Outcomes of Combined Antegrade–Retrograde Dilations for Radiation-Induced Esophageal Strictures in Head and Neck Cancer Patients

Abstract

The purpose of this study is to analyze outcomes of combined antegrade–retrograde dilations (CARD). This retrospective study was conducted on 14 patients with a history of head and neck cancer, treated with radiation therapy that was complicated by either complete or near-complete esophageal stenosis. All patients had minimal oral intake and depended on a gastrostomy tube for nutrition. Swallow function before and after CARD was assessed using the Functional Oral Intake Scale, originally developed for stroke patients and applied to head and neck cancer patients. Patients undergoing CARD demonstrated a quantifiable improvement in swallow function (p = 0.007) that persisted at last known follow-up (p = 0.015) but only a minority (23.1%) achieved oral intake sufficient to obviate the need for tube feeds. Complication rates were 24% per procedure or 36% per patient, almost all complications required procedural intervention, and all complications occurred in patients with complete stenosis. Our study suggests further caution when considering CARD, careful patient selection, and close post-operative monitoring.

Introduction

Head and neck cancers currently account for approximately 4% of all cancers in the United States, with approximately 65,630 new diagnoses and 14,500 deaths this year [1]. Radiotherapy is a mainstay of treatment, often used as either adjuvant therapy following surgical resection or as a definitive treatment, depending on the tumor site, subtype, and stage [2, 3]. One common complication as a result of radiotherapy is dysphagia due to fibrosis and stenosis of the pharynx and esophagus [4]. Rarely, the stenosis can progress to a complete occlusion, resulting in an occluded esophagus in approximately 3% of head and neck cancer patients receiving radiotherapy [5, 6].

Partial esophageal strictures are managed by antegrade dilation techniques utilizing Maloney bougie dilators, rigid Savary dilators, or balloon dilators that are passed through the mouth into the esophagus to serially increase the esophageal lumen diameter [7]. Patients often require multiple sessions to achieve adequate symptomatic relief [8]. However, these techniques are not applicable to either a complete stenosis without a lumen or a pinhole-sized lumen that cannot be located with traditional endoscopy. Without alternative options for treatment, patients are often left completely G-tube dependent with some unable to tolerate even their own saliva.

Multiple groups have reported small series analyzing a new endoscopic technique, commonly referred to as combined antegrade and retrograde esophageal stenosis dilation (CARD) [9–13]. This procedure utilizes retrograde endoscopy via the patient’s G-tube to transilluminate the stricture which allows for passage of a guidewire or traditional antegrade endoscopy to safely traverse the stricture as a “rendezvous” method [14, 15]. Our study is the first to use the Functional Oral Intake Scale (FOIS), a validated method for documenting changes in dysphagia and swallow function [16–18], to quantify long-term outcomes in this setting. This study presents our experience with CARD including patients with both complete and incomplete stenosis, focusing on long-term functional outcomes in addition to safety and efficacy of the procedure.

Patients and Methods

This study has been approved by the University of Southern California Institutional Review Board. All patients who underwent combined antegrade/retrograde dilation between January 1, 2005, and September 1, 2019, at either LAC + USC Hospital or Keck Hospital of USC were retrospectively reviewed. Patients were identified by searching for all patients who underwent esophageal dilation and then reviewing operative reports for patients who underwent the CARD procedure. Patients were included if they had a history of upper aerodigestive tract cancer, received either adjuvant or definitive radiotherapy, and had a G-tube in place prior to surgery. Patients were excluded if their stricture was discovered to be malignant.

A total of 14 patients were included in this study and summary characteristics are shown in Table 1. A number of patients had additional surgeries related to their cancer treatment, and all patients had received a combination of chemotherapy and radiation either as adjuvant or definitive therapy (Supplemental Table 1). Many of the patients received cancer treatment at outside hospitals prior to their presentation at our institution, and thus information about staging and radiation therapy was limited. All patients presented with dysphagia and were reliant on a gastrostomy tube for nutrition. The majority of patients (71%) had at least one unsuccessful antegrade dilation attempt that was terminated due to inability to pass a bougie dilator, flexible esophagoscope, or guidewire. After an unsuccessful antegrade attempt, the patient was rescheduled for a CARD procedure. Length of follow-up after CARD procedure varied widely (mean = 173 weeks, median = 29.5, SD = 235, range 0–643).

