Punthakee X, Zaghi S, Nabili V, Knott PD, Blackwell KE. Effects of Salivary Bypass Tubes on Fistula and Stricture Formation. JAMA Facial Plast Surg. 2013;15(3):219-225. doi:10.1001/jamafacial.2013.791
Author Affiliations: Department of Otolaryngology–Head and Neck Surgery, University of Toronto, Toronto, Ontario, Canada (Dr Punthakee); Department of Head and Neck Surgery, David Geffen School of Medicine at UCLA (University of California, Los Angeles), Los Angeles, California (Drs Zaghi, Nabili, and Blackwell); and Department of Otolaryngology–Head and Neck Surgery, UCSF School of Medicine, University of California, San Francisco (Dr Knott).
Importance Stricture and fistula formation are two of the most common long-term complications of free flap reconstruction of hypopharyngeal defects.
Objective To examine the effects of salivary bypass tubes (SBTs) on fistula and stricture formation after free flap reconstruction of hypopharyngeal defects.
Design Retrospective cohort study.
Setting Academic tertiary care medical center.
Participants A total of 103 consecutive patients who underwent hypopharyngeal free flap reconstruction.
Intervention Use of salivary bypass tube.
Main Outcome Measure Fistula and stricture formation.
Results The overall fistula and stricture rates were 14.6% and 27.2%, respectively. Subgroup analysis revealed fistula rates of 7.4% in patients who received SBTs and 22.4% in those who did not (P = .048). However, no statistically significant difference between the two groups was shown with multivariate analysis. The unadjusted stricture rate was 30.6% without vs 24.1% with SBT placement (P = .51).
Conclusions and Relevance Univariate analysis showed that SBT placement significantly reduced the risk of fistula in this population. Larger multicenter studies are needed to further explore the benefits of SBT use in preventing fistula and stricture formation.
Level of Evidence 3.
Pharyngoesophageal reconstruction is considered one of the most challenging areas of head and neck reconstructive surgery. Its goals fall into two general categories: minimizing the potential for postoperative life-threatening complications (eg, protection of great vessels) and restoring pharyngoesophageal function.1
Two well-known complications related to pharyngeal reconstruction are postoperative fistula and stricture formation. Many attempts have been made to reduce the rates of these complications over the years. The salivary bypass tube (SBT) was described in 1955 by Montgomery2; its initial use was after laryngoesophagectomy before reconstruction of the cervical esophagus. In 1978, Montgomery further described the SBT as a tool to decrease fistula and stricture rates in advanced carcinoma of the pharyngeal segment.3 Despite the common use of SBTs in current pharyngeal reconstruction, to our knowledge, no studies have provided substantial evidence supporting their use in reducing complications. Our objective was to review our experience in free flap reconstruction involving hypopharyngeal defects and to identify factors related to perioperative fistula and stricture formation, with particular attention to the effects of SBTs.
We conducted a multicenter retrospective analysis on the effects of SBT placement on postoperative fistula and stricture outcomes among 103 patients who underwent hypopharyngeal resection and free flap reconstruction from October 13, 1995, through May 14, 2010. The study was performed at the David Geffen School of Medicine at UCLA, the MetroHealth Hospital system in Cleveland, and the Cleveland Clinic Foundation, all academic tertiary care medical referral centers. The study was reviewed and approved by the UCLA Office for the Protection of Research Subjects and the Cleveland Health System Institutional Review Board.
The study included patients undergoing free flap reconstruction as a component of their management of a hypopharyngeal defect, identified by reviewing medical records at 3 institutions (teaching hospitals affiliated with the David Geffen School of Medicine, MetroHealth, and the Cleveland Clinic Foundation) to identify patients who met the inclusion criteria. All patients were followed up for at least 4 weeks after operation. Data were extracted prospectively, including patient age, sex, American Society of Anesthesiologists class, indication for surgery, TNM staging category (if applicable), and treatment variables, including prior head and neck surgery, radiation therapy, chemoradiation, location of tumor or lesion, defect classification, and flap reconstruction type. The data were then entered into a worksheet (Excel 2011; Microsoft).
