OUP user menu

Gastrointestinal Pathology in Celiac Disease
A Case Series of 150 Consecutive Newly Diagnosed Patients

Ian S. Brown MBBS, FRCPA, Jason Smith MBBS, Christophe Rosty MD, PhD, FRCPA
DOI: http://dx.doi.org/10.1309/AJCPE89ZPVJTSPWL 42-49 First published online: 1 July 2012


The main histologic feature of celiac disease is increased intraepithelial lymphocytes (IELs) with or without villous atrophy of the duodenal mucosa. The aim of this study was to document a broad range of additional morphologic changes in intestinal mucosa biopsy specimens from patients with celiac disease. Our cohort comprised 150 patients with positive tissue transglutaminase serologic findings; 7 were at Corazza stage A1, 58 at stage B1, and 85 at stage B2. IEL counts per 100 epithelial cells ranged from 34 to 156 (mean, 88.6); a significant neutrophilic infiltrate was present in 85 cases (56.7%); eosinophil count ranged from 3 to 50 per high-power field (mean, 14.6). Additional findings included morphologic changes in enterocytes in 68.7%, subepithelial collagen thickening in 45.3%, and associated lymphocytic gastritis in 30.4% of patients. We demonstrated that these underrecognized features, which can be misleading, are not uncommon in celiac disease and were positively associated with more advanced stages of the disease (P < .0001).

Key Words:
  • Celiac disease
  • Gluten sensitivity
  • Villous atrophy
  • Mucosal eosinophilia
  • Collagen band
  • Lymphocytic gastritis
  • Lymphocytic colitis

Celiac disease is a chronic immune-mediated disorder induced by dietary exposure to gluten in genetically predisposed individuals. It may affect as much as 1% of the population, with many cases remaining undiagnosed.13 The proximal small intestine is the major site of disease. Increased intraepithelial lymphocytosis, with or without concomitant villous atrophy, is the characteristic histologic finding. While histologic examination remains the “gold standard” for diagnosis of celiac disease, changes can be subtle when duodenal villous architecture is intact,4 and villous atrophy may also be encountered in various other conditions.5 Similarly, serologic testing using an enzyme-linked immunosorbent assay for antitissue transglutaminase (tTG) antibodies, which recognizes an important enzyme in celiac disease pathogenesis, is not fully sensitive or specific for the disease.6 In practice, a combination of clinical suspicion, morphologic abnormality, and positive serologic findings is used for the initial diagnosis of most patients with celiac disease. Subsequent clinical, serologic, and/or biopsy demonstration of improvement with a gluten-free diet is necessary for a definitive diagnosis of celiac disease.1,3,7

Interest in the morphologic features of celiac disease has renewed as we endeavor to identify cases with atypical biopsy findings that may lead to misdiagnosis. The recognition of disease patchiness, variability in villous architectural change, and cases with no architectural disturbance are among these atypical patterns.4,811 Likewise, while much attention has focused on intraepithelial lymphocytosis in celiac disease, there is less quantified information on the presence and significance of the lamina propria inflammatory infiltrate that is a common part of the disease.

This study was undertaken to document a large range of histologic features noted on initial gastrointestinal tract biopsy in a series of consecutive patients. These patients were diagnosed as having celiac disease on the basis of clinical suspicion, raised tTG level, and intraepithelial lymphocytosis in the proximal small intestine. The study focused on the presence and significance of morphologic changes in proximal small intestinal biopsy specimens and their correlation to the presence of extraduodenal lymphocytic inflammation.

Materials and Methods

The primary aim of this study was to examine the histopathologic features in initial index small-bowel biopsy specimens from patients with celiac disease. The study cohort was obtained by searching the laboratory database of Sullivan Nicolaides Pathology (Taringa, Australia) to identify histologic specimens containing the words “celiac disease” over a 12-month period, in the clinical notes, body of report, or final diagnosis. All pathology reports were reviewed. Levels of tTG for each patient displaying intraepithelial lymphocytosis with or without villous atrophy were checked and only those with an elevated level were included. Patients who had had a previous diagnosis of celiac disease were excluded. The ultimate study group comprised a consecutive cohort of patients with elevated tTG levels and abnormal proximal small intestinal biopsy findings.