Table 1.

Functional Oral Intake Scale developed by Crary et al. [16]

FOIS score	Intake level
1	Nothing by mouth
2	Tube dependent with minimal attempts of food or liquid
3	Tube dependent with consistent oral intake of food or liquid
4	Total oral diet of a single consistency
5	Total oral diet with multiple consistencies but requiring special preparation or compensations
6	Total oral diet with multiple consistencies without special preparation, but with specific food limitations
7	Total oral diet with no restrictions

CARD-specific outcomes included technical success rate, complications, frequency and timing of procedures, and maximum dilation achieved. Technical success of the procedure was defined as establishing a lumen that allowed the subsequent antegrade passage of dilators. Potential intraoperative complications included sequelae of esophageal perforation such as pneumothorax, pneumomediastinum, or pneumoperitoneum, common surgical complications such as bleeding, and G-tube malfunctions. Post-operative complications included surgical site infections or aspiration pneumonia occurring within 30 days after CARD. Vocal fold paralysis was reported in one other cohort and was also included as a post-operative complication [12]. Frequency and timing of antegrade dilations were also recorded.

Long-term swallow function was primarily assessed using the Functional Oral Intake Scale (FOIS), which measures oral dietary tolerance on a 7-point scale (Table 2) [16]. FOIS scores were determined by reviewing descriptions of patients’ swallow function documented by physicians or speech language pathologists (SLP) during swallow evaluations. Scores before CARD were compared to maximal scores achieved during follow-up and to scores at the last follow-up. Swallow function was also assessed by G-tube independence, defined as documented history of at least 1 month of either PO tolerance without G-tube supplementation or removal of an unneeded G-tube. The first attempted CARD procedure was used as the primary time point for assessing timing of subsequent procedures and FOIS scores. Length of follow-up was defined as the time between the first CARD and the last encounter with a physician or SLP at our institution.

Table 2.

Patient demographics and characteristics

Age, years, median (IQR)	62.5 (55.0–70.8)
Gender
Male	71.4%
Female	28.6%
Primary tumor site
Nasopharynx	7.1%
Oral cavity	35.7%
Oropharynx	28.6%
Hypopharynx	7.1%
Larynx	14.3%
Cervical esophagus	7.1%
Treatment modality
Surgical resection with adjuvant chemoradiation	64.3%
Definitive chemoradiation	35.7%
Stricture type
Incomplete	42.9%
Complete	57.1%
Length of follow-up, weeks, median (IQR)	29.6 (8.1–330.2)

To investigate associations of stenosis severity with key clinical outcomes, patients were subdivided based on whether they had complete or near-complete stenosis of the esophagus at the time of their first CARD attempt. Complete stenosis was defined as total absence of a lumen, confirmed by either failure of any contrast on barium swallow to pass the stenosis or inability to pass a retrograde guidewire during a dilation procedure.

Study data were collected and managed using Research Electronic Data Capture (REDCap), a secure, web-based software platform designed to support data capture for research studies [19, 20]. Data analysis was performed in SAS 9.4. Student’s t-test was used for continuous variables, Wilcoxon matched pairs signed-rank or Wilcoxon rank sum test for FOIS scores, and Fisher’s exact test for categorical variables, given the small sample size. Statistical significance was defined as p < 0.05.

Procedure in Detail

All CARD procedures were performed under general anesthesia with an otolaryngologist and general surgeon. The otolaryngologist approached the proximal side of the stricture using either a direct laryngoscope or a rigid esophagoscope to first visualize the stenosis. Since patients were PO intolerant, they had existing G-tubes that the general surgeon used as access for retrograde flexible endoscopy in order to visualize the distal side of the stricture. The G-tube was removed, and a pediatric flexible esophagoscope was introduced through the G-tube site and directed retrograde through the esophagus. With direct visualization from above and below the stenosis, a determination was made whether there was 100% stenosis or a pinpoint lumen. In patients with a near-total stenosis with a pinpoint lumen, a guidewire could then be passed retrograde by the general surgeon and grasped by the otolaryngologist and brought through the mouth. In cases of complete stenosis, the general surgeon provided transillumination or pressure from a guidewire from below the stricture, which was used to direct and guide the otolaryngologist in blunt dissection, sharp incision, or direct guidewire puncture of the stricture.

