Eosinophilic Esophagitis: A Prevalent Disease in the United States That Affects All Age Groups

Article Outline

• Abstract
• Materials and Methods
  • Study Design and Database
  • Pathology Review
  • Inclusion and Exclusion Criteria
  • Statistical Methods and Analysis
• Results
• Discussion
• Acknowledgment
• References
• Copyright

Background & Aims: Most reports on eosinophilic esophagitis (EE) are limited to small series from single institutions. This study describes features of EE in a broader population. Methods: A national pathology database (Caris Diagnostics, Irving, TX) was used to identify EE cases from a cohort of upper endoscopies. Slides from potential cases were reviewed by pathologists. Study inclusion required a mean of (1) 20 eosinophils or more in 5 high-power fields, or (2) 30 eosinophils in 2–4 high-power fields. Endoscopists provided demographic and clinical information. Results: There were 363 cases identified from 74,162 patients and 26 states. EE had a male predominance (odds ratio, 3.0; 95% confidence interval, 2.4–3.8). Ages ranged from 1 to 98 years, including 42 children and 321 adults. The most common endoscopy indications in adults were dysphagia (70.1%) and gastroesophageal reflux disease (GERD)/heartburn (27.1%). Children most frequently reported GERD/heartburn (38.1%) and abdominal pain/dyspepsia (31.0%). A total of 25.1% of cases had a peak mucosal eosinophil count of 20–59, 29.2% had a peak mucosal eosinophil count of 60–100, and 45.7% had a peak mucosal eosinophil count of more than 100. There was no difference in the peak counts between age groups or sexes, but patients with dysphagia had higher counts (P < .001). The prevalence of EE increased during the study period (P < .001). Conclusions: EE is a national disease found in all age groups, more frequently in males. Dysphagia and GERD symptoms are common indications for endoscopy. The degree of eosinophilic infiltration is high throughout all ages and may be related to patients' symptoms. Our series highlights important elements of this disease, the prevalence and/or recognition of which is increasing.

Abbreviations used in this paper: EE, eosinophilic esophagitis, GERD, gastroesophageal reflux disease, HPF, high-power field

Recently there has been a high level of research and clinical interest in the entity eosinophilic esophagitis (EE). Although it initially was recognized that eosinophilic infiltration of esophageal mucosa could arise in the setting of gastroesophageal reflux disease (GERD),¹ more recent literature, first in case reports and series², ³ and subsequently in basic science work, has focused on a potential allergic and genetic causes.⁴, ⁵

Multiple studies have described epidemiologic, clinical, endoscopic, and histologic characteristics of the disease. These studies suggest that EE is more predominant in males, young adults, and children, and is associated commonly with a personal and family history of other allergic conditions.⁶, ⁷, ⁸, ⁹, ¹⁰ Clinically, the most common presentations are dysphagia and heartburn, but symptoms such as nausea and vomiting often are noted in children.⁶, ⁹, ¹¹, ¹² Unfortunately, the relatively few number of patients with EE reported in the literature has resulted in limitations on the conclusions one can reach from these studies. Many important epidemiologic and clinical questions remain to be validated fully. For example, because many EE studies examine patients from a single institution, is there any geographic variability in the disease? Is EE limited to specific age groups? How do previously described clinical, histologic, and epidemiologic characteristics vary across a broader population of patients? Has there been an increase in the incidence of EE? Finally, is the finding of EE in an older patient more likely to suggest gastroesophageal reflux as a cause and does the density of eosinophilic mucosal infiltration support this hypothesis? Overall, additional cases from a more diverse group of patients and physicians are necessary both to support current ideas on the disease and to answer some of these important questions.

By using a national pathology database, we were able to identify a large series of patients with EE from a large cohort of patients undergoing upper endoscopy with biopsy. The goal of this study was to better characterize demographic, clinical, and pathologic features of EE in this diverse patient population.