For evaluating each level, 3 sections from each tissue block were routinely stained with H&E. Further levels were evaluated in cases in which the biopsy orientation was suboptimal in the initial sections and when additional staining was performed. Patient age and sex were documented. The serologic status of tTG was recorded from the laboratory information system; tTG levels greater than 100 were reported as “>100.” All biopsy specimens were reviewed by one of the authors (I.S.B.) to assess for the following 10 features: (1) number of biopsy fragments received; (2) modified Marsh-Oberhuber stage12 and Corazza stage13 for each biopsy fragment Table 1 and Table 2; (3) intraepithelial lymphocyte (IEL) density, counted on H&E-stained sections as number of IELs per 100 epithelial cells in the biopsy fragment with the highest density of IELs (the IELs were counted at ×400 magnification along the luminal border, avoiding the crypts); (4) lymphoplasmacytic infiltration, graded in 3 tiers: grade 0 for inflammatory cells in less than 1/3 of the lamina propria surface, grade 1 for inflammatory cells in up to 2/3 of the lamina propria surface, and grade 2 for full thickness of lamina propria occupied by inflammatory cells14; (5) neutrophilic infiltration, graded by the finding of any neutrophil collection (≥5 cells) in lamina propria only for grade 1, or multifocal infiltration including neutrophilic exocytosis, cryptitis, and crypt abscess for grade 2; (6) eosinophilic infiltration, counted as maximum number of eosinophils per high-power field (hpf; ×400; Olympus BX 41 microscope [Olympus America, Center Valley, PA]); (7) presence of abnormal surface enterocytes defined as a loss of the normal columnar appearance leading to a more cuboidal shape (“cuboidalization”) with loss of cytoplasmic mucin (the presence of prominent cytoplasmic vacuolization (“lipid hang up”), defined as present in most of the cells in at least 1 biopsy specimen, was also considered an enterocyte abnormality); (8) Paneth cell density, counted as maximum number of Paneth cells per high-power field on H&E sections (×400; Olympus BX 41 microscope); (9) appreciable thickening of the subepithelial basement membrane in well-oriented biopsy specimens (Trichrome stain and HVG stain were used in cases with this feature only; maximum thickness of the collagen band was assessed via an ocular micrometer); and (10) concomitant biopsy findings in other parts of the gastrointestinal tract were also recorded where available, with respect to the presence of increased IELs and diagnostic features of lymphocytic gastritis and lymphocytic colitis. Criteria for significantly increased IELs were more than 20 IELs/hpf in the esophagus,15 more than 25 IELs/100 epithelial cells in the stomach,16 more than 25 IELs/100 epithelial cells in the ileum,17 more than 10 IELs/100 epithelial cells in the colon.18 Gastric biopsy specimens with nonspecific chronic gastritis or lymphocytic gastritis were examined for Helicobacter pylori using a Romanowsky stain and/or immunohistochemical examination.

View this table:
Table 1
View this table:
Table 2

Statistical analyses were performed with SPSS statistics software version 17.0 (SPSS, Chicago, IL). Categorical variables were compared using the χ2 test. Means of 2 or more groups were compared using the Mann-Whitney U test or analysis of variance 1-way test as appropriate. A 2-tailed P value was used in all analyses and a P value of less than .05 was used to determine statistical significance.


The study group of 150 patients consisted of 103 female and 47 male patients (female-male ratio, 2:1). Patients ranged in age from 1 to 84 years (mean, 37.9 years). Twenty-two patients (14.7%) were 60 years of age or older while 29 patients (19.3%) were 20 years or younger. The number of biopsy fragments received per case ranged from 1 to 8, with a median of 4, comprising a total of 576 biopsy fragments in this series. In 38 cases (25.3%), fewer than 3 biopsy fragments were received.

