Background

Symptomatic benign esophageal strictures are considered “refractory” when there is a failure to maintain luminal patency after at least 5 endoscopic dilations [1]. Phenotypically, these refractory benign esophageal strictures (RBES) are often fibrotic and cicatricle. They are most commonly secondary to radiation, surgery, and the ingestion of caustic agents [2, 3].

Despite aggressive and rigorous dilations, 30 to 40% of RBES recur [4]. Recurrent hypertrophy of the fibrosis is the mechanism thought to underlie re-stenosis. Yet interventions to inhibit the fibrotic response to dilation or to promote re-modeling of fibrosis after dilation, including corticosteroid or mitomycin C injections, incisional therapy, and/ or temporary stent placement, offer limited benefit above dilation alone [5,6,7,8,9,10,11]. The largest study on the natural history of RBES, in which multiple techniques were used, found that the mean dysphagia-free interval was 3 months [2]. Whatever therapy is offered, patients with RBES continue to require endoscopic dilations, may progress to requiring enteral nutritional support, and may succumb to underlying co-morbidities or complications from repeated interventions [2].

Esophageal self-dilation therapy (ESDT) is a technique in which the patient learns to pass a polyvinyl dilator orally on a routine basis [3, 12,13,14,15,16,17,18]. The technique offers patients the potential for greater independence from endoscopic procedures. Although practiced since the 1970s, literature relating to this therapy is limited. The largest retrospective study described 52 patients and found that ESDT reduced the median number of endoscopic dilations from 9.5 (range 5–30) in the 12 months before to 0 (range 0–3) in the 12 months after self-dilation was commenced [3]. No prospective studies have evaluated the outcomes of ESDT. Consequently, it remains uncertain whether ESDT is effective and safe in the management of RBES.

We hypothesized that for patients with RBES: (1) ESDT would decrease the need for endoscopic dilation, (2) ESDT is non-inferior for treatment of dysphagia symptoms in comparison to standard endoscopic care, and (3) ESDT is safe. Secondary aims were to determine the length of intervention-free interval, change in dysphagia score, and evaluate factors which could influence clinical success of ESDT, including etiology and location of the stricture. We also aimed to assess the safety of ESDT by identifying clinically significant adverse events, such as perforation and bleeding.

Methods

Aware that the incidence of RBES is low, that prior studies have been limited by relatively small sample size, and that some trials have terminated recruitment due to slow accrual [19], we attempted to capture outcomes among as many patients with RBES as possible by offering participation in two concurrent studies: a randomized controlled trial (NCT03738566) and a prospective observational study. Both studies were approved by the Institutional Review Board at Mayo Clinic in Rochester and Arizona. All participants provided informed consent.

Patients

Patients with RBES presenting to the esophageal clinic at Mayo Clinic in Rochester MN or Mayo Clinic in Scottsdale AZ between November 2018 and June 2021 were offered participation in either study after completing standard clinical care consisting of serial endoscopic dilation under monitored anesthesia care using propofol in order to achieve an esophageal diameter ≥ 10 mm. RBES was defined as an esophageal stricture with persistent dysphagia despite undergoing 5 endoscopic dilations within a 1-year period. Patients who agreed to participate in the randomized trial were randomized to standard clinical care consisting of endoscopic dilation as needed versus esophageal self-dilation therapy in a 1:1 ratio using a block randomization approach. Patients who agreed to take part in the observational prospective study were invited to learn ESDT. Both studies recruited patients who were 18 years of age or older. Exclusion criteria included malignant esophageal strictures, angulated strictures which might prevent safe passage of a polyvinyl dilator, failure to achieve an esophageal diameter of ≥ 10 mm through endoscopic dilatation, inability to learn self-dilation secondary to blindness or cognitive dysfunction, or the use of chronic anticoagulants.