Once the stricture was successfully canalized with the guidewire, wire-guided Savary dilators were passed antegrade. After passing the smallest dilator (typically 18–20 Fr), dilators were passed sequentially in increments of either 2 or 3 Fr until either the dilator was met with moderate resistance or the lumen was sufficiently large. Depending on surgeon preference, a nasogastric tube was placed in a portion of patients as a means of maintaining patency of the esophagus by acting as a temporary stent. Following extubation, chest X-ray was obtained to evaluate for signs of perforation such as pneumothorax or pneumoperitoneum. Bedside swallow evaluations were performed prior to discharge. Patients were typically discharged the same day or kept overnight for monitoring. Following discharge, patients were seen in clinic within 2 weeks to assess swallowing, with fiberoptic endoscopic evaluation of swallowing (FEES), or a modified barium swallow study, depending on surgeon preference and the patient’s symptoms. Patients were then evaluated for repeat CARD, antegrade dilation, a previously reported home self-dilation program [21], additional swallow evaluations, and additional follow-up on an individual basis. Briefly, the home dilation program includes 10 sessions of exercise-based swallow therapy followed by twice-weekly self-dilation with a 22- to 24-Fr Bougie dilator. Patients were assessed every 4 weeks to potentially increase the dilator size.

Results

A total of 25 CARD procedures were performed on 14 patients, of which 20 were successful, yielding an overall 80% (95% CI 59.3%, 93.2%) technical success rate of the procedure (Table 3). Eleven of 14 (78.6%, 95% CI 49.2%, 95.3%) patients had at least one successful CARD procedure. Among these patients, the first successful CARD procedure achieved a median maximum dilation of 36 Fr (range 27–39, IQR 33.8–36). This represents a median increase of 18 Fr (range 12–24, IQR 17–18.8) compared to the initial dilator used during the same CARD.

Table 3.

CARD outcomes Total number CARD

Total number CARD	25
Successful CARD	20 (80%)
No. CARD/patient	Median = 2, range 1–3, IQR 1–2.25
NG tube stenting	9 (36%)
Total number antegrade dilations	19
Successful antegrade dilations	19 (100%)
No. antegrade dilations/patient	Mean = 0.7, range 0–9, IQR 0–2
Weeks between dilations	Median = 8.5, range 1–540, IQR 4.5–31.25
Initial dilation achieved, Fr	Median = 36, range 27–39, IQR 33.8–36
Max dilation achieved, Fr	Median = 40.5, range 27–58, IQR 36.0–44.5
Self-dilation program	7 (50%)
FOIS score, median (IQR)
Initial	(1–1)
Highest	2(1–5.3)
End	2 (1–2.8)
G-tube independence	28.6%

Eight patients (57%) did not undergo any antegrade dilations following the initial CARD attempt, although four of them (29%) underwent more than one CARD. Five other patients had four or fewer antegrade dilations, and a sixth patient was treated with nine antegrade dilations. Timing between subsequent dilations was variable, with most patients receiving repeat dilations a month or two apart. Since repeat dilations were performed for worsening or recurring dysphagia, frequent dilations indicate both refractory disease and consistent follow-up. Through the course of their treatments, patients with at least one successful CARD achieved a median maximum dilation of 40.5 Fr (range 27–58, IQR 36.0–44.5).

Overall, pre-CARD FOIS scores were low (median = 1, range 1–2, IQR 0). With one exception, all patients had a pre-operative FOIS score of 1, indicating zero PO intake (Fig. 1a). Nine of 14 (64.3%, 95% CI 35.1%, 87.2%) patients included in this study and 8 of 11 patients (72.7%, 95% CI 39.0%, 94.0%) with at least one successful CARD had a quantifiable improvement in FOIS score at any point during follow-up. Including only patients with at least one successful CARD, maximal FOIS scores were significantly increased compared to initial scores (Wilcoxon signed-rank, z = 2.54, p = 0.011, r = 0.77). For all patients, maximal FOIS scores (median = 2, range 1–7, IQR 4) were also significantly increased (Fig. 1b) compared to initial FOIS scores (z = 2.68, p = 0.007, r = 0.72).

