Control of inflammation decreases the need for subsequent esophageal dilation in patients with eosinophilic esophagitis

SUMMARY

It is unknown if successful control of esophageal inflammation in eosinophilic esophagitis (EoE) decreases the need for subsequent esophageal dilation. We aimed to determine whether histologic response to topical steroid treatment decreases the likelihood and frequency of subsequent esophageal dilation. We conducted a retrospective cohort study. Patients with an incident diagnosis of EoE were included if they had an initial esophageal dilation, received topical steroids, and had a subsequent endoscopy with biopsies. The number of dilations performed in each group was determined, and histologic responders (<15 eos/hpf) were compared to nonresponders. The 55 EoE patients included (27 responders and 28 nonresponders) underwent a mean of 3.0 dilations over a median follow-up of 19 months. Responders required fewer dilations than nonresponders (1.6 vs. 4.6, P = 0.03), after adjusting for potential confounders. Despite undergoing significantly fewer dilations, responders achieved a similar increase in esophageal diameter with dilation (4.9 vs. 5.0 mm; P = 0.92). In EoE patients undergoing esophageal dilation at baseline, control of inflammation with topical steroids was associated with a 65% decrease in the number of subsequent dilations to maintain the same esophageal caliber. This suggests that inflammation control is an important goal in patients with fibrostenotic changes of EoE.

INTRODUCTION

Eosinophilic esophagitis (EoE) manifests as two major phenotypes: inflammatory and fibrostenotic.^1–3 Inflammation in EoE is measured histologically by epithelial eosinophil counts on esophageal biopsy^4–6 and can be appreciated endoscopically by certain findings such as white plaques/exudates, linear furrows, and edema.^2,7–9 Fibrostenotic EoE can be observed on biopsy with collagen deposition in the lamina propria if subepithelial tissue is sampled,^10–12 and on endoscopy by strictures and narrowing.^2,7,13–15 Recent natural history data suggest that chronic inflammation can progress to esophageal remodeling and fibrosis if there is prolonged diagnostic delay with associated lack of treatment.^2,16–18

The first line pharmacologic treatments for EoE are topical corticosteroid medications, which are swallowed to coat the esophagus.¹ These medications act as anti-inflammatory/anti-eosinophilic agents. Both fluticasone and budesonide have been demonstrated to be effective in randomized control trials.^19–28 While these medications decrease esophageal eosinophilia, they appear to be less effective for resolving strictures, narrowing, and fibrosis, particularly in adults.^14,24,25,29 These fibrotic complications require separate mechanical treatment with esophageal dilation.^1,30,31 However, these endoscopic therapies add cost and potential risk to the procedure, as serial dilations are often required to achieve an esophageal caliber large enough to minimize symptoms of dysphagia.^13,32–36 While it is known that dilation treatment alone has no effect on the underlying eosinophilic inflammation,³⁷ it is unknown whether response to anti-inflammatory treatment with topical steroids in EoE decreases the need for subsequent esophageal dilation. Such data are vitally important to guide clinical practice.

The aim of this study is to determine whether histologic response to topical steroid treatment decreases the need for subsequent esophageal dilation in EoE patients with fibrostenotic features requiring baseline dilation. We hypothesized that patients with histologic response to topical steroid therapy would require fewer subsequent future dilations.

MATERIALS AND METHODS

Data source, case definitions, and outcomes

We conducted a retrospective cohort study using the University of North Carolina (UNC) EoE Clinicopathologic database from 2001 to 2014, for which details have previously been described.^38–41 In brief, this resource contains information on patients of all ages who had an incident diagnosis of EoE as per consensus guidelines.^5,6,42 Included patients were required to have symptoms of esophageal dysfunction (such as dysphagia, food impaction, heartburn, or feeding intolerance), an esophageal biopsy with at least 15 eosinophils in at least one high-power field (eos/hpf) after a high-dose 8-week trial of a proton pump inhibitor (PPI), and other causes of esophageal eosinophilia excluded. Patients diagnosed with EoE prior to the first consensus guidelines of 2007 were only retained if they were documented to have met all of the diagnostic criteria, including a PPI trial.

