Risk Factors for Stricture Formation After Esophageal Atresia Repair

Abstract

Purpose:

Esophageal stricture remains a common morbidity of esophageal atresia (EA) repair. The purpose of this study was to examine the association of multiple pre- and postoperative variables with stricture formation after EA repair.

Methods:

A retrospective review of all patients who underwent EA repair was performed from June 1999 to January 2014, excluding patients who died prior to discharge. Data were collected on patient demographics, disease specifics, treatment, and outcomes. A clinically significant esophageal stricture was defined as those requiring more than three esophageal dilations. Univariate analysis and multivariate analysis was performed to determine associations with stricture formation.

Results:

The study included 121 infants. On univariate analysis, tracheoesophageal fistula (TEF) Gross classification (P = .046), method of repair (P = .0099), surgery staging (P = .0211), and development of leak (P = .0479) had a statistically significant association with stricture formation. Most (81%) underwent open repair with a stricture rate of 16.3%, while 20 patients underwent thoracoscopic repair with a 40% stricture rate. Multivariate analysis showed that patients undergoing a staged repair had increased risk of stricture formation over primary repair (odd ratio [OR] 6.360; P = .0008). Thoracoscopic surgery also increased the risk of stricture (OR 7.409; P = .0014). Cardiovascular anomalies were found to be associated with decreased stricture formation (OR 0.251; P = .0083).

Conclusion:

Thoracoscopic repair and staged repair were both associated with increased risk of clinically significant stricture formation after TEF repair. However, the presence of cardiovascular anomalies was associated with decreased stricture formation. TEF Gross classification also affects stricture risk.

Introduction

Esophageal atresia (EA) has been a challenging problem for pediatric surgeons since the early 1900s. It occurs with an estimated incidence of 1/3500 live births, with some estimates as high as 1/4500 for births in the United States.¹ Most commonly, EA occurs in association with tracheoesophageal fistula (TEF), but may occur separately in 7% of cases.² Even with advances in surgical materials and techniques, esophageal strictures remain a common morbidity after EA repair. These strictures are significant because they will require dilation and sometimes require additional operative interventions, thereby increasing patient risks and care costs. The purpose of this study was to examine the association of multiple pre- and postoperative variables with stricture formation after EA repair.

Methods

Study cohort

A single institution retrospective review of patients who underwent EA repair was performed from June 1999 to January 2014. Institutional Review Board approval was obtained (X120314018). All classes of TEF were included. Patients who died prior to discharge were excluded. Data were collected on patient demographics (sex, race, gestational age, birth weight, mother's age), disease specifics (Gross type, Waterson classification, associated anomalies), treatment (operation type, operation staging, blood loss, and operative time), and outcome (length of stay, time to extubation, time until oral intake, esophageal leak, need for gastrostomy tube, need for fundoplication, and number of esophageal dilations needed). All treatments were based on surgeon preference at the time of patient care.

Definitions

Operation staging was defined as primary or staged, depending on whether the TEF ligation and EA repair were both performed at the initial operation (primary) or in two separate procedures (staged). Method of repair included thoracotomy, thoracoscopy, or conversions from thoracoscopy to thoracotomy. Esophageal leak was determined based on postoperative esophagram, usually performed on postoperative day 7. A clinically significant esophageal stricture was defined as those requiring more than three esophageal dilations. Need for dilation was determined by clinical symptoms with stricture evidence on esophagram either preoperatively or at the time of scheduled dilation.

Statistical analysis

Univariate analysis for stricture formation used Pearson's chi-square test for categorical variables and binomial logistic regression for continuous variables. Due to sample size, seven variables were selected for multivariate analysis, and a multivariate logistic regression model was created. Variables were sequentially removed until all remaining variables held statistical significance. To estimate the relative risk better, a Poisson regression model was created, and incidence rate ratios were calculated.³ Statistical significance was determined at an alpha level of 0.05. JMP Pro software v10 (SAS Institute, Cary, NC) was utilized for descriptive statistics, univariate, and logistic multivariate analysis. Stata v13.1 special edition (StataCorp; College Station, TX) was used for Poisson regression analysis.

