Mast Cell Infiltration Is Associated With Persistent Symptoms and Endoscopic Abnormalities Despite Resolution of Eosinophilia in Pediatric Eosinophilic Esophagitis

Abstract

OBJECTIVES:

Mast cells (MCs) are increased in eosinophilic esophagitis (EoE). Endoscopic abnormalities, symptoms, and epithelial changes can persist after treatment despite a reduction of esophageal eosinophilia. It is unknown whether this could be due to persistent MC infiltration. We aimed to determine whether patients with histologically inactive (HI) EoE (defined as <15 eosinophils per high-powered field) with persistent symptoms, endoscopic, or epithelial abnormalities after treatment have increased MCs.

METHODS:

Secondary analysis of prospective data from 93 children with EoE undergoing post-treatment endoscopy between 2011 and 2015. Thirty-five non-EoE controls were included. Immunohistochemistry for tryptase, an MC marker, was performed on mid and distal esophageal biopsies. Total and degranulated intraepithelial MCs per high-powered field (MC/hpf) were quantified. Symptoms and endoscopic findings were recorded at time of endoscopy. MC/hpf were compared between HI-EoE and control, and among HI-EoE based on endoscopic and histologic findings, and symptoms. Nine clinical remission (CR) patients were identified, with absence of endoscopic abnormalities and symptoms.

RESULTS:

MC/hpf were increased in HI-EoE compared with control (17 ± 11 vs 8 ± 6, P < 0.0). Patients with persistent endoscopic abnormalities had increased total (20 ± 12 vs 13 ± 10, P = 0.001) and degranulated (8 ± 6 vs 5 ± 4, P = 0.002) MC/hpf, with no difference in eosinophils. MC/hpf predicted furrowing (odds ratio = 1.06, P = 0.01) and rings (odds ratio = 1.05, P = 0.03) after controlling for treatment type, proton-pump inhibitor, eosinophils, and duration of therapy. Patients with persistent basal zone hyperplasia and dilated intercellular spaces had increased MC/hpf. Eosinophils were weakly correlated with MC/hpf in the mid (r = 0.30, P < 0.001) and distal (r = 0.29, P < 0.001) esophagus. Clinical remission patients had lower MC/hpf compared with patients with persistent symptoms and/or endoscopic abnormalities.

DISCUSSION:

MC density is increased in patients with endoscopic and epithelial abnormalities, as well as a few symptoms, despite resolution of esophageal eosinophilia after treatment. This association warrants further study to ascertain whether MCs play an eosinophil independent role in EoE.

INTRODUCTION

Eosinophilic esophagitis (EoE) is a chronic immune-mediated disease characterized by symptoms of esophageal dysfunction and at least 15 eosinophils per high-powered field (eos/hpf) on biopsy. Treatment with diet elimination and swallowed “topical” steroids can induce histological response measured by peak eosinophil count, with less than 15 eos/hpf considered inactive EoE (1,2). Despite histologic improvement, abnormal endoscopic findings and symptoms of esophageal dysfunction can persist (3–5). The factors associated with this poor clinical response remain unclear.

Mast cells (MCs) are tissue-resident immune cells that increase in density and activation in EoE (6,7) compared with conditions such as gastroesophageal reflux disease (8,9). The MC-associated response to therapy varies among patients (10–13) and dissociates from eosinophilia (14,15). MCs are filled with granules that contain preformed proinflammatory mediators such as tryptase (7). Activation occurs secondary to diverse stimuli with release of these mediators through degranulation. Based on animal models, this drives smooth muscle contractility and hyperplasia along with fibroblast activation (6,11,15–17). A role for esophageal MCs in driving mucosal abnormalities, endoscopically or histologically, remains unclear. Recently, MCs were reported to be elevated in a subgroup of adults with EoE and normalization of intraepithelial eosinophilia (18). Although MCs promote symptoms and remodeling, the extent to which this occurs in histologically inactive (HI) EoE and associates with persistent clinical findings is unknown.

The hypothesis of this study is that MC counts are increased in children with HI EoE with persistent symptoms, endoscopic, and epithelial abnormalities. The aim of the study was to determine the MC-associated clinical phenotype of patients with EoE.

