OUP user menu

The Liver in Celiac Disease
Clinical Manifestations, Histologic Features, and Response to Gluten-Free Diet in 30 Patients

Taofic Mounajjed MD, Amy Oxentenko MD, Eugenia Shmidt, Thomas Smyrk MD
DOI: http://dx.doi.org/10.1309/AJCPDOMY5RI5TPMN 128-137 First published online: 1 July 2011


Descriptive reports of liver histologic features in celiac disease (CD) are sparse, and the effect of a gluten-free diet (GFD) on the course of liver injury is poorly understood. We reviewed liver biopsy specimens in 30 patients with CD and performed immunostains for IgG, IgG4, IgM, and IgA. Subsequent liver biochemical tests and compliance with the GFD were recorded. Of the patients, 19 had autoimmune-mediated liver disease (AILD; autoimmune hepatitis, 9; primary sclerosing cholangitis, 7; and primary biliary cirrhosis, 3). The remaining 11 patients had cryptogenic hepatitis (5), hepatitis C (2), steatohepatitis (2), sarcoidosis (1), and T-cell lymphoma (1). The liver disease diagnosis preceded the CD diagnosis in all groups except steatohepatitis. Although 82% of patients without AILD had symptomatic CD, only 26% of patients with AILD had such symptoms. The pathology of the specific liver disease was not atypical in histologic features or IgG/IgM ratios. While GFD improved cryptogenic hepatitis, it did not seem to affect AILD. We propose that AILD and cryptogenic hepatitis in patients with CD represent distinct clinical, histologic, and immunohistochemical entities rather than 2 ends of a spectrum of liver injury.

Key Words:
  • Liver diseases
  • Celiac disease
  • Gluten-free diet
  • Autoimmune-mediated liver disease
  • Hepatitis
  • Autoimmune hepatitis
  • Primary sclerosing cholangitis
  • Primary biliary cirrhosis

At least 20% to 30% of patients with celiac disease (CD) have extraintestinal manifestations1; hepatic injury, first described in 1977,2 is one of the most common.3 The prevalence of hypertransaminasemia in patients with CD ranges between 15% and 61%,412 with the highest prevalence observed in children.7 Conversely, the prevalence of CD in patients with unexplained hypertransaminasemia is 10%, a finding that justifies screening for CD in all patients with abnormal liver biochemical test results.13

There are 2 main forms of liver damage in CD: cryptogenic and autoimmune-mediated.14 Cryptogenic liver damage is more frequent; it is typically asymptomatic and is characterized by mild hypertransaminasemia. It is not associated with autoantibodies other than CD autoantibodies. The hypertransaminasemia of cryptogenic liver damage usually resolves within 6 to 12 months of beginning a gluten-free diet.15,16 The CD-associated autoimmune-mediated liver disorders, in contrast, can cause chronic progressive liver injury. Autoimmune hepatitis (AIH),1627 primary biliary cirrhosis (PBC),2835 and primary sclerosing cholangitis (PSC) have all been described.24,3641 The effect of a gluten-free diet on the course of CD-associated autoimmune-mediated liver disease (AILD) is variable, with most reports concluding no beneficial effect14,19,23,2729,36,40,42,43 and a few describing significant improvement.20,3739,44

Rarely, CD-associated liver disease can manifest as acute liver failure, cryptogenic chronic hepatitis, or cryptogenic cirrhosis.20,45,46 Viral hepatitis and fatty liver disease are also described in patients with CD, but they likely represent chance associations.4,10,15,47,48

Descriptive reports of liver histologic features in patients with CD are sparse.2,5,8 In this study, we characterized the histologic features of liver disease in 30 patients with CD who had a liver biopsy. We also assessed the effect of a gluten-free diet on the course of liver disease in these patients.

Materials and Methods

We searched our electronic pathology database for cases with a small intestinal biopsy sample showing histologic features of CD between the years 1990 and 2009. Among this group, we identified 48 cases that also had liver biopsies. We then reviewed the medical records and recorded the clinical manifestations, date of diagnosis (of CD and liver disease), indication for liver biopsy, serum gamma globulins, antiendomysial antibodies, anti–tissue transglutaminase antibodies, antinuclear antibodies (ANA), anti–smooth muscle antibodies (ASMA), antimitochondrial antibodies (AMA), anti–liver-kidney-microsome antibodies, HLA phenotype (when available), and viral hepatitis serology. CD was defined as a malabsorption pattern on a small bowel biopsy sample with a positive antiendomysial antibody and/or anti–tissue transglutaminase antibody result. In 18 cases there was no serologic evidence for CD, despite the small bowel morphologic findings, and the cases were excluded. The remaining 30 cases comprise our study group Figure 1.