Results

The study included 121 infants. Table 1 shows the study population demographics. Table 2 summarizes treatments for EA at the authors' institution, while Table 3 presents outcomes after EA repair. Forty-three percent of patients developed a stricture requiring at least one dilation: 36.7% of open repairs, 65% of thoracoscopic repairs, and 100% of conversions. Furthermore, 21.5% of patients developed a clinically significant stricture requiring more than three dilations: 16.3% of open cases, 40% of thoracoscopic cases, and 66.7% of conversions (see Table 4). Univariate analysis findings are summarized in Table 4. Demographic variables such as sex, race, birth weight, gestational age, mother's age, or Waterson class were not found to have a significant association with a clinically significant stricture requiring more than three dilations. However, TEF Gross classification (P = .046), operation staging (P = .0211), method of repair (P = .0099), time to oral feeds (P = .0206), and development of a leak (P = .0479) all had a statistically significant association with stricture formation. Eighty-four percent of the repairs were for TEF type C, of which 21.6% developed a stricture. Both patients with type B TEF developed a stricture, whereas none of the patients with type E TEF had this outcome. Patients who could be repaired primarily had a lower stricture rate (15.7%) compared with those who required staged repair (34.2%). Ten percent of patients developed a leak, and these patients had an increased stricture rate (45.5%) compared with those without a leak (19.4%). Associated anomalies and other outcome variables such as blood loss, operation time, length of stay, time to extubation, or future gastrostomy or fundoplication requirements were not found to be associated with stricture formation on univariate analysis.

Table 1.

Study Population Demographics

Variable	Population
n	121
Sex
Male	71 (58.7%)
Female	50 (41.3%)
Race/ethnicity
White	84 (70%)
Black	32 (26.7%)
Hispanic	4 (3.3%)
Mother's age (years)	Mean 27.1	SD 6.85
Birth weight (g)	Mean 2407.2	SD 762.5
Gestational age (weeks)	Mean 35.9	SD 3.4
Waterson class
A	20 (19.6%)
B	45 (44.1%)
C	37 (36.3%)
TEF Gross type
A	7 (5.8%)
B	2 (1.6%)
C	102 (84.3%)
D	3 (2.5%)
E	7 (5.8%)
Associated anomalies
Cardiovascular	55 (45.5%)
Genitourinary	26 (21.5%)
Central nervous system	17 (14.1%)
Vertebral	17 (14.1%)
Skeletal	15 (12.4%)
Gastrointestinal	10 (8.8%)

TEF, tracheoesophageal fistula; SD, standard deviation.

Table 2.

Treatment Factors for Study Population

Variable	Population
Operation type
Open thoracotomy	98 (81%)
Thoracoscopic repair	20 (16.5%)
Conversion	3 (2.5%)
Operation staging
Primary repair	83 (68.6%)
Staged repair	38 (31.4%)
Blood loss (mL)	Median 5	IQR 5–10
Operation time (minutes)	Median 150	IQR 115–200

IQR, interquartile range.

Table 3.

Outcomes After Repair of EA

Variable	Population
Stricture	26 (21.5%)
Length of stay (days)	Median: 23	IQR: 14–57
Time to extubate (days)	Median: 3	IQR: 2–6
Time to oral intake (days)	Median: 9	IQR: 7–25
Leak on esophagram	11 (10.1%)
Need for future fundoplication	21 (18.6%)
Need for gastrostomy tube	54 (47%)
Tracheomalacia	22 (18.2%)

EA, esophageal atresia.

Table 4.

