Utility of repeated therapeutic endoscopies for pediatric esophageal anastomotic strictures

Anastomotic stricture is a common complication of esophageal atresia (EA) repair. Such strictures are managed with dilation or other therapeutic endoscopic techniques such as steroid injections, stenting, or endoscopic incisional therapy (EIT). In situations where endoscopic therapy is unsuccessful, patients with refractory strictures may require surgical stricture resection; however, the point at which endoscopic therapy should be abandoned in favor of repeat thoracotomy is unclear. We hypothesized that increasing numbers of therapeutic endoscopies are associated with increased likelihood of stricture resection. We retrospectively reviewed the records of patients with EA who had an initial surgery at our institution resulting in an esophago-esophageal anastomosis between August 2005 and May 2019. Up to 2 years of post-surgery endoscopy data were collected, including exposure to balloon dilation, intralesional steroid injection, stenting, and EIT. Primary outcome was need for stricture resection. Receiver operating characteristic (ROC) curve analysis and univariate and multivariable Cox proportional hazards regression analyses were performed. There were 171 patients who met inclusion criteria. The number of therapeutic endoscopies was a moderate predictor of stricture resection by ROC curve analysis (AUC = 0.720, 95% CI 0.617-0.823). With increasing number of therapeutic endoscopies, the probability of remaining free from stricture resection decreased. By Youden's J index, a cutoff of ≥7 therapeutic endoscopies was optimal for discriminating between patients who had versus did not have stricture resection, though an absolute majority of patients (≥50%) remained free of stricture resection at each number of therapeutic endoscopies through 12 endoscopies. Significant predictors of needing stricture resection by univariate regression included ≥7 therapeutic endoscopies, Foker surgery for long-gap EA, fundoplication, history of esophageal leak, and length of stricture ≥10 mm. Multivariate analysis identified only history of leak as statistically significant, though this regression was underpowered. The utility of repeated therapeutic endoscopies may diminish with increasing numbers of endoscopic therapeutic attempts, with a cutoff of ≥7 endoscopies identified by our single-center experience as our statistically optimal discriminator between having stricture resection versus not; however, a majority of patients remained free of stricture resection well beyond 7 therapeutic endoscopies. Though retrospective, this study supports that repeated therapeutic endoscopies may have clinical utility in sparing surgical stricture resection. Esophageal leak is identified as a significant predictor of needing subsequent stricture resection. Prospective study is needed.