Heartburn and dyspepsia symptom severity improves after treatment and correlates with histology in adults with eosinophilic esophagitis

SUMMARY

Non-dysphagia symptoms, such as heartburn and dyspepsia, are poorly characterized in adults with eosinophilic esophagitis (EoE). It is unclear if treatment improves these symptoms. The aim of this paper was to assess (i) heartburn and dyspepsia symptom severity in adult EoE patients using validated symptom measures; (ii) change in symptoms after treatment; and (iii) symptom association with endoscopic and histologic features. In a prospective cohort of adult EoE patients who were not responsive to proton pump inhibitor therapy, non-dysphagia symptoms were assessed with heartburn items from the validated GERD-HRQL (gastroesophageal reflux disease health-related quality of life) and SODA (severity of dyspepsia assessment) instruments. Subjects completed the questionnaires at baseline and after treatment. Association of baseline symptoms with endoscopic and histologic features, and before and after treatment with diet or topical steroids, was assessed. Eighty-six EoE patients (mean age 39 years, 57% male, 95% white) completed a baseline questionnaire and 62 completed the follow-up questionnaire. The mean baseline GERD-HRQL score was 4.5 ± 6.5 and the mean total SODA score was 41.0 ± 12.6. At baseline, there was a weak but significant correlation between peak eosinophils and the SODA score (r = 0.28; p = 0.03) and no association between heartburn and SODA scores and endoscopic or other histologic findings. After treatment, there was a decrease in GERD-HRQL heartburn (4.3 vs. 2.6; p = 0.04) and SODA (49.5 vs. 35.5; p = 0.04) scores in histologic responders, but not in nonresponders. In a prospective cohort of EoE patients, baseline eosinophils positively correlated with dyspepsia severity. Heartburn and dyspepsia symptoms improved after treatment in histologic responders.

INTRODUCTION

Eosinophilic esophagitis (EoE) is a chronic immune-and antigen-mediated condition characterized by esophageal inflammation resulting in symptoms of esophageal dysfunction.¹ While dysphagia and food impaction are the most common presentations in patients with EoE,^2,3 especially in adults, other esophageal symptoms, such as heartburn, chest pain, and regurgitation, are also reported. In a European multicenter study, 33% of EoE patients reported symptoms of reflux, retrosternal pain, or odynophagia,⁴ and other studies have ranged in reporting reflux symptoms in this group from 9% to 94%.^5–7 While symptoms other than dysphagia are commonly factored into assessing symptom response in pediatric EoE patients,^8–10 nondysphagia symptoms are poorly characterized in adult EoE patients.

Further understanding of nondysphagia symptoms in adults is important as it is unclear if current treatments improve these clinical parameters in addition to dysphagia. A majority of existing patient-reported outcomes and disease activity indices for adults with EoE, such as the EoE activity index¹¹ and the dysphagia symptom questionnaire,¹² primarily focus on dysphagia,¹³ whose severity can be confounded by the consistency of the food ingested as EoE patients adapt their eating habits to progressive esophageal fibrosis. In this setting, dysphagia symptoms have only moderately correlated with other measures of disease activity.^14,15 In addition to dysphagia, integrating common esophageal symptoms in EoE, such as heartburn, reflux, or dyspepsia, to assess symptom severity and treatment response can potentially add value in providing a more accurate measure of EoE symptom burden and better correlation with histologic and endoscopic disease activity. Assessment of these symptoms may also better characterize patients who have improvement in esophageal eosinophilia but who have persistent gastrointestinal complaints.

The specific aims of this study were to assess heartburn and dyspepsia symptom severity in adult EoE patients using validated symptom measures, evaluate whether baseline symptom scores are associated with baseline endoscopic and histologic features, and determine if there is a change in symptom severity following EoE treatment.

MATERIALS AND METHODS

Study design, patients, and data collection

We analyzed data collected during a prospective study conducted at the University of North Carolina (UNC) from 2009 to 2015, and full study details have been previously reported.^16–21 Participants were consecutive adults scheduled to undergo outpatient upper endoscopy for symptoms of esophageal dysfunction, such as dysphagia, heartburn, reflux, chest pain, and food impaction. Participants were enrolled prior to undergoing endoscopy and were excluded if they had a known diagnosis of EoE or eosinophilic gastrointestinal disorder, esophageal varices, prior esophageal surgery, bleeding or on anticoagulation, medical instability, or multiple comorbidities.