The protocols for use of SBTs were similar at all participating institutions. Among patients undergoing prophylactic SBT placement, the site of pharyngoesophageal reconstruction was stented with an SBT (10, 12, or 14 mm in diameter) that was placed at the time of pharyngoesophageal reconstruction and left in place after operation for approximately 2 weeks. The reconstructive surgeon selected the SBT size by matching the estimated native internal diameter of the cervical esophagus. The SBT was secured by using a 12F red rubber catheter sutured to its proximal end, brought up through the pharynx and out through the nose, and sutured again to the caudal nasal septum to prevent dislodgement. It was removed transorally at a postoperative office visit after approximately 2 weeks.
The primary intervention was placement of an SBT, and the primary outcome measures were the incidences of fistula and stricture after operation. These outcomes were assessed on the basis of clinical history and physical examination findings and results of diagnostic endoscopy and/or contrast swallowing studies. Data were examined by using the PivotTable function of Excel 2011 and analyzed for significance with JMP 10 statistical software (SAS Institute, Inc). Unadjusted proportions were calculated for the outcomes, including fistula and stricture, of two conditions: SBT and no SBT. Two-tailed univariate statistical analysis of variables associated with the primary outcome measures was performed. The Fisher exact test was used for 2 × 2 contingency table analysis when 1 cell had a value less than 5, and Pearson χ2 analysis was performed in all other instances. Multivariate analysis using nominal logistic regression was performed for the primary intervention and for variables that approached or achieved statistical significance in univariate analysis. Covariates of SBT placement were tested for significance by using Kruskal-Wallis, Pearson χ2, and Fisher exact tests. Results were considered statistically significant at P < .05. Adjusted odds ratios were calculated by means of multivariate logistic regression controlling for the significant covariate factors.
A total of 103 patients were included in the final analysis. Their median age was 64 years (range, 36-87 years), and their mean (SD) age, 63.0 (12.1) years. Patient characteristics, including sex, comorbidity level, primary vs secondary reconstruction, prior treatments, and defect classification, are listed in Table 1. Defects were classified according to the Blackwell and Urken1 classification system for pharyngoesophageal defects (Table 2). Because there were no type 0 defects, a skin or mucosal component of the flap was incorporated into the pharyngeal repair in all cases, and no flaps were used as an overlay graft to augment a primary mucosal repair.
In our sample of 103 patients, we identified 15 postoperative fistulas and 28 postoperative strictures (Table 1). The overall fistula and stricture rates were 14.6% (SE, 3.5%) and 27.2% (4.4%), respectively. Univariate analysis showed that the fistula rate was significantly associated with SBT placement (P = .048), primary vs secondary reconstruction (P = .02) and flap type (P = .02); prior chemoradiation (P = .053) approached statistical significance. Univariate analysis showed that the stricture rate was significantly associated with primary vs secondary reconstruction (P = .04), defect classification (P = .02), and tumor location (P = .03). The unadjusted fistula rate was higher in the no-SBT group (22.4% [SE, 6.0%]) than in the SBT group (7.4% [3.6%]), demonstrating an absolute 3-fold reduction in the risk of fistulas in patients who received SBT placement. The unadjusted stricture rate was 30.6% (SE, 6.6%) in the no-SBT group vs 24.1% (5.8%) in the SBT group, but this difference was not significant (P = .51).
The results of multivariate analysis are summarized in Table 3 and Table 4. Defect classification was significantly associated with stricture rate (P = .02). Multivariate analysis of factors associated with fistula showed that this study was insufficiently powered to demonstrate which of the 4 factors analyzed remained significant after controlling for covariables.
We then examined covariates of SBT placement to determine whether the SBT vs no-SBT groups differed significantly with respect to population demographics. The analysis (Table 5) revealed a significant difference between the SBT and no-SBT groups for the following factors: (1) proportion of primary vs secondary reconstructions (P = .002; Fisher exact test), (2) proportion of patients with previous chemoradiation (P < .001; Fisher exact test), and (3) proportion of patients with a radial forearm free flap (RFFF) vs other flap types (P < .001; Fisher exact test). Other covariate factors, including age, sex, American Society of Anesthesiologists class, tumor and node categories, radiation therapy, prior head and neck surgery, defect classification, and tumor location, did not differ significantly between the groups.
To control for the 3 significant covariate factors, we used multivariate logistic regression to identify the adjusted stricture and fistula rates with associated P values in the SBT vs no-SBT groups, in addition to the associated odds ratios (Table 6 and Table 7). We were unable to show a statistically significant association between SBT use and fistula or stricture rate after adjusting for the 3 covariate factors (primary vs secondary reconstruction, prior chemoradiation, and free flap type) in this multivariate logistic regression model.
