Celiac disease (CD) is a systemic immune-mediated disorder elicited by gluten in genetically susceptible individuals. The common factor for all patients with CD is the presence of a variable combination of gluten-dependent clinical manifestations, specific serum autoantibodies (anti-tissue transglutaminase/anti-endomysium), human leukocyte antigen (HLA)-DQ2 and/or DQ8 haplotypes, and different degrees of enteropathy (1). Its overall prevalence in the general population is 0.7% to 2% in adults and 0.4% to 1.3% in children (2). The clinical spectrum of CD is remarkably varied, and the disease can affect many extraintestinal organs and systems, including the liver. An association between CD and cryptogenic liver damage, which usually resolved after a gluten-free diet (GFD), was first reported in 1977 in adults (3,4) and 1 year later in pediatric patients (5,6). The association has since been reported in several studies (7). The prevalence of CD in adults with cryptogenic hypertransaminasemia (HTS) is 3% to 4% (8), slightly lower than the percentage reported in a pediatric study center (12%) (9). In children, abnormal serum transaminases levels return to normal with a GFD in a percentage between 77% (10) and 100% (6,11,12). Similar figures have been reported for adult patients (8,13).
In addition to this benign entity (also referred to as celiac hepatitis), CD may be associated with steroid-dependent autoimmune hepatitis (AIH) (14–17). Most of the studies included in a systematic review reported a prevalence of CD of 4% or higher in adult AIH (16). An higher prevalence of primary biliary cirrhosis (PBC) in adults with CD has been reported (18,19). PBC is extremely rare in the pediatric age group and no systematic data regarding the association with CD in children are available. Other rare types of liver damage in CD are primary sclerosing cholangitis (PSC) (19) and fatty liver disease (20,21). Finally, extrahepatic HTS caused by rhabdomyolysis should be taken into account before labeling it a cryptogenic transaminases elevation (22,23).
We conducted a systematic review and meta-analysis to evaluate the prevalence of CD in children with cryptogenic serum HTS or AIH, and vice versa (namely, the prevalence of HTS or AIH in children with CD).
Research Strategy, Eligibility Criteria, and Study Selection
MEDLINE/PubMed, the Cochrane Library, Web of Science, and MD Consult databases from 1977, that is, when the associated CD-liver damage was first reported, to May 1, 2012 were searched for case-control studies, case series, cross-sectional studies, and case reports of children with either CD and HTS or AIH. No publication status or language restrictions were imposed.
Only reports of CD in children (ages 0–18 years) diagnosed by distal duodenal biopsy and/or serological tests were considered eligible. Studies were included in the analysis if CD was diagnosed based on anti-endomysium IgA (and IgG antibodies in case of IgA deficiency), and anti-tissue transglutaminase antibodies. Early pediatric studies in which CD diagnosis was based on duodenal biopsy alone were included in the analysis as well. Similarly, studies were considered eligible if the following serum liver tests had been performed: alanine aminotransferase, aspartate aminotransferase, and/or γ-glutamyl transpeptidase. Lastly, we considered eligible studies in which AIH was diagnosed based on abnormal values of total serum immunoglobulins, positivity of anti-nuclear antibodies, and/or smooth muscle antibodies, or liver kidney microsomal 1 antibodies, and/or liver cytosol 1. Liver biopsy was not considered essential.
We searched the literature using the following free test terms: ([transaminases/aminotransferases OR hyper-transaminasemia/aminotransferasemia OR cryptogenic hyper-transaminasemia/aminotransferasemia OR celiac hepatitis] OR [autoimmune hepatitis OR autoimmune liver disease OR overlap syndrome]) AND (celiac disease OR gluten sensitive enteropathy OR celiac sprue), AND (children OR pediatric patients) (Appendix 1, http://links.lww.com/MPG/A202). In addition to the electronic search, we checked cross-references from original articles.
Two reviewers (P.V. and G.P.) independently screened databases for abstracts and full-text articles that matched the inclusion criteria. All of the potentially relevant abstracts and articles were retrieved and evaluated. In case of disagreement, a third independent reviewer (G.M.) made the final decision. We planned to contact the authors in case additional unpublished data were needed.
The 2 reviewers (P.V. and G.P.) independently retrieved data from studies using a standard data extraction form specifically adapted to our meta-analysis. Also here, disagreement was resolved by consensus after consulting a third reviewer (G.M.), if necessary. For each study, the following information were recorded: country, setting, number of centers, study design, age and number of patients, diagnostic criteria for CD and AIH, timing of response to GFD, and study results, including statistical indicators.
Other relevant articles that did not contain the indication of the denominator for calculating prevalence and case reports were evaluated just for subsequent biographical hand searches.
Statistical Analysis and Heterogeneity Investigation
Meta-analysis with pooled prevalences and 95% confidence intervals (CI) and relative risk (RR) was performed by means of the meta package of the R system (24). RR was used because of its simple interpretation and because we were not referring to a logistic model.
RRs for CD in children with cryptogenic HTS and vice versa were calculated based on the prevalences of CD and HTS in the general pediatric population listed in Table 1(25–29).
