ORIGINAL ARTICLES ALIMENTARY TRACT

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2011;9: ORIGINAL ARTICLES ALIMENTARY TRACT Natural History of Eosinophilic Gastroenteritis GUILLAUME PINETON DE CHAMBRUN,*,,, FLORENT GONZALEZ,*,, JEAN YVES CANVA,,, SAMIA GONZALEZ,*,# LUCIE HOUSSIN,*,, PIERRE DESREUMAUX,*,,, ANTOINE CORTOT,*,,, AND JEAN FRÉDÉRIC COLOMBEL*,,, *Univ Lille Nord de France, F-59000, Lille; Hepato-Gastroenterology Department, CHU Lille, F-59000, Lille; UDSL, F-59000, Lille; Inserm U995, F-59000, Lille; Hepato-Gastroenterology Department, Douai General Hospital, Douai; and # Pathology Department, CHU Lille, F-59000, Lille, France See related article, Assa ad AH et al, on page 1593 of Gastroenterology. BACKGROUND & AIMS: Eosinophilic gastroenteritis (EGE) is a rare gastrointestinal disorder; little is known about its natural history. We determined the clinical features and long-term outcomes of patients with EGE. METHODS: We reviewed files from 43 patients diagnosed with EGE who were followed from January 1988 to April The diagnosis was made according to standard criteria after other eosinophilic gastrointestinal disorders were excluded. We analyzed data on initial clinical presentation and long-term outcomes. RESULTS: EGE was classified as mucosal, subserosal, or muscular in 44%, 39%, and 12% of cases, respectively. Disease location was mostly duodenal (62%), ileal (72%), or colonic (88%); it was less frequently esophageal (30%) or gastric (38%). Blood eosinophilia (numbers 500/mm 3 ) was observed in 74% of cases. Spontaneous remission occurred in 40% of patients; the majority of treated patients (74%) received oral corticosteroids, which were effective in most cases. After a median follow-up period of 13 years ( years), we identified 3 different courses of disease progression: 18 patients (42%; 9 with subserosal disease) had an initial flare of the disease without relapse, 16 (37%) had multiple flares that were separated by periods of full remission (recurring disease), and 9 (21%) had chronic disease. CONCLUSIONS: The clinical presentation of EGE is heterogeneous and varies in histologic pattern; about 40% of patients resolve the disease spontaneously, without relapse. Approximately 50% have a more complex disease, which is characterized by unpredictable relapses and a chronic course. Keywords: Eosinophilic Gastrointestinal Disorders; Treatment; Long-Term Outcome; Stomach. Eosinophilic gastroenteritis (EGE) is a rare primary gastrointestinal disorder of unknown etiology characterized by the presence of an intense eosinophilic infiltrate on histopathology of 1 or multiple segments from the esophagus to the rectum. 1,2 The Klein classification 3 arbitrarily divided patients with EGE into those with predominantly mucosal, muscle layer, or subserosal disease, relying on the concept that clinical presentation is dependent on the predominant involved layer of the gastrointestinal tract. EGE causes a wide array of gastrointestinal symptoms, and its diagnosis requires a high degree of clinical likelihood, given the nonspecific presentation and physical examination findings. 1,4 Since its first description by Kaijser 5 in 1937, less than 300 cases have been reported, most of them as single case reports or rather small case series The etiology of EGE remains obscure. An underlying allergic mechanism is generally considered, although firm demonstration is lacking. 4 Main therapeutic options are represented by oral corticosteroids for a short period with good efficacy. 6 EGE usually appears as a benign disease, but its natural history is almost unknown because previous series reported a limited number of patients with a short-term follow-up. The aim of our study was to evaluate the clinical spectrum of EGE at diagnosis, to assess the value of clinicopathologic correlations, and to report the natural history and the long-term outcome of the disease in 43 adult patients diagnosed at major French referral centers during the past 2 decades. Methods From January 1988 to April 2009, all patients with EGE criteria referred to the Hepato-Gastroenterology Department at the Lille University Hospital, France were prospectively identified and recorded with the diagnosis of EGE. Files from 38 patients were retrospectively reviewed by a single operator. During the same period, files from 7 patients followed in other French academic medical centers were also reviewed. Data from 2 patients have been previously published as case reports. 11,12 Case Definition EGE was defined by the presence of gastrointestinal symptoms associated with a pathologic eosinophilic infiltrate of the intestine wall without evidence of parasitic infection or other causes of secondary eosinophilic infiltration. 10 Slides from all patients were analyzed at time of diagnosis by experienced histopathologists to confirm the presence of a pathologic eosinophilic infiltrate of the intestine. Pathologic eosinophilic Abbreviation used in this paper: EGE, eosinophilic gastroenteritis; IQ, interquartile. Jean-Yves Canva died in May We remember our esteemed colleague and recognize his contribution to this article by the AGA Institute /$36.00 doi: /j.cgh

