Article Text
Abstract
Objective To determine the incidence of refractory anastomotic strictures after oesophageal atresia (OA) repair and to identify risk factors associated with refractory strictures.
Methods Retrospective national multicentre study in patients with OA born between 1999 and 2013. Exclusion criteria were isolated fistula, inability to obtain oesophageal continuity, death prior to discharge and follow-up <6 months. A refractory oesophageal stricture was defined as an anastomotic stricture requiring ≥5 dilations at maximally 4-week intervals. Risk factors for development of refractory anastomotic strictures after OA repair were identified with multivariable logistic regression analysis.
Results We included 454 children (61% male, 7% isolated OA (Gross type A)). End-to-end anastomosis was performed in 436 (96%) children. Anastomotic leakage occurred in 13%. Fifty-eight per cent of children with an end-to-end anastomosis developed an anastomotic stricture, requiring a median of 3 (range 1–34) dilations. Refractory strictures were found in 32/436 (7%) children and required a median of 10 (range 5–34) dilations. Isolated OA (OR 5.7; p=0.012), anastomotic leakage (OR 5.0; p=0.001) and the need for oesophageal dilation ≤28 days after anastomosis (OR 15.9; p<0.001) were risk factors for development of a refractory stricture.
Conclusions The incidence of refractory strictures of the end-to-end anastomosis in children treated for OA was 7%. Risk factors were isolated OA, anastomotic leakage and the need for oesophageal dilation less than 1 month after OA repair.
- oesophageal atresia
- outcome
- oesophageal strictures
- oesophageal dilation procedures
- risk factors
Statistics from Altmetric.com
What is already known on this topic?
Anastomotic stricture formation is a frequent postoperative complication after oesophageal atresia (OA) repair (up to 59%), occurring mostly in the first year of life.
A uniform definition of an oesophageal stricture after OA repair is lacking, which makes comparing different studies difficult.
What this study adds?
This is the first multicentre study in patients with OA focusing on refractory strictures with a high dilation burden only.
Refractory anastomotic strictures requiring at least five dilations occurred in 7.3% of children.
Risk factors for refractory anastomotic strictures were: isolated OA, anastomotic leakage and the need for oesophageal dilation within 28 days after OA repair.
Introduction
Oesophageal atresia (OA) with or without a tracheo-oesophageal fistula (TOF) is a rare anatomical anomaly (2.43:10 000 births).1 Dedicated centres have reported survival rates up to 95%.2 3 However, anastomotic stricture formation is still the most frequent postoperative complication (17%–59%), occurring mostly in the first year of life.4 5
Data on oesophageal stricture formation after OA repair and potential risk factors are scarce.6–12 A recent single-centre retrospective study found postoperative oesophageal strictures (≥4 dilations, no interval specified) in 21.5% of patients with OA (26/121), thoracoscopic and staged OA repairs were both associated with stricture formation.6
A uniform definition of an oesophageal stricture after OA repair is lacking, which makes comparing different studies difficult. Most studies define an oesophageal stricture as any narrowing of oesophageal lumen requiring at least one dilation.5 7 9 However, in current literature definitions vary based on frequency of dilations,6 13–17 luminal diameter12 or symptomatology.8 10 In some centres, patients with OA are routinely subjected to a series of three dilations, even when symptoms and luminal narrowing have disappeared.
Refractory strictures require frequent dilations and therefore result in a high burden for both child and parents, including frequent anaesthesia, hospital stay and risk of perforation. The newest European Society of Gastrointestinal Endoscopy-European Society for Paediatric Gastroenterology Hepatology and Nutrition Guideline on Pediatric Endoscopy suggests the following definition of a benign refractory oesophageal stricture in children: an anatomic restriction without endoscopic inflammation that results in dysphagia after a minimum of five dilations at maximally 4-week intervals.18
We hypothesised that anastomotic leakage and thoracoscopic OA repair will increase refractory oesophageal stricture formation. In a large multicentre cohort of children born with OA in the Netherlands, we retrospectively determined the incidence of postoperative dilations and refractory anastomotic strictures. We studied possible determinants of refractory stricture formation with focus on strictures of end-to-end anastomoses.
Methods
Patients
We included all patients with OA born between 1999 and 2013 who were treated for OA in one of the participating centres. Data until June 2016 were included, ensuring minimum follow-up of 2.5 years. Exclusion criteria were: isolated TOF (Gross type E19), inability to obtain oesophageal continuity, death prior to discharge and follow-up less than 6 months.
Five of the six university hospitals involved in neonatal surgery—all members of the Dutch Consortium of Esophageal Atresia Study Group—participated (see online supplementary appendix). The Medical Ethics Committee of Erasmus MC concluded that the Dutch Medical Research Involving Human Subjects Act did not apply to the study protocol (protocol ID MEC-2016-570). Ethics approval from the local committees in the other participating centres was obtained.