The main clinical and pathologic characteristics are summarized in Table 3. The distribution of the histologic stages was as follows: 7 patients had Corazza stage A1 lesions, comprising 5 with modified Marsh stage 1 and 2 with modified Marsh stage 2 lesions; 58 patients had Corazza stage B1 lesions comprising 22 with modified Marsh stage 3a and 36 with modified Marsh stage 3b lesions; and 85 patients had Corazza stage B2 lesions, all corresponding to modified Marsh stage 3c. Of 146 patients with more than 1 small bowel biopsy specimen, 35% had modified Marsh stages: variability of 1 stage in 30 cases (20.6%), 2 stages in 15 (10.3%), and 3 or more stages in 6 cases (4.1%). In 1 biopsy set, a Marsh stage 0 was encountered in 1 fragment while other fragments displayed Marsh 3b and 3c patterns. A Marsh stage 1 lesion was encountered in at least 1 biopsy specimen in 11 cases (7.3%).

Levels of tTG ranged from 7 to more than 100, with 68 patients having more than 100 tTG levels. Patients with a tTG level of 20 or more were significantly older than patients with a tTG level less than 20 (mean age, 47.3 vs 35.2 years, P = .002). A positive correlation was found between the level of tTG and the severity of architectural changes as measured by the Corazza staging scheme (P < .0001).

View this table:
Table 3

The degree of inflammation in the duodenal mucosa was assessed by the number of IELs per 100 epithelial cells, the density of lymphoplasmacytic infiltration, the grade of neutrophilic infiltration, and eosinophil counts. The number of IELs ranged from 34 to 156 (mean, 88.6). A higher IEL density was positively associated with increasing Corazza stage (P < .0001) Table 4. A significant increase in lamina propria infiltration by lymphoplasmacytic cells was observed in 137 cases (91.3%), with 114 cases showing mild grade 1 infiltration and 23 cases showing grade 2 infiltration, which correlated with advanced Corazza stage (P < .0001). Of the 13 cases without appreciable chronic inflammation, 5 were Marsh stage 1 lesions. Neutrophil collections were observed in 85 cases (56.7%), with 62 cases (41.3%) showing focal infiltration in the lamina propria and 23 cases (15.3%) with multifocal infiltration, including crypt abscess in 2 cases (1.3%) Image 1A. No ulceration was seen in any case. Eosinophilic infiltration ranged in number from a maximum of 3 eosinophils up to 50/hpf (mean, 14.6) Image 1B. An eosinophil count of more than 20/hpf was found in 37 cases overall (24.7%). No increased eosinophil counts in peripheral blood or in the mucosa of other intestinal parts were found in these patients. Higher eosinophil count or increased neutrophil density had a strong statistical association with advanced Corazza stage (P < .0001) (Table 4).

View this table:
Table 4
Image 1

A, Small-bowel mucosa showing abnormal enterocytes in the superficial epithelium and inflammation in the lamina propria comprising numerous neutrophils with cryptitis and crypt abscess (H&E); B, Dense eosinophilic infiltrate in the lamina propria of small-bowel mucosa (H&E); C and D, Thick patchy subepithelial collagen band associated with total villous atrophy of the small-bowel mucosa (H&E and Masson trichrome).

Surface enterocytes displayed abnormal morphologic features, such as cuboidalization in all biopsy fragments, in 103 cases (68.7%). None of the Corazza stage 1 cases showed this feature. Cytoplasmic vacuolization was a widespread, prominent feature in 3 cases. Most biopsy specimens displaying enterocyte abnormalities showed at least focal cytoplasmic vacuolization. Surface enterocyte abnormalities were more frequently found in Corazza stage B2 cases than in stage B1 cases (P < .0001) (Table 4). Paneth cell count ranged from 0 to 44/hpf (mean, 14.9/hpf). The number of Paneth cells in advanced stages of celiac disease decreased nonsignificantly (mean, 19.1/hpf in stage A1, 15.7/hpf in stage B1, and 14.0/hpf in stage B2; P = .17) (Table 4). No correlation was found between Paneth cell count and the density of any inflammatory infiltrate.