2 November 2011 NATURAL HISTORY OF EGE 951 infiltrate was defined as a dense and diffuse eosinophilic infiltration of the lamina propria ( 20 eosinophils per high-power field [ 400] on optical microscopy 8,10 ) associated with other histopathologic findings such as eosinophilic infiltration of the epithelium, eosinophilic cryptitis, presence of degranulating eosinophils, or extension of the eosinophilic infiltrate into the muscularis mucosae and submucosa. When present, eosinophilrich ascites and eosinophilic pancreatic infiltrate were also regarded as diagnostic criteria. Patients were excluded if they had a history or diagnosis of any of the following: parasitic or bacterial infections, inflammatory bowel diseases, celiac disease, any kind of neoplasia, or hypereosinophilic syndrome as defined as peripheral blood eosinophilia, more than 1500 eosinophils per mm 3 for at least 6 months associated with infiltration, and end-organ damage related to hypereosinophilia. Data Collection Data were collected on demographic characteristics, presenting symptoms, number of years of symptoms before diagnosis, number of years of follow-up, history or family history of allergy, a history of specific food intolerance or food allergy, tobacco and alcohol consumption, and past treatments. Serum total immunoglobulin E level, absolute blood eosinophils count, C-reactive protein, and biological evidence of intestinal malabsorption or protein-losing enteropathy were collected when available. Results of upper and lower gastrointestinal endoscopies, abdominal sonography and/or computed tomography, barium swallow, and esophageal manometry were recorded. All clinical and laboratory data were transcribed from the medical record onto a computer coding sheet for analysis. Biopsy specimens were available from all patients. Classifications of Eosinophilic Gastroenteritis Patients were divided into 3 groups according to Klein classification 3,10 : (1) patients with predominantly mucosal disease, defined as infiltration of the mucosa by eosinophils with no evidence of infiltration of the muscular layer and/or gastrointestinal obstruction and eosinophilic ascites; (2) patients with predominant disease of the muscle layer, defined as documented complete or incomplete bowel obstruction and/or eosinophilic infiltration of the muscle layer with no evidence of eosinophilic ascites; and (3) patients with predominant subserosal disease, defined by the presence of eosinophilic infiltration of the gut and eosinophilic ascites. A fourth group was identified as a predominant pancreatic disease, defined by the presence of eosinophilic infiltration of the gut and pancreas leading to biliary obstruction and jaundice. Patients with mucosal involvement who also had muscle layer infiltration were included in the group of muscle layer disease of EGE. Patients with transmucosal disease and ascites were included in the subserosal group. Patients with EGE and mucosal involvement were also divided into groups according to predominantly involved intestinal segment(s). Predominantly involved intestinal segment(s) was defined by the presence of macroscopic lesions (assessed during endoscopic or radiologic examinations) and histologic eosinophilic infiltration of the mucosa. Follow-up After initial diagnosis, data were collected on disease course and long-term outcome. To appreciate the long-term outcome of the disease, a standardized questionnaire was completed during phone contact with patient s family care physician and, if needed, with the patient himself. Presence of clinical remission after first episode, need for treatment at diagnosis, type of drug used, need for steroid therapy, steroid-dependent or steroid-resistant status of the disease, number of relapses, and occurrence of complications such as neoplasia or lymphoproliferative disorders were reviewed. Relapse was defined as a recurrence of a similar clinical presentation with or without recurrence of blood eosinophilia. Statistical Analysis The number of cases of EGE during the study period determined the sample size. A descriptive analysis was performed for the whole sample. Groups of EGE were compared for the qualitative variables by using Pearson 2 test or Fisher exact test and for quantitative variables by using Wilcoxon Mann Whitney test or Kruskal Wallis test, when necessary. Two-tailed significance tests were used. Kaplan