Summary

To assess the incidence of esophageal intra‐epithelial eosinophilic infiltration following endoscopic ablation of Barrett’s esophagus (BE), a retrospective study of consecutive cases of endoscopic ablation of BE with dysplasia or cancer using radiofrequency ablation (RFA) and spray cryotherapy at two centers in the United States was performed. Post‐ablation eosinophilia was defined as ≥5 eosinophils per high power field during post‐treatment surveillance. Twenty of 122 patients (16%) undergoing ablation developed esophageal eosinophilia after ablation, including 8/77 (10%) treated with RFA and 12/44 (27%) treated with cryotherapy. No patient had clinical or endoscopic findings of or risk factors for eosinophilic esophagitis. Esophageal eosinophilia persisted in 30% over a median of 20.2 months. On multivariate analysis, post‐ablation eosinophilia was independently associated with increasing BE segment length (adjusted odds ratio 1.46 for every 2‐cm increase, 95% confidence interval 1.24–1.71) and cryotherapy as the ablation modality (adjusted odds ratio 5.23, 95% confidence interval 1.67–16.39). Esophageal eosinophilic infiltration after endoscopic ablation with RFA and cryotherapy is common and is associated with the BE segment length and treatment modality. The clinical significance of post‐ablation eosinophilia is unclear.

Eosinophilic infiltration of the esophagus is an invariably abnormal finding recently described in association with Barrett’s esophagus (BE) and noted to occur after photodynamic therapy (PDT).1–3 Radiofrequency ablation (RFA) and cryotherapy are ablation modalities that eradicate BE through superficial tissue injury similar to PDT but with superior safety and efficacy.4,5 To date, esophageal eosinophilia has not been reported following treatment with RFA or cryotherapy, and risk factors for this finding have not been identified. Further, the clinical implications of post‐ablation eosinophilia are unknown.

Esophageal eosinophilia is most frequently associated with eosinophilic esophagitis (EoE) or untreated gastroesophageal reflux disease (GERD).6,7 The pathogenesis of esophageal eosinophilia is incompletely understood. EoE has been ascribed to an atopic mechanism, while GERD‐related eosinophilia is thought secondary to chemokine activation through acid peptic injury.8–10 In this mechanism, derangement of the esophageal mucosal barrier may allow altered antigen presentation to the subepithelial tissues.

This is the first report of esophageal eosinophilia after RFA or cryotherapy. In a consecutive series of patients treated with these ablation modalities, we describe the characteristics of patients with and without post‐ablation eosinophilia. We also identify risk factors for this finding, which may provide insight into the mechanism of pathogenesis.

Methods

We performed a retrospective study of post‐ablation (RFA or cryotherapy) eosinophilia. Consecutive patients with dysplastic BE or T1 N0M0 adenocarcinoma treated with RFA or cryotherapy were identified at two American academic medical centers (University of Maryland Medical Center, University of North Carolina Hospitals).

The peri‐ablation clinical protocols at each institution were notably similar. The same device model was used for each ablation modality (RFA, BÂRRX Medical, Sunnyvale, CA, USA; spray cryotherapy, CryoSpray Ablation, CSA Medical, Inc., Baltimore, MD, USA). Prior to ablation, all patients underwent sampling of the Barrett’s tissue with four‐quadrant biopsies every 1–2 cm throughout the Barrett’s segment. Ablation was performed according to previously described methods.4,5 Post‐ablation surveillance was performed every 3–6 months for 3 years, with four‐quadrant biopsies performed every 1–2 cm in the esophagus in areas previously treated with ablation. All patients remained on proton pump inhibitors (PPIs) at least once daily throughout the study period. The dose of PPI was adjusted prior to initiation of ablation for elimination of all GERD symptoms. At each encounter, all patients were asked about GERD symptoms, and the dose of PPI was increased as needed.

Post‐ablation eosinophilia was defined by the presence of five or more eosinophils/high power field (hpf) on any single post‐ablation biopsy specimen (Fig. 1). Other pertinent clinical findings were collected by chart review for all study subjects: age, gender, pre‐ablation histology, maximum BE segment length, ablation modality, and number of ablation treatment sessions. Subjects with eosinophilia were contacted by telephone and underwent a structured interview or chart review for EoE risk factors, including history of asthma and seasonal, pet, or food allergies. Outcome measures included number of eosinophils/hpf, interval to onset and duration of eosinophilic infiltration, and treatment success.

Fig. 1

Histopathology of endoscopic biopsy demonstrating eosinophilic esophagitis after ablation of Barrett’s esophagus (H&E stain, original magnification ×40).

Fig. 1

Histopathology of endoscopic biopsy demonstrating eosinophilic esophagitis after ablation of Barrett’s esophagus (H&E stain, original magnification ×40).