Once the CD diagnosis was confirmed, we reviewed histologic sections from all liver biopsies. In addition to routine H&E staining, all liver biopsy specimens were also stained for trichrome, iron, periodic acid–Schiff with diastase, and reticulin stains. Chronic hepatitis (autoimmune and viral) was graded and staged using the Batts-Ludwig grading and staging system.49 Chronic biliary disease (PBC and PSC) was also staged using the Ludwig criteria.50,51

Immunohistochemical stains for IgG, IgM, and IgA were performed on sections of paraffin-embedded hepatic tissue from patients with CD and on a control group consisting of 15 patients with AILD (5 each of AIH, PBC, and PSC) matched for age, sex, stage, and, if applicable, grade, to our patients with CD with AIH, PBC, and PSC. Immunohistochemical staining was done using standard streptavidin-biotin-per-oxidase techniques with commercially available antibodies against IgG, IgM, and IgA. The source of the antibodies and the reaction conditions are listed in Table 1. We counted the number of IgG+, IgM+, and IgA+ plasma cells in portal tracts and recorded the maximum count per high-power field.

We reviewed liver biochemical test results, including alanine transaminase, aspartate transaminase, alkaline phosphatase, and bilirubin levels, before and after the diagnosis of CD. We reviewed the results of imaging studies, including endoscopic retrograde cholangiopancreatography and magnetic resonance cholangiopancreatography, when available. We documented any treatment received for the liver disease and reviewed the medical record for information regarding patient compliance with a gluten-free diet. Changes in liver biochemical test results over time were also recorded.

View this table:
Table 1
Figure 1

In 48 cases, there was a malabsorption pattern in the small bowel biopsy specimens; only 30 of these met the criteria for a diagnosis of celiac disease (CD). EMA, antiendomysial antibody; GFD, gluten-free diet; TTG, anti–tissue transglutaminase antibody.


The liver diseases found in the 30 patients with CD are listed in Table 2. As shown, the temporal relationship between the diagnosis of CD and liver disease was variable; in 33% of patients, both conditions were diagnosed simultaneously. There was also variability in the prevalence of CD-related symptoms at the time of diagnosis; while 82% of patients without AILD had symptomatic CD at diagnosis, only 26% of patients with AILD had such symptoms Figure 2.

All but 4 patients had elevated liver biochemical test results (aspartate transaminase, alanine transaminase, bilirubin, and/or alkaline phosphatase) at the time of liver biopsy. The indication for the liver biopsy in the 4 patients included nonspecific abdominal symptoms, staging of possible pancreatic adenocarcinoma, abnormal appearance of the liver noted during bariatric surgery, and grading/staging of hepatitis C infection. Serologic evidence of hepatitis C infection was present in 2 patients; chronic hepatitis was confirmed by liver biopsy in both.

Eight patients had a body mass index of more than 30, but the diagnosis of CD was confirmed in all. Our 2 patients with steatohepatitis, for example, had body mass indexes of 31 and 48. Of these 2 patients, 1 had nonspecific symptoms, but CD serology and small intestinal biopsy results were positive, and the patient had an HLA-DQ2 phenotype. Furthermore, symptoms improved with a gluten-free diet. The second patient had diarrhea and positive CD serology and small intestinal biopsy results. The CD diagnosis preceded the liver disease diagnosis in the first patient, and both were diagnosed simultaneously in the second patient. Three patients had substantial alcohol use; none had steatohepatitis shown by liver biopsy.

Five cases lacked a specific clinical diagnosis for liver disease and had nonspecific findings on liver biopsy. The liver biopsies in this group showed minimal changes, including Kupffer cell hyperplasia (3 cases), minimal macrovesicular steatosis (2 cases), and focal ductular proliferation (1 case). There was no lymphocytic cholangitis Image 1. By immunohistochemical analysis, the liver contained minimal or absent IgG+, IgM+, or IgA+ plasma cells Table 3.