Patients were eligible for inclusion in the present study if they were an incident case of EoE (treatment naïve with the exception of a PPI), had an esophageal dilation performed at their index endoscopy for strictures and/or narrowing, then underwent treatment with topical steroids and had a subsequent upper endoscopy (EGD) with biopsies. Data were extracted using standardized forms and included demographics, presenting symptoms, endoscopic features (such as rings, strictures, white plaques/exudates, linear furrows, and edema), and histologic findings (peak eosinophil count, measured in eosinophils per high-power field [eos/hpf]; hpf area = 0.24 mm²). Dilation frequency and dilator sizes were abstracted from the electronic medical record. Patients underwent dilation at UNC as clinically indicated for symptoms of dysphagia in the setting of an esophageal stricture and/or narrowing. Either a through-the-scope balloon or a wire-guided bougie was used, as determined by the endoscopist. Esophageal lumen measurements were based on the size of the dilators used, measured in millimeters. The initial stricture caliber was defined as the first dilator size used at the time of index dilation, and the final stricture caliber was the dilator size used in the last procedure in our records.

After diagnosis of EoE, all included patients were treated with topical steroids at UNC as prescribed by their gastroenterologist at UNC for clinical purposes. At our institution, initial treatment was with either a total of 2 mg/day of a budesonide slurry or 1760 mcg/day of fluticasone dispensed via a multidose inhaler, for an 8-week course. At this time point, repeat endoscopy with biopsy was performed. We defined histologic responders as patients who had a peak eosinophil count of <15 eos/hpf on biopsy after their initial treatment, and non-responders as patients who had ≥15 eos/hpf post-treatment.⁴³

Statistical analysis

Descriptive statistics were used to summarize data for the study population. We compared histologic responders and nonresponders (after their initial treatment course) with bivariate analyses using Student's t-test for means, Wilcoxon rank sum for medians, and chi-square for proportions. Next, the total number of dilations performed in each group was determined, and we used logistic regression to calculate the odds of requiring subsequent dilation based on histologic response. We also used multiple linear regression to estimate the number of dilations predicated based on histologic response. We adjusted for potential confounding factors that differed between the responder and nonresponder groups at the P < 0.05 level, and removed factors that did not change the estimate with a backwards elimination strategy. Next, we repeated this analysis to examine rates of dilation (median dilations/month). Finally, we performed a Kaplan-Meier analysis to examine time to first dilation after the endoscopy assessing histologic response. Responders and nonresponders were compared with the log-rank test, and we performed a Cox proportional model to determine the hazard ratio for dilation, censuring patients either when they had a subsequent dilation or when they had no additional follow-up time. Patients with missing data were excluded from multivariate analyses. Statistical analysis was performed with Stata version 13 (Statacorp, College Station, TX). This study was approved by the University of North Carolina Institutional Review Board.

RESULTS

Patient characteristics

A total of 55 patients met study eligibility criteria of baseline esophageal dilation, topical steroid treatment, and at least one follow-up endoscopy. Of these, 28 (51%) were nonresponders to topical steroids and 27 (49%) were responders. In general, the responders and nonresponders did not differ by age, sex, or baseline symptoms (Table 1). There was also no significant difference in median length of follow-up time or mean number of follow-up clinic visits. At baseline, there were no differences in the typical endoscopic findings of EoE, in esophageal diameter, or in peak eosinophil counts prior to treatment (86 ± 67 eos/hpf in the nonresponders, and 79 ± 87 eso/hpf in the responders; P = 0.75) (Table 2). The mean initial doses of budesonide (72% of subjects) and fluticasone (28%) were 2105 ± 764 mcg and 1276 ± 553 mcg, respectively (Table 1). As expected after treatment, the eosinophil counts were significantly higher in nonresponders than responders (61 ± 30 vs. 3 ± 5 eos/hpf; P < 0.001). Some endoscopic findings were also more common post-treatment in the nonresponders, including rings (86% vs. 48%; P = 0.003) and furrows (82% vs. 30%; P < 0.001), but not strictures (39% vs. 41%; P = 0.70) (Table 2).

Table 1.