Results of Univariate Analysis for Stricture Formation

Variable	Stricture	P value
n	26 (21.5%)
Sex
Male	14 (19.7%)	.5737
Female	12 (24%)
Race/ethnicity
White	18 (21.4%)	.9304
Black	6 (18.8%)
Hispanic	1 (25%)
Mother's age (years)		.3499
Birth weight (grams)		.4513
Gestational age (weeks)		.8897
Waterson class
A	6 (30%)	.6302
B	11 (24.4%)
C	7 (18.9%)
TEF Gross type
A	1 (14.3%)	.0460
B	2 (100%)
C	22 (21.6%)
D	1 (33.3%)
E	0 (0%)
Operation type
Open rhoracotomy	16 (16.3%)	.0099
Thoracoscopic repair	8 (40%)
Conversion	2 (66.7%)
Operation staging
Primary repair	13 (15.7%)	.0211
Staged repair	13 (34.2%)
Blood loss (mL)		.9675
Operation time (minutes)		.0869
Length of stay (days)		.0731
Time to extubate (days)		.8079
Time to oral intake (days)		.0206
Leak on esophagram
No	19 (19.4%)	.0479
Yes	5 (45.5%)
Need for future fundoplication
No	20 (21.7%)	.5021
Yes	6 (28.6%)
Need for gastrostomy tube
No	10 (16.4%)	.0904
Yes	16 (39.6%)
Tracheomalacia	5 (22.7%)	.8756
No tracheomalacia	21 (21.2%)
Cardiovascular anomaly	8 (14.6%)	.0897
No cardiovascular anomaly	18 (27.3%)
Genitourinary anomaly	6 (31.1%)	.8238
No genitourinary anomaly	20 (21.1%)
Central nervous system anomalies	5 (29.4%)	.3909
No central nervous system anomalies	21 (20.2%)
Vertebral anomalies	2 (11.8%)	.2924
No vertebral anomalies	24 (23.1%)
Skeletal anomalies	1 (6.7%)	.1354
No skeletal anomalies	25 (23.6%)
Gastrointestinal anomalies	1 (10%)	.3398
No gastrointestinal anomalies	24 (23.1%)

Statistically significant values shown in bold italic.

Waterson classification, operation type, operation staging, time to oral feeds, esophageal leak, gastrostomy tube requirement, and cardiovascular anomalies were then selected for multivariate analysis using logistic regression. TEF type was not included due to small numbers of less common types. Variables were then sequentially removed until all remaining variables were significant. This left a final model containing operation type, operation staging, and presence of cardiovascular anomalies. Infants undergoing a staged repair had an increased risk of stricture formation compared with those who could be repaired primarily (odds ratio [OR] 6.360; P = .0008). Thoracoscopic surgery also increased the risk of stricture (OR 7.409; P = .0014). Cardiovascular anomalies were found to be protective from stricture formation (OR 0.251; P = .0083).

Due to the high incidence of stricture in the population, the odds ratio becomes a poor approximation for the relative risk. As such, the sign and significance of the odds ratio remain valuable, but the magnitude is less reliable. A Poisson regression model was created and incidence rate ratios (IRR) calculated in order to approximate the magnitude of the true relative risk better. Staged operations (IRR 3.701; P = .004) and thoracoscopic repair (IRR 4.075; P = .004) were both associated with increased risk of stricture, while the presence of cardiovascular anomalies was again associated with a decreased risk of significant stricture formation (IRR 0.378; P = .029). The final multivariate model is summarized in Table 5.

Table 5.

Results of Multivariate Analysis for Stricture Formation

Variable	OR [95% CI]	P value	IRR [95% CI]	P Value
Operation type
Open thoracotomy	Reference		Reference
Thoracoscopic repair	7.41 [2.14–25.55]	.002	4.075 [1.57–10.526]	.004
Conversion	7.38 [0.54–99.12]	.131	2.640 [0.597–11.663]	.200
Operation staging
Primary repair	Reference		Reference
Staged repair	6.36 [2.05–19.72]	.001	3.706 [1.507–9.112]	.004
Cardiovascular anomaly
No	Reference		Reference
Yes	0.25 [0.06–0.33]	.012	0.378 [0.059–0.907]	.029

Discussion

Esophageal strictures have represented a significant morbidity since the dawn of EA repairs. Older studies examining outcomes from open thoracotomy repair report stricture rates anywhere from 17% to 33.3%.^4–7 More recent studies report similar rates of 5%–58% for strictures after thoracotomy repair.^8–11 Series of thoracoscopic repairs also report similar rates between 17% and 48.3%.^12–18 Indeed, a meta-analysis by Borruto et al. found no statistical difference in stricture rate comparing the two surgical techniques.¹⁹ While the stricture rate might have been expected to improve over time, it remains comparable to the reports from older studies.