Rates of fistula and stricture formation associated with free flap reconstruction of the hypopharynx have been well reported. Anthony et al4 used RFFFs in the reconstruction of both partial and circumferential hypopharyngeal defects, without the use of SBTs, and reported a fistula rate of 32%. Azizzadeh et al5 subsequently also used RFFFs in the reconstruction of both partial and circumferential hypopharyngeal defects, without the use of SBTs, and reported both fistula and stricture rates of 20%. For circumferential hypopharyngeal defects, Scharpf and Esclamado6 used the RFFF for reconstructions, again without the use of SBTs, and reported fistula and stricture rates of 28% and 36%, respectively. In a larger series in which anterolateral thigh free flaps were used for partial and circumferential hypopharyngeal reconstructions, Yu et al7 reported fistula and stricture rates of 9% and 6%, respectively, also without the use of SBTs. The overall rates of fistula (14.6% [SE, 3.6%]) and stricture (27.2% [4.4%]) formation in our study are comparable to those in the aforementioned studies.
Some studies have specifically used SBTs for all of their hypopharyngeal reconstructions to identify their utility in reducing complications. Varvares et al8 and Genden and Jacobson9 used RFFFs with SBTs for partial and circumferential reconstructions, and both groups identified fistula and stricture rates of 20% and 10%, respectively, in their individual populations. Murray et al10 reported no fistulas and a 6% stricture rate in circumferential hypopharyngeal reconstructions using only anterolateral thigh free flaps in a group of 14 patients in whom SBT was used. Owing to heterogeneity, comparisons across studies are challenging. To our knowledge, no studies have compared SBT and no-SBT groups and their associated rates of fistula and stricture formation.
Our univariate analysis shows a statistically significant reduction in fistula rates among patients receiving prophylactic SBT treatment, from 22.4% (SE, 6.0%) to 7.4% (3.6%), with an absolute reduction in the fistula rate of 15.0% (9.6%). This finding is clinically significant because it illustrates differences between the two groups for one of the most common and severe complications of free flap reconstruction involving the hypopharynx. The analysis comparing the no-SBT and SBT groups showed that the statistically significant covariate factors were primary vs secondary reconstruction, prior chemotherapy, and flap type. The differences in our multivariate analysis showed a trend but were not statistically significant. Although our study is, to our knowledge, the largest of its kind comparing these treatment groups, a post hoc power calculation showed that for a power of 80% with a type I error of 0.05, the sample size would need to be 188 to demonstrate this outcome in a multivariate analysis.
There was an increased risk of fistula formation among patients undergoing secondary rather than primary reconstruction, which was significant with univariate analysis but not with the multivariate model. A confounding problem is the large discrepancy in the sample size and fistula rate of patients with primary vs secondary reconstruction. The sample included 89 patients with primary reconstruction (fistula rate, 11.2% [SE, 3.3%]) and 14 with secondary reconstruction (fistula rate, 35.7% [12.8%]). Secondary reconstruction is often more challenging, and, despite the small sample size, we can hypothesize that the fistula risk may be higher in these patients, given the increased complexities.
Some recent articles7,10 have described very low rates of fistula formation (0% and 9%) and strictures (6% and 14%) with use of the anterolateral thigh flap for total laryngopharyngectomy defects. Our results support the contention that anterolateral thigh free flaps may be associated with a reduced rate of fistula formation (4% vs 17.9% for other flap types). Many centers are moving toward increased use of the anterolateral thigh flap because of the large amount of donor site tissue available for harvest and minimal donor site morbidity, making it the best option especially for patients with large defects.11 However, at least 1 prior publication reported an increased risk of fistula and stricture formation in patients who undergo pharyngoesophageal reconstruction with anterolateral thigh free flaps.12 As the use of anterolateral thigh flaps increases, the use of SBT with these flaps, compared with other flap types, will need to be explored.