EoE cases were diagnosed per consensus guidelines,^1,22 with at least 15 eosinophils per high-power field (eos/hpf) on esophageal biopsy after an 8-week course of high-dose proton pump inhibitor (PPI) therapy and exclusion of alternate causes of esophageal eosinophilia. Because the study was designed and conducted from 2009 to 2015, during which time failure of a PPI trial was necessary to establish a diagnosis of EoE, subjects with PPI-responsive esophageal eosinophilia were excluded. Therefore, the data presented in this study are from the endoscopy that was performed to confirm a diagnosis of EoE while on a high-dose PPI. The study was approved by the UNC Institutional Review Board.

Clinical data were prospectively collected using standardized case report forms as outlined previously.^16–18 Current and prior use of PPIs, H2-receptor antagonists, antihistamines, inhaled and systemic corticosteroids, leukotriene antagonists, mast cell stabilizers, and nonsteroidal anti-inflammatory medications was collected, but these were not specifically prohibited. Because there was no validated system to measure endoscopic severity at the time of study design, endoscopic findings were recorded as absent or present: white plaques/exudates, rings, decreased vascularity/edema, linear furrows, strictures, narrowing, and crepe-paper mucosa. A subset of more recently enrolled patients had endoscopic findings graded according to the EoE endoscopic reference score (EREFS) system.²³ All endoscopic biopsies were obtained per research protocol. Two biopsies from the proximal esophagus, one from the mid-esophagus, and two from the distal esophagus were collected. Additionally, gastric and duodenal biopsies were collected to exclude more diffuse eosinophilic disorders. From each biopsy, the maximum eosinophil count (eosinophils per high-power field [eos/hpf]) was quantified using a microscopic field size of 0.24mm² and according to previously described methods.²⁴

Nondysphagia symptom assessment

Nondysphagia symptom assessment was performed with validated instruments that were available at the time of study design: Heartburn severity was measured using the gastroesophageal reflux disease health-related quality of life (GERD-HRQL) instrument,^25,26 and abdominal pain severity was quantified using the severity of dyspepsia assessment (SODA) instrument.^27–30 The heartburn-specific items from the GERD-HRQL were utilized for the purposes of this study and were scored from 0 to 30, with ‘0’ indicating no heartburn symptoms and ‘30’ indicating the worst heartburn symptoms. SODA, a multidimensional instrument that is responsive to clinically meaningful change in symptoms, measures three domains of dyspepsia: pain intensity, nonpain symptoms, and satisfaction with dyspepsia-related health. SODA scores range from 11 to 105, with higher scores indicating more severe symptoms. Participants completed symptom questionnaires at baseline and after treatment with diet or an 8-week course of either oral viscous budesonide (1 mg twice a day) or fluticasone from a multidose inhaler (880 mcg twice a day) at the discretion of the clinical provider.³¹

Statistical analysis

We used descriptive statistics to summarize the demographic, endoscopic, and histologic characteristics of the participants. A subanalysis was performed for patients who had histologic findings graded per the EREFS scoring system. The mean heartburn severity and the SODA scores were calculated at baseline, and mean scores were compared at baseline and after treatment using a paired t-test stratified by histologic response defined as <15 eos/hpf.³² Association between mean heartburn and SODA severity scores and baseline endoscopic and histologic features were assessed using a 2-sample t-test and Kruskal–Wallis one-way analysis of variance, as appropriate. Finally, dyspepsia and heartburn scores were correlated with eosinophil density using Spearman's correlation.

RESULTS

There were 86 PPI-non-responsive EoE patients who completed the baseline symptom questionnaire and were included for analysis (Table 1). Mean age was 38.8 ± 12.8 years, 57% were men, and 97% of the patients were treated with topical steroids versus dietary therapy. At the time of baseline endoscopy after enrolling in the study, the mean baseline GERD-HRQL score was 4.5 ± 6.5 and the mean total SODA score was 41.0 ± 12.6. PPI doses used at baseline were the equivalent of at least omeprazole 20 mg twice daily, but during the follow-up period PPI dosing was variable and at the discretion of the treating clinical provider.

Table 1.