Supplementary file 1
Data collection
Baseline demographics and outcome data at follow-up were retrieved from patient records. Major associated anomalies included Ravitch’ paediatric surgical index diagnoses,20 major cardiac anomalies (cardiac malformations requiring surgical correction or cardiological follow-up), other congenital malformations requiring major surgical interventions and malformations seriously affecting normal function (eg, tethered cord with neurogenic bladder function). All other anomalies were considered minor (eg, small atrial septal defect closing spontaneously). Prematurity was defined as gestational age <37 weeks. VACTERL (vertebral defects, anal atresia, tracheo-oesophageal fistula with oesophageal atresia, cardiac anomalies, renal anomalies and limb anomalies) association was defined according to Solomon et al.21 Type of OA was classified according to Gross.19 A child was considered to have gastro-oesophageal reflux disease (GORD) if pH monitoring showed pathological reflux, if upper endoscopy showed oesophagitis or if antireflux surgery was performed. Furthermore, frequent aspiration, typical symptoms with spontaneous reflux present at contrast oesophagography, or symptom relief using antireflux drugs were considered to indicate GORD. Anastomotic leakage was defined as leakage visible at contrast oesophagography or necessitating placement of a chest tube postoperatively.
Definitions of strictures
A stricture was defined as a symptomatic stenosis of an anastomosis for which dilation was indicated. A stricture requiring dilation ≤28 days after OA repair was considered ‘early’. A refractory stricture was defined as an anastomotic (end-to-end, oesophagojejunal or oesophagogastric) stricture requiring ≥5 dilations at maximally 4-week intervals.18
Data analysis
Data are presented as frequencies or as medians (minimum; maximum; IQR). Pearson’s Χ2 test, Fisher’s exact test and the non-parametric Mann-Whitney U test were used for statistical comparison between children with and without a refractory stricture of the end-to-end anastomosis. Multivariable logistic regression analysis was performed to identify risk factors for refractory strictures of the end-to-end anastomosis. Predefined risk factors for refractory strictures were (based on previous literature): gestational age, isolated OA (Gross type A), thoracoscopic correction, anastomotic leakage and early stricture (≤28 days after anastomosis). Centre was also included as an independent variable in the logistic regression model. The Hosmer-Lemeshow goodness-of-fit test was used to assess whether the model adequately describes the data. Because in several studies a clinically significant stricture—to be distinguished from a refractory stricture—is defined as a stricture requiring ≥3 dilations, findings from children with and without a stricture requiring ≥3 dilations were also compared including identification of risk factors. Data were analysed with SPSS V.21.0 (SPSS).
Results
We included 454/563 (80.6%) children treated in one of the participating centres. Reasons for exclusion were: isolated TOF (n=31), no oesophageal continuity obtained (n=2), early death (n=59), follow-up <6 months (n=15) and missing data (n=2). Six (1.3%) children who died after discharge and after the age of 6 months (median of 576 (range 207–1729) days) were included.
Demographics
Thoracotomy with primary anastomosis was performed most frequently (357/454; 78.6%) (table 1). Oesophageal continuity was obtained with a jejunal interposition or gastric pull-up in 18/454 (4.0%) children. Stricture formation was the reason for oesophageal replacement at an older age in three children with isolated OA.
Anastomotic strictures
An anastomotic stricture was documented for 262/454 (57.7%) children: 251/436 (57.6%) children with an end-to-end anastomosis (median (minimum; maximum; IQR) of 3 (1; 34; 1–5) dilations); 9/13 (69.2%) children with a jejunal interposition (median of 5 (3; 32; 4–16) dilations); and 2/5 (40.0%) children with a gastric pull-up (two and five dilations) (table 2). Figure 1 illustrates the number of dilations performed in children with an end-to-end anastomosis.
Refractory strictures developed in 32/436 (7.3%) children with an end-to-end anastomosis (varying between centres: 1.6%–13.3%, see online supplementary figure 1). A median of 10 (5; 34; 8–15) dilations were performed; the fifth dilation at a median of 110 (49; 158; 81–124) days after anastomosis. Refractory strictures of the oesophagojejunal anastomosis developed in 2/13 (15.4%) jejunal interpositions: one stricture of a proximal anastomosis requiring 32 dilations (fifth dilation at day 86 after anastomosis) and one stricture of a distal anastomosis requiring 15 dilations (fifth dilation at day 72 after anastomosis).