Meier analysis with log-rank statistics was performed to assess for differences in time to relapse of the disease. A P value.05 was considered statistically significant. Results Clinical Presentation and Data at Diagnosis During the study period, 43 patients with EGE were identified, 24 male and 19 female. Main characteristics of patients at diagnosis are presented in Table 1. All patients but one were adults at diagnosis. Age at diagnosis ranged from years old. According to Klein s classification, distribution between predominantly mucosal, muscle layer, and subserosal disease was 44% (19/43), 12% (5/43), and 39% (17/43), respectively. Two patients (5%) had a pancreatic predominant disease, characterized by jaundice and cephalic pancreatic mass. Most frequent presenting symptoms were permanent or episodic abdominal pain in 88% of cases. Other frequent presenting symptoms were diarrhea, nausea/vomiting, bloating, and ascites, reported in 58%, 46%, 35%, and 21% of cases, respectively. History of allergy was present in 44% of cases (19/43), and a family history of allergy was recorded in 28% of cases (12/43). Peripheral blood eosinophilia (eosinophils 500/mm 3 ) was present in 74% of cases (32/42) with a median eosinophil count of 1900/mm 3 (600 24,530/mm 3 ). Gastroscopy, colonoscopy, and ileoscopy were performed in 93% (40/43), 81% (35/43), and 65% (28/43) of cases, respectively. Macroscopic lesions were found at gastroscopy in 53% of cases. Most patients had nonspecific erythema, edema, or superficial ulcerations; 1 patient had stenotic esophagitis; 3 patients had gastric polyps; and 1 had eosinophilic infiltration of the stomach mimicking peptic ulcer of the antrum. Macroscopic lesions were found during colonoscopy in 20% of cases. Most patients had nonspecific mucosal edema, erythema, inflammatory polyps, and ulcerations; 1 patient had proctitis; 1 had ulcerated pancolitis; and 1 had ulcerated left-sided colitis with stenosis. Seven patients had macroscopic abnormalities at ileoscopy such as mucosal edema and erythema without ulcerations; 1 patient had inflammatory ileal stenosis. Histologic samples were available for all patients. Fourteen patients (33%) had biopsies of all intestinal segments (esophagus, stomach, duodenum, ileum, and colon), and the majority of patients had 3 or more sites with biopsies (85%). When

3 952 PINETON DE CHAMBRUN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 9, No. 11 Table 1. Clinical and Biological Characteristics of Patients With EGE at Diagnosis (n 43) No. of patients, n (%) 26 (70) Male 24 (56) Female 19 (44) Male/female ratio 1.3 History of allergy 19 (44) Asthma 8 (19) Pollen allergic rhinitis 8 (19) Food allergy 5 (12) Chronic sinusitis 2 (5) Atopic eczema 1 (2) Urticaria 1 (2) Insect bite 1 (2) Family history of allergy 12 (28) Median age at first symptoms (IQ range) (y) 32.3 ( ) Median age at diagnosis (IQ range) (y) 40 ( ) Mean time between first symptoms and 3.8 ( 1.05) diagnosis (y) ( SEM) General signs a 21 (49) Mean body weight loss (kg) ( SEM) 5.36 ( 0.92) Blood eosinophilia 32 (74) Median blood eosinophil count (IQ range) 1300/mm 3 ( ) Elevated total immunoglobulin E blood count b Median immunoglobulin E level (IQ range) KU/L ( ) Protein-losing enteropathy 1 (2) Iron deficiency (low ferritin and iron blood 4 (9) levels) Anemia (caused by iron deficiency) 2 (5) SEM, standard error of mean. a Asthenia, anorexia, body weight loss. b Analysis was only available for 37 patients, normal 91 KU/L. biopsy results were available, a pathologic eosinophilic infiltrate was found in 30%, 38%, 62%, 72%, and 88% of cases in esophagus, stomach, duodenum, ileum, and colon, respectively (Supplementary Table 1). Most patients had multiple areas of disease, with at least 2 intestinal segments involved in 70% of cases. Among patients with pathologic eosinophilic infiltrate of the esophagus (6/20), diagnosis of eosinophilic esophagitis was ruled out, because all patients presented another involved intestinal segment. Of these patients, 2 had dysphagia, and 1 had gastroesophageal reflux disease. Peritoneal fluid was available in 12 patients with subserosal disease and was mostly protein-rich (median, 43 g/l; range, g/l), with numerous elements (median, 6200/mm 3 ; range, ,500/mm 3 ) and a large proportion of eosinophils (median, 78%; range, 39% 96%). Clinicopathologic Correlation Comparisons between the 3 anatomical patterns of EGE (predominant mucosal, muscle layer, and subserosal diseases) and clinical and biological characteristics at diagnosis are summarized in Table 2. Median time between first symptoms and diagnosis was shorter in subserosal disease as compared with mucosal disease and muscle layer disease (0.2 years, interquartile [IQ]1-3, vs 1.3 years, IQ1-3, , P.048 and vs 9 