Back to Article Outline

Materials and Methods 

Study Design and Database 

This study identified patients diagnosed with EE using the Caris Diagnostics Database (based in Irving, TX). This database is derived from all patients referred to Caris Diagnostics, a provider of gastrointestinal pathology services for physicians from community-based freestanding endoscopy centers throughout the United States. Over the study period, the database contained 414,598 patient gastrointestinal pathology cases from 34 different states. This is a predominantly adult database, with 98.5% of patients age 18 or older. The cohort of patients included all adults and children who had upper endoscopies with biopsy of the esophagus and/or gastroesophageal junction from January 2002 through May 2006.

WinSURGE anatomic pathology software (Computer Trust Corporation, Boston, MA) was used to perform all database queries. An initial free-text search was performed to find upper-endoscopy cases between January 1, 2002, and May 31, 2006, in which the word “eosinophilic” was present in the diagnosis and/or comment text. The query results were exported into Microsoft Office Excel 2003 (Microsoft Corporation, Redmond, WA), where manual review of the queried pathology reports was used to eliminate cases for which EE was in fact not originally diagnosed, suggested, or part of the differential diagnosis, histologically. Clinical data for each patient, as provided by the endoscopist, were entered into the database. These included demographic information and the procedure indications for all patients. When available, additional clinical information from endoscopy reports or pathology requisitions also was recorded. These data were provided via various electronic medical record platforms, as well as handwritten formats.

Pathology Review 

Once the potential EE cases were identified, a pathology review was performed. All slides were retrieved for each case, and all cases were reviewed by 1 of 3 gastrointestinal pathologists (R.L., C.T., and S.A.). Each case was assessed for the number of eosinophils per 400× high-power fields (HPFs, 2.388 mm²), selecting the 5 most densely populated fields of esophageal squamous epithelium. When tissue was limited to less than 5 fields, assessment was made for those available. A field required that tissue fill at least 50% of the 400× view. Eosinophils were not counted beyond 100/HPF (and in that case were recorded as >100). When gastric or duodenal biopsies were included, prominent eosinophilia (more than rare, scattered eosinophils) was noted to evaluate for the diagnosis of eosinophilic gastroenteritis. The pathologists were blinded to the original pathology reports.

Inclusion and Exclusion Criteria 

Because there is no clear consensus on the definition of EE, we used the following 2 criteria to define our cases: (1) a mean eosinophil count of 20 or more per HPF in 5 HPFs, or (2) when 5 HPFs were not available, a mean of 30 eosinophils or more in 2–4 HPFs. Patients were excluded if there was a prior history of EE or if biopsies identified a prominent eosinophilic infiltrate in the stomach or duodenum.

Statistical Methods and Analysis 

Statistical analysis was performed with Stata version 9.1 (StataCorp, College Station, TX).

Differences between means were tested using a t test. We calculated unadjusted odds ratios with 95% confidence intervals as a measure of association when comparing EE in men vs women. A 2-sample test of proportions was performed to examine differences in sex between adults and children. A Pearson chi-squared statistic was calculated when comparing patients across groups on categoric variables and a chi-square test for trend was used when the order of the groups was believed to be important. The presence of an increasing prevalence of EE from 2002 to 2005 was assessed using the more conservative nonparametric Cuzick test for trend.¹³ All patient data were de-identified before analysis, and Health Insurance Portability and Accountability Act (HIPAA) guidelines were followed by the Caris Diagnostics Corporation.

Back to Article Outline

Results 

During the study period, esophageal and/or gastroesophageal biopsy specimens were obtained from upper endoscopies in 74,162 unique patients. In this cohort of patients, roughly 98% were adults, and 2% were younger than 18 years. The cohort represented 217 different clients (≥1 physician at each client endoscopy center) in 34 different states for adult cases, and 74 clients in 23 states for pediatric cases. Over the 5-year study period, the age and sex were very similar year to year, with the yearly average age ranging from 55.4 to 56.3 years, and the yearly male percentage ranging from 49.8% to 51.3%.