General population prevalences of AIH are also listed in Table 1(27–29). RRs for AIH in children with CD were calculated based on the average value of general population AIH reported prevalences (Table 1).
Fixed and/or random effect methods were used, according to the heterogeneity test results. I2 statistics were also determined to define a statistically significant degree of heterogeneity in studies selected for the meta-analysis. Funnel plot was used to assess small studies, to detect biases in the identification and selection of studies and publication bias. Percentages of patients who obtained normalization of serum transaminase levels after starting a GFD were also determined (6,11,12,30).
Validity and Study Quality Assessment
The validity and quality of the selected studies were assessed by 2 authors (P.V. and G.P.) according to the guidelines of the Cochrane Consumers and Communication Review Group (31). Disagreements were resolved by consulting a third reviewer (G.M.). Further information or clarification were obtained by study authors if necessary.
Study quality was assessed with the following a priori criteria:
1. Study design: cross-sectional studies, case-control studies, and case series specifically designed to assess the prevalence of HTS in children with CD or vice versa and/or the prevalence of AIH in children with CD and vice versa were considered to be of higher quality than others with different study design and/or with mixed-age population.
2. Incomplete outcome reporting: studies with complete outcome reporting have been defined as being of better quality than those with incomplete outcome reporting.
3. Selective outcome reporting: reported prespecified outcomes of interest were considered of better accuracy in comparison with outcomes not reported as prespecified or expected.
The search strategy on MEDLINE/PubMed, MD Consult, and Web of Science generated 105, 121, and 129 citations regarding pediatric liver disease and CD, respectively, and 68, 50, and 72 citations regarding pediatric autoimmune liver disease and CD, respectively. The search on the Cochrane Library provided no results.
The full texts of 25 potentially relevant articles with the eligible criteria reported in the “research strategy, eligibility criteria, and study selection” section were retrieved and further analyzed independently by 2 reviewers. Only 9 studies (10 series of patients because 1 study (12) analyzes children with CD with either HTS or AIH for a total of 350 patients) (6,9,11,12,28,32–35) met the full criteria for inclusion in the meta-analysis (Fig. 1). Results of risks of bias assessment are shown in online-only Table S2 (http://links.lww.com/MPG/A203). One study was excluded from the meta-analysis because only the hazard ratio was reported (36). All of the studies were written in English (n = 8) or Italian (n = 1). In all of the included studies, which satisfied the guidelines of the Cochrane Consumers and Communication Review Group for validity and quality study assessment, exclusion of other causes of HTS had been assessed. Data for more precise characterization of AIH patients were requested from authors of 2 studies (33,35).
Prevalence of CD in Pediatric Patients With Cryptogenic HTS
Only 1 study on CD in pediatric patients with cryptogenic HTS was eligible for this analysis (9). Iorio et al (9) reported data on serologically and histologically proven CD in 25 children with persistent cryptogenic HTS observed during a period of 13 years (Table 2). All were clinically asymptomatic. Tests for the most common hepatic inborn errors of metabolism associated with isolated HTS, hepatitis A, B, and C, Epstein-Barr virus, cytomegalovirus, toxoplasmosis, HIV, AIH, biliary diseases, obesity-related hepatopathy, and drug toxicity were negative in all of the patients. Twelve percent (95% CI 4.17%–29.96%) of children with cryptogenic HTS had positive CD serology and positive jejunal biopsy.
Prevalence of Cryptogenic HTS in Children With Untreated CD
Four studies (6,11,12,30) reported data on cryptogenic HTS in 556 pediatric patients with untreated CD (Table 2). The pooled proportion of cryptogenic HTS in children with CD was 36.0% (95% CI 32.15–40.11). In all children with cryptogenic HTS, transaminase levels normalized after 4 to 8 months on a GFD. Most of these patients (150/206) had overt CD (6,11,30) and a mean age at clinical onset of 4.7 years. Because none of these studies is randomized case-control based, we have to refer to prevalence of HTS in the general population as reported by Nobili et al (25).
Knowing the common prevalence of HTS in the general population (179/3280) (25) and the 4 specific prevalences of the selected studies and the pooled one, RR may be computed. Individual RRs and 95% CI depicted in Figure 2 show that RR in all series of pediatric patients with untreated CD is significantly higher than in the general population. Pooled results are even more significant because of the Di Biase et al study (12), reporting a prevalence of 140 of 350, because it is the larger sample under analysis. Heterogeneity evaluation giving a P value >0.05 allows us to consider the 4 studies as homogeneous (heterogeneity test: Q = 5.31; df = 3; P = 0.1508; quantifying heterogeneity: τ2 = 0.0231; H = 1.33 [1; 2.3]; I2 = 43.5% [0%; 81.1%]). The meta-analysis results are shown by the Forest plot (Fig. 2) and the funnel plot (online-only Figure 3, http://links.lww.com/MPG/A204).