years, IQ1-3, , P.021, respectively). Regarding presenting symptoms, nausea/vomiting were more frequent in subserosal disease (73.7%), as compared with mucosal disease (31.6%) and muscle layer disease (20%) (P.029). The proportion of patients with peripheral blood eosinophilia was similar in the 3 groups. However, patients with subserosal disease presented a higher median blood eosinophil count (3833/mm 3 ; IQ1 3, ) as compared to patients with mucosal (1100/mm 3 ; IQ1 3, ) and muscle layer disease (875/mm 3 ; IQ1 3: ) (P.0022 vs mucosal disease and P.19 vs muscle layer disease). In patients with EGE and mucosal involvement (39/43), 5 types of disease localization were defined, according to predominantly involved intestinal segment(s): gastric disease (8%), proximal small intestine disease (13%), terminal ileitis (20%), isolated colitis (8%), and extensive disease (38%). Extensive disease was defined as the involvement of 2 or more intestinal Table 2. Clinical and Biological Characteristics of Patients With EGE According to Klein s Classification 3a Mucosal disease (n 19) Muscle layer disease (n 5) Subserosal disease (n 17) P value Symptoms (%) Nausea/vomiting Dysphagia NS Reflux NS Bloating NS Abdominal pain NS Diarrhea NS Gastrointestinal bleeding NS Sex ratio (M/F) NS History of allergy (%) NS Age at diagnosis (y) (mean SEM) NS Median time between first symptoms 1.3 ( ) 9 ( ) 0.2 ( ).05 and diagnosis (y) (IQ range) Blood eosinophilia (%) NS Median blood eosinophil count (IQ 1100/mm 3 ( ) 875/mm 3 ( ) 3833/mm 3 ( ).012 range) Elevated serum immunoglobulin E level (%) NS SEM, standard error of mean. a Patients with pancreatic disease were not included in the analysis.

4 November 2011 NATURAL HISTORY OF EGE 953 Table 3. Clinical and Biological Characteristics and Natural History of Patients With EGE and Mucosal Involvement According to Disease Localization Variables Gastric disease (n 3) Proximal small intestine disease (n 5) Terminal ileitis (n 8) Isolated colitis (n 3) Extensive disease (n 15) Eosinophilic infiltration without macroscopic lesion (n 5) Median age at diagnosis (y) 50.7 ( ) 32.5 ( ) 27.4 ( ) 28.5 ( ) 40.4 ( ) 33.5 ( ) (IQ range) Years of symptoms before 1.9 ( ) 0.1 (0.1 4) 0.4 ( ) 0.1 ( ) 2.1 ( ) 0.4 (0.3 1) diagnosis (IQ range) History of allergy (%) Symptoms(%) Nausea/vomiting Dysphagia Reflux Bloating Abdominal pain Diarrhea Gastrointestinal bleeding Median blood eosinophil 1300 ( ) 2800 ( ) 1300 ( ) 1300 ( ) 1300 ( ) 435 ( ) count (eosinophils/ mm 3 ) (IQ range) Disease course patterns (%) Single flare Recurring course Continuous course Spontaneous remission at diagnosis (%) Disease recurrence after diagnosis (%) a a P.01 by Fisher exact test. segments. Five patients (13%) presented a pathologic eosinophilic infiltrate of the intestine without any macroscopic lesion. Table 3 summarizes clinical and biological characteristics at diagnosis according to disease localization. There were no significant differences in clinical presentation or history of allergy between disease localizations. Treatment The proportion of patients with EGE treated at diagnosis was 63% (27/43). Spontaneous remission after diagnosis was observed in 40% of cases (17/43). In 2 patients, specific food allergens were identified (milk and cheese for 1 patient and citrus fruits for the other). Avoidance of incriminated allergens was inefficient at resolving EGE symptoms. The most frequent treatment used was oral corticosteroid therapy in 74% of treated patients (20/27). Corticosteroids were given orally at the dose of mg/d for a short duration (1 week to 4 months) and were rapidly tapered. Corticosteroid therapy was efficient in 95% of patients (18/19). Only 1 patient did not respond to corticosteroids and presented a few weeks later as spontaneous remission. Antihistamines such as ketotifen and cetirizine (H 1 - receptor antagonists) were administered to only 5 patients. Sodium chromoglycate was used in 3 patients at diagnosis and was efficient for 1 patient. Long-term Follow-up and Natural History The median duration of follow-up was 13.1 years (IQ1-3, ). Relapse of the disease after diagnosis was observed in 33% (14/43) of all cases and in 60% (12/20) of patients who were treated with corticosteroids at diagnosis. Blood eosinophil counts were available during relapse in 6 patients and were elevated in all. Among patients treated with corticosteroids Figure 1. Kaplan Meier analysis for clinical relapse of EGE according to presence of spontaneous remission or hypereosinophilia. (A) Proportion of patients free from clinical relapse during follow-up according to presence of spontaneous remission after diagnosis. (B) Proportion of patients free from clinical relapse during follow-up according to presence of hypereosinophilia at diagnosis.