Eighteen of 81 (22.2%) children who underwent fundoplication surgery had developed a refractory stricture prior to antireflux surgery. After antireflux surgery 31 (39.3%) children still needed oesophageal dilation, with a median of 3 (1; 29; 1–7) dilations per child. Two (2.5%) children developed a refractory stricture after antireflux surgery (Nissen fundoplication and Boerema anterior gastropexy with hiatoplasty, respectively).
Online supplementary table 1 summarises details on other treatments of strictures (ie, stent placement, mitomycin application and stricture resection).
Determinants of refractory anastomotic strictures
Table 3 summarises characteristics of children with and without a refractory stricture of the end-to-end anastomosis. Results from multivariable logistic regression analysis (adjusted for centre) in children with an end-to-end anastomosis demonstrated that isolated OA (OR 5.7; p=0.012), anastomotic leakage (OR 5.0; p=0.001) and an early stricture (OR 15.9; p<0.001) were associated with refractory stricture formation (table 4). Thoracoscopic OA repair was not significantly associated with refractory strictures. In multivariable logistic regression analysis 6/436 children were excluded (missing values for gestational age).
Online supplementary tables 2–4 summarise results from multivariable logistic regression analysis for strictures requiring ≥3 dilations (definition of a clinically significant stricture in several studies).
Discussion
In this multicentre national cohort of 454 children born with OA the incidence of anastomotic strictures after end-to-end anastomosis was 57.6%. Refractory stricture of an end-to-end anastomosis requiring ≥5 dilations developed in 7.3% of cases. Isolated OA, anastomotic leakage and early stricture (≤28 days after anastomosis) were associated with refractory stricture development.
The incidence of anastomotic strictures after OA repair in our study (57.7%) is concordant with previous literature (9%–79%).22 This wide range reflects the lack of a uniform definition of refractory strictures. Many clinicians decide to surgically intervene after three consecutive dilations. Thus, in several studies a clinically significant stricture is defined as a stricture requiring ≥3 or ≥4 dilations.6 16 17 We recommend to use a uniform definition for refractory anastomotic strictures in future studies: an anastomotic stricture requiring ≥5 dilations at maximally 4-week intervals. This definition distinguishes refractory strictures from so called recurrent strictures.
Nice et al found that both thoracoscopic and staged OA repairs were associated with stricture formation.6 These authors considered strictures refractory and clinically significant after ≥4 dilations. Unfortunately, information on the interval between dilations and the duration of follow-up was not provided, which impedes comparison to our study.
Despite better visualisation and usually a more limited dissection during thoracoscopic OA repair, opening of the upper pouch can be less than in open surgery which might lead to increased stricture formation. We could however not confirm our hypothesis that thoracoscopic OA repair is associated with refractory stricture formation. The relatively low number of thoracoscopic corrections (n=54 performed in two centres only) may explain this. Two recent literature reviews concluded that the incidence of strictures after both thoracoscopic and open OA repairs is comparable.23–25
Both in the present and in earlier studies leakage was predictive of anastomotic stricture formation.26 27 We assume that leakage enhances inflammation and scarring of the anastomotic area.
Our finding that isolated OA was a risk factor is supported by others.8 11 12 The long gap in isolated OA often requires staged anastomosis or oesophageal replacement. Although correction of a large gap is thought to result in anastomotic tension with subsequent stricture formation, contradictory results have been reported.15 28 We included only the uniformly objective variable isolated OA as a potential explanatory variable, since staged repair, long gap OA and oesophageal replacement are correlated to each other.
In children with an isolated OA, the current practice to restore oesophageal continuity is performing a primary anastomosis, either immediately or delayed. In our study, only 15/30 children with an isolated OA underwent an oesophageal anastomosis. Children with an isolated OA in whom a primary anastomosis is performed may be at higher risk of developing a refractory stricture than children with other types of OA. Indeed we found that a refractory stricture had developed in 26.7% (4/15). This information can be shared in preoperative parental counselling.
We also found that an early anastomotic stricture predicted the development of a refractory stricture. More severe strictures may occur in the first weeks postoperatively, but early dilation in a still vulnerable anastomosis might be an independent risk factor for refractory strictures. Our data do not allow to draw any conclusions on this subject. Most of the refractory strictures developed within 4 months postoperatively. We assume that the ‘late onset’ oesophageal strictures are related to altered food consistency as more solid formulas can cause dysphagia, food impaction, stasis, aspiration or vomiting.
Although others have identified anastomotic tension as a risk factor for strictures, we chose to not include tension as a potential determinant.7–11 26 Anastomotic tension is a subjective observation which is usually poorly recorded. Anastomotic leakage is a more objective finding and was found to be a predictor for stricture formation in our study.