The initial free-text search identified 1,052 upper-endoscopy cases in which the word “eosinophilic” was present in the diagnosis and/or comment text. After reviewing all 1052 pathology reports, a total of 360 cases were excluded for the following reasons: 341 cases in which the term “eosinophilic esophagitis” was mentioned as a pertinent negative finding, 13 cases of eosinophilic gastritis, 3 cases of eosinophilic colitis (when upper and lower endoscopies were performed simultaneously), and 1 case of eosinophilic duodenitis. In the other 692 cases, EE was either diagnosed, suggested, or part of the pathologist's differential diagnosis. After the histologic re-review, 329 cases did not fulfill our histologic inclusion criteria. Therefore, we identified a total of 363 cases, 308 according to criterion 1 and 55 according to criterion 2. In 345 patients a single esophageal specimen jar was submitted. In 17 patients, 2 jars specifying 2 specific esophageal locations, and in 1 patient 3 jars specifying 3 esophageal locations were submitted.

Of the cases meeting the inclusion criteria, 270 (74.4%) were men and 93 (25.6%) were women. When compared with the gender characteristics of the entire cohort, patients with EE were much more likely to be male than female (odds ratio, 3.0; 95% confidence interval, 2.4–3.8; P < .001). The mean age of cases was 37.6 years (range, 14 mo to 98 y). Forty-two children (age, <18 y) were identified from 12 states and 15 different client endoscopy centers, and 321 adults met criteria from 26 states and 82 different endoscopy centers. Figure 1 shows the age distribution of the cases. Among children, there were 9 females (21%) and 33 males (79%), and among the adult cases there were 84 females (26%) and 237 males (74%). There was no statistically significant difference in the proportion of males between adults and children (P > .1). The diagnosis of EE was made in a higher proportion of the pediatric cases (2.95%; 42 of 1424) than adult cases (0.44%; 321 of 72,738).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 1. 

  Age distribution of EE cases.

Biopsy specimens were received from 34 states with cases being identified in 26 of the 34 states (Figure 2). Table 1 shows the breakdown of upper endoscopies from each state and also the number of cases from each state. The median number of procedures with biopsies received from each state was 1196 (range, 14–10,348). States in which EE was identified had an average of 2812 endoscopies submitted to the database (range, 61–10,368) vs an average of 130 in those without EE identified (range, 14–534; P < .001).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 2. 

  Map of the United States showing states in which biopsy specimens were received and cases of EE were identified. , States in which EE was diagnosed; , states in which EE was not diagnosed; , states in which no data were collected.

Table 1. States Receiving Biopsy Specimens and Cases of EE From Each State

The peak eosinophil count represents the highest number of eosinophils counted in 1 HPF for each patient. To allow for additional analysis of our cases we divided them into 3 groups based on the peak mucosal eosinophil count (peak mucosal eosinophil counts of 20–59, 60–100, and >100). Many of the cases had very high peak eosinophil counts. A total of 25.1% had a peak mucosal eosinophil count of 20–59, 29.2% had a peak mucosal eosinophil count of 60–100, and 45.7% had a peak mucosal eosinophil count of greater than 100. We also examined the distribution of peak counts in children (age, <18 y: n = 42), young adults (age, 18–39 y; n = 162), and older adults (age, >39 y; n = 159). There was not a statistically significant difference across these age groups (Pearson chi square (4) = 6.9, P = .14) (Figure 3). In addition, there was no statistically significant difference in the peak mucosal eosinophil count between men and women (Pearson chi square (2) = 1.2249, P = .54). We also analyzed patients whose indication for endoscopy was dysphagia vs those who did not have dysphagia as an indication and found that patients with dysphagia were more likely to have higher peak mucosal eosinophil counts (chi-square test for trend (1) = 14.5; P < .001).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 3. 

  Peak mucosal eosinophil count by age group. The proportion of patients within each eosinophil count group did not differ by age group, χ² (4, n = 363) = 6.9, P = .14).