METHODS

Study design

We performed a secondary analysis of prospectively collected symptom and endoscopy data in the Lurie Children’s Eosinophilic Gastrointestinal Diseases Database (referred to as “database”). Pediatric EoE patients (2–17 years old) underwent endoscopy with biopsy at Ann & Robert H. Lurie Children’s Hospital of Chicago between January 2011 and December 2015 to assess disease activity in response to diet elimination or swallowed steroids. Non-EoE controls from the database underwent diagnostic endoscopy during this period and had archived slides available for analysis. This study was approved by the Lurie Children’s Hospital Institutional Review Board.

Case definitions

EoE cases were diagnosed as per 2011 consensus guidelines: one or more symptoms of esophageal dysfunction and demonstration of ≥15 eosinophils per high-powered field (eos/hpf) on esophageal biopsy after 8 weeks of high-dose, twice-daily proton-pump inhibitor (PPI) therapy (1). Other causes of esophageal eosinophilia were excluded. EoE with histologic inactivity (HI EoE) was defined as peak eosinophil count <15 eos/hpf after elimination diet or swallowed steroid therapy (19,20). Patients were managed with standard clinical treatment protocols by pediatric gastroenterologists (JBW, BKW, and AFK). Non-EoE controls were patients with symptoms of esophageal dysfunction with <15 eos/hpf and normal endoscopic visual findings. A consort diagram of reasons for exclusion for non-EoE controls is depicted in Figure 1, Supplementary Digital Content 1, http://links.lww.com/AJG/B325. Subjects were excluded if on a biologic, immunomodula-tory, or immune-suppressive drug at time of esophagogastroduodenoscopy (EGD).

Clinical data

Demographics, atopic/medical history, and medications were collected at enrollment. Surveys were completed by caregivers at time of endoscopy in patients with EoE to assess the presence of abdominal pain, chest pain, dysphagia, early satiety, feeding aversion, gag, heartburn, nausea, odynophagia, pockets/spits out food, poor appetite, regurgitation, and vomiting. Endoscopic findings, as previously described (5,21), were collected at time of endoscopy for presence of edema, rings, exudates, furrows, and strictures.

Histologic data collection

At least 3 biopsies each from distal and mid esophagus were obtained during standard of care endoscopy performed under general anesthesia. Peak eosinophil counts were determined by a GI pathologist. Each H&E sample underwent blinded, independent review by a GI pathologist and was scored using a validated scoring system (22).

Tryptase immunohistochemistry and quantification

Archived unstained slides were obtained from the mid and distal esophagus and blinded to case/control and treatment status. Sections were dehydrated then quenched with 3% hydrogen peroxide/methanol for 10 minutes at room temperature. Next, sections were blocked in 5% horse serum for 30 minutes at room temperature then incubated with mouse monoclonal antitryptase (clone AA-1, Cat#MS-1216-P, Fisher Scientific, Hampton, NH) at 1:5,000 overnight at 4°C. Sections were incubated in biotinylated horse anti-mouse secondary (Cat#BA-2000, Vector Labs, Burlingame, CA) for 1 hour at room temperature. Finally, sections were developed with DAB (Cat#SK-4100, Vector Labs), rehydrated, and counterstained with hematoxylin. As previously described (23), peak MCs per high-powered field (MC/hpf at 400× magnification; hpf area = 0.26 mm²) were quantified in all epithelial regions, including the basal zone and excluding intravascular MC within papillae; lamina propria was excluded. Degranulated MCs were identified as having >15% granules outside a well-defined cell body (23,24). Representative examples of tryptase immunohistochemistry from non-EoE controls and EoE patients with low (<15) and high (>15) MC/hpf are shown in Figure 1.

Figure 1.

Representative tryptase immunohistochemistry. 4003 high-powered field images from non-EoE controls (a, d, g) and EoE subjects with low (b, e, h), and high (c, f, i) mast cell infiltrate. Arrow indicates intact mast cell; star indicates degranulated mast cell. EoE, eosinophilic esophagitis; LP, lamina propria, BZ, basal zone, SE, squamous epithelium.

Statistical analysis

The Fisher exact test was used to compare categorical variables. Nonparametric Mann-Whitney (2 groups) or Kruskal-Wallis rank sign tests with Dunn post hoc testing (3 groups) were used to compare eos/hpf, MC/hpf, and degranulated MC/hpf. Spearman correlation was used to correlate MC and eosinophil counts. Statistical analysis was performed on GraphPad Prism v6.0 (La Jolla, CA) and SPSS v22.0 (Armonk, NY). See Supplemental Methods for description of biostatistical models (Supplementary Digital Content 2, http://links.lww.com/AJG/B344).