Characteristics of patients with treated eosinophilic esophagitis, comparing those with initial steroid response to those without initial steroid response^†

	All (n = 55)	Nonresponders (n = 28)	Responders (n = 27)	p*
Age at diagnosis (mean years ± SD)	35.4 ± 12.0	33.5 ± 12.6	37.5 ± 11.3	0.22
Symptom length prior to diagnosis (mean years ± SD)	12.7 ± 10.7	12.9 ± 11.2	12.5 ± 11.2	0.92
Males, n (%)	34 (62)	15 (54)	19 (70)	0.20
White, n (%)	49 (91)	24 (89)	25 (93)	1.0
Symptoms, n (%)
Dysphagia	53 (98)	26 (96)	27 (100)	1.0
Food impaction	27 (51)	14 (52)	26 (50)	0.89
Heartburn	19 (36)	7 (26)	12 (46)	0.13
Chest pain	9 (17)	3 (11)	6 (23)	0.29
Abdominal pain	5 (9)	2 (7)	3 (12)	0.67
Vomiting		3 (11)	8 (31)	0.10
Initial dose of topical steroid (mean daily mcg ± SD)
Fluticasone	1276 ± 533	1509 ± 429	1073 ± 593	0.13
Budesonide	2105 ± 764	2017 ± 498	2176 ± 1014	0.61
Follow-up time (median months, IQR)	18.6 (6-46)	20.9 (7-49)	18.4 (6-35)	0.66
Clinic visits during follow-up (mean number ± SD)	2.4 ± 2.2	2.5 ± 2.6	2.3 ± 1.8	0.69

*Steroid response defined as <15 eos/hpf after treatment.

^†Means compared with t-test, medians compared with Wilcoxon rank-sum test, and proportions compared with chi-square.

Table 2.

Baseline and post-treatment endoscopic and histologic findings

	All (n = 55)	Nonresponders (n = 28)	Responders (n = 27)	p*
Endoscopic findings, n (%)
Baseline
Rings	43 (78)	24 (86)	19 (70)	0.21
Furrows	36 (65)	21 (75)	15 (56)	0.13
Stricture	41 (74)	18 (64)	23 (85)	0.12
Narrowing	24 (44)	16 (57)	8 (30)	0.06
Crepe-paper mucosa	2 (4)	2 (7)	0 (0)	0.49
White plaques	20 (36)	8 (29)	12 (44)	0.22
Decreased vascularity	20 (36)	11 (39)	9 (33)	0.65
Post-treatment
Rings	37 (67)	24 (86)	13 (48)	0.003
Furrows	31 (56)	23 (82)	8 (30)	<0.001
Stricture	22 (40)	11 (39)	11 (41)	0.91
Narrowing	17 (31)	10 (36)	7 (26)	0.43
Crepe-paper mucosa	0 (0)	0 (0)	0 (0)	N/A
White plaques	15 (27)	7 (25)	8 (30)	0.70
Decreased vascularity	13 (24)	8 (29)	5 (19)	0.38
Peak eosinophil counts (mean eos/HPF ± SD)
Baseline	82.5 ± 76.6	85.8 ± 66.9	79.1 ± 86.7	0.75
Post-treatment	34.7 ± 37.0	61.1 ± 30.2	2.5 ± 4.5	<0.001
Esophageal diameter† (mean mm ± SD)
Before dilation	11.2 ± 3.4	10.8 ± 3.5	11.7 ± 3.3	0.35
After dilation	16.2 ± 2.5	15.8 ± 2.7	16.6 ± 2.1	0.19
Increase in esophageal diameter after dilation (mean mm ± SD)	4.9 ± 2.7	5.0 ± 2.8	4.9 ± 2.7	0.92
Number of total dilations needed, after treatment (mean ± SD)
Unadjusted	3.0 ± 4.3	4.1 ± 4.8	1.9 ± 3.4	0.05
Adjusted‡	–	4.6 ± 4.6	1.6 ± 4.6	0.03

*Means compared with t-test and proportions compared with chi-square.

^†Diameter reflects the initial dilator size used before dilation, and the last dilation size used for the final dilation on record.

^‡Adjustment for age, gender, presence of a focal stricture, initial esophageal diameter, and presence of dysphagia using multiple linear regression.