The most common indications for endoscopy in adult EE cases were dysphagia (70.1%), GERD/heartburn (27.1%), and abdominal pain/dyspepsia (13.1%), with additional indications listed in Table 2. The most common indications in children (age, <18 y) were GERD/heartburn (38%), abdominal pain dyspepsia (31%), and dysphagia (26%), with additional indications listed in Table 3. There is a statistically significant difference between the proportion of children and adult cases with dysphagia (P < .001). To explore potential differences in indications between men and women we combined several similar indication categories and found that the distribution of the major indications did not differ by sex (Figure 4).

Table 2. Indications for Upper Endoscopy in Adult Cases (n = 321)

Table 3. Indication for Upper Endoscopy in Cases Younger Than 18 Years (n = 42)

• 
  • View Large Image
  • Download to PowerPoint

• Figure 4. 

  Indication by sex. P value for the Pearson chi-square comparing men and women for each indication. ^aPatients could have more than one indication. Nausea/vomiting (N/V).

Patients with dysphagia listed as the primary indication for endoscopy were identified from the entire database and the prevalence of EE was determined in this subset. The year 2006 is excluded from this analysis because we do not have data from the full 12 months. Overall, there were 12,465 upper endoscopies for the indication of dysphagia with biopsies submitted to the database between January 1, 2002, and December 31, 2005. From this group there were 174 cases of EE. The prevalence of EE in this cohort increased significantly from 2002 to 2005 (Cuzick test for trend: P < .001) with a prevalence of 0.1% (1 of 726) in 2002, 0.9% (20 of 2226) in 2003, 1.2% (43 of 3621) in 2004, and 1.9% (110 of 5892) in 2005 (Figure 5).

• 
  • View Large Image
  • Download to PowerPoint

• Figure 5. 

  Prevalence of EE in patients with dysphagia.

Back to Article Outline

Discussion 

This study examined a large population of patients with EE derived from a national database. There are several strengths of this study, most notable is that it is one of the largest series of patients with EE to date. Because cases were derived from samples submitted by community practices in 34 of the 50 states it is both geographically diverse and less prone to referral center bias. Cases were nested within the larger cohort of all patients who had undergone an upper endoscopy with biopsy samples submitted to this pathology company. We believe that there was no significant selection bias in this cohort because essentially no patients were derived from tertiary referral centers, and it is our understanding that the vast majority of centers send almost all, if not all, specimens to this facility. Furthermore, as subspecialty gastrointestinal pathologists, we believe that no cases are diverted to other institutions by clinicians based on perceived complexity or difficulty. Although we had limited patient level data on this larger cohort, we were able to perform several unadjusted calculations to explore some of the basic differences between the cases and noncases as well as look at some trends in prevalence among the cases received at the Caris Diagnostics laboratory. As a result, several important conclusions can be made and prior observations in smaller series of patients can be corroborated.

Much like other case series, we found an approximately 3:1 male to female ratio within our cases in both adults and children.⁶, ¹⁴ This association remained when we examined the cases in comparison with the entire cohort of endoscopies submitted to the database. This lends further statistical strength to the notion that EE is a disease with a significant male predominance. The reasons for this are unclear but genetic factors may be important as outlined in recent studies.¹⁵

Many physicians often think of EE as a disease diagnosed in children and young adults. Our series shows that EE is a disease that spares no age group in that a large proportion of patients were older than age 40 and multiple cases were identified in patients older than age 80. In fact, the largest proportion of patients with EE was found in the 30- to 40-year-old range, not in younger adults and children. There was no difference in gender ratio or peak eosinophil counts across age groups; therefore, there is no need to vary histologic definitions between adults and children.

Although our study was not a true population-based epidemiologic analysis of disease prevalence, by using a national database we were able to confirm that EE truly is a national disease found in all regions of the United States. Cases were identified only in 26 of the 34 states from which biopsies were submitted, but this likely is owing to the relatively small number of samples that were submitted from the other states.

