This is the first prospective RCT examining the safety and efficacy of administering MFR to preterm neonates. Our study showed that MFR is beneficial for increasing the bowel diameter and growth rate. Moreover, citrulline levels tended to be higher in the MFR group than for infants in the control group, which was not statistically significant.
Premature infants are capable of intestinal growth and adaptation after bowel resection . Higher volumes of ostomy output adversely affect the growth and the body fluid and electrolyte status. Previous studies show that a stoma discharge < 40 mL/kg/day is considered ideal [12, 13]. To manage this situation and promote intestinal adaptation, a previous study showed the usefulness of MFR for growth and PN discontinuation because it artificially maintains the bowel flow to help absorption [10, 14, 15]. Furthermore, MFR improved the tolerance for EN after reanastomosis . Therefore, we classified the infants with an ostomy output > 40 mL/kg/day into the high-output group and performed MFR to them. In the comparison of the outcomes of the normal-output MFR group and control group with an ostomy output < 40 mL/kg/day, no significant differences were observed in the PN duration, number of days required to reach full-feeding after reanastomosis, and the timing of reanastomosis.
The z-score of bodyweight at the stoma formation in the normal-output MFR group was significantly lower than that of the control group; however, at the times of reanastomosis operation and three months after the reanastomosis, the z-scores of bodyweight were not significantly different, thus meaning that MFR facilitated weight gain.
In the loopograms taken just before reanastomosis, colon diameters were significantly larger in the transverse, descending, and sigmoid colon of infants in the normal-output MFR group than those in the control group. Lau et al.  reported that MFR could decrease the risk of anastomotic complication. In our study, the anastomosis procedure had to be rescheduled for one infant in the control group due to a bowel-end size discrepancy. The infant subsequently underwent MFR and reanastomosis 75 days later.
Histopathological findings of the distal ileum at the stoma closure showed that chronic inflammation and destruction of villous structures were more frequent in the control group than in the normal-output MFR group. Similar to our findings, Yabe et al.  reported that MFR helped increase intestinal mucosal thickness and maintain the villous structure of the distal ileum. Intestinal maturation and rehabilitation are reinforced by exposure to enteral nutrients and enterotrophic factors [17, 18], which can be simulated by MFR.
Serum citrulline is a non-protein amino acid that is synthesized from glutamine and glutamine-related components within enterocytes [19, 20]. Serum citrulline is a widely used marker in infants to evaluate length and absorption capabilities of the small bowel and prognosis for PN weaning [21, 22]. We compared serum citrulline levels between groups and found no significant difference between the normal-output MFR and control groups. The median citrulline level at 4 weeks, just before reanastomosis, and 12 weeks after reanastomosis tended to be higher in the normal-output MFR group. This may indicate that MFR has a persistent positive effect on bowel rehabilitation after reanastomosis.
The amount of weight gain in premature infants naturally increases as PMA advances. Therefore, we compared the growth velocity before and after MFR after adjusting for PMA, by which the positive impact of MFR on weight gain would be verified more appropriately. Both of the daily weight and length gain were significantly larger after MFR in high-output + normal-output MFR group after adjusting for PMA.
In our study, several complications had occurred. MFR was terminated in one case due to perforation during a manual stoma reduction and minor complications of stoma prolapse, bowel distension and skin erosion were reported. Haddock et al.  conducted a study on MFR using a 12-Fr catheter that constantly implanted and reported that 17% of patients who underwent MFR experenced serious complications including bowel perforation or bleeding. In recent studies evaluating the safety of MFR, several factors including catheter size, personnel responsible for insertion, and insertion method were standardized, and no major MFR-related complications (perforation, stricture, or death) were reported [7, 9]. By standardizing the protocol, a more refined process could be performed using a weight-appropriate catheter. During the MFR procedure, there were two cases of culture-proven sepsis, each from the normal-output and high-output group. However, the intervention was not discontinued because the direct relationship between MFR and sepsis could not be determined. Even without any intervention, four additional cases of culture-proven sepsis had occurred. The conditions of extremely low gestational age and the presence of a central venous catheter and stoma alone confer a very high risk for sepsis to this study population. In infants with enterostomies, the intestinal epithelial barrier function is decreased due to the mucosal inflammatory response and villous atrophy, which facilitate the movement of luminal bacteria and its constituents into the underlying tissue and blood, which increases the susceptibility of bloodstream infection . Approximately 13% of recurrent sepsis cases occur as a complication in newborns with enterostomies . Pataki et al studied the microbiological safety of recycling bowel contents and reported that the stoma effluent was colonized by commensal facultative pathogenic enteral and skin flora including coagulase-negative Staphylococcus after 120 minutes . The microorganisms cultured during MFR in this study were Staphylococcus epidermidis and Klebsiella aerogenes, which are stomal pathogens too. On the contrary, Yabe et al cultured the stoma output 3 hours after its ejection and detected no pathogenic bacteria . In a recent review, infection was not described as a MFR-related complication . In our study, stoma contents were collected and recycled every 4 hours. Two cases of sepsis occurred during MFR, but the relationship between MFR and sepsis could not be determined in this study. Therefore, we ascertain the necessity to probe whether this recycling interval is associated with a risk of infection. Further research is necessary on the infectious complications in the future studies and careful clinical monitoring for infection is necessary during the refeeding period.
Several limitations of this study need to be acknowledged. Due to the vulnerability of the patient group comprising premature infants, achieving a sufficient sample size through power calculation was difficult. Nevertheless, this was the first RCT for MFR that explored the benefits and safety of MFR with a controlled protocol. Hence, it can provide the basis for further RCTs for MFR.