Figure 2

Percentage of symptomatic CD in all liver disease categories. AIH, autoimmune hepatitis; AILD, autoimmune-mediated liver diseases; CD, celiac disease; NS, nonspecific changes; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis.

In 9 cases, there were clinical and histologic features of AIH. All met the revised diagnostic criteria for a definite diagnosis set by the International Autoimmune Hepatitis Group.52 Hypergammaglobulinemia was present in 7 cases, ANA was positive in 8, and ASMA was positive in 5. None had a positive anti–liver-kidney-microsome antibody. The histologic findings were characteristic of AIH, featuring lymphoplasmacytic inflammation with interface activity Image 2. In 4 cases, activity was grade 1; in 2, it was grade 2; in 2, grade 3; and in 1, grade 4. The disease was stage 1 in 3 cases, and it was stage 2, 3, and 4 in 2 cases each. Lymphocytic cholangitis was prominent in 1 case (Image 2). In 2 cases, there were granulomas, 1 parenchymal and 1 portal, but both cases had positive ANA and ASMA, negative AMA, and normal alkaline phosphatase results. As shown in Table 3 and Image 3, immunohistochemical stains for immunoglobulin subsets showed an IgG-dominant pattern, similar to that seen in our AIH control samples and reported in the literature for AIH.53,54

View this table:
Table 2
Image 1

Minimal nonspecific changes observed in 5 cases of cryptogenic/celiac hepatitis. A, Minimal macrovesicular steatosis (H&E, ×100). B, Kupffer cell hyperplasia (H&E, ×100).

In 3 cases, there was PBC; AMA was positive in 2 and the third had the characteristic clinical and histologic picture of PBC and a positive ANA (autoimmune cholangitis). All 3 had granulomatous cholangitis accompanied by portal and lobular inflammation Image 4. The stage at diagnosis was 1 in 1 case and 3 in 2 cases. As expected from the literature and from our PBC control samples, IgM was the dominant immunoglobulin in tissue sections (Table 3 and Image 3).

View this table:
Table 3
Image 2

Autoimmune hepatitis (AIH). A, Classic AIH histologic characteristics featuring a portal plasma cell–rich lymphoplasmacytic infiltrate with interface activity (H&E, ×200). B, One patient with otherwise classic AIH had prominent lymphocytic cholangitis. Antimitochondrial antibody was negative and alkaline phosphatase was normal, excluding an overlap syndrome (H&E, ×50).

In 7 cases, the diagnosis was PSC, with elevation in serum alkaline phosphatase in all 7. Endoscopic retrograde cholangiopancreatography and/or magnetic resonance cholangiopancreatography showed characteristic multifocal stricturing and dilation of intrahepatic and/or extrahepatic bile ducts in 5 cases. The remaining 2 were diagnosed with small-duct PSC based on the clinical findings, cholestatic biochemical profile, and findings on liver biopsy. In 4 cases, inflammatory bowel disease was found, including 3 with chronic ulcerative colitis and 1 with Crohn disease. Histologic features were characteristic of PSC: portal fibrosis, often with periductal distribution, bile duct injury, and ductopenia Image 5. The stage at diagnosis was 1 in 1 case, 2 in 3 cases, 3 in 2 cases, and 4 in 1 case. IgG was more prevalent than IgM, a finding similar to that for our control samples (Table 3). Our CD-associated cases of PSC had more IgA+ cells than did the control cases, an unexplained finding.

Follow-up liver biochemical test results were available 6 to 60 months after the diagnosis of CD (mean, 39 months; median, 37 months). Follow-up was available for 4 of 5 cases with liver biopsies showing minimal nonspecific changes. All patients were compliant with a gluten-free diet and had normal follow-up liver biochemical test results. Roughly half of the patients with AILDs seemed to comply with a gluten-free diet. Figure 3 shows the complicated relationship of disease state, diet, and outcome. All patients except 1 with AIH were treated with immunosuppressive therapy. The only patient with AIH who did not receive immunosuppression showed slight improvement of liver biochemical results on a gluten-free diet, but the test results did not normalize. All patients with PSC received treatment, including liver transplantation in 3 patients and ursodeoxycholic acid in 5. Of the 3 patients with PBC, 1 was lost to follow-up, and the 2 remaining patients received ursodeoxycholic acid treatment. In patients with AILD, the grade and stage of disease did not seem to influence the effect of diet on liver biochemical test results (Figure 3).