RESULTS

Clinical and demographic characteristics

Approximately 161 patients in the database had prospectively obtained symptom surveys and endoscopic findings at the time of standard of care endoscopy to assess treatment response to diet elimination or swallowed steroid treatment. Among these patients, 93 subjects treated with diet elimination or swallowed steroids had HI EoE (<15 eos/hpf, HI EoE) and were included for analysis, along with 35 non-EoE controls from the database (Figure 2 and see Figure 1, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). No significant differences in age, sex, ethnicity, or eczema were observed between non-EoE controls and subjects with HI EoE (Table 1). Patients with HI EoE were noted to have increased prevalence of asthma and food allergies. Although all patients with HI EoE had <15 eos/HPF based on inclusion criteria, we compared the intraepithelial eosinophil counts between groups and found the peak eos/HPF (mean ± SD) were 3.4 ± 3.9 vs 1.4 ± 3.3 (P < 0.001) in patients with HI EoE compared with non-EoE controls, respectively. The overall peak MC/hpf in the HI EoE cohort were 16.6 ± 11.2 vs 7.6 ± 5.7 in non-EoE controls (P < 0.001), while the maximal degranulated MC/hpf were 6.7 ± 5.4 vs 3.2 ± 3.6 (P < 0.001) (Table 1 and Figure 1). Thus, MCs and eosinophils were present at a higher frequency in HI EoE biopsies than in non-EoE controls.

Figure 2.

Consort diagram for HI EoE patient cohort. Reasons for exclusion noted. EoE, eosinophilic esophagitis; HI, histologically inactive.

Table 1.

Patient characteristics

	Non-EoE control (N = 35)	HI EoE (N = 93)	Pa
Demographics (n, %)
Age (mean ± SD)	10.9 ± 4.2	9.7 ± 4.4	0.40
Male	18 (51.4)	66 (71)	0.06
White	31 (88.6)	86 (92.5)	0.49
Hispanic	6 (17.1)	9 (9.7)	0.35
Comorbidities (n, %)
Asthma	8 (22.9)	40 (43)	0.042
Eczema	12 (34.2)	38 (40.9)	0.55
Allergic rhinitis	16 (45.7)	53 (57)	0.32
Food allergy	4 (11.4)	30 (32.3)	0.024
GERD	6 (17.1)	20 (21.5)	0.81
Visual findings (n, %)
Edema	0	23 (24.7)	<0.001
Exudate	0	12 (12.9)	0.036
Furrows	0	28 (30.1)	<0.001
Rings	0	19 (20.4)	0.002
Stricture	0	0	ND
Peak cells/hpf (mean ± SD)
Mast cell	7.6 ± 5.7	16.7 ± 11.3	<0.001
Degranulated mast cell	3.2 ± 3.6	6.7 ± 5.4	<0.001
Eosinophil	1.4 ± 3.3	3.4 ± 3.9	0.001
Histologic scoring (mean ± SD)
Epithelial score	0.6 ± 1.6	2.2 ± 2.2	<0.001
Eosinophilia score	0 ± 0	0.7 ± 0.7	<0.001

^aCategorical variables compared with the Fisher exact test; continuous variables compared with unpaired 2-sample t-test.

EoE, eosinophilic esophagitis; GERD, gastroesophageal reflux disease; HI, histologically inactive; ND, not determined.