5 954 PINETON DE CHAMBRUN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 9, No. 11 Figure 2. Course of eosinophilic gastroenteritis. After median follow-up of 13 years, we identified 3 different eosinophilic gastroenteritis courses. Forty-two percent (n 18) of patients presented a single flare with no relapse (A), 37% (n 16) a recurring course with multiple flares and period of full remission (2 months to several years) (B), and 21% (n 9) a continuous course (C). Seven of 9 patients with continuous course, 12 of 16 patients with recurring course, and 8 of 18 patients with single flare received treatment at diagnosis. at diagnosis, 3 (15%) were steroid-dependent because of relapsing disease during a tapered course of treatment, and 9 (45%) presented a clinical relapse after treatment withdrawal. Of these 9 patients, all were retreated by corticosteroids with good efficacy, and one needed maintenance treatment because of frequent relapse of the disease. During follow-up, 26% of EGE patients (11/43) were on maintenance therapy: 5 with corticosteroids, 2 with ketotifen, 2 with cetirizine, and 2 with sodium chromoglycate. At the maximum follow-up, 21% of patients remained on treatment. No myeloproliferative transformation of EGE was observed. A lower risk of clinical relapse was observed in patients with initial spontaneous remission as compared with patients who needed treatment at diagnosis (Figure 1A). Hypereosinophilia at diagnosis was associated with an increased risk of clinical relapse (Figure 1B). Although there was a trend for more relapses in patients with a history of allergy as compared with those without, the difference was not significant. Patients with clinical relapse after initial diagnosis also had higher blood eosinophil counts at diagnosis as compared with patients without relapse, with a median blood eosinophil count of 3035/ mm 3 (IQ1-3, ) versus 1100/mm 3 (IQ1-3, , P.042). Three different patterns of disease course were observed (Figure 2): (1) 18 of 43 patients (42%) presented a single flare of EGE, defined by clinical symptoms present for 6 months associated with the absence of any relapse after initial flare; (2) 16 of 43 patients (37%) presented a recurring course of EGE, defined by at least 2 flares of the disease separated by a period without digestive symptoms and without blood eosinophilia; (3) 9 of 43 patients (21%) presented a continuous course of the disease, defined by chronic persistent gastrointestinal symptoms for 6 months without period of remission. In the recurring course group, the mean number of flares of EGE was 5.2 (range, 2 15). Intervals between disease flares were extremely variable, ranging from 2 months to several years. Median time between first symptoms and diagnosis was significantly lower in single flare (0.2 years, IQ1-3, ) compared with continuous (2.2 years, IQ1-3, 1.9-9, P.0001) and recurring (1.5 years, IQ1-3, , P.0039) courses of EGE. The distributions of EGE anatomical patterns were significantly different between the 3 course groups (P.025) (Figure 3A). Continuous course was principally represented by predominant mucosal disease (8/9, 89%), whereas predominant subserosal disease represented 50% of recurring course (8/16) and single flare (9/18). According to Klein classification, the predominant mucosal disease group presented a majority of continuous courses (8/19, 42%), whereas the predominant muscle layer disease group presented a majority of recurring courses (3/5, 60%). Subserosal disease evolved as a single flare in 53% of cases (9/17) and with a recurring course in 47% of cases (8/17) without evolving as continuous course (Figure 3B). Distribution of disease course patterns, according to predominantly involved intestinal segment in patients with EGE and mucosal involvement, is summarized in Table 3. Spontaneous remission was observed in only 20% of patients with extensive disease (3/15). Proximal small intestine and extensive diseases presented a higher risk of clinical recurrence after diagnosis as compared with other disease localizations (P.01). Figure 3. Correlation between Klein classification and natural history of EGE. (A) Distribution of EGE anatomical patterns in function of disease course. (B) Distribution of disease course in function of EGE anatomical patterns according to Klein. 3