Image 3

Immunohistochemical stains for IgG and IgM in primary biliary cirrhosis and autoimmune hepatitis. IgM+ plasma cells (A, ×200) predominate over IgG+ plasma cells (B, ×200) in primary biliary cirrhosis. In contrast, IgG+ plasma cells (D, ×200) predominate over IgM+ plasma cells (C, ×200) in autoimmune hepatitis.

Patients with hepatitis C had a typical course. One patient died with end-stage liver disease, and a second patient received a liver transplant. Of the 2 patients with steatohepatitis, 1 underwent gastric bypass surgery and had significant improvement in liver biochemical test results. The second patient’s liver biochemical test results also improved over time, possibly owing to weight loss. The patient with sarcoidosis had significant liver fibrosis and continued to have abnormal liver biochemical test results. That patient was awaiting liver transplant 8 years after diagnosis. The patient with T-cell lymphoma was lost to follow-up. Of all 30 patients, 5 eventually received a liver transplant (1 with AIH, 3 with PSC, and 1 with hepatitis C).

Image 4

Classic histologic features of primary biliary cirrhosis were observed in patients with celiac disease. A, Portal lymphoplasmacytic infiltrate with lymphocytic cholangitis (H&E, ×200). B, Periportal noncaseating granuloma (H&E, ×200).

Image 5

Classic histologic features of primary sclerosing cholangitis were observed in patients with celiac disease. A, Typical concentric periductal fibrosis (H&E, ×50). B, Portal and periportal fibrosis with loss of native interlobular bile duct (ductopenia) (H&E, ×100).


Despite the large body of literature documenting CD-associated liver disease, descriptions of liver histologic features in this setting are sparse. The effect of a gluten-free diet on the course of liver disease, particularly the autoimmune-mediated diseases, also remains uncertain. Finally, there is some question as to whether cryptogenic hepatitis and the AILDs that arise in patients with CD are distinct entities or part of a broad spectrum of “celiac liver disease.” We describe the liver histologic features and clinical features in 30 patients with CD who had liver biopsies.

In 5 patients with nonspecific findings on liver biopsy, there was no obvious cause other than CD for the liver disease. Liver biochemical test results improved on a gluten-free diet. We suspect that these cases represent examples of cryptogenic (“celiac”) hepatitis. In our patients, the histologic findings were minimal (including minimal macrovesicular steatosis, Kupffer cell hyperplasia, and focal ductular proliferation), particularly when compared with what has been described in the literature.2,5,8,13,55 There were no histologic features that might raise consideration for CD (lymphocytic cholangitis, for example), nor were there clues that might elucidate the pathogenesis of liver injury in this form of CD-associated liver disease.

Previous reports have indicated that steatosis is a feature of celiac-related liver injury.2,5,56 Our study did not reproduce this finding: Only 1 of the patients with celiac hepatitis had minimal steatosis. It may be that previous descriptions of celiac hepatitis included other liver diseases in which steatosis is more common. In this series, 2 patients had steatohepatitis, but both were obese and one had diabetes. Despite the apparent paradox of obesity in patients with CD, recent data indicate that a significant number (39%) of patients with CD are overweight and up to 13% are obese.5759 This “changing face” of CD in recent decades may be secondary to the serologic approach in diagnosis, probably resulting in early detection in overweight patients before the classic clinical characteristics of CD have evolved.

An association between CD and AILD has been recognized since 1978, when Logan et al28 described 4 patients with CD and PBC. Multiple subsequent series, including a few large population studies, have established an epidemiologic association between CD and AILD, particularly PBC3034 and AIH.1624,26,27 For example, a study that included 8,631 patients with CD from Sweden and Denmark reported a significantly higher prevalence of PBC compared with the general population.32 Likewise, Volta et al19 found CD in about 4% of 181 patients with AIH, whereas other studies report a CD incidence ranging between 2% and 20% in patients with AIH.1624,26,27 An association between CD and PSC has also been described in multiple studies,24,3641 although this association is less well established because most studies are limited by small samples, and 2 studies failed to reproduce the association.14,60

Figure 3

Effect of a gluten-free diet on subsequent liver biochemical test results (LBs). AIH, autoimmune hepatitis; AILD, autoimmune-mediated liver disease; CD, celiac disease; G, grade; PBC, primary biliary cirrhosis; PSC, primary sclerosing cholangitis; S, stage.