Fig. 1.

Histogram of FOIS scores, a baseline prior to first CARD attempt, b maximum achieved, and c at the end of follow-up. Initial median FOIS score was 1 (range 1–2, IQR 0), improved to a median maximum FOIS score of 2 (range 1–7, IQR 4), and remained at a median FOIS score of 2 (range 1–7, IQR 1). Both maximum FOIS scores and FOIS at the end of follow-up were significantly different compared to initial FOIS scores (p = 0.007 and p = 0.015, respectively)

Excluding one patient because they had zero follow-up, 7 of the remaining 13 (53.9%, 95% CI 25.1%, 80.8%) patients and 6 of the remaining 10 (60.0%, 95% CI 26.2%, 87.8%) patients that had at least one successful CARD procedure maintained a higher FOIS score at last known follow-up compared to their baseline FOIS scores. Including only patients with at least one successful CARD and with follow-up, FOIS scores at the end of follow-up were significantly increased compared to baseline (Wilcoxon signed-rank, z = 2.26, p = 0.024, r = 0.71). Including all patients, FOIS scores at the end of follow-up (median = 2, range 1–7, IQR 1) were also significantly increased (Fig. 1c) compared to baseline (z = 2.41, p = 0.015, r = 0.64).

Four patients achieved G-tube independence, indicated by a FOIS score of 4 or higher. One patient’s first CARD attempt failed to establish a lumen after repeated attempts to pass a guidewire and was converted to transcervical esophageal reconstruction. This patient achieved a decade of G-tube independence which could not be attributed to CARD, resulting in three patients achieving G-tube independence after a successful CARD procedure (23.1%, 95% CI 5.0%, 53.8%). Two of these patients maintained G-tube independence for the duration of follow-up, and both had documented histories of consistent follow-up with frequent operative or in-office dilations. The third patient maintained G-tube independence until they were no longer adherent to a home dilation program, and the stenosis recurred.

A total of 5 out of 14 (35.7%, 95% CI 12.8%, 64.9%) patients and 6 out of 25 (24.0%, 95% CI 9.4%, 45.1%) CARD procedures had complications (Table 4). The most common complication was esophageal perforation, occurring four times, followed by G-tube site problems. Three of the perforations occurred in procedures that failed to establish a lumen. All complications except one required invasive management. Only one perforation was treated medically, and the remaining complications required either chest tubes or additional surgery. One particularly severe complication included an esophageal perforation which possibly contributed to an intraoperative arrest, although there was also evidence of pulmonary embolism. Details regarding individual complications are included in Table 5. No patients died as a result of CARD.

Table 4.

CARD complications

Complication	No. (%) (N = 14 patients; N = 25 CARD)
Esophageal perforation, pneumothorax	4/14 (29%); 4/25 (16%)
w/pneumomediastinum	2 patients
w/pneumoperitoneum	2 patients
Intraoperative arrest w/PE + PTX	1/14 (7%); 1/25 (4%)
Vocal fold paralysis	1/14 (7%); 1/25 (4%)
G-tube site problems	2/14 (14%); 2/25(8%)
G-tube dislodgment	1 patient
Stomach fell away from anterior abdominal wall	1 patient
Totala	5/14 (36%); 6/25 (24%)

^aOne patient had CARD with both perforation and G-tube dislodgment, followed by vocal fold paralysis during a subsequent CARD. A second patient had CARD with both perforation and arrest

Table 5.