Study Procedures and Data Collection

Randomized Trial Group

Patients randomized to ESDT were taught ESDT within 24 h following endoscopic dilation achieving an esophageal diameter of at least 10 mm. ESDT training consisted of 1–3 in person training sessions by one of three esophageal physicians (MH, AK, JA) and an esophageal nurse using methods previously described [3, 18]. In brief, an esophageal physician and nurse collaborated to support the patient during training. After topical anesthesia is applied to the oropharynx, the patient is instructed to begin with the head in a neutral position. Once the polyvinyl dilator (Maloney Dilator) reaches the upper esophageal sphincter, the patient is instructed to swallow to facilitate passage of the dilator. Patients are reassured and encouraged to engage in relaxed breathing as this aids in suppressing the gag reflex. Once the dilator has reached the esophagus, the patient may extend their head and advance the dilator downward or maintain a neutral position and advance it hand-over-hand. The dilator can be passed as quickly as the patient is comfortable. Once the target depth has been achieved, marked with a tape on the dilator to give tactile feedback to the patient, the dilator is removed quickly. Patients were instructed to start ESDT twice a day for at least 1 week. If dysphagia was adequately controlled and there was no resistance with passing the dilator, patients were asked to decrease the frequency of ESDT to daily. All patients underwent close clinical supervision; follow-up was performed via weekly phone calls from a member of the esophageal care team. Repeat endoscopy and dilation was offered if patients reported dysphagia of severity sufficient to request endoscopic treatment. Patients in the standard clinical care group who required two further endoscopic dilations within 3 months of randomization were considered to have failed standard care and offered cross-over to the self-dilation therapy. This endpoint was selected by consensus among the study investigators. The Mayo Dysphagia Questionnaire (MDQ-30) was used at baseline and monthly for 6 months to measure dysphagia [20].

Prospective Observational Study

Patients who agreed to participate in the prospective observational study underwent identical training to participants in the randomized study. Participants were monitored via weekly phone calls by esophageal nurses and offered endoscopic dilation in an identical manner to patients in the randomized study. In the observational study, the number of endoscopic dilations in the 6 months prior to enrollment was compared with the 6-month period of self-dilation as the primary endpoint. The Mayo Dysphagia Questionnaire (MDQ-30) was used at baseline and monthly for 6 months to measure dysphagia [20]. Paired responses to the MDQ-30 from the first and final responses were evaluated.

Statistical Analysis

Descriptive data are presented as mean and standard deviations, medians and interquartile ranges (IQR), or proportions. Comparative analyses were performed with Mann–Whitney U tests, Chi-squared tests, or paired T-tests as appropriate. A two-tailed α < 0.05 was considered significant. Analyses were performed using JMP Pro Version 14.1.0 (SAS Institute, Cary, NC, USA).

Results

Outcomes of the Randomized Trial

Twelve patients participated in the randomized trial; 6 randomized to ESDT and 6 to standard clinical care. Baseline demographic and clinical characteristics are shown in Table 1. Because 5 patients in the standard of care group were allowed to cross over to self-dilation if they reached the predefined endpoint of two endoscopic dilations within 3 months of randomization, the median (IQR) length of follow-up was longer in the self-dilation group (5 [3,4,5,6] months versus 2 [1, 2] months, P = 0.008).

Table 1 Randomized study: demographic and clinical features
Full size table

Ability to Learn Self-Dilation

Four of the 6 patients (66%) randomized to ESDT were able to learn the technique and pass the dilator to adequate depth. All five patients (100%) who crossed over from standard care to ESDT per protocol were able to learn the technique.

Effect of ESDT on Need for Endoscopic Interventions

In an intention to treat analysis, patients randomized to self-dilation had a lower likelihood of requiring endoscopic dilation due to recurrent symptoms within 6 months of randomization (50% versus 100%, P = 0.02). The median (IQR) interval from randomization to first dilation was significantly longer in the self-dilation group (43 [39–124) days versus 7 [2–33] days, P = 0.03, Fig. 1). The median (IQR) number of procedures did not differ between the 2 groups (0.5 [0–3.5] versus 2 [2], P = 0.13). Among all participants in the randomized trial who undertook ESDT, the initial median (IQR) dilator size used was 14 mm (13–15). There was no difference in median dilator size achieved based on stricture location (P = 0.06) or on randomization cohort at study commencement (P = 0.3). In an intention to treat analysis, patients who successfully continued self-dilation achieved a median (IQR) dilator size of 14 (13–18) mm.

Fig. 1
figure 1

Kaplan–Meier curve demonstrating time to 1st endoscopic dilation in days in patients receiving standard of care (dashed line) or practicing ESDT among participants in the randomized trial (Solid line, P value for difference = 0.0005)

Full size image

A per-protocol analysis, excluding the 2 patients randomized to the ESDT cohort who were unable to perform the technique, demonstrated a lower likelihood of requiring endoscopic dilation (50% versus 100%, P = 0.03) and fewer procedures (median [IQR], 0.5 [0–2] versus 2 [2], P = 0.03) in the ESDT cohort in the 6 months after randomization. The median interval (IQR) from randomization to first dilation was longer in the ESDT cohort (82 [39–124] days versus 7 days [2–33] days, P = 0.06).