In our study, 19 patients had an AILD. The clinical features in this group were different from the rest of our study patients in that they tended not to have the classic symptoms of CD. Instead, these patients typically had nonspecific symptoms such as fatigue and malaise. Previous studies have shown that the classic symptoms of CD are lacking in most patients with CD-associated AIH,19,21,61 but the absence of such symptoms has not been reported in patients with PSC and PBC. Our findings reinforce the existing recommendation of screening patients with cryptogenic hypertransaminasemia13 and AILD for CD.14,33,34

The histologic features of CD-associated AIH, PBC, and PSC were not atypical. As expected, plasma cells expressing IgG, IgM, and IgA were present in relatively large numbers in AIH, PBC, and PSC; the IgG/IgM ratios in our CD-associated cases were similar to those in our control samples and similar to reports in the literature.53,54

Almost all patients with CD-associated AIH, PBC, or PSC received treatment for the liver disease, making it difficult to detect any effect of the gluten-free diet on the progression of liver disease. One patient with AIH thought to be secondary to CD was managed with dietary change only. Although liver biochemical test results improved slightly, they did not normalize. In the other patients, all of whom received treatment for liver disease, compliance with diet did not seem to influence the progression of liver disease. It has been argued that early-stage PSC or PBC in patients with CD may respond to a gluten-free diet.37,38,43,44 This was not our experience with PSC; of 4 PSC patients who were compliant with diet, there was some improvement of liver biochemical test results in 3 patients with stage 2 or 3 disease, but not in a patient with stage 1 disease. The 2 patients with PBC for whom follow-up was available were not compliant with a gluten-free diet; 1 showed improvement in liver biochemical test results and 1 did not. These findings, although difficult to interpret owing to the confounding effect of medical treatment, argue that a gluten-free diet has little influence on liver disease progression in CD-associated AILDs.

The mechanism of liver injury in CD is uncertain, but suggested causes include increased intestinal permeability,11 systemic autoimmunity,62 mucosal damage and inflammation,6 malnutrition,63 and intestinal bacterial overgrowth.64 These causes may explain liver damage in cryptogenic hepatitis. One opinion holds these causes responsible for CD-associated AILD as well. This view regards cryptogenic hepatitis and autoimmune-mediated liver disease as different ends of a spectrum of liver injury wherein individual factors (genetic predisposition, duration of gluten exposure) may influence the reversibility of liver damage.16,27,65 The other view is that cryptogenic hepatitis and AILD are distinct entities in which the association of AIH, PBC, and PSC and CD likely represents clustering of diseases in patients who are prone to autoimmune conditions.66 Our study supports the latter view. The histologic features in our patients with CD-associated AIH, PBC, or PSC were typical of their respective liver disease (as opposed to a nonspecific morphologic picture seen in cryptogenic hepatitis), and we could not document any consistent response of liver disease to a gluten-free diet. The AILDs also differed from the cryptogenic hepatitis cases in terms of immunoglobulin-laden infiltrates and in the prevalence of CD-related symptoms. In fact, the paucity of classic CD symptoms in our patients with AILD could lead one to speculate that an atypical form of CD develops in some patients with AIH, PBC, and PSC. Our initial search for patients yielded 18 others who had liver disease and a malabsorption pattern shown on biopsy, but no serologic evidence of CD, perhaps supporting the notion that there is a liver disease–associated form of villous injury.

We describe the clinical, histologic, and immunohistochemical findings in 30 patients with CD who had a liver biopsy. Our study provides additional data regarding the clinical manifestations of CD in association with liver disease and reiterates the need to screen patients with cryptogenic hypertransaminasemia and AILDs. Our data suggest that a gluten-free diet has little influence on AILDs associated with CD. While there seems to be a CD-associated hypertransaminasemia (which responds to gluten-free diet), we suggest that the association between CD and AILDs is based on a shared underlying predisposition rather than CD-associated liver injury.


  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.
  41. 41.
  42. 42.
  43. 43.
  44. 44.
  45. 45.
  46. 46.
  47. 47.
  48. 48.
  49. 49.
  50. 50.
  51. 51.
  52. 52.
  53. 53.
  54. 54.
  55. 55.
  56. 56.
  57. 57.
  58. 58.
  59. 59.
  60. 60.
  61. 61.
  62. 62.
  63. 63.
  64. 64.
  65. 65.
  66. 66.
View Abstract