The lack of a consistent definition for anastomotic stricture complicates comparisons across studies. Combinations of narrowing seen on endoscopy, narrowing seen on contrast study, need for dilation, and clinical symptoms are used in the various definitions. Furthermore, many studies define an anastomotic stricture as a stricture requiring more than a certain number of dilations, anywhere from one to five, depending on the study. For this study, anastomotic stricture was defined as those requiring more than three dilations. This definition was chosen for two reasons. First, it was in the midrange of definitions from other studies. More importantly, a recent survey of International Pediatric Endosurgery Group members found that while most members had no maximum number of dilations they would perform on a patient before considering more invasive operative procedures, some would only try up to a maximum of three dilations prior to operative intervention.²⁰

Even though the TEF Gross type was found to be significant on univariate analysis, most patients in this study had a type C TEF. Therefore, TEF type was excluded from the multivariate analysis due to the low numbers of other types. Percentages of TEF types found in this study are similar to those reported in the literature.² The 0% stricture rate in type E TEF seems intuitive due to the lack of esophageal anastomosis, but the reason for the 100% stricture rate of repairs of type B TEF is less obvious, although this was only found in two cases. The low stricture rate after the repair of type A TEF (14.3%) was also surprising. Type A is often associated with a larger gap between pouch and distal esophagus.² Longer gaps often require either a staged procedure or increased tension on the repair, a factor that has been shown to be associated with increased stricture formation.²¹ However, in the present study, the type A TEF repairs had a lower stricture rate than even the type C TEF.

Esophageal leak after EA repair is often thought to lead to an increased risk of stricture. This may be explained by the increased inflammation, leading to fibrosis and stricture. While an association between leak and stricture was seen on univariate analysis, this significance did not hold up in this multivariate model.

Gastroesophageal reflux has been associated with persistent strictures.²² An attempt was made to account for this by including need for fundoplication as a variable in the model. However, this may have been a poor approximation for actual gastroesophageal reflux, and this may account for the finding that fundoplication was not significant in the model for stricture formation.

The current study did find that both a staged operation and thoracoscopic repair were associated with increased stricture rates. A staged repair indicates that the esophageal gap was large enough that it could not be repaired safely primarily. It also portends a more difficult repair with possibly higher tension. Therefore, it seems intuitive that a staged repair may be associated with stricture. Contrary to other studies that showed no difference in stricture rates between thoracoscopy and thoracotomy for repair of EA, the present study shows that thoracoscopy was associated with a higher stricture rate at the authors' institution.^19,23 The study time frame includes the initiation of thoracoscopic repair at the authors' institution, and cases are performed by training fellows under guidance by attending surgeons. Both these factors may contribute to a higher stricture rate after thoracoscopic repair due to a “learning curve” for this advanced technique. This learning curve has been suggested based on improved outcomes at one institution when the first 5-year period of thoracoscopic repair was compared with the second 5-year period.¹⁶ Additionally, Rothenberg notes that his stricture rate decreased to 10% in the last 5 years of his study period.¹⁴ However, the learning curve of a group of surgeons may be different from that of a single surgeon, especially when technical aspects of cases are performed mainly by fellows in training, rather than the attending surgeons themselves. Different study periods were not compared in our analysis.

Over the study period, several different attending surgeons performed thoracoscopic repairs. The number of thoracoscopic repairs attempted by each surgeon is depicted in Table 6. The surgeons with lower numbers tended to be at the institution for less time during the study period, and no surgeon abandoned future thoracoscopic repair attempts.

Table 6.