6 November 2011 NATURAL HISTORY OF EGE 955 Discussion We report on a large series of adult patients with EGE. A total of 43 patients were identified in our center and other French referral centers during a 2 decade period. Since the first description in 1937, less than 300 cases of EGE were reported in limited case series or case reports. 3,7 9 Talley et al 10 reported the first largest case series of EGE, with 40 patients followed at the Mayo Clinic (Rochester, Minnesota) from In 2010, Chang et al 6 updated the Mayo Clinic experience of EGE by reporting 59 new cases since Overall, these data suggested that unlike eosinophilic esophagitis, EGE remains a rare intestinal disorder, with an estimated prevalence of 1 case per 100,000 patients. 6,10 Our results confirmed several characteristics of EGE that have already been reported. 3,6 10 EGE occurs mostly in the third and fourth decades of life, with a male predominance. 6,7,10 In our cohort, the median age at diagnosis was 40 years, and all patients but one were adults at diagnosis. In children, most frequent eosinophilic gastrointestinal disorders are allergic colitis and eosinophilic esophagitis, 2 and the number of reported cases of EGE is too small to allow meaningful comparisons with adults. Diagnosis of EGE might be elusive, and nonspecific gastrointestinal symptoms such as abdominal pain, diarrhea, nausea/vomiting, and bloating were inaugural in 88%, 58%, 46%, and 35% of our patients, respectively. In patients presenting unspecific abdominal symptoms, a high eosinophil blood count ( 500/mm 3 ), which was present in 74% of our patients, should be considered suggestive of EGE and lead to upper and lower endoscopy with systematic biopsies. As in previous studies, we arbitrarily divided our patients according to Klein classification. 3 In our cohort, predominant mucosal, muscle layer, and subserosal diseases were present in 44%, 12%, and 39% of cases, respectively. These results contrast with the report by Talley et al 10 that showed muscle layer disease was more frequent with a small number of subserosal diseases (30% and 12.5%, respectively) and with the more recent report by Chang et al 6 that demonstrated a large majority of predominant mucosal diseases (52 mucosal, 3 muscle layer, and 4 subserosal diseases). Chang et al suggested a shift in the clinical spectrum of EGE toward mucosal disease that was not apparent in our series. The natural history of EGE remains largely unknown. Indeed, only few studies reported follow-up after treatment that was often limited and always described in small cases series. 6 8,10 Therapy of EGE is not well established because this disease is rare, and there are a large proportion of patients who will present spontaneous remission, as we showed in our series with 40% of cases. Corticosteroids remain the mainstay of therapy for EGE, with a dramatic response in the majority of patients being reported and confirmed by our results (95%). 3,7,8,10,13,14 This study reports the longest follow-up in a large case series of EGE, with a median follow-up duration of 13 years. Disease relapsed in 33% of all patients and in 60% of patients who were treated with corticosteroids at diagnosis. Reduction of corticosteroid dose resulted in 15% of clinical recurrence, whereas 45% of patients relapsed after withdrawal of corticosteroid therapy, confirming the results of Naylor. 9 Clinical recurrences were treated with corticosteroids with good efficacy. However, at the maximum follow-up, 21% of our patients remained on treatment. Regarding the risk of recurrence, we found that absence of spontaneous remission and high blood eosinophil count at diagnosis were significantly associated with clinical relapse of EGE. We also observed that among patients with mucosal disease, proximal small intestine and extensive localizations were associated with a higher risk of relapse. We were eventually able to identify 3 different courses of EGE: single flare, recurring course, and continuous course in 42%, 37%, and 21%, respectively. The disease course was different between the 