Effect of histologic response on number and rates of dilations

Overall, patients dilated at index endoscopy subsequently underwent a mean of 3 ± 4 dilations per patient over the median of 19 months of the study follow-up period (Table 2). After stratification by histologic response status, responders required less than half the absolute number of dilations after treatment than nonresponders (1.9 vs. 4.1; P = 0.05). This response was more prominent (1.6 vs. 4.6, P = 0.03) after adjusting for age, gender, presence of stricture, initial esophageal diameter, and presence of dysphagia. Despite undergoing significantly fewer dilations per patient, responders achieved a similar increase in esophageal diameter with dilation from the first to the last recorded procedure (4.9 vs. 5.0 mm; P = 0.92). The odds ratio for requiring a subsequent dilation in responders versus nonresponders was 0.14 (95% CI: 0.03–0.67) after adjusting for the same set of confounders. Overall, 13 of 27 responders (48%) did not require any dilation during the follow-up period, compared with 7 of 28 nonresponders (25%) (P = 0.07).

Similar results were seen when we examined rates of dilation. Responders required a median of 0.01 dilations/month (interquartile ratio: 0–0.16) while nonresponders required a median of 0.14 dilations/month (IQR: 0.01–0.21) (P = 0.10). The odds ratio for requiring an additional dilation per month for responders compared to nonresponders was 0.13 (95% CI: 0.03–0.63) after adjusting for age, gender, presence of stricture, initial esophageal diameter, and presence of dysphagia.

On Kaplan–Meier analysis, responders had a lower likelihood of requiring repeat dilation compared to non-responders (P = 0.04 by log-rank test) (Fig. 1). A Cox proportional hazards model, again adjusting for age, gender, presence of stricture, initial esophageal diameter, and presence of dysphagia, showed a lower likelihood of repeat dilation in responders as well (HR 0.48, 95% CI: 0.24–0.99).

Fig. 1.

Kaplan–Meier curve comparing time until first dilation after initial histologic assessment in responders (dashed line) and nonresponders (solid line). By the log-rank test comparing the two groups, P = 0.04.

DISCUSSION

EoE has both inflammatory and fibrostenotic components, requiring different treatments to target these manifestations.^1,2,4 Topical steroids decrease inflammation from esophageal eosinophilia, and esophageal dilation mechanically treats esophageal strictures and narrowing. However, it was not known whether suppressing inflammation with topical steroids would lead to a decreased requirement for dilation, particularly because strictures may not respond to topical steroids. We performed a retrospective cohort study to address this question. Our findings were consistent with our hypothesis, in that patients with an initial histologic response to topical steroids needed on average less than half the number of subsequent esophageal dilations despite starting with equal esophageal calibers, had greatly reduced odds of subsequent dilation, and had a longer time until another dilation was needed. These relationships also held after multivariate analyses accounting for potential confounders. These findings not only impact patients clinically in that medical therapy may halt or slow the progression of fibrosis, but provide a strong rationale for treating the inflammatory component of EoE. An approach that does not include anti-inflammatory treatment might ultimately accrue a higher number of endoscopic dilations, adding substantially higher costs and some additional attendant risks.

Why might control of inflammation lead to decreased need for future dilation? It has been demonstrated that in addition to local tissue injury, esophageal eosinophilia is profibrotic.^44,45 Eosinophils recruit and activate mast cells in the esophagus, and both of these cell types can produce TGF-β, which in turn activates fibroblasts and promotes epithelial to mesenchymal transition.^46–54 This manifests histologically as proliferative changes such as basal cell hyperplasia and lamina propria fibrosis on mucosal biopsy, and endoscopically with diffuse esophageal narrowing and focal strictures. However, EoE is likely a transmural process as evidenced by esophageal wall thickening on endoscopic ultrasound,^55,56 decreased compliance and distensibility as measured by the functional luminal imaging probe (FLIP),⁵⁷ and intramural collagen deposition on full thickness esophageal resection specimens.^58,59 Also important is the concept that unchecked inflammation progresses to fibrostenosis over time, which is supported by several studies showing increased stricture prevalence with longer duration of symptoms prior to EoE diagnosis.^2,16,18