Mast cells are increased in patients with persistent endoscopic abnormalities

We next assessed the frequency by which endoscopic and histologic abnormalities persist despite eosinophilic remission. Forty-nine percent of the subjects in the HI EoE cohort had an abnormal endoscopy. Within the cohort, 25% had persistent edema, 13% had exudate, 30% had furrows, and 20% had rings (Table 1, examples shown in Figure 2, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). Similar to our four-food elimination diet study, no strictures were recorded (4). Patients with persistent endoscopic abnormalities had increased total (20.0 ± 11.5 vs 13.3 ± 10.0, P = 0.001) and degranulated (8.4 ± 6.2 vs 5.0 ± 3.8, P = 0.002) MC counts, despite no significant difference in eosinophil counts between the groups (Table 2). We then examined whether MC counts were increased in specific areas of the esophagus where visual abnormalities were present. Abnormal vs normal endoscopy showed significant differences in MC/hpf in both mid (16.6 ± 10.1 vs 9.9 ± 7.5, P = 0.002) and distal (18.5 ± 11.5 vs 13.2 ± 10.3, P = 0.01) specimens. Increased MC/hpf were observed in patients with mid (16.1 ± 9.0 vs 11.2 ± 8.7, P = 0.03) and distal (19.4 ± 11.3 vs 14.0 ± 10.8, P = 0.02) edema; mid (18.4 ± 9.9 vs 10.3 ± 7.8, P < 0.001) and distal (20.4 ± 12.3 vs 13.3 ± 10.0, P = 0.003) furrows; and mid (22.5 ± 10.9 vs 11.3 ± 8.4, P = 0.02) rings (Figure 3a,c). Only 2 patients in the cohort exhibited distal rings, precluding an analysis of this finding. Notably, exudates, commonly associated with eosinophilic microabscess, were the only finding not associated with MCs. We further examined whether degranulated MCs were elevated with abnormal visual findings in specific esophageal areas. No differences were seen in the distal esophagus; however, in the mid esophagus, degranulated MC/hpf were increased with abnormal endoscopic findings (6.9 ± 5.0 vs 3.6 ± 2.8, P < 0.01); in particular, edema (7.7 ± 5.0 vs 4.0 ± 3.4, P < 0.01) and furrows (8.3 ± 5.0 vs 3.6 ± 2.9, P < 0.001) had significantly elevated degranulated MC/hpf (Figure 3b). There was no significant difference in MC counts in patients with or without exudate. We further determined the frequency of endoscopic findings in low (≤15) MC/hpf and high (>15) MC/hpf patients (see Table 1, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). Receiver operating characteristic curves for MC/hpf were generated for those endoscopic findings where significant associations were noted and determination of the area under the curve (AUC) (see Figure 3, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). We found a threshold of 14 MC/hpf provided optimal combined sensitivity and specificity (68%/64%, AUC 0.70, P < 0.01) for distal furrows, whereas 16 MC/hpf (60%/86%, AUC 0.75, P < 0.001) were more optimal for furrows in the mid and 20 MC/hpf for mid rings (83%/87%, AUC 0.79, P < 0.05). Higher frequency of edema, furrows, and rings was found in the high MC group. MC counts were also compared between patients with 0 eos/hpf and 1–15 eos/hpf. Interestingly, MC counts were significantly reduced, but degranulated MCs were not (see Table 2, Supplementary Digital Content 1, http://links.lww.com/AJG/B325).

Table 2.

Comparison of mast cell and eosinophil density by endoscopic appearance in patients with inactive EoE

Cell/hpf (mean ± SD)	Normal (N = 47)	Abnormal (N = 46)	Pa
Mast cell	13.3 ± 10	20 ± 11.5	0.001
Degranulated mast cell	5 ± 3.8	8.4 ± 6.2	0.002
Eosinophil	2.5 ± 3.1	4.3 ± 4.4	0.067

EoE, eosinophilic esophagitis.

^aComparison with the Mann-Whitney test.

Figure 3.

Increased mast cells in patients with abnormal endoscopic findings. Comparison of peak mast cell counts in mid (a and b) and distal (c and d) biopsies in EoE subjects with edema, exudate, furrows, and rings. Total (a and c) and degranulated (b and d) mast cell counts compared using nonparametric Mann-Whitney test (*P < 0.05, **P < 0.01, ***P < 0.001). EoE, eosinophilic esophagitis.

To better understand which endoscopic abnormality independently associated with higher MC counts, linear regression was performed with MC count as the dependent variable. Treatment type (diet elimination vs swallowed steroids) and duration (from last abnormal EGD), eosinophils/hpf, and PPI use were included as potential confounders. This model identified a significant independent relationship between increased MC count and mid furrows (β = 7.5, P = 0.006) and mid rings (β = 7.7, P = 0.02) (see Table 3, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). To evaluate the predictive nature of elevated MC/hpf with individual endoscopic abnormalities, we performed binomial logistic regression with endoscopic findings as the dependent variable. Treatment type and duration, eosinophils/hpf, and PPI use were again included as potential confounders. MC counts predicted furrowing (odds ratio = 1.06, P = 0.01) and rings (odds ratio = 1.05, P = 0.03) after controlling for treatment type, PPI, eosinophils, and duration of therapy (see Table 4, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). Use of PPI was not a significant predictor for the persistence of endoscopic features within inactive EoE, and additional analysis of patients from the HI EoE cohort demonstrated no significant patient differences between patients on PPI (n = 16) and not on PPI (n = 77) except for allergic rhinitis (see Table 5, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). These results suggest that MC inflammation and activation persists despite therapy in association with persistent endoscopic abnormalities, in particular furrows.