Indications for Endoscopic Evaluations and Interventions

Overall, among the 6 patients randomized to ESDT, a total of 11 EGDs were performed during the 6 months following enrollment. Eight of these occurred in a single patient with a long stricture caused by caustic injury who was unable to successfully learn to self-dilate during the study period. This patient eventually had a stent placed for 4 weeks and, following stent removal, was able to learn self-dilation and did not require any further EGDs for stricture management for over 6 months. One patient with an anastomotic stricture needed two needle-knife procedures and then resumed successful self-dilation without the need for subsequent endoscopic dilations. The final patient required one dilation after having ceased ESDT due to the need to commence anticoagulation therapy.

All patients randomized to standard of care required more than two or more endoscopic dilations due to recurrent dysphagia within 3 months of randomization. These patients were deemed to have failed standard therapy. Having reached this pre-determined endpoint, five of these patients were then offered, and elected to pursue, ESDT. The sixth patient withdrew from the study due to recurrent cancer. Four of the five patients who crossed over to ESDT completed 6 months of follow-up. The fifth patient developed an esophageal perforation and withdrew from the study. Of the four patients who completed follow-up, the outcomes were as follows. One patient did not require any endoscopic dilation during the follow-up period. One patient had one EGD within 3 days of starting self-dilation and then did not require any further EGDs during the follow-up period. One patient had two EGDs performed during the 6-month follow-up period; the esophagus was patent, no dilation was performed, and the dysphagia which triggered the endoscopy was eventually determined to be oropharyngeal in nature. In the final patient, who had a diffusely narrowed esophagus secondary to esophageal lichen planus, two EGDs with dilation were performed shortly after starting self-dilation and then none in the subsequent 6 months.

Effect of ESDT on Dysphagia

In order to reduce the probability of statistical overreach (i.e., type 1 error) in the analyses of the small number of patients recruited, only two questions from the MDQ-30 were analyzed. These were “In the past 30 days, have you had trouble swallowing, not associated with other cold symptoms (such as strep throat or mono)? 1 = Yes and 2 = No,” and “How would you rate the severity of your trouble swallowing over the past 30 days? 1 = Doesn't bother me at all 2 = Mild 3 = Moderate 4 = Severe 5 = Very severe 6 = I don't know.” In the latter question, no patient responded, “I don’t know.” At baseline, all patients were experiencing dysphagia and consequently needed dilation. There was no difference in the severity or frequency of symptoms experienced by patients in the two groups. In the last completed questionnaire, either at the end of the study or prior to being deemed a “standard-care treatment failure,” a greater proportion of patients in the self-dilation group reported no trouble with swallowing in the ITT (66% versus 0%) and per-protocol (75% versus 0%) analysis with further details in Table 2

Table 2 Effects of self-dilation therapy on the MDQ among Randomized Patients
Full size table

Adverse Events and Withdrawals

Two patients (one in each group) experienced a recurrence of their laryngeal cancer, one of whom succumbed from complications following total laryngectomy. One patient who crossed over to ESDT developed esophageal perforation with an associated tracheoesophageal fistula within days of commencing self-dilation. It was unclear whether the injury occurred during the last endoscopic dilation after which the patient reported more pain than usual or from initiation of self-dilation as both occurred on the same day. The patient, who had received radiation therapy twice, 10 years apart, was then placed on enteral nutrition and died unexpectedly of an unknown cause 3 months later. No bleeding occurred because of self-dilation in any patient.

Outcomes of the Observational Study

Fourteen patients were offered participation in the prospective observational study. Thirteen patients were able to perform the technique of self-dilation and participate. Baseline demographic and clinical characteristics are shown in Table 3.

Table 3 Observational study: demographic and clinical features
Full size table

Effect of ESDT on Need for Endoscopic Interventions

Fewer endoscopic dilations were performed in the 6 months after ESDT compared to the 6 months prior (median [IQR], 1 [0–2] versus 7 [7,8,9,10], P < 0.001). In the 6 months after ESDT, 6 patients had complete relief of dysphagia and did not require endoscopic dilations. Among the 7 patients who required endoscopic dilation for recurrent dysphagia, the median (IQR) time to first dilation was 46 (32–59) days and the median (IQR) number of procedures was 2 (2–2). Two patients experienced tumor recurrence and ESDT was discontinued. There was no difference in age, gender, or maximum dilator sized reach in those who did and did not require endoscopic dilation. Patients with a proximal esophageal stricture achieved a lower dilator size that those with a distal stricture (median [IQR], 12 mm [10,11,12,13] versus 14 mm [14,15,16], P = 0.018).

