346 / 412
The
American Journal
of
GASTROENTEROLOGY
white plaques, mucosal tearing/friability, strictures, Schatzki
rings, and erosive esophagitis. The MDQ was completed by all
patients at the end of the study.
Histopathology
All biopsy samples were embedded in formalin and stained with
hematoxylin and eosin. Samples to include slides from index
endoscopy and following treatment were reviewed and read by
a single-blinded expert gastrointestinal pathologist. Each slide
had three separate sections and all were reviewed. Eosinophils
were counted in all available fields limited by the size of the biopsy
specimens. A high-powered field was considered ×40 magnifica-
tion on our microscopes, which measured 0.19mm
2
. The total
number of eosinophils in all fields was counted and the peak
eosinophil count per hpf was reported. Findings of eosinophilic
microabscesses, intercellular edema, evidence of eosinophilic
degranulation, epithelial basilar hyperplasia, and whether eosi-
nophils were confined to the epithelial basal layer or extended to
the epithelial surface (full thickness involvement) were also noted.
End points
The primary end point measuredwas histological response defined
as achieving <7 eos/hpf in both proximal and distal esophageal
biopsies following 8 weeks of treatment. Secondary outcomes
measured included symptomatic change in dysphagia on the basis
of the score from the MDQ and interval change in endoscopic and
other histological findings.
Statistical analysis
Sample size estimation was based on the following assumptions:
10% of patients will be GERD positive and respond to PPIs
compared with 55% of the patients treated with topical steroids.
Controlling the probability of a Type I error at
α
= 0.05, a sample
of 38 patients in the treatment groups (19 in each arm) will have
80% power to detect a difference in treatment response of 45%.
Data were collated and analyzed with SPSS 15.0 statistical analysis
package (SPSS Inc, Chicago, IL). Categorical data are expressed as
frequency and percentage, and continuous data as means and stand-
ard deviation (s.d.). Histological response, comparison of histologi-
cal markers, and endoscopic features between the two treatment
groupswere analyzedusing Fisher’s exact test.Within each treatment
arm, the change in MDQ score after treatment was compared with
Wilcoxon signed rank test. Adherence to treatment was compared
using Mann–Whitney
U
test. Spearman rank correlation coefficient
(
r
s
) was used to assess the relationship between reflux impedance
episodes in the proximal and distal esophagus to eosinophil counts,
as well as the relationship of change in MDQ and eosinophils count
post treatment. A
P
value of <0.05 was considered statistically
significant. Analysis was performed as intention to treat.
RESULTS
Forty-two patients with esophageal eosinophilia were enrolled
into the study. The mean age ± s.d. was 38 years old ±10, 81%
were Caucasian, 10% Hispanic, 7% African American, and 1%
other. Sixty-two percent of patients had a history of coexisting
atopic diathesis (33% seasonal allergies, 29% food allergies, 10%
asthma, and 5% eczema). The primary indication for endoscopy
was dysphagia (69%), followed by food impaction (19%), heart-
burn (12%), and other (2%). Nineteen percent (8/42) had GERD
by Johnson-DeMeester score and were equally stratified into each
treatment arm. Endoscopy revealed Los Angeles Grade A erosive
esophagitis in seven patients, all of whom had GERD by John-
son-DeMeester score. There were significantly more acid reflux
episodes on 24-h pH monitor in the FP arm compared with the
ESO arm (45.8±40.6 vs. 25.5±19.3,
P
=0.045), as well as imped-
ance reflux episodes (63.6±23.1, 44.8±21.4,
P
=0.012) (
Table 1
).
There was no significant change in mean eosinophil counts
before and after treatment in either arm (FP: 55.9±25 vs. 39.2±29.4,
P
=0.102; ESO: 42.9±18.9 vs. 30.5±33.7,
P
=0.174) (
Figure 1
). His-
tological response was achieved in 33% (7/21) of ESO patients vs.
19% (4/21) of FP patients,
P
=0.484. Among the eight patients
with GERD, all four patients randomized to ESO achieved histo-
logical response, whereas none of the four patients randomized to
FP achieved a histological response,
P
=0.029. In the 34 GERD-
negative patients, response was achieved in 18% (3/17) of ESO
patients vs. 24% (4/17) of FP patients,
P
=1.000. Among patients
with coexisting allergies, response was similar between the two
treatment groups, FP: 27% (3/11) vs. ESO: 33% (5/15),
P
=1.000.
The histological response for FP vs. ESOwas similar in the proxi-
mal esophagus (29 vs. 55%,
P
=0.118) and in the distal esopha-
gus (19 vs. 40%,
P
=0.181). In two patients of the FP arm and four
patients of the ESO arm, histological response was achieved in the
proximal but not distal esophagus. Improvement in other histo-
logical markers of EoE following treatment (basal cell hyperpla-
sia, intercellular edema, eosinophilic microabscess, eosinophilic
degranulation, and eosinophilic distribution within the epithe-
lium) was similar between the two treatment arms (
Table 2
).
Frequency and severity of dysphagia were similar at baseline
between the two treatment arms. The majority of patients reported
moderate to severe dysphagia on question no. 3 of the MDQ with
no significant difference between the two groups, (FP 77% vs. ESO
83%,
P
=0.512). In terms of frequency of dysphagia (question no.
4), the majority of patients reported symptoms from less than once
per week to several times per week with no difference between
the two treatment arms (FP 89% vs. ESO 83%,
P
=0.646). Only
three patients indicated symptoms with every meal. On baseline
questionnaire, 30% of patients randomized to FP avoided fibrous
foods (meat, chicken, bread, celery, salad) compared with 42% of
patients randomized to ESO,
P
=0.381.
The MDQ score before and after therapy significantly improved
in the ESO group (19±21 vs. 1.4±4.5
P
=0.001), but not in the FP
group (17±18 vs. 12±16,
P
=0.162) (
Figure 2
). A similar finding
was noted among GERD-negative patients: there was a significant
difference in MDQ score before and after treatment with ESO
(16±14 to 1.7±5.0,
P
=0.001) but not with FP (18±19 to 10±16,
P
=0.086). Overall, there was no significant correlation between
the change in symptoms by MDQ and the change in eosinophil
count in the proximal (
r
s
=0.001,
P
=0.996) and distal (
r
s
=0.101,
P
=0.558) esophagus.
121