., different animal models, experimental conditions [ischemia, radiation, colitis]). No statistically significant difference in AL rates was identified (OR 0.34, 95% CI: 0.04–3.15, p = 0.248, I² = 34.1%, 95% CI: 0–75.2%, Q = 6.07, df = 4, p = 0.194, Fig. 1). Eleven out of 12 studies reported AL rates at any time point. Random-effects meta-analysis showed a trend for lower AL rates among treated animals at the latest time point reported (OR 0.28, 95% CI: 0.08–1.01, p = 0.052, I² = 34%, 95% CI: 0–70.8%, Q = 10.6, df = 7, p = 0.157, Fig. 2).

Due to heterogeneity between different studies, we decided to perform subanalyses among studies with similar experimental conditions. Among studies evaluating high-risk anastomoses (e.g., ischemia, radiation etc.), there was a trend for lower AL rates reported at the latest time point reported (OR 0.27, 95% CI: 0.06–1.23, p = 0.079, I² = 43.4%, 95% CI: 0–76.2%, Q = 10.6, df = 6, p = 0.102, Fig. 3). A random-effects meta-analysis was performed among studies that used local stem cell administration (i.e., local injection, local instillation, biosutures, local scaffold). A trend was found for lower AL rates among these studies at the latest time point reported (OR 0.20, 95% CI: 0.03–1.25, p = 0.074, I² = 47.1%, 95% CI: 0–79%, Q = 9.4, df = 5, p = 0.093, online suppl. Fig. 2). As stated above, the most common study design among included studies was to investigate adipose-derived stem cells for high-risk colorectal anastomoses in rats. A random-effects meta-analysis was performed among these studies. No statistically significant differences were identified with regards to AL rates reported at the time of latest follow-up (OR 0.38, 95% CI: 0.08–1.78, p = 0.166, I² = 25.1%, 95% CI: 0–68.5%, Q = 6.7, df = 5, p = 0.246).

A random-effects meta-analysis was performed among studies reporting anastomotic bursting pressures in the treatment and control groups. Only studies on rats were included as bursting pressures in swine are expected to differ considerably. Additionally, only studies where the mean and standard deviation could be extracted were included in the meta-analysis. A significantly higher anastomotic bursting pressure was identified among treated animals compared to controls (mean difference 35.1 mm Hg, standardized mean difference 1.29, 95% CI: 0.43–2.16, p = 0.003, I² = 76%, 95% CI: 33.8–91.3%, Q = 12.48, df = 3, p = 0.006, online suppl. Fig. 3 and Table 3).

A wide discrepancy of reported outcomes was observed among different studies. Our meta-analysis found no differences in ALs on POD 7, but a nonsignificant trend for decreased ALs at the latest time point reported. These findings are limited by heterogeneous study design among studies, partially offset by subanalyses. High-risk anastomotic models and local stem cell administration also showed nonsignificant trends for lower AL rates. Several, but not all, studies examined stem cell delivery in perianastomotic tissues and their in vitro differentiation potential.

A wide heterogeneity of outcomes was found among different studies. Some studies found lower AL rates among treated animals, but others did not. Interestingly, there were no differences in AL rates on POD 7, but a nonsignificant trend for lower ALs was found at the time of latest follow-up, suggesting the need for examining longer postoperative follow-up to better capture any therapeutic potential of stem cells. It is possible that short postoperative follow-up is not adequate to capture the beneficial healing effects of stem cells since stem cell differentiation into different cell types may take several days. High-risk anastomotic models may also be better at capturing differences in healing, though this was again not significant on meta-analysis. It is also possible that discrepancies in AL rates among different studies may be attributed to variations in diagnostic and reporting methods for ALs. Several studies did not report AL rates in the two groups. Although it is possible that no ALs were observed, it should nevertheless be explicitly stated as part of the study results. Most studies did not explicitly state in their methodology how ALs were assessed (e.g., macroscopical inspection, air leak test), and thereby, some studies may have missed small or asymptomatic ALs. Furthermore, some studies used proxy terms for ALs, such as sepsis, intra-abdominal abscesses, and mortality. While those outcomes are of interest, it can be argued that the overall goal of these research efforts is to prevent ALs and as such explicit reporting of ALs is of utmost importance.