Individual complications of CARD

ID	Event	Successful dilation?	Adverse outcome	Intervention
3	First attempted CARD	No	Stomach fell away from anterior abdominal wall, required open G-tube revision; after revision, flex scope could not be passed into esophagus	Converted to transcervical esophageal reconstruction with sternocleidomastoid flap and Alloderm
4	First attempted CARD	Yes	Pneumomediastinum, pneumoperitoneum, left  pneumothorax	Pigtail catheter
6	First attempted CARD	No	Dislodged G-tube with pneumoperitoneum left-sided pneumothorax with almost complete collapse of the left lung	Laparoscopic g-tube replacement, pigtail catheter for pneumothorax
6	Second repeat CARD	Yes	Laryngeal edema and stridor 2 days later, eventually developed bilateral TVF paralysis 1 week after the procedure	Edema was monitored and improved without intervention. Bilateral TVF paralysis treated with tracheostomy which remained in place as of last follow-up 6 months later
10	First attempted CARD	No	Intraoperative arrest with evidence of small pulmonary embolism, bilateral pneumothorax, pneumomediastinum	CPR initiated with chest compressions for 10 min before achieving ROSC; bilateral chest tubes, hypothermia protocol, prolonged ICU stay
12	First repeat CARD	No	Pneumothorax	Medical management, high FiO2 for improved resorption

TVF true vocal fold, ROSC return of spontaneous circulation

Three patients had zero successful CARD procedures after four total attempts (Supplemental Table 2). The failed CARD that was converted to a transcervical esophageal reconstruction resulted in substantial long-term improvements in swallow function such that the patient did not require any additional CARD attempts. The other two included the patient who experienced intraoperative arrest and was lost to follow-up upon discharge, and a patient with a long stricture that persisted despite two extensive attempts at dissection. Four other patients (28.6%, 95% CI 8.4%, 58.1%), despite at least one successful CARD, had either no improvement in FOIS score or a return to a baseline FOIS of 1 by the time of analysis.

The operative findings at the time of initial CARD attempt indicated eight patients (57.1%) had complete obliteration of the esophagus, confirmed visually during endoscopy, and the other six (42.9%) had an existing pinpoint lumen. Comparing the groups, complications per procedure and per patient were the only outcomes with a statistically significant difference (Table 6). All complications occurred in patients with complete stenosis (Fisher’s exact, p = 0.020 and p = 0.031). Swallow outcomes, including FOIS scores, maximum dilation achieved, and rates of g-tube independence, were all higher in patients with complete stenosis, although these differences were not significant. All patients with incomplete stenosis had at least one successful CARD, while three patients with complete stenosis had zero successful CARDs.

Table 6.

Outcomes of patients with complete vs incomplete stenosis

Outcome	Complete (n = 8)	Incomplete (n = 6)	p-value
Length of follow-up, weeks, median (range)	56 (2–643)	10 (0–413)	0.491
≥ 1 Successful CARD	5/8 (63%)	6/6 (100%)	0.209
Max dilation achieved, Fr, median (range)	40.5 (27–58)	39 (33–45)	0.515
FOIS scores, median (range)
Initial	1 (1–1)	1 (1–2)	0.312
Maximal	2.5 (1–7)	2(1–6)	0.643
End	2(1–7)	1.5 (1–6)	0.582
G-tube independence	2/8 (25%)	1/6 (17%)	1.000
Complication rate
Per procedure	6/14 (43%)	0/11 (0%)	0.020
Per patient	5/8 (63%)	0/6 (0%)	0.031

Discussion

Dysphagia is a common complication among head and neck cancer patients receiving radiation therapy, with a significant percentage requiring prolonged tube feeds and experiencing weight loss or aspiration pneumonia [22, 23]. Furthermore, dysphagia profoundly affects patients’ quality-of-life in multiple aspects and is associated with anxiety and depression [24]. While most strictures are managed by antegrade dilations, there is a subset of patients where antegrade dilation is not feasible due to complete stricture or inability to identify a pinpoint lumen.

Antegrade esophageal dilations are generally considered safe and effective, achieving symptomatic improvement in 85–93% of patients, with rates of esophageal perforation, the most common serious complication, ranging from 0.1 to 0.4% [25]. However, these studies typically included strictures of all etiologies, with a predominant majority of them caused by gastroesophageal reflux. Many of the same studies have suggested that radiation-induced strictures have higher recurrence rates, require frequent repeat dilations, and have higher rates of perforation [8, 25–28].

A meta-analysis and systematic review of esophageal dilations specifically in 449 patients with head and neck cancer noted a 72.9% overall success rate per patient, complication rates of 7.4% per dilation or 10.6% per patient, and a perforation rate of 5.4% per patient. While esophageal perforation accounted for approximately half of the observed complications, other complications included G-tube issues, pharyngo-cutaneous fistulas, bleeding and infection, and two deaths. This study also noted a significantly higher complication rate of 23.3% among patients undergoing CARD compared to 4.4% of patients undergoing standard antegrade dilations [29].