Characteristics of study subjects were described using the median with interquartile range (IQR) for continuous variables and proportions for categorical variables. Subjects with and without post‐ablation eosinophilia were compared using non‐parametric statistics (Wilcoxon rank‐sum test for continuous variables, Fisher’s exact test for categorical variables). Characteristics associated with post‐ablation eosinophilia on bivariable analysis (P≤ 0.20) were included in a logistic regression model to identify independent predictors of esophageal eosinophilia. The likelihood ratio test was used to reduce the model to statistically significant independent predictors. Statistical significance was defined by a P‐value <0.05.

Results

The study cohort included 122 patients with dysplastic BE or early cancer ablated at the two academic medical centers (77 RFA, 44 cryotherapy, 1 both). The median pre‐ablation eosinophil count among cases was 1 eosinophil/hpf (IQR 1, range 0–14). Two patients previously treated with PDT had pre‐ablation eosinophil counts of 6 and 14/hpf.

Twenty patients (16%) developed post‐ablation esophageal eosinophilia a median of 12.9 months (IQR 11.8) after ablation had been initiated (Table 1). The median post‐ablation eosinophil count among the cases was 15/hpf (IQR 39, range 5–103). Eosinophils were most frequently localized to the neosquamous epithelium (18/20, 90%), although in two cases, the highest eosinophil count was found in recurrent or persistent BE. On average, patients with post‐ablation eosinophilia were 65 years old (IQR 13.5), male (85%), and white (95%) with high‐grade dysplasia (13/20, 65%) or superficial adenocarcinoma (6/20, 30%). On endoscopy, no subject was noted to have mucosal furrowing, rings, microabscesses, or other endoscopic stigmata associated with EoE. Post‐ablation eosinophilia spontaneously resolved in 14/20 patients (70%) but persisted in six patients over a median follow‐up of 20.2 months after the initial diagnosis (IQR 21.4, range 0–44). Among the patients with prior PDT, both patients had increased eosinophil counts following ablation (6–38/hpf after cryo, 14–92/hpf after RFA).

Table 1

Characteristics of patients with and without intra‐epithelial eosinophilia after ablation for Barrett’s esophagus

Parameter Eosinophilia present (n= 20) No eosinophilia (n= 102) All patients (n= 122) 
Age (years), median (IQR) 65 (13.5) 67 (15) 66 (14) 
Male gender, no. (%) 17 (85) 79 (77) 96 (79) 
Length of Barrett’s esophagus segment, cm, median (IQR) 6 (4.5) 4 (5) 4 (6) 
Pre‐ablation histology, no. (%)    
 Low‐grade dysplasia 1 (5) 14 (14) 15 (12) 
 High‐grade dysplasia 13 (65) 74 (72) 87 (71) 
 Adenocarcinoma 6 (30) 14 (14) 20 (16) 
Mode of ablation, no. (%)*    
 Radiofrequency ablation 8 (40) 69 (68) 77 (63) 
 Cryotherapy 12 (60) 32 (31) 44 (36) 
 Both 1 (1) 1 (1) 
Number of ablation sessions* 4 (2.5) 3 (3) 3.5 (3) 
Esophageal eosinophil count after ablation, median (IQR, range) 15 (39, 5–103) – – 
Time to onset of eosinophilic infiltration 12.9 (11.8) – – 
Follow‐up, months, median (IQR)* 34.2 (16) 20.0 (19.4) 22.3 (20.3) 
Parameter Eosinophilia present (n= 20) No eosinophilia (n= 102) All patients (n= 122) 
Age (years), median (IQR) 65 (13.5) 67 (15) 66 (14) 
Male gender, no. (%) 17 (85) 79 (77) 96 (79) 
Length of Barrett’s esophagus segment, cm, median (IQR) 6 (4.5) 4 (5) 4 (6) 
Pre‐ablation histology, no. (%)    
 Low‐grade dysplasia 1 (5) 14 (14) 15 (12) 
 High‐grade dysplasia 13 (65) 74 (72) 87 (71) 
 Adenocarcinoma 6 (30) 14 (14) 20 (16) 
Mode of ablation, no. (%)*    
 Radiofrequency ablation 8 (40) 69 (68) 77 (63) 
 Cryotherapy 12 (60) 32 (31) 44 (36) 
 Both 1 (1) 1 (1) 
Number of ablation sessions* 4 (2.5) 3 (3) 3.5 (3) 
Esophageal eosinophil count after ablation, median (IQR, range) 15 (39, 5–103) – – 
Time to onset of eosinophilic infiltration 12.9 (11.8) – – 
Follow‐up, months, median (IQR)* 34.2 (16) 20.0 (19.4) 22.3 (20.3) 
*

P < 0.05 comparing patients with eosinophilia to those without. –, not applicable.