Mast cells are increased in patients with persistent histologic abnormalities

We next examined epithelial reactive markers associated with EoE, using the validated EoE Histologic Scoring System (HSS) (22). Among the HI EoE group, 48.4% of distal biopsies and 45.1% of mid biopsies demonstrated persistent basal zone hyperplasia (BZH, grade 1 or greater). Comparison of biopsies with and without BZH demonstrated increased MC counts in the mid (P = 0.01) and distal (P < 0.001) regions (Figure 4a,c). Degranulated MCs were also significantly increased in distal (P = 0.01) and mid (P = 0.02) biopsies with BZH (Figure 4b,d). Moreover, distal (P < 0.001) but not mid biopsies with dilated intercellular spaces (DIS) demonstrated significantly elevated total and degranulated MC counts over those without DIS (Figure 4c,d). Neither differences were seen with the remaining measures of the EoE-HSS. Comparison of low (≤15) MC/hpf and high (>15) MC/hpf patients also demonstrated significant differences in HSS epithelial score (see Table 1, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). Thus, peak MC counts correlated with histologic changes of epithelial reactivity associated with EoE in these patients with HI EoE.

Figure 4.

Mast cells are increased in biopsies with basal zone hyperplasia and dilated intercellular spaces. Comparison of total (a and c) and degranulated (b and d) mast cell counts in mid (a and b) and distal (c and d) biopsies with nonparametric Mann-Whitney test (*P < 0.05, **P < 0.01, ***P < 0.001).

Patients with clinical remission have reduced mast cells

We next examined patients with no symptoms nor endoscopic abnormalities (defined as clinical remission [CR]). Symptomatic patients had significantly increased MC counts (16.8 ± 11.4 vs 5.6 ± 1.4, P = 0.008). Increased MC counts were found for patients with several specific symptoms (Table 3) when compared with CR patients. Of note, there were no significant differences in eos/hpf between symptomatic patients with or without abnormal endoscopy compared with CR, while these patients demonstrated elevated MC/hpf after adjusting for multiple comparisons (Figure 5). However, for symptoms alone, we did not identify a significant receiver operating characteristic curve. In addition, there were no significant differences in peak eosinophilia between patients with individual symptoms vs CR, except for a trend in distal biopsies for heartburn (P = 0.056; see Table 6, Supplementary Digital Content 1, http://links.lww.com/AJG/B325). Spearman correlation of MC density demonstrated a weak correlation with eosinophil density in the mid (r = 0.30, P < 0.001) and distal (r = 0.29, P < 0.001) (see Figure 4, Supplementary Digital Content 1, http://links.lww.com/AJG/B325), consistent with the dissociation that has previously been described (15).

Table 3.

Comparison of mast cells counts in patients with clinical remission (asymptomatic, visually normal endoscopy) and specific symptoms

Symptom (n)	Peak MC/hpf	Pa
Clinical remission (9)	5.6 ± 1.4	ND
Any symptom (66)	16.8 ± 11.4	0.008
Abdominal pain (29)	10.8 ± 1.3	0.04
Chest pain (7)	12.6 ± 3.0	0.04
Dysphagia (14)	9.3 ± 1.8	NS
Early satiety (25)	12.0 ± 1.6	0.03
Feeding aversion (18)	14.7 ± 2.7	0.03
Food impaction (13)	13.5 ± 3.3	NS
Gag (9)	12.3 ± 2.3	0.02
Heartburn (14)	9.5 ± 1.9	NS
Nausea (17)	12.1 ± 2.1	0.04
Odynophagia (7)	17.4 ± 4.5	0.01
Pockets or spits out food (6)	13.7 ± 5.6	0.007
Poor appetite (17)	9.3 ± 1.2	NS
Regurgitation (20)	11.5 ± 2.1	NS
Slow eating (29)	10.8 ± 3.0	NS
Vomiting (13)	13.2 ± 3.1	NS

^aComparison of specific symptoms with clinical remission group by the Mann-Whitney test.