Baseline clinical characteristics and symptom scores

	Sample (n = 86)
Age (mean years ± SD)	38.8 ± 12.8
Male, n (%)	49 (57)
White, n (%)	82 (95)
Body mass index (mean ± SD)	25.5 ± 4.6
PPI therapy, n (%)	80 (93)
Any atopy, n (%)	64 (80)
Rhinitis/Sinusitis	64 (80)
Asthma	62 (72)
Eczema	28 (33)
Food allergies	8 (9)
Endoscopic findings, n (%)
Rings	73 (85)
Stricture	23 (27)
Narrowing	28 (33)
Furrows	75 (87)
Exudates/White plaques	43 (50)
Edema/Decreased vascularity	43 (50)
Dilation	25 (30)
Hiatal hernia	11 (13)
Erosive esophagitis	2 (3)
Baseline peak eosinophil counts (mean eos/hpf ± SD)	128.2 ± 118.6
Baseline histologic features, n (%)
Degranulation	77 (95)
Microabscesses	54 (67)
Basal zone hyperplasia	37 (47)
Spongiosis	70 (84)
Lamina propria fibrosis	18 (35)
Symptom scores (mean ± SD)
GERD-HRQL (GERD health-related quality of life) (range 0–30)	4.5 ± 6.5
SODA (severity of dyspepsia assessment)
Pain intensity (range 2–47)	17.1 ± 8.5
Non-pain symptoms (range 7–35)	12.9 ± 4.8
Satisfaction with dyspepsia-related health (range 2–23)	11.2 ± 2.1
Total score (range 11–105)	41.0 ± 12.6
Treatments, n (%)
Diet, n (%)	3 (3)
Topical corticosteroid, n (%)	83 (97)
Budesonide, n (%)	71 (86)
Mean dose (mcg ± SD)	2107 ± 664
Fluticasone, n (%)	12 (14)
Mean dose (mcg ± SD)	1707 ± 269

Correlation with baseline endoscopic and histologic findings

At baseline, the most common endoscopic findings were furrows (87%) and rings (85%). Half of the patients had exudates and edema, and about one-third had esophageal stricture or narrowing at baseline. Mean baseline peak eosinophil counts were 128.2 ± 118.6 eosinophils per high-power field (eos/hpf) prior to treatment and 35.1 ± 65.3 eos/hpf after treatment. Mean peak baseline eosinophil counts were 88.4 ± 108.0 eos/hpf in the distal, 63.1 ± 102.4 eos/hpf in the mid-, and 80.0 ± 80.7 eos/hpf in the proximal esophagus. Most common histologic findings were degranulation (97%), spongiosis (84%), and microabscesses (67%). None has gastric or duodenal eosinophilia. There was no association between location of peak baseline eosinophilia in the esophagus and SODA and GERD-HRQL scores, p = 0.56 and p = 0.41, respectively. There was a positive but weak correlation between the baseline peak eosinophil count (eos/hpf) and total SODA score (r = 0.28; p = 0.03), specifically in the pain intensity domain (r = 0.27; p = 0.04). There was no correlation with baseline peak eosinophil counts and heartburn severity score (r = 0.01; p = 0.94). There were no significant associations between heartburn and SODA scores and the presence or absence of individual endoscopic and other histologic findings at baseline (Fig. 1). Out of the 86 EoE patients, 55 had histologic findings graded according to the EREFS scoring system. In this group, baseline findings included: exudates (55%), rings (82%), edema (59%), furrows (89%), and strictures (27%). The mean EREFS score was 4.2 ± 2.1. In this subgroup, there was a positive correlation between baseline EREFS score and total SODA score (r = 0.47; p = 0.02). There was no correlation between EREFS and GERD-HRQL scores (r = 0.02; p = 0.85).

Fig. 1.

Mean baseline GERD-HRQL heartburn severity and SODA (severity of dyspepsia assessment) scores based on baseline endoscopic findings and baseline histologic findings. GERD-HRQL findings are displayed in panel (A), and SODA findings are displayed in panel (B). SODA scores range from 11 to 105, with higher scores indicating more severe symptoms. GERD-HRQL heartburn severity scores range from 0 to 30, with ‘0’ indicating no heartburn symptoms and ‘30’ indicating worst heartburn symptoms.