Twelve of the 20 patients with post‐ablation eosinophilia were successfully contacted and underwent structured interviews. No interviewed subject had risk factors for EoE. The other eight subjects who were unable to be contacted had no risk factors by chart review. If the criteria for abnormal eosinophil count were set to ≥15/hpf, then 11 patients (9%) would be identified as abnormal. No significant difference was seen between the cohorts of 11 versus 20 patients in terms of age, Barrett’s length, time to eosinophilia, and duration of follow‐up.

Comparison of those with and without post‐ablation esophageal eosinophilia did not reveal clinical or statistically significant differences in terms of age or gender (Table 1). Post‐ablation eosinophilia was more frequently seen after cryotherapy (12/44, 27%) than RFA (8/77, 10%) (P= 0.02). The number of treatment sessions was significantly greater among those with eosinophilia (median 4, IQR 2.5) compared with those without (median 3, IQR 3) (P= 0.03). Patients with eosinophilia also had significantly more follow‐up time (median 34.2 months, IQR 16) than those without (median 20.0, IQR 19.4) (P < 0.001).

Clinically meaningful differences for BE segment length and pre‐ablation histology were noted between those with and without post‐ablation eosinophilia. On average, patients with eosinophilia had a longer BE segment (median 6.0 cm, IQR 4.5) than those without (median 4.0 cm, IQR 5.0) (P= 0.20). Post‐ablation eosinophilia was more frequent among those with superficial adenocarcinoma (6/20, 30%) than high‐grade dysplasia (13/87, 15%) or low‐grade dysplasia (1/15, 7%) (P= 0.06). Although neither of these trends reached statistical significance, both met criteria for inclusion in the logistic regression model.

Barrett’s segment length, treatment modality, number of treatment sessions, and pre‐ablation histology were evaluated as independent predictors of post‐ablation eosinophilia by logistic regression modeling (Table 2, model 1). Number of treatment sessions and pre‐ablation histology did not demonstrate an independent association with post‐ablation eosinophilia (P > 0.05) and were removed from the model. Of note, pre‐ablation histology did trend toward a dose–response relationship; however, the confidence intervals (CIs) were very wide indicating that statistical power was likely a limiting factor. In the final reduced model (model 2), BE segment length and treatment modality were both independently associated with post‐ablation eosinophilia. The odds of developing eosinophilia increased by 1.46 (P= 0.02) for every 2 cm increase in Barrett’s segment length (P= 0.02). Patients treated with cryotherapy had 5.2 times the odds of developing eosinophilia (95% CI 1.7–16.4) compared with those treated with RFA.

Table 2

Predictive model of post‐ablation esophageal eosinophilia

Odds ratio (95% CI) Model 1† Model 2‡ 
Length of Barrett’s segment (2‐cm increase) 1.53 (1.28–1.84) 1.46 (1.24–1.71) 
Treatment with cryotherapy 4.66 (1.22–17.77) 5.23 (1.67–16.39) 
Number of treatment sessions (one‐session increase) 0.94 (0.73–1.22) – 
Pre‐ablation histology   
 Low‐grade dysplasia 1 (reference) – 
 High‐grade dysplasia 1.52 (0.16–13.97) – 
 Adenocarcinoma 3.13 (0.27–36.74) – 
Odds ratio (95% CI) Model 1† Model 2‡ 
Length of Barrett’s segment (2‐cm increase) 1.53 (1.28–1.84) 1.46 (1.24–1.71) 
Treatment with cryotherapy 4.66 (1.22–17.77) 5.23 (1.67–16.39) 
Number of treatment sessions (one‐session increase) 0.94 (0.73–1.22) – 
Pre‐ablation histology   
 Low‐grade dysplasia 1 (reference) – 
 High‐grade dysplasia 1.52 (0.16–13.97) – 
 Adenocarcinoma 3.13 (0.27–36.74) – 

Model 1 covariates include length of Barrett’s segment, treatment with cryotherapy, number of treatment sessions, and pre‐ablation histology.

Model 2 covariates include length of Barrett’s segment and treatment with cryotherapy. Other covariates were excluded using the likelihood ratio test. CI, confidence interval; –, not applicable.