Esophageal dilation is now considered an important therapy in the EoE armamentarium, which is a major shift from when EoE was first described and dilation was felt to represent a high-risk procedure.^60,61 Since that time, multiple studies have shown that dilation can be safely performed in EoE, with a perforation risk similar to that of dilation for other esophageal conditions.^{13,30,33–37,62,63} However, it has also been documented that dilation does not impact underlying inflammation in EoE.³⁷ Moreover, several clinical trials have shown that while endoscopic findings improve overall with topical steroid therapy, the improvements tend to be limited to esophageal exudates, edema, plaques, and rings; strictures have not generally improved with this treatment.^14,24,25,29 Interestingly, our current study also found that strictures did not resolve after anti-inflammatory therapy, highlighting the importance of combining this treatment with esophageal dilation when fibrosis is also present. While there has been one prospective trial comparing dilation and fluticasone treatment,⁶⁴ we are aware of no other study assessing long-term outcome results for dilation by initial histologic response.

This study has limitations. It was conducted at a single referral center in adults. While our population reflects a typical EoE population and our findings can likely be applied to patients in community practice, they may not be applicable to children. However, children are far less likely to present with fibrostenotic complications of EoE.^7,38 The sample size is also relatively small. However, this was in part due to the stringent inclusion criteria, which limited our overall large EoE population to just those patients who had esophageal dilation at baseline, were treated with topical steroids, and had at least on additional endoscopy for follow-up. Because the study was retrospective, dilation technique, frequency, and regular follow-up were not standardized, were at the clinical discretion of the provider, and could differ between groups during the follow-up period. All of these factors could impact on subsequent therapies or need for additional dilation. However, these limitations and subsequent treatments in initial nonresponders would be expected to bias towards the null, and we saw prominent results even with this potential heterogeneity of practice. The retrospective cohort design also yielded a minimum follow-up of 6 months with more than half the patients having a follow-up of 18 months or more, and follow-up times were similar between responder groups. Additionally, although we made substantial efforts to capture all of the care given to the patients, procedures or medications occurring external to UNC might not have been captured in our database. We also could not use an advanced technique such as FLIP to quantify esophageal compliance or formally measure esophageal caliber. Instead, we used initial and final dilator size as a proxy for stricture caliber. While this may introduce some inaccuracy in the absolute values for diameter, this should be nondifferential between the treatment responders and nonresponders, and it should not change the conclusion that histologic responders had an equal improvement in esophageal caliber despite requiring fewer dilations. We also did not have data on dysphagia severity or dietary behavior details such as careful eating or avoiding symptomatic foods. Finally, we only assessed treatment with topical steroids, so we cannot comment on whether the same results would be seen with dietary elimination. While we would hypothesize that any treatment that reduces inflammation would also decrease the need for subsequent dilation, this would need to be confirmed in future investigations.

There are also a number of strengths to this study, which is the first of its kind to assess long-term need for dilation by initial treatment response. The cohort design with strict eligibility requirements led to a homogenous study group specifically constructed to answer the research question. We had standardized and comprehensive data collection protocols for extracting information from the electronic medical record. We had a substantial follow-up period during which dilations could be assessed, and this time period was similar for both the treatment responders and nonresponders. Our results were also consistent across a number of analytic techniques, including examination of number of dilations, rates of dilation, and time to first dilation after response assessment.

In conclusion, this retrospective cohort study found that in EoE patients requiring esophageal dilation at baseline, histologic response to topical steroids resulted in a need for less than half as many dilations to achieve the same increase in esophageal caliber when compared to those without a histologic response to steroids. This result was further accentuated after accounting for possible confounding factors, and persisted on time-dependent analyses. Control of inflammation appears to halt or slow the progression of fibrosis and provides a strong rationale for treatment of esophageal eosinophilia. Moreover, the data suggest that suppression of inflammation to a level of <15 eos/hpf is an important goal in patients with fibrostenotic changes of EoE who require baseline dilation. An important implication of our study is that the combination of steroids and dilation may be able to increase the esophageal caliber as much as multiple additional dilations in the absence of inflammation control.