Another important observation was that AL rates among controls were relatively low, with some studies reporting zero leaks among controls. In this way, even if stem cell therapy prevented ALs in these models, this could not be shown due to the low AL incidence among controls. High-risk anastomotic models may be more suitable to investigate anastomotic healing, and we found a nonsignificant trend for lower ALs with stem cell treatment in such models. Ischemia and radiation are commonly cited as factors predisposing to ALs, but clinical evidence is at best conflicting [4, 38‒42]. Another study used colitis, which could be contributing to ALs through changes in the intestinal microbiome [43], a recently recognized culprit in AL pathogenesis [7, 8]. Future studies should focus on high-risk experimental models, and it would also be interesting to investigate stem cells in high-risk models that target the intestinal microbiome.

We also observed a considerable diversity of experimental methodology among different studies. As stated above, variations were observed in postoperative follow-up, with several studies reporting outcomes up to 1 week and others extending follow-up to several weeks. Future studies should consider using additional experimental groups with extended postoperative follow-up. Furthermore, most studies administered stem cells locally, using needle injection, scaffolding, simple instillation, or embedding into sutures. Among studies using systemic administration, none reported a significant reduction of leak rates on POD 7. After meta-analysis, there was a trend for lower ALs with local stem cell administration, though this was not statistically significant. Among studies using systemic or a combination of local and systemic administration, none reported a drop in ALs. In terms of reported outcomes, several studies reported in vitro differentiation potential and stem cell delivery to perianastomotic tissues, but these should be further encouraged in future studies.

This study has several limitations. The study is limited by lack of reporting of ALs by some studies, as well as heterogeneity of different methods for diagnosing ALs. Study designs are greatly heterogeneous among included studies. Different studies use different time points for termination, stem cells, administration techniques, animal models, experimental conditions, and reported outcomes. In addition, although we did not find a statistically significant drop in ALs, this could be due to a lack of power and the presence of additional studies in the analysis could tip the scale in favor of stem cell therapies. The meta-analysis on anastomotic bursting pressures is also limited as there is great inconsistency among different operators, animal models, and measurement methods.

In conclusion, stem cell treatment was not associated with drops in ALs on POD 7, but a nonsignificant trend for lower ALs at the latest time point reported. Meta-analysis is severely limited by heterogeneity in design among different studies. Different studies employed a wide variety of methods for capturing and reporting ALs and a great heterogeneity of anastomotic models. More studies are needed to better establish the efficacy of stem cell therapy in preventing ALs. In terms of study design, longer postoperative follow-up, use of high-risk anastomotic models, and local stem cell administration may have promising potential in capturing any therapeutic potential of stem cell treatment. Explicit reporting of AL rates and examination of stem cell delivery in perianastomotic tissues are also advised to improve study quality.

All the authors that contributed in the preparation of this manuscript have been mentioned.

An ethics statement is not applicable because this study is based exclusively on published literature.

The authors have no conflicts of interest to declare.

This study was supported by the General Secretariat for Research and Technology (GSRT) and the Hellenic Foundation for Research and Innovation (HFRI).

Conception – Michail Pitiakoudis, George Kolios, and Apostolos Gaitanidis. Design – Apostolos Gaitanidis, Alexandra Tsaroucha, George Kolios, and Michail Pitiakoudis. Acquisition – Apostolos Gaitanidis, Leonidas Kandilogiannakis, and Eirini Filidou. Analysis: Apostolos Gaitanidis, Leonidas Kandilogiannakis, and Eirini Filidou. Interpretation: Apostolos Gaitanidis, Alexandra Tsaroucha, George Kolios, and Michail Pitiakoudis. Drafting: Apostolos Gaitanidis, Eirini Filidou, and Alexandra Tsaroucha. Revisions: Leonidas Kandilogiannakis, Michail Pitiakoudis, and George Kolios. Final approval: Apostolos Gaitanidis, Leonidas Kandilogiannakis, Eirini Filidou, Alexandra Tsaroucha, George Kolios, and Michail Pitiakoudis.

All data generated or analyzed during this study are included in this article and its online supplementary material. Further inquiries can be directed to the corresponding author.