A subepithelial band thickening greater than 1.5 μm was seen in at least part of 1 biopsy fragment of 68 cases (45.3%) Image 1C and Image 1D. The band measured up to 20 μm in thickness, but was less than 10 μm in all but 4 cases. This feature was mostly encountered in B2 rather than in B1 stage cases (63.6% vs 24.1%, P < .0001). The appearance of the subepithelial thickening on the special stains was consistent with minimal collagen deposition. Capillary vascular congestion did appear to exaggerate the impression of an expanded subepithelial collagen plate in most cases.

Image 2

Examples of lymphocytic gastritis (A) and lymphocytic colitis (B) in patients with celiac disease (H&E).

Biopsy specimens from at least 1 other gastrointestinal tract site (esophagus, stomach, ileum, and colon) were available in 99 patients. A significant increase in IELs was found in 7 (28%) of 25 esophageal biopsy specimens, most of which were from the distal esophagus and likely to be related to reflux disease. Twenty-four (30%) of 79 gastric biopsy specimens demonstrated increased IELs with chronic inflammation in the lamina propria in keeping with lymphocytic gastritis Image 2A. No case of Helicobacter pylori infection was found. In 1 (17%) of 6 ileal biopsy specimens, and in 7 (54%) of 13 colonic biopsy specimens, IEL count was found to be increased. Of the 7 positive colonic biopsy specimens, 4 had additional inflammatory changes of lymphocytic colitis (4 [31%] of 13) Image 2B while 3 were considered intraepithelial lymphocytosis only. Interestingly, a significant positive association was found between the presence of lymphocytic gastritis and the degree of neutrophilic inflammation in the duodenal biopsy specimen (P = .006) Table 5. A trend for a positive association between the presence of lymphocytic gastritis and IEL count (P = .08) or eosinophilic infiltrate (P = .1) was also noted. In addition, lymphocytic gastritis was more likely to be found in patients with more advanced stage celiac disease (0% of Corazza stage A1, 14.3% of Corazza stage B1, 47.6% of Corazza stage B2 patients, P = .002). No association was found between histologic variables in the duodenal mucosa and the presence or absence of lymphocytic colitis.


This study examined a comprehensive range of histologic features of celiac disease in duodenal biopsy specimens and associated changes in specimens taken from other gastrointestinal tract sites, in a series of 150 consecutive, serology-positive, first-time cases seen in a community-based laboratory setting. We chose these criteria to focus on additional inflammatory changes not commonly described in the duodenal mucosa of patients with celiac disease. Our findings confirm previous reports that celiac disease is at least twice as commonly seen in female patients, though the reason for this is uncertain.2 The age range at the time of diagnosis was 1 to 84 years, with 14.7% of patients aged 60 years or more, confirming that presentation in later life is not uncommon.19

View this table:
Table 5

Recent studies have highlighted histologic variability in celiac disease severity. Up to 25% of adults and 50% of children have been reported to exhibit variable degrees of villous architectural changes in multiple biopsy fragments.20,21 We found variability of at least 1 Marsh stage in 35% of biopsy specimens we examined. This figure may underestimate the true degree of variability in the proximal small intestine in celiac disease because fewer than 3 biopsy specimens were received in 25.3% of cases. In a retrospective study, Pais et al22 reported 25% of duodenal biopsy specimens to have at least 1 Marsh stage difference between biopsy fragments. The problem of variability has led to calls for at least 3 or 4 biopsy specimens20,21 to be taken to avoid possible underdiagnosis. A previous series reported a histologically normal biopsy finding in up to 36% of otherwise abnormal specimens.11 In our series, the rate of normal biopsy specimens or specimens with only intraepithelial lymphocytosis was low. This is possibly caused by our study selection criteria because we only included cases with positive serologic findings and abnormal duodenal biopsy findings suggestive of celiac disease.