Another factor thought to increase stricture development is GORD.5 11 26 27 Studies are hard to compare as different definitions of GORD are used. Besides, most studies are retrospective without use of standardised protocols to diagnose GORD. Interestingly, prophylactic antireflux drugs did not always reduce stricture formation in patients with OA.29–31 We did not include GORD as a possible determinant in our study as participating centres used different protocols for diagnosis. In our study, antireflux surgery was more frequent in children with a refractory stricture than in those without. It was typically performed after refractory stricture development and should therefore be considered as therapeutic management rather than being a risk factor for stricture formation.
Prematurity, birth weight and cardiac anomalies have been associated with stricture formation, but this was not the case in our study.6 8 30
Several adjuvant treatments are currently available for the treatment of refractory strictures, such as stent placement, intra-lesional steroid injection, mitomycin C application, endoscopic needle knife incision and resection surgery. Studying the effectiveness of these treatments was outside the scope of our study.
The strengths of our study are the large cohort (covering ~80% of all newborns born with OA in the Netherlands), the long follow-up period and the small number of missing data. Still, some limitations need to be addressed. First, risk factors for refractory strictures after jejunal interposition (2/13) or gastric pull-up (0/5) could not be identified due to the limited sample size. Second, indications for dilations were not recorded reliably in all cases. One of the problems with OA is the associated dysmotility, so some level of stasis is always present. Whether feeding can be ameliorated with dilations is usually decided based on the combination of clinical symptoms and by radiographic findings in selected cases. Third, the method used to open the proximal pouch (monopolar/bipolar electrocautery, knife, scissors) and suture techniques for anastomosis were not recorded. Since anastomotic tension and ischaemia are thought to play a role in anastomotic stricture formation, it would be interesting to use innovative new optical techniques to address these aspects in future studies (eg, deformation sensors in (endoscopic) instruments to quantify tension and spectroscopic assessment of tissue oxygenation and perfusion). Fourth, due to the retrospective design and the absence of uniform protocols for dilations and GORD diagnosis, we were unable to reliably study the relation between these factors and stricture development. Besides, eosinophilic oesophagitis—known to be associated with oesophageal stricture development—could not be included as a risk factor as no standardised upper endoscopies with biopsies were performed in our study cohort.32 Last, the participating centres differed with regard to the frequencies of thoracoscopic corrections, chest tube placement postoperatively, anastomotic leakage, recurrent TOFs and antireflux procedures. Studying a large cohort with a higher number of thoracoscopic surgeries is needed to examine whether a learning curve is present.33 34 Numbers of antireflux procedures performed in different centres might reflect differences in dilation management and GORD protocols between centres.
We observed an alarmingly high number of refractory strictures after end-to-end anastomosis in children with isolated OA (26.7%). Whether these children, those with anastomotic leakage and those with an early stricture may benefit from supportive care (eg, adequate acid suppression) aiming to protect the anastomotic area is still unknown.
In conclusion, refractory anastomotic strictures requiring ≥5 dilations had developed in 7.3% of 436 patients with OA with an end-to-end anastomosis. A high number (26.7%) was observed after end-to-end anastomosis in children with isolated OA. We observed that isolated OA, anastomotic leakage and early anastomotic stricture (≤28 days after anastomosis) are associated with refractory anastomotic stricture formation. Future prospective multicentre studies should focus on uniform recording of baseline data—including surgical techniques used—standardised protocols for diagnosis of GORD, eosinophilic oesophagitis32 and indications and techniques for oesophageal dilations.
Acknowledgments
The authors thank Catelijne Coppens, MD and Jos Draaisma, MD, PhD (Department of Paediatrics, Radboud University Medical Center-Amalia Children’s Hospital, Nijmegen) for the use of a research database, and Ajay Bakas (medical student) for retrieval of clinical data. Ko Hagoort provided editorial advice.
References
Footnotes
Contributors Study conception and design: FWTV, JV, HI, MCWS, JvR and RMHW. Acquisition of data: FWTV and MS. Analysis and interpretation of data: FWTV, JV, HI, MCWS, JvR, MWNO, JBFH, MD, MJB, MS and RMHW. Drafting of manuscript: FWTV, JV, HI, MCWS, MJB and RMHW. Critical revision: FWTV, JV, HI, MCWS, JvR, MWNO, JBFH, MD, MJB, MS and RMHW. All authors (FWTV, JV, HI, MCWS, JvR, MWNO, JBFH, MD, MJB, MS and RMHW) gave their final approval of the version to be published.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Ethics approval The Medical Ethics Committee of Erasmus MC Rotterdam
Provenance and peer review Not commissioned; externally peer reviewed.
Collaborators See the online supplementary appendix for a list of the DCEA Study Group members of the participating centres.