Effect of ESDT on Dysphagia

Ten patients had dysphagia questionnaires available for analysis both from recruitment and from follow-up. Among these, the median (IQR) length of follow-up was 146 (61–195) days. At recruitment, 90% of responders reported difficulty swallowing. In the last available questionnaire, 50% reported difficulty. The difference in dysphagia severity between recruitment and last available data was significant (Median [IQR], 4 [3, 4] versus 1.5 [1,2,3], P = 0.01).

Adverse Events and Study Withdrawals

Two patients withdrew due to recurrent cancer. No patient in the observational study experienced esophageal perforation or bleeding.

Discussion

For many patients with benign esophageal strictures, endoscopic interventions may be required frequently, are of limited efficacy, are associated with risk, and come with significant burden [2]. A safe, effective treatment which concurrently promotes patient independence is required. ESDT potentially offers this [3, 18]. However, prospective data evaluating ESDT for the management of RBES have been lacking. The two prospective studies presented in this paper, one prospective randomized and one prospective observational, examined the outcomes of ESDT for the treatment of RBES. The main findings from the studies are that for patients with RBES, (1) most patients can perform the technique of ESDT, (2) ESDT is effective in reducing the need for endoscopic intervention, (3) ESDT is effective in managing dysphagia, presumably by maintaining luminal patency, and (4) ESDT is relatively safe.

The primary aim of these studies was to evaluate the effect of ESDT on the need for endoscopic dilation therapy for patients with RBES. Although limited by small sample sizes, both the randomized and the prospective study clearly demonstrated that ESDT reduces the likelihood of requiring endoscopic dilation. Although there was a high intervention rate in the randomized study, the majority of these patients (4/6) simply needed additional procedures to achieve adequate luminal patency; once this was achieved, ESDT provided adequate symptomatic relief during the follow-up period. This may relate to the Mayo model of teaching patients self-dilation in the 3 days after achieving minimal luminal patency. Other groups offering ESDT have taught patients the technique, with a small caliber dilator, while progressive dilations were occurring [16]. Thus, when adequate luminal patency was achieved, the patients were immediately capable of maintaining patency with ESDT. Unfortunately, some patients who experience recurrent dysphagia after ESDT had tumor recurrence. Our experience suggests that when previously successful ESDT begins to fail in patients with a stricture of cancer-related etiology, urgent endoscopy should be performed with concern for a recurrent malignant process.

Regarding secondary aims, the interval to first endoscopic dilation was substantially longer in those practicing ESDT than those receiving standard care. Strengthening this finding, the time to first endoscopic intervention was almost identical in patients randomized to ESDT and in practicing ESDT as part of the observational study.

Moreover, because self-dilation maintains luminal patency and prevents re-stenosis, ESDT relieved dysphagia and improved dysphagia scores in both studies. These findings are consistent with prior retrospective studies [3, 16]. However, even had ESDT had no effect on dysphagia scores, the reduction in the need for endoscopic interventions in itself would have been a clinically meaningful outcome.

In our observational study, ESDT also appeared safe and effective. In particular, the 13 patients who self-dilated required 14 EGDs cumulatively in the 6 months during therapy compared to 102 in the 6 months prior. These are remarkable outcomes given the challenging nature of treating RBES in patients who often have co-morbidities and are at risk of cancer recurrence.

Nevertheless, treating RBES, whether with endoscopic therapies or ESDT, is not without risk. Among 24 patients who were exposed to ESDT, one patient developed an esophageal perforation with tracheoesophageal fistula. However, overall, our complication rate appears significantly lower than among patients who undergo repeated EGDs with dilations/stents or other advanced endoscopic therapies [2, 11].

Strengths of our studies include the randomized design coupled with a prospective observational study to capture patients who chose not to participate in the RCT. We assessed esophageal diameter and used a validated questionnaire for measurement of dysphagia and followed patients closely. However, we also acknowledge further limitations. Despite being a tertiary referral center and using two study sites, accruing patients with RBES was extremely challenging. This limitation has been described by other groups studying RBES [19]. During our study period, the COVID-19 pandemic further hampered recruitment. Consequently, the sample sizes were small. To mitigate this limitation, we performed comparative analyses on only 2 questions from the MDQ-30. Finally, the extensive expertise and established clinical protocol of teaching ESDT in our practices is a great strength of the study, but limits generalizability of these outcomes to other centers with less expertise in this area.

In summary, ESDT appears to be a safe and effective tool in the armamentarium to help patients with RBES. With outcomes in several aspects superior to ongoing endoscopic dilation, the potential role of ESDT should be widely considered a viable strategy among providers who care for patients with RBES.