An important aspect of our study is the careful description of various types of inflammatory infiltrate encountered in celiac disease. Lymphocytes and plasma cells are typically conspicuous in the lamina propria of the duodenal mucosa; however, as highlighted in our series, cases without villous change usually do not display this feature. A neutrophilic infiltration is also a common feature in duodenal biopsy specimens from patients with celiac disease.23 We found at least focal lamina propria neutrophilic infiltrate in 56.7%. The presence of neutrophils may cause confusion with peptic duodenitis, particularly in biopsy specimens from the duodenal bulb.24,25 Because crypt abscess formation was rarely seen and true erosion was not encountered, we believe the presence of these features should suggest an alternate diagnosis such as peptic duodenitis or nonsteroidal anti-inflammatory drug injury. Accumulation of eosinophils in the gastrointestinal tract is known to be associated with allergy, inflammatory bowel disease, infection, and neoplasms and is the main feature of eosinophilic gastroenteritis.26 Celiac disease is underrecognized as a cause of duodenal mucosa eosinophilia, probably because of the associated architectural and other inflammatory changes. An eosinophilic infiltrate of 20/hpf or more in the duodenal mucosa was encountered in 24.7% of our cases and correlated with advanced Corazza stage. Patients younger than 20 years old were found to have a particular predilection for this finding, as seen in 47% of our cases. To our knowledge, this is the first time an association between eosinophil density and stage of celiac disease is shown in a large case series. The role of eosinophils in celiac disease may involve the release of toxic cationic proteins, accounting for more severe histologic findings in patients with high eosinophil density on the small bowel mucosa.27,28 Recently, De Re et al29 found that galectin-10 expression correlated with celiac disease stage and the density of eosinophil infiltrate, suggesting that galectin-10 may be a marker of disease severity and that eosinophils can be a therapeutic target for severe celiac disease.

It has long been recognized that minimal thickening (≤5 μm) of the subepithelial collagen layer occurs in about 40% of the duodenal mucosa in celiac disease, and up to 3% of cases display collagen thickening beyond this.14,3032 A significant proportion of our cases demonstrated thickened subepithelial collagen band, a common feature in patients with Corazza B2 stage lesions (63.6%). Celiac disease is a common underlying condition of collagenous sprue; however, there are no uniform criteria for this diagnosis.31 Because subepithelial collagen deposition is a common feature of celiac disease, it would seem prudent to require a more rigorous definition for a diagnosis of collagenous sprue.

The role of Paneth cells in immune-mediated disease of the intestines is now receiving attention.33 Data on the role of Paneth cells in celiac disease have been conflicting. Some studies reported a decrease in the number of duodenal mucosa Paneth cells in patients with celiac disease while others have not found any changes compared with normal mucosa.3436 Other studies using different methods to quantify Paneth cells are necessary to validate this observation.

Celiac disease may be associated with lymphocytic, collagenous, and neoplastic disease in the luminal gastrointestinal tract beyond the small intestine. In our series, a gastric biopsy was performed in 79 patients, with lymphocytic gastritis found in 30.4%. This is in agreement with previous studies reporting this association in 10% to 45% of patients, with an antral predominance for gastric lesions and regression of lesions after patients followed a gluten-free diet.16,37,38 Interestingly, the presence of concurrent lymphocytic gastritis was associated with the severity of inflammation in the duodenal biopsy specimen in terms of IEL density and neutrophil and eosinophil infiltration. Bhatti et al39 also found that lymphocytic gastritis was commonly associated with severe celiac disease at the time of diagnosis.15,40 We also observed lymphocytic colitis in 31% and lymphocytic ileitis in 17% of concurrent biopsy specimens. However, only a small number of biopsy specimens were available from these sites and a definitive statement for these associations cannot be made.

In summary, histologic characteristics of celiac disease are not limited to villous atrophy and intraepithelial lymphocytosis. We report the occurrence of commonly associated features that may lead to misdiagnosis, such as prominent neutrophilic and eosinophilic infiltration, surface enterocyte changes, and subepithelial collagen thickening, all of which are more frequently found in advanced stage disease.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 38.
  39. 39.
  40. 40.
View Abstract