In our study, 79% of patients had at least one successful CARD procedure, 80% of CARD procedures were technically successful, and 23% of patients achieved G-tube independence during their treatment course. Approximately 64% of patients undergoing CARD had at least a temporary improvement in FOIS, and 54% of patients maintained long-term improvement compared to baseline. A number of other similar series of head and neck cancer patients undergoing CARD for radiation-induced esophageal stricture reported similar success rates of 83–87% of CARD procedures and G-tube independence in 23–60% of patients [9–13].

Our lower rate G-tube independence may, in part, be due to our lack of standardized protocol following a successful CARD. Performing regular antegrade dilations in the weeks and months immediately following a successful CARD may produce greater improvements in swallow function. Additionally, our maximum dilation achieved was smaller than certain cohorts, further supporting the importance of performing regular antegrade dilations following a successful CARD. Although we implemented a self-dilation program in half of our patients, compliance to the program was often poor or difficult to monitor. Given the severity of the strictures in these patients, consistently implementing a self-dilation program may be challenging. To achieve desired improvements in swallow function, a standardized protocol that includes frequent and regular operative dilations may be effective.

In our study, 24% of all CARD procedures had complications and 36% of patients had at least one complication from a CARD procedure, similar to the 23.3% reported in the meta-analysis. Though individual CARD studies varied in definition and reporting of complication rates, they noted complication rates that ranged from 11 to 29% of CARD procedures or 13–38% of patients [9, 11, 13]. One study reported three complications including one intraoperative death attributed to cardiac arrest caused by air embolism and two additional G-tube-associated complications that required exploratory laparotomies [9]. Another CARD study also observed a similar case of bilateral vocal fold paralysis requiring tracheostomy [12]. Nearly all of our complications required invasive intervention.

All complications in our study occurred in patients with complete stenosis. Given that other groups have reported longer strictures as a risk factor for complication [10, 12, 13], one plausible explanation is that extensive strictures lead to increasingly traumatic attempts at recanalization, which increases risk of unintentional perforation. This may explain the finding that one study had simultaneously the highest rate of g-tube independence, fewest dilations per patient, and the highest complication rate [10].

Given the concerns regarding complications, our study suggests caution is necessary for treatment of complete strictures with CARD. Post-operative CXR should be used to evaluate for signs of pneumothorax, a common complication of CARD. Additionally, patients should be advised that routine follow-up and repeat dilations are important components to achieving satisfactory outcomes.

One other group, Francis et al. performed a similar subgroup analysis of complete vs near-incomplete stenosis and found no significant difference in complication rates [12]. Whereas we exclusively used CARD for initial dilation, their initial approach to incomplete stenosis was to use antegrade endoscopy followed by balloon dilation unless resistance was encountered, in which case CARD was used. Because complication rate was reported as the percentage of patients experiencing any complication over the course of serial dilations, we could not discern the safety of balloon dilation versus CARD from their study.

Two alternative methods for approaching a near-complete stenosis include antegrade dilation with fluoroscopy assistance and, as previously mentioned, endoscopic balloon dilation. A number of other studies explored these two methods, typically with a 100% rate of successful recanalization [26, 27, 30–32]. However, these studies often included strictures of varying severity. Given our 100% success rate using CARD in near-complete stenosis, a direct comparison of the safety and efficacy of these methods may be warranted. We believe that when the lumen is particularly small, CARD’s retrograde approach may be safer and more reliable due to the ability to directly visualize the guidewire and both sides of the stricture.

Key limitations to our study include its retrospective nature, small sample size, and highly variable follow-up. The infrequent nature of CARD is at least partially due to these patients having significant comorbidities that preclude them from surgery. Similarly, these patients may be less likely to have consistent follow-up due to tumor recurrences, comorbidities that require prioritization, significant costs associated with their care, or other unspecified reasons. The lack of a standardized protocol following initial dilation also contributed to highly variable treatment courses that likely affected outcomes. Prospective studies or meta-analyses are warranted to overcome these challenges.