One concern with our findings is that there might be misclassification bias, specifically that patients may have EE from gastroesophageal reflux as opposed to allergy. Several factors in our study design and results mitigate this bias. For example, we opted to use a conservative pathologic definition of EE to define our cases, which would ensure minimal misclassification. The majority of the cases did not fall near this lower cut-off level but tended to have much higher mucosal eosinophilic counts, a pattern not typically seen with reflux-induced eosinophilic infiltrate in which counts of less than 10 per HPF are the norm.⁷, ¹⁶ Furthermore, the distribution of peak mucosal eosinophil counts was similar across age groups, to the older extremes, which casts doubt on reflux in the elderly as a cause of EE.

As with other series, adults with EE had a high prevalence of dysphagia and heartburn, whereas children had a more subtle and varied presentation.⁶, ⁸, ⁹, ¹² In fact, adults reported dysphagia more than all other indications combined whereas in children it was only the third most common indication, suggesting that these groups differ in their clinical presentations. Although other studies such as the study by Remedios et al¹¹ had a high percentage of cases presenting with food impaction (42.3%), our study had only 7 of the 363 cases listing food impaction as an indication. The reason for this difference is not known but we postulate that many patients receiving upper endoscopy with biopsy after an episode of food impaction are listed as having dysphagia as their indication. An alternative explanation would be that series from referral centers tend to oversample these emergency-type presentations. Another interesting observation from our data is that patients with dysphagia tended to have higher peak mucosal eosinophil counts than those without dysphagia, suggesting that the amount of infiltration of the mucosa is important in the development of this symptom.

In the 4 complete years of data that were analyzed, the prevalence of EE within the subset of patients who underwent a biopsy for dysphagia increased. Cherian et al¹⁷ showed a similar increase in a pediatric population. There are 2 possible explanations for this trend. The first is that the prevalence truly is increasing, representing an epidemic of sorts. The second is that gastroenterologists have become more knowledgeable about the disease and their choice of who should undergo a biopsy has become more targeted. Population-based studies from geographically diverse populations certainly would help to answer this and other questions.

There were several limitations to this study. As with any database study we were limited by the information available in the database. Although we had endoscopy indications for all cases, we had access to a limited number of full endoscopy reports, and additional clinical history was available only when it was provided on the pathology requisitions forms. Thus, we cannot comment in detail with regard to disease presentation or other clinical factors. Our criteria to identify cases of EE were somewhat conservative. According to criteria used in other studies we may have overlooked more subtle cases of EE. Also, by searching our database for cases referencing the term “eosinophilic,” it would necessarily underreport some cases, especially in the earlier years of the study. In addition, some cases might have been regarded as severe reflux esophagitis, based on the limited understanding of this disease and the criteria for its diagnosis at that time. Another reason for likely under-recognition of cases of true EE is sampling error, an issue with all biopsies for any disease, but especially with one known to have a patchy distribution.¹⁸ We had some patients in whom only a single small biopsy specimen was obtained; nevertheless, a diagnosis of EE was confirmed in some of these, even with the stricter criteria used in this study for such limited samples. Finally, the fact that we did not have detailed patient level data on the non-EE cases within the database meant that we were not able to perform multivariable regression or look at other comparisons of EE cases vs noncases. As suggested earlier, the study is not an epidemiologic evaluation of the national population; rather, it is a report of the demographic, clinical, and pathologic characteristics of this condition across a geographically diverse group of patients with EE.

In conclusion, this study confirmed several well-known characteristics of EE, such as the male predominance and the clinical presentation on a much broader geographic scale. More importantly, our study highlights the need to consider this disease in patients with presentations other than dysphagia as well as in older patients. Finally, our observations support further investigation into the increasing prevalence of the disease and how the geographic variations may play into the pathophysiology of the disease. Certainly, we are a step closer to fully understanding EE and to increasing awareness in physicians who could identify and thereby treat patients with this condition.

Back to Article Outline

The authors are very grateful to Mike Elam for his technical assistance with the database.