Discussion

The finding of increased esophageal eosinophilia after ablation of BE is a new and interesting observation in the era of increasing incidence of eosinophilic gastrointestinal diseases including EoE. A recent study evaluated the incidence of intra‐epithelial eosinophils >15/hpf in patients treated with PDT or RFA.3 They found eosinophilia in 3/385 (3.4%) treated patients, all of whom were treated with PDT. Like that cohort, none of the patients in this study demonstrated symptoms or clinical findings suggestive of EoE. Unlike that cohort, patients in this study were all maintained on at least once‐daily therapy with PPIs, making acid reflux injury a less likely etiology for the eosinophilia. In our cohort, two patients previously treated with PDT showed intra‐epithelial eosinophils, but these counts increased significantly after ablation. Interestingly, recent literature suggests a possible role of PPIs in the pathogenesis of EoE, and this area merits further research.11 The etiology of the higher prevalence of intra‐epithelial eosinophils in our study, including those treated with RFA, is unclear. Physiologic evaluation for GERD in this cohort using pH/impedance testing may further delineate the role of reflux in this population.

Patients in this cohort had no history of atopy or food allergy, suggesting a non‐allergic pathogenesis. However, a limitation of this study is that this information could only be obtained by post‐hoc interview (n= 12) or by chart review (n= 8). It is possible that these patients may develop symptoms of EoE at a later date, but currently none fulfill the disease definition of both histological findings and appropriate symptomatology. Repetitive injury from endoscopic ablation, with resulting chronic inflammation, could be an explanation for eosinophilic infiltration.12 Another possible explanation is increased sensitivity of denuded and inflamed esophageal epithelia to otherwise inert food antigens, inciting a T‐helper cellular type response, which can promote chronic inflammation and eosinophilic infiltration through a chemotactic cascade.13 The finding of eosinophilia after ablation may not be clinically relevant, but long‐term follow‐up is needed to further characterize any changes in the degree of esophageal eosinophilia and development of symptoms of EoE in this cohort.

References

1
Ravi
K
,
Katzka
D A
,
Smyrk
T C
et al.  
Prevalence of esophageal eosinophils in patients with Barrett’s esophagus
.
Am J Gastroenterol
 
2011
;
106
:
851
7
.
2
Genevay
M
,
Rubbia‐Brandt
L
,
Rougemont
A L
.
Do eosinophil numbers differentiate eosinophilic esophagitis from gastroesophageal reflux disease?
Arch Pathol Lab Med
 
2010
;
134
:
815
25
.
3
Owens
V L
,
Katzka
D A
,
Lutzke
L S
et al.  
Endoscopic ablative therapy for Barrett’s esophagus: a potential cause of eosinophilic esophagitis
.
Dis Esophagus
 
2012
;
25
:
33
9
.
4
Shaheen
N J
,
Greenwald
B D
,
Peery
A F
et al.  
Safety and efficacy of endoscopic spray cryotherapy for Barrett’s esophagus with high‐grade dysplasia
.
Gastrointest Endosc
 
2010
;
71
:
680
85
.
5
Shaheen
N J
,
Sharma
P
,
Overholt
B F
et al.  
Radiofrequency ablation in Barrett’s esophagus with dysplasia
.
N Engl J Med
 
2009
;
360
:
2277
88
.
6
Rodrigo
S
,
Abboud
G
,
Oh
D
et al.  
High intraepithelial eosinophil counts in esophageal squamous epithelium are not specific for eosinophilic esophagitis in adults
.
Am J Gastroenterol
 
2008
;
103
:
435
42
.
7
Rothenberg
M E
.
Biology and treatment of eosinophilic esophagitis
.
Gastroenterology
 
2009
;
137
:
1238
49
.
8
Isomoto
H
,
Wang
A
,
Mizuta
Y
et al.  
Elevated levels of chemokines in esophageal mucosa of patients with reflux esophagitis
.
Am J Gastroenterol
 
2003
;
98
:
551
6
.
9
Kanazawa
Y
,
Isomoto
H
,
Wen
C Y
et al.  
Impact of endoscopically minimal involvement on IL‐8 mRNA expression in esophageal mucosa of patients with non‐erosive reflux disease
.
World J Gastroenterol
 
2003
;
9
:
2801
4
.
10
Lucendo
A J
,
Lucendo
B
.
An update on the immunopathogenesis of eosinophilic esophagitis
.
Expert Rev Gastroenterol Hepatol
 
2010
;
4
:
141
8
.
11
Molina‐Infante
J
,
Zamorano
J
.
Acid‐suppressive therapy and eosinophilic esophagitis: friends or foes?
Am J Gastroenterol
 
2010
;
105
:
699
700
.
12
De Giorgi
F
,
Palmiero
M
,
Esposito
I
et al.  
Pathophysiology of gastro‐oesophageal reflux disease
.
Acta Otorhinolaryngol Ital
 
2006
;
26
:
241
6
.
13
Chehade
M
,
Aceves
S S
.
Food allergy and eosinophilic esophagitis
.
Curr Opin Allergy Clin Immunol
 
2010
;
10
:
231
7
.

Author notes

*
Contributed equally as the first author.