MC, mast cell; ND, not determined; NS, not significant.

Figure 5.

Total and degranulated mast cells are increased in patients with persistent symptoms and/or endoscopic abnormalities despite inactive EoE. Comparison of total (a) and degranulated (b) mast cell and eosinophil (c) counts in clinical remission (CR: no endoscopic findings/no symptoms) compared to subjects with symptoms and/or endoscopic abnormalities (EREFS). Comparison by nonparametric Kruskal-Wallis with the Bonferroni multiple comparison test (*P < 0.05, **P < 0.01). EoE, eosinophilic esophagitis.

DISCUSSION

We identified a novel elevation of total and degranulated MC counts in esophageal biopsies from EoE patients with histologic eosinophilic remission (<15 eos/hpf) and a poor clinical treatment response. In EoE, clinical symptoms of esophageal dysfunction can dissociate from histological inflammation (3); moreover, persistent and unexplained endoscopic abnormalities occur in patients who histologically respond to treatment based on peak eosinophil count (4,5,25). Although MCs have been suggested to play a functional role in EoE, their role in these clinical features and as an outcome metric for treatment is unclear. This novel finding may address a gap in the understanding of pathophysiological aspects of EoE associated with a poor clinical response to therapy.

MCs express mediators in the epithelium (12) and activate fibroblasts (16,17), believed to be involved in remodeling. We found EoE patients with persistent rings or furrows to have significantly elevated MC counts compared to those without. Regression models further supported this relationship as MC counts predicted furrows and rings after controlling for treatment type/duration and PPI. A similar association between MCs and furrows was found in an adult cohort with active EoE (26). In addition, a study in adults demonstrated increased diagnostic sensitivity when incorporating MCs into a cell count algorithm (27). Our work supports a diagnostic role for MC counts in EoE as an outcome measure, particularly in patients with unexplained endoscopic abnormalities despite eosinophilic remission.

The interaction between esophageal MCs and the epithelium is not well understood. We observed increased MC counts in association with epithelial reactivity, characterized by BZH and DIS. This finding provides insight into the recent observation of persistent BZH despite inactive EoE in association with endoscopic abnormalities and symptoms (28). This suggests MCs may be interacting with the epithelium, despite low eosinophil levels. The functions and signaling for this remain unclear, and mechanistic studies are needed. MCs may alter epithelial barrier function through the release of proteases. Tryptase activates and cleaves protease-activated receptor 2 (PAR-2) in intestinal epithelium by altering tight junction integrity (29). Chymase increases intestinal epithelial permeability in colonic epithelial cells (30), while the role of proteases such as carboxypeptidase A3 is entirely unknown. Interestingly, acid exposure activates MCs with subsequent esophageal nodose C fiber activation and disruption of the esophageal epithelial barrier (31). Although our statistical models controlled for PPI use, this was not a significant factor. Further work is needed to dissect the biological mechanisms underlying the pathogenic role of MCs in the esophageal mucosa.

Symptoms of esophageal dysfunction and endoscopic abnormalities can dissociate from esophageal eosinophilia (3). Our study identifies 2 key subgroups in EoE: endoscopic remission (absence of endoscopic abnormalities) and CR (absence of symptoms and endoscopic abnormalities, CR EoE). Total and degranulated MCs were significantly elevated in patients with any endoscopic abnormalities compared with endoscopic remission, whereas no differences in peak eosinophilia were appreciated. A significant difference in MC counts was noted in individuals with specific clinical symptoms when compared with patients with EoE in CR. Total MC counts were significantly lower in patients in CR compared to patients with either symptoms or endoscopic findings. Patients with various symptoms including abdominal pain, nausea, odynophagia, early satiety, and gag had significantly increased MC counts in both mid and distal biopsies compared with the CR group. Individual symptoms did not have overall increased MC/hpf, however, clearly limiting the extent to which conclusions can be made in this regard. This is likely limited by the small size of the CR cohort and suggests longitudinal research with validated symptom and quality of life instruments is necessary to better understand this relationship.