Stricture rates in the literature for free flap reconstruction involving the hypopharynx vary from 6% to 36%.4- 10 We found that defect classification was significantly associated with the rate of stricture formation at multivariate analysis, with an increased risk of structure formation (37.7%) in patients with circumferential defects of the pharyngoesophageal segment compared with the risk in patients with noncircumferential defects (15.8%). We have observed that most strictures occur at the distal flap-to-esophagus enteric anastomosis, and we believe that the small circumference of this annular anastomosis may lead to an increased risk of stricture formation through a process of circumferential scar tissue formation.
In our study, stricture rates did not differ significantly between patients with SBT and those without SBT placement (univariate analysis, 30.6% [SE, 6.6%] for the no-SBT group vs 24.1% [5.8%] for the SBT group; P = .51). Although there was an insignificant reduction in the incidence of strictures seen with SBT placement, this finding may be confounded by changing patterns of practice at the institution where most patients were treated during the 15-year study period. Salivary bypass tubes were not used in any patients before 2006, and this was at about the same time when transnasal esophagoscopy (TNE) became available for use in the outpatient clinic. During this time, our group began using TNE to assess most patients on a routine basis at approximately 3 weeks after their operation to allow for early detection and dilation of strictures when they would occur. We have observed that early detection and balloon dilation of postoperative strictures are safe and may reduce the risk of long-term dysphagia secondary to stricture formation. However, routine use of TNE in patients who received SBT placement may have increased our ability to detect subclinical strictures that were not detected in patients who did not receive an SBT and therefore did not undergo routine postoperative TNE; therefore, the incidence of strictures observed in patients with SBT placement may not be comparable to that observed in patients without SBT placement. Another weakness of our analysis arises from the minimum follow-up (4 weeks) required for inclusion in this series, because stricture formation can occur beyond that time, especially in patients who have received heavy doses of radiation.
Since its inception, SBT use has been thought to help diminish fistula and stricture formation, and our study further strengthens the arguments for its use. Nevertheless, our study unfortunately has some shortcomings in that it was insufficiently powered to demonstrate a beneficial effect of using SBTs to reduce the risk of fistulas and strictures when other potentially confounding variables were included in the analysis. We strongly advocate an ongoing effort of the Microvascular Committee of the American Academy of Otolaryngology–Head and Neck Surgery13 to study the outcomes of pharyngoesophageal reconstructive surgery; its analysis of a large multi-institutional sample may be sufficiently powered to enable definitive conclusions about effects of the variables that we have touched on in our study. In the interim, our promising results, the minimal complications of SBTs, and the severe sequelae of fistulas make the use of SBTs worth serious consideration in high-risk patients undergoing microvascular flap hypopharyngeal reconstructive surgery, including those undergoing secondary reconstruction or reconstruction with flaps other than anterolateral thigh free flaps and those who have previously received chemoradiation therapy.
In conclusion, the use of prophylactic SBTs for patients undergoing microvascular hypopharyngeal reconstruction was associated with a significant decrease in the rate of fistula formation. Our study was a multi-institutional collaboration and the largest series analyzed to date, but an even larger population is needed to test the significance of other factors in fistula formation and SBT use. Although larger multicenter studies are needed to further explore the potential benefits of SBTs, our promising results suggest that SBT placement should be considered to reduce the risk of fistula formation in high-risk patients undergoing microvascular flap reconstruction of the hypopharynx.
Correspondence: Keith E. Blackwell, MD, Department of Head and Neck Surgery, David Geffen School of Medicine, UCLA, Los Angeles, CA 90095 (firstname.lastname@example.org).
Accepted for Publication: November 18, 2012.
Published Online: March 21, 2013. doi:10.1001/jamafacial.2013.791
Author Contributions:Study concept and design: Nabili and Blackwell. Acquisition of data: Zaghi, Nabili, Knott, and Blackwell. Analysis and interpretation of data: All authors. Drafting of the manuscript: Punthakee, Zaghi, Nabili, and Knott. Critical revision of the manuscript for important intellectual content: All authors. Statistical analysis: Zaghi. Administrative, technical, and material support: Punthakee, Nabili, and Knott. Study supervision: Nabili, Knott, and Blackwell.
Conflict of Interest Disclosures: Dr Punthakee was an American Academy of Facial Plastic and Reconstructive Surgery Fellow at UCLA, and Dr Knott was a faculty member at the Cleveland Clinic when data presented in this article were collected.
Additional Contributions: We thank Jeffrey Gornbein, DrPH, for guidance with statistical analysis.