SODA and heartburn severity scores before and after treatment

Among the 86 EoE patients, 62 completed the follow-up questionnaire. Of these, 3 (5%) were treated with dietary therapy and 59 (95%) were treated with topical corticosteroids. The mean post-treatment peak eosinophil count was 35.1 ± 65.3 eos/hpf. There was no correlation between follow-up peak eosinophil counts and total SODA score (r = 0.22; p = 0.20) or heartburn severity score (r = 0.09; p = 0.49). Similarly, there was no correlation between post-treatment EREFS score and SODA (r = 0.16; p = 0.55) and GERD-HRQL (r = 0.04; p = 0.82) scores. Forty-four of the 62 (71%) had histologic response (<15 eos/hpf) after treatment with a peak eosinophil count of 2.0 ± 3.6 eos/hpf versus 114.3 ± 88.5 eos/hpf in the post-treatment group not achieving histologic response. In the histologic responders, there was a decrease in both mean GERD-HRQL heartburn (4.3 ± 6.5 vs. 2.6 ± 4.8; p = 0.04) and total SODA score (39.9 ± 12.0 vs. 35.5 ± 11.6; p = 0.04) from baseline to after treatment (Table 2). In the histologic non-responders, there was no decrease in GERD-HRQL heartburn (5.2 ± 6.5 vs. 3.1 ± 4.5; p = 0.18) or change in SODA score (41.7 ± 12.1 vs. 45.1 ± 12.5; p = 0.13).

Table 2.

Baseline and posttreatment GERD-HRQOL heartburn and SODA dyspepsia scores before and after treatment for EoE in histologic responders (<15 eos/hpf) and non-responders (≥15 eos/hpf)

	Baseline	Post-Treatment
EoE cases with histologic remission posttreatment	(n = 44)	(n = 44)	P-value
GERD-HRQL (GERD health-related quality of life)
Heartburn score (range 0–30)	4.3 ± 6.5	2.6 ± 4.8	0.04
SODA (severity of dyspepsia assessment)
Pain intensity (range 2–47)	16.9 ± 8.2	14.3 ± 6.7	0.06
Non-pain symptoms (range 7–35)	12.4 ± 4.1	11.3 ± 4.3	0.04
Satisfaction with dyspepsia-related health (range 2–23)	11.0 ± 1.3	10.8 ± 2.2	0.42
Total score (range 11–105)	39.9 ± 12.0	35.5 ± 11.6	0.04
	Baseline	Posttreatment
EoE cases without histologic remission posttreatment
	(n = 18)	(n = 18)	P-value
GERD-HRQL (GERD health-related quality of life)
Heartburn score (range 0–30)	5.2 ± 6.5	3.1 ± 4.5	0.18
SODA (severity of dyspepsia assessment)
Pain intensity (range 2–47)	17.0 ± 8.2	19.7 ± 8.0	0.06
Non-pain symptoms (range 7–35)	13.5 ± 4.8	12.4 ± 4.5	0.28
Satisfaction with dyspepsia-related health (range 2–23)	11.6 ± 1.7	10.6 ± 2.3	0.28
Total score (range 11–105)	41.7 ± 12.2	45.1 ± 12.5	0.13

DISCUSSION

Very few studies that have used validated symptom measures have assessed non-dysphagia symptom profiles in adult EoE patients. Therefore, in this study, we aimed to characterize heartburn and dyspepsia severity using validated GERD and dyspepsia instruments in a cohort of adult EoE patients and determine if non-dysphagia symptom severity correlated with baseline endoscopic and histologic features or treatment response. We found that the mean eosinophil count before treatment had a weak positive correlation with increased dyspepsia severity and higher SODA scores. There was no correlation with nondysphagia symptom severity and baseline endoscopic and histologic features, although, in the subset of patients with standardized reporting of endoscopic findings with EREFS, an increased dyspepsia score correlated with a higher EREFS score. Additionally, EoE patients with successful histologic response (<15 eos/hpf) had a decrease in SODA scores reflecting decreased dyspepsia from baseline, compared to the treatment nonresponsive group. GERD-HRQL scores were low at baseline, and this is likely due to the EoE cases being on PPI therapy at baseline to confirm the EoE diagnosis. There was a mild decrease in GERD-HRQL scores in the histologic responders compared to non-responders. Our finding that dyspepsia severity correlates with histologic response suggests that it may be able to serve as an adjunctive symptom marker to dysphagia when assessing histologic disease activity and treatment response.