Number of Thoracoscopic Attempts by Surgeon

Surgeon	Thoracoscopic cases
A	3
B	1
C	3
D	11
E	3
F	1
G	1

The decision for thoracoscopic versus open thoracotomy for repair of the EA was left up to surgeon preference. Table 7 shows patient characteristics by repair type. Patients in the thoracoscopic group tended to have a higher birth weight and gestational age. Associated anomalies, such as cardiac defects, were similar across repair types. At one point, a 2-kg minimum weight limit was supported for thoracoscopic esophageal repair. However, this bias has now been dropped by several surgeons, and infants weighing <2 kg have been repaired thoracoscopically.

Table 7.

Comparison of Patient Characteristics by Surgery Type

Variable	Thoracotomy	Thoracoscopic	Conversion	P value
N	98	20	3
Mean birth weight (g)	2306.2	2765.85	3313.3	.0048
Mean gestational age (weeks)	35.6	37.4	40	.0111
Cardiovascular anomaly	44 (44.9%)	10 (50%)	1 (33.3%)	.837
Genitourinary anomaly	21 (21.4%)	4 (20%)	1 (33.3%)	.871
Central nervous system anomalies	14 (14.3%)	3 (15%)	0 (0%)	.775
Vertebral anomalies	12 (12.2%)	5 (25%)	0 (0%)	.254
Skeletal anomalies	14 (14.3%)	1 (5%)	0 (0%)	.416
Gastrointestinal anomalies	9 (9.9%)	1 (5%)	0 (0%)	.675

One of the more interesting findings from this multivariate model was the association of cardiovascular anomalies with decreased stricture formation. Certainly, cardiovascular anomalies are associated with increased mortality in infants with EA.²⁴ However, the reason behind the association with decreased stricture formation is unclear. Were the infants with cardiovascular anomalies and higher stricture risks more likely to die and be excluded from the study? For this study, that would have selected infants with cardiovascular anomalies and lower stricture risks, giving the appearance of a protective association. Alternatively, is the association related to the severity of the anomaly, the response to the immune system, or some medication these infants are given? More research is needed in this area to understand the significance.

This study suffers from the inherent weaknesses common to all retrospective studies, including the reliance of data found in medical records. Additionally, other variables important to stricture formation, such as gap length or tension on repair, could not be reliably gathered from the medical records and could not be included in the study. Finally, the sample size adds several limitations to our study. Although 121 patients is a fairly large number for such a rare condition, the sample size limited the number of variables that could be included in the multivariate analysis. Interactions between variables could also not be included in the model.

In conclusion, thoracoscopic repair and a staged approach were both associated with an increased risk of clinically significant stricture formation after EA repair, while the presence of cardiovascular anomalies may decrease the risk of stricture formation. TEF Gross classification also affects stricture risk. This information is important in the consultation of parents regarding prognosis and the possible need for future dilations, and also provides direction for future quality improvement initiatives to decrease stricture formation after EA repair.

Footnotes

Disclosure Statement

No competing financial interests exist.

References

Harmon

, Coran

. Congenital anomalies of the esophagus. In: Coran

, Adzick

, Krummel

, Laberge

, Shamberger

, Caldamone

(eds). Pediatric Surgery. 7th ed. Philadelphia, PA: Elsevier Saunders, 2012, pp. 893–918.

Rothenberg

. Esophageal atresia and tracheoesophageal fistula malformations. In: Holcomb

, Murphy

, Ostlie

(eds). Ashcraft's Pediatric Surgery. 6th ed. Philadelphia, PA: Elsevier Saunders, 2014, pp. 365–384.

Zou

. A modified poisson regression approach to prospective studies with binary data. Am J Epidemiol, 2004; 159:702–706.

Yanchar

, Gordon

, Cooper

, et al. Significance of the clinical course and early upper gastrointestinal studies in predicting complications associated with repair of esophageal atresia. J Pediatr Surg, 2001; 36:815–822.

Spitz

, Kiely

, Brereton

. Esophageal atresia: five year experience with 148 cases. J Pediatr Surg, 1987; 22:103–108.

Engum

, Grosfeld

, West

, et al. Analysis of morbidity and mortality in 227 cases of esophageal atresia and/or tracheoesophageal fistula over two decades. Arch Surg, 1995; 130:502–508; discussion 508–509.