3 anatomical patterns of EGE according to Klein et al. 3 Predominant subserosal disease presented a majority of single flare and no continuous chronic course, whereas predominant mucosal disease presented mostly a continuous course, and predominant muscle layer disease presented a recurring course. It is worthy to note that in our cohort almost half of patients presented with a unique flare, which was rapidly diagnosed because of intense symptoms and did not present relapse during the follow-up. However, more than one-third of our patients presented at least 1 clinical relapse of EGE that required a new course of corticosteroid therapy. Although EGE is generally considered a harmless disease, its course might be more complex with refractory disease and need for maintenance therapy in some patients. In conclusion, this study reports the natural history and the long-term outcome of EGE in a case series of 43 patients. Clinical presentation of EGE was heterogeneous, depending on its histologic pattern. The evolution of the disease was often benign with no relapse. However, half of patients presented with a more complex natural history characterized by unpredictable relapses and a chronic course. Supplementary Material Note: To access the supplementary material accompanying this article, visit the online version of Clinical Gastroenterology and Hepatology at and at doi: / j.cgh References 1. Khan S. Eosinophilic gastroenteritis. Best Pract Res Clin Gastroenterol 2005;19: Yan BM, Shaffer EA. Primary eosinophilic disorders of the gastrointestinal tract. Gut 2009;58: Klein NC, Hargrove RL, Sleisenger MH, et al. Eosinophilic gastroenteritis. Medicine (Baltimore) 1970;49: Rothenberg ME. Eosinophilic gastrointestinal disorders (EGID). J Allergy Clin Immunol 2004;113:11 28, quiz Kaijser R. Zur Kenntnis der allergischen Affektionen des Verdauungs-Kanals vom Standpunkt des Chirurgien aus. Arch Klin Chir 1937;188: Chang JY, Choung RS, Lee RM, et al. A shift in the clinical spectrum of eosinophilic gastroenteritis toward the mucosal disease type. Clin Gastroenterol Hepatol 2010;8: Chen MJ, Chu CH, Lin SC, et al. Eosinophilic gastroenteritis: clinical experience with 15 patients. World J Gastroenterol 2003; 9: Lee CM, Changchien CS, Chen PC, et al. Eosinophilic gastroenteritis: 10 years experience. Am J Gastroenterol 1993;88: Naylor AR. Eosinophilic gastroenteritis. Scott Med J 1990;35: Talley NJ, Shorter RG, Phillips SF, et al. Eosinophilic gastroenteritis: a clinicopathological study of patients with disease of the

7 956 PINETON DE CHAMBRUN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 9, No. 11 mucosa, muscle layer, and subserosal tissues. Gut 1990;31: Beaugerie L. [Recurrent edema of the small intestine with ascites]. Ann Gastroenterol Hepatol (Paris) 1989;25: Torpier G, Colombel JF, Mathieu-Chandelier C, et al. Eosinophilic gastroenteritis: ultrastructural evidence for a selective release of eosinophil major basic protein. Clin Exp Immunol 1988;74: Cello JP. Eosinophilic gastroenteritis: a complex disease entity. Am J Med 1979;67: Leinbach GE, Rubin CE. Eosinophilic gastroenteritis: a simple reaction to food allergens? Gastroenterology 1970;59: Reprint requests Address requests for reprints to: Jean-Frédéric Colombel, MD, Clinique des maladies de l appareil digestif et de la nutrition, Hôpital Claude Huriez CHRU, 1 Place de Verdun, Lille, France. jean-frederic.colombel@wanadoo.fr; fax: 33 (0) Acknowledgments The authors thank Dr Jane C. Harrington, University of California, San Diego for her critical review of the paper and helpful advice. Conflicts of interest The authors disclose no conflicts.

8 956.e1 PINETON DE CHAMBRUN ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 9, No. 11 Supplementary Table 1. Distribution of Biopsy Specimens and Eosinophilic Infiltration in Patients With EGE (n 43) Intestinal segments No. of patients biopsied at each intestinal segment (n) Patients with eosinophilic infiltration, n (%) Esophagus 20 6 (30) Stomach (38) Duodenum (62) Ileum (72) Colon (88) Mean no. of intestinal segments biopsied per patient ( SD) Mean no. of intestinal segments with eosinophilic infiltration per patient ( SD) SD, standard deviation.