Multiple studies in children with EoE evaluating abdominal pain symptoms have shown that abdominal pain persists despite histologic response after treatment,^8,10,33–35 with one study⁸ demonstrating that eosinophil count directly correlated with improvement in dysphagia and anorexia/early satiety but not with abdominal pain. The results from these pediatric studies differ from our results in an adult EoE population, which shows that there is a correlation between baseline eosinophil count and dyspepsia severity as well as a decrease in dyspepsia severity with histologic response. One potential explanation for the difference in results could be that in the pediatric studies, abdominal pain/dyspepsia was primarily reported as a dichotomous outcome or as mild/moderate/severe, in contrast to the SODA instrument used in this study. SODA is a multidimensional instrument that is not only reliable and valid, but also responsive and detects clinically meaningful symptom changes.²⁷ A pediatric study by Martin et al. supports this theory. Using a content-validated metric (PEESS^® v2.0) to capture EoE-specific symptoms, the authors found that eosinophil levels were associated with pain rather than dysphagia.¹⁰ Another potential explanation is that there may be an overlapping component of functional abdominal pain that is more predominant in children with EoE given a higher prevalence of psychosocial issues in this group.³⁶ There are limited non-dysphagia symptom data in adults with EoE to help contextualize our findings. In an abstract report, pain with swallowing was found to be common, and there was a strong trend toward improvement of odynophagia after topical steroid therapy.³⁷

Mechanistically, contents of eosinophil granules such as major basic protein can increase smooth muscle reactivity, leading to pain.³⁸ Eosinophils also play a role in the pain pathway, leading to visceral sensitization,^39,40 and can also potentiate smooth muscle contraction, causing upper gastrointestinal symptoms. Therefore, it is possible to speculate that an eosinophil-mediated pathway can contribute to symptoms of dyspepsia and abdominal pain and that these symptoms can potentially be used as a correlate to histologic disease severity, if measured with a valid symptom severity measurement such as SODA. Prior studies have shown that there can be a disassociation between symptoms and histologic response in EoE,^41–43 and part of this could be due to utilization of unvalidated symptom measurement indices, or focusing primarily on dysphagia and swallowing discomfort without accounting for dietary modification, severity of underlying strictures, or performance of esophageal dilation.^14,15 Incorporating nondysphagia symptom outcomes such as dyspepsia may have the potential to provide a more multifaceted symptom measurement that corresponds to eosinophilic inflammation.

This study has some limitations. This was a single-center study, which can limit the generalizability of the results. There was also some loss to follow-up with 62 of the initial 86 EoE subjects completing the posttreatment symptom questionnaire. However, we do not think this affected the study results as only the 62 subjects with pre- and post-treatment data were included in the treatment response analysis. In addition, approximately a third of the EoE patients did not have endoscopic severity graded according to the EREFS system because the study initiation predated the scoring system, and if we had more granular endoscopic data on all the cases, a correlation between histologic findings and symptom severity may have been detected. Baseline GERD-HRQL scores were low likely due to a majority of the subjects being on PPI therapy, making the difference in scores after treatment small in a clinical context. The study also has multiple strengths. It is one of the few studies in adults assessing non-dysphagia symptoms using validated symptom measurement instruments. Data and biospecimens were collected prospectively with a standardized case report form and protocols for endoscopies and biopsies resulting in well-characterized clinical, endoscopic, and histologic data.

In summary, in a prospectively assessed cohort of EoE patients, there was no correlation between non-dysphagia symptoms and baseline endoscopic findings. We did find a weakly positive correlation between baseline eosinophil count and dyspepsia severity, which decreased after treatment with dietary elimination or topical steroids in the group with histologic response versus the non-responders. There was not a clinically significant change in heartburn severity after treatment in those with histologic response. Our study suggests that nondysphagia symptoms may improve with treatment. Therefore, including nondysphagia symptoms when developing EoE patient reported outcomes and symptom measurement instruments in adults may have some utility in correlating with histologic disease activity in EoE. Moreover, it is likely beneficial to elicit and monitor non-dysphagia symptoms to fully assess clinical response in adult patients with EoE after treatment.

Guarantor of the article: Evan S. Dellon.

Specific author contributions: Study design, data interpretation, manuscript drafting, critical revision: Swathi Eluri; Data acquisition, critical revision: Irina Perjar, Renee Betancourt, Cara Randall, Soha Raja; Study design, data interpretation, critical revision: John T. Woosley; Project conception, study design, data interpretation, critical revision: Nicholas J. Shaheen; Project conception, study design, data interpretation, manuscript drafting, critical revision: Evan S. Dellon; all authors approved the final draft submitted.

Financial support: ESD has received research funding from Adare, Allakos, Meritage, Miraca, Nutricia, Celgene/Receptos, Regeneron, and Shire; and has received educational grants from Banner and Holoclara. The other authors report no potential related competing interests.

Potential competing interests: ESD has consulted for Adare, Alivio, Allakos, AstraZeneca, Banner, Enumeral, EsoCap, Calypso, Celgene/Receptos, GSK, Regeneron, Robarts, and Shire.