Fibrostenotic changes persist in a large percentage of patients despite response to therapy (25,32,33), and chronic symptoms are associated with increasing stricture rate over time (34). This questions the need for MC-targeted therapy. We found no differences in MC counts between patients treated with swallowed steroids and diet elimination, suggesting the relationship between endoscopic findings and MCs is independent of treatment type. Although our findings necessitate further work to better understanding of the mechanistic role of MCs in the mucosa, novel molecular treatment targets are needed as cromolyn sodium (35) and montelukast (36,37) as well as omalizumab (38) have not demonstrated clear effects of therapy.

Our study has several strengths. This is the first study to evaluate peak total and degranulated MC counts in a large pediatric EoE cohort. Although this was a secondary analysis, all clinical data were collected prospectively regarding demographics, treatment, visual findings, and symptoms before knowing the eosinophil response, thus enhancing the generalizability of the study and limiting recall/selection bias associated with retrospective chart review. Validated histologic scoring was used to determine epithelial reactivity. Finally, models were used to identify independent associations and control for potential confounders.

Several limitations exist for our study. Pretreatment endoscopies were not assessed, and thus, we are unable to determine whether MC counts changed with therapy in association with persistent endoscopic abnormalities and symptoms; future studies will address this. This cohort was derived from routine clinical practice, with variation in interval to follow-up endoscopy, in addition to differing diets and steroid doses. This was addressed in the models as potential confounders. Adherence to therapy was not measured; however, chart review did not identify compliance concerns. A further limitation was the lack of a “cutpoint” MC number above which endoscopic abnormalities were present. This may be due to biopsy variation and patchiness, presence of histologic without endoscopic abnormalities, or tissue-resident nature of MCs, limiting absolute resolution. We addressed this with regression models that strengthened the association between furrows and MC counts. Immunohistochemical staining for eosinophil degranulation such as through eosinophil peroxidase was not performed to more thoroughly rule out eosinophil involvement, as has been examined for patients with EoE-like symptoms despite absence of tissue eosinophilia (39–41). Nor did we assess for lymphocytic infiltration, which may account for the persistent pathologic findings (39). Finally, the lack of a validated symptom questionnaire is a limitation; however, the survey used captured a clear CR group. Further studies will capture the relationship between MC counts, symptom frequency, and quality of life using validated measures. The small size of the CR group additionally necessitates larger cohort studies to further define CR and the relationship to MC inflammation and epithelial reactivity. Although the pathogenicity of MCs remains unclear, increased MC counts may be a bystander of incompletely resolved inflammation, perhaps a reflection of need for longer therapy, rather than driving the inflammation, and further mechanistic studies are needed. Indeed, MCs but not degranulated MCs were significantly reduced when comparing patients with 0 vs 1–15 eos/hpf. This may suggest that those patients within our cohort with 0 eosinophils identify a separate group of complete healing including MCs.

Our study identifies a novel eosinophil-independent association of mucosal MC activation with persistent symptoms and endoscopic abnormalities in patients with EoE despite a histologic response at the 15 eos/hpf threshold. Our work supports a role for MC quantification in addition to eosinophil count as part of the evaluation of an anti-inflammatory therapeutic intervention. Longitudinal prospective studies are warranted to validate these findings and establish outcome metrics. The utility of MC-targeted therapy remains to be determined with validated clinical, endoscopic, and histologic outcome metrics. Our work supports further studies into the pathogenesis of MC activation in EoE, including investigation of MC phenotypes and mechanisms driving epithelial reactivity.

Study Highlights.

WHAT IS KNOWN

• Symptoms of esophageal dysfunction and endoscopic abnormalities such as edema, furrows, rings, and strictures are present at diagnosis of EoE and can persist despite resolution of eosinophilia with treatment.

• The EREFS and EoE-HSS are validated markers of disease activity for adults and children with EoE.

• MCs are increased in density and activation at EoE diagnosis and have a variable response to treatment.

WHAT IS NEW HERE

• Mast cells are increased in biopsies from children with EoE with persistent endoscopic and epithelial abnormalities despite resolution of esophageal eosinophilia with treatment.

• In patients with eosinophilic remission, mast cell counts predict endoscopic furrows, histologic basal zone hyperplasia, and dilated intercellular spaces independent of treatment type, duration, and proton-pump inhibitor therapy.

• Patients in clinical remission, without symptoms nor endoscopic abnormalities, have lower mast cell counts compared to patients with persistent symptoms and endoscopic abnormalities.