Back to Article Outline

References 

1. Winter HS, Madara JL, Stafford RJ, et al. Intraepithelial eosinophils: a new diagnostic criterion for reflux esophagitis. Gastroenterology. 1982;83:818–823
   • View In Article
   • Abstract
2. Attwood SE, Smyrk TC, Demeester TR, et al. Esophageal eosinophilia with dysphagia (A distinct clinicopathologic syndrome). Dig Dis Sci. 1993;38:109–116
   • View In Article
   • MEDLINE
   • CrossRef
3. Vitellas KM, Bennett WF, Bova JG, et al. Idiopathic eosinophilic esophagitis. Radiology. 1993;186:789–793
   • View In Article
   • MEDLINE
4. Akei HS, Mishra A, Blanchard C, et al. Epicutaneous antigen exposure primes for experimental eosinophilic esophagitis in mice. Gastroenterology. 2005;129:985–994
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (250 KB)
   • CrossRef
5. Blanchard C, Wang N, Rothenberg ME. Eosinophilic esophagitis: pathogenesis, genetics, and therapy. J Allergy Clin Immunol. 2006;118:1054–1059
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (546 KB)
   • CrossRef
6. Sgouros SN, Bergele C, Mantides A. Eosinophilic esophagitis in adults: a systematic review. Eur J Gastroenterol Hepatol. 2006;18:211–217
   • View In Article
   • CrossRef
7. Parfitt JR, Gregor JC, Suskin NG, et al. Eosinophilic esophagitis in adults: distinguishing features from gastroesophageal reflux disease: a study of 41 patients. Mod Pathol. 2006;19:90–96
   • View In Article
   • MEDLINE
   • CrossRef
8. Potter JW, Saeian K, Staff D, et al. Eosinophilic esophagitis in adults: an emerging problem with unique esophageal features. Gastrointest Endosc. 2004;59:355–361
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (247 KB)
   • CrossRef
9. Liacouras CA, Spergel JM, Ruchelli E, et al. Eosinophilic esophagitis: a 10-year experience in 381 children. Clin Gastroenterol Hepatol. 2005;3:1198–1206
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (516 KB)
   • CrossRef
10. Croese J, Fairley SK, Masson JW, et al. Clinical and endoscopic features of eosinophilic esophagitis in adults. Gastrointest Endosc. 2003;58:516–522
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (1229 KB)
    • CrossRef
11. Remedios M, Campbell C, Jones DM, et al. Eosinophilic esophagitis in adults: clinical, endoscopic, histologic findings, and response to treatment with fluticasone propionate. Gastrointest Endosc. 2006;63:3–12
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (320 KB)
    • CrossRef
12. Sant'Anna AM, Rolland S, Fournet JC, et al. Eosinophilic esophagitis in children: symptoms, histology and pH probe results. J Pediatr Gastroenterol Nutr. 2004;39:373–377
    • View In Article
    • CrossRef
13. Cuzick J. A Wilcoxon-type test for trend. Stat Med. 1985;4:87–89
    • View In Article
    • MEDLINE
    • CrossRef
14. Assa'ad AH, Putnam PE, Collins MH, et al. Pediatric patients with eosinophilic esophagitis: an 8-year follow-up. J Allergy Clin Immunol. 2007;119:731–738
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (254 KB)
    • CrossRef
15. Blanchard C, Wang N, Stringer KF, et al. Eotaxin-3 and a uniquely conserved gene-expression profile in eosinophilic esophagitis. Am Soc Clin Invest. 2006;116:536–547
    • View In Article
16. Steiner SJ, Gupta SK, Croffie JM, et al. Correlation between number of eosinophils and reflux index on same day esophageal biopsy and 24 hour esophageal pH monitoring. Am J Gastroenterol. 2004;99:801–805
    • View In Article
    • MEDLINE
    • CrossRef
17. Cherian S, Smith NM, Forbes DA. Rapidly increasing prevalence of eosinophilic oesophagitis in Western Australia. Arch Dis Childhood. 2006;91:1000–1004
    • View In Article
18. Gonsalves N, Pollicarpio-Nicolas M, Zhang Q, et al. Histopathologic variability and endoscopic correlates in adults with eosinophilic esophagitis. Gastrointest Endosc. 2006;64:313–319
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (505 KB)
    • CrossRef