Randolph

, Newman

, Anderson

. Current results in repair of esophageal atresia with tracheoesophageal fistula using physiologic status as a guide to therapy. Ann Surg, 1989; 209:526–530; discussion 530–531.

Burford

, Dassinger

, Copeland

, et al. Repair of esophageal atresia with tracheoesophageal fistula via thoracotomy: a contemporary series. Am J Surg, 2011; 202:203–206.

Mortell

, Azizkhan

. Esophageal atresia repair with thoracotomy: the Cincinnati contemporary experience. Semin Pediatr Surg, 2009; 18:12–19.

10.

Thakkar

, Cooney

, Kumar

, et al. Measured gap length and outcomes in oesophageal atresia. J Pediatr Surg, 2014; 49:1343–1346.

11.

Jawaid

, Chan

, Jesudason

. Subspecialization may improve an esophageal atresia service but has not addressed declining trainee experience. J Pediatr Surg, 2012; 47:1363–1368.

12.

Patkowsk

, Rysiakiewic

, Jaworski

, et al. Thoracoscopic repair of tracheoesophageal fistula and esophageal atresia. J Laparoendosc Adv Surg Tech A, 2009; 19:S19–22.

13.

Dingemann

, Zoeller

, Ure

. Thoracoscopic repair of oesophageal atresia: results of a selective approach. Eur J Pediatr Surg, 2013; 23:14–18.

14.

Rothenberg

. Thoracoscopic repair of esophageal atresia and tracheo-esophageal fistula in neonates: evolution of a technique. J Laparoendosc Adv Surg Tech A, 2012; 22:195–199.

15.

Holcomb

, Rothenberg

, Bax

, et al. Thoracoscopic repair of esophageal atresia and tracheoesophageal fistula: a multi-institutional analysis. Ann Surg, 2005; 242:422–428; discussion 428–430.

16.

van der Zee

, Tytgat

, Zwaveling

, et al. Learning curve of thoracoscopic repair of esophageal atresia. World J Surg, 2012; 36:2093–2097.

17.

Huang

, Tao

, Chen

, et al. Thoracoscopic repair of oesophageal atresia: experience of 33 patients from two tertiary referral centres. J Pediatr Surg, 2012; 47:2224–2227.

18.

Okuyama

, Koga

, Ishimaru

, et al. Current practice and outcomes of thoracoscopic esophageal atresia and tracheoesophageal fistula repair: a multi-institutional analysis in Japan. J Laparoendosc Adv Surg Tech A, 2015; 25:441–444.

19.

Borruto

, Impellizzeria

, Montalto

, et al. Thoracoscopy versus thoracotomy for esophageal atresia and tracheoesophageal fistula repair: review of the literature and meta-analysis. Eur J Pediatr Surg, 2012; 22:415–419.

20.

Lal

, Miyano

, Juang

, et al. Current patterns of practice and technique in the repair of esophageal atresia and tracheoesophageal fistua: an IPEG survey. J Laparoendosc Adv Surg Tech A, 2013; 23:635–638.

21.

Serhal

, Gottrand

, Sfeir

, et al. Anastomotic stricture after surgical repair of esophageal atresia: frequency, risk factors, and efficacy of esophageal bougie dilatations. J Pediatr Surg, 2010; 45:1459–1462.

22.

Ashcraft

, Goodwin

, Amoury

, et al. Early recognition and aggressive treatment of gastroesophageal reflux following repair of esophageal atresia. J Pediatr Surg, 1977; 12:317–321.

23.

Al Tokhais

, Zamakhshary

, Aldekhayel

, et al. Thoracoscopic repair of tracheoesophageal fistulas: a case-control matched study. J Pediatr Surg, 2008; 43:805–809.

24.

Wang

, Tashiro

, Allan

, et al. A nationwide analysis of clinical outcomes among newborns with esophageal atresia and tracheoesophageal fistulas in the United States. J Surg Res, 2014; 190:604–612.