Prevalence of CD in Pediatric Patients With AIH
Three studies (206 patients) reported data on CD in children with AIH (32,34,35). CD was diagnosed in all by serological tests and duodenal biopsy (Table 3). The prevalence of AIH pediatric patients with CD in individual studies ranged from 3.6% to 12.5%, with a pooled prevalence of 6.3% (95% CI 3.87–11.73). In the 10/180 AIH children with CD reported in 2 studies (32,34), the AIH was type 1 in 7/10 and serum negative in 3. The results have been reported in Figure 2. Heterogeneity evaluation (heterogeneity test: Q = 6.36; df = 2; P value 0.0416; quantifying heterogeneity: τ2 = 0.4854; H = 1.78 [1; 3.31]; I2 = 68.6% [0%; 90.9%]) shows to be significant at 5% level. The funnel plot representation of results is depicted in online-only Figure 4 (http://links.lww.com/MPG/A205).
Prevalence of AIH in Pediatric Patients With CD
Two studies (1259 patients with CD) (12,33) provided data on AIH in children with CD. The pooled prevalence of AIH in CD was 1.4% (95% CI 0.84%–2.15%). As shown in Table 4, the prevalence of AIH in the Ventura et al study (33) was higher in children ages 2 to 10 years than in children older than 10 years. In the study by De Biase et al (12), 5 of 7 children had type 1 AIH and 2 of 7 had type 2 AIH. Transaminase levels normalized in these 7 patients as the effect of a GFD associated with prolonged immunosuppressive therapy. One study on 18 cases of AIH and CD did not report information on the total number of patients with CD, so that prevalence of AIH in children with CD could not be calculated (32).
Consistent with the extreme ranges of AIH prevalence in the general population (Table 1), the 2 studies show homogeneous results (Q statistics; P = 0.39). We considered the fixed effect model for which the results show evidence of higher AIH prevalence for pediatric patients with CD (Fig. 2). Given the limited number of studies, the funnel plot has not been reported. Heterogeneity evaluation showed the following results: heterogeneity test Q = 0.73; df = 1; P = 0.3937; quantifying heterogeneity: τ2 < 0.0001; H = 1; I2 = 0%. A synopsis of results is provided in Table 5.
Other Pediatric Autoimmune Hepatobiliary Disorders and CD
Only 2 studies (Caprai et al (32) and Lacaille et al (37)) reported children presenting the association CD and PSC or AIH/sclerosing cholangitis overlap syndrome or AI cholangitis. Increased prevalence of PBC in patients with CD was reported only in adults. This association cannot be systematically reviewed in children because of PBC rarity at this age (38,39), and no meta-analysis is therefore feasible.
This meta-analysis shows that the development of liver damage or the discovery of preexisting liver damage in children with newly diagnosed CD is extremely frequent (approximately 1 in 3 patients). In most cases, HTS appears in patients with classical CD and responds rapidly to a GFD. Symptoms of gluten-dependent celiac hepatitis are nonspecific. Associated HTS may occur also in a smaller subgroup of pediatric atypical patients with CD presenting without overt gastrointestinal symptoms but with possible subtle signs of nutritional abnormalities (eg, iron-deficiency anemia or low levels of transferrin saturation, short stature), usually around/before puberty (mean age 8.5 years) (14,40); however, data concerning this category of patients could not be included in the meta-analysis because they were available only as case reports or case series without denominator.
HTS in patients with CD has possibly been attributed to a dysfunction of the gut-liver axis. Optimal function of the gut-liver axis depends on an intact intestine and healthy liver that ensure a normal immunological response to toxins and resident microbiota. Liver damage in these patients may result from disruption of the integrity and function of the intestinal epithelium, because, in turn, of a cytokine-mediated gliadin toxic effect acting on the liver. Upregulation of zonulin seems to cause the increased intestinal permeability of aggressors that target the liver in CD (41).
Conversely, autoimmune liver damage seems to be caused by a complex immune-mediated attack diverted toward liver antigens in genetically predisposed individuals who often present other autoimmune diseases (eg, thyroiditis, diabetes) before, during, or after the diagnosis of CD (42,43).
The histological correlates of such attacks may vary from a mild nonspecific reactive hepatitis to a severe interface hepatitis in GFD-responsive cryptogenic HTS and in AIH, respectively (Table 6) (12,14,32,40). Steatosis and Kupffer cell hyperplasia are described as well (14,18,20,37,44). Autoimmune liver disease and CD share specific HLA class II haplotypes, namely, HLA-DR3 and HLA-DR4. HLA-DR3- and DR4-positive patients are considered to be more susceptible to CD (45).
The prevalence of the association between CD and cryptogenic HTS in adults is generally comparable with that observed in the pediatric age group, although its pooled prevalence is lower than that in children. Specifically, 11% to 42% of patients with CD develop HTS (pooled prevalence 27% with 95% CI 13–44), and 1% to 9% develop HTS because of CD (pooled prevalence 4% with 95% CI 1–7) (8). These wide intervals may most likely depend also on differences in the laboratory screening panels adopted for the diagnostic definition of cryptogenic HTS in several settings. Different respective disease prevalences in the general population may explain why in our meta-analysis the development of AIH in children with CD was 4 times less frequent than vice versa (1.4% vs 6.3%). Other facets need to be considered as well. As an example, CD develops in AIH more frequently in early childhood (2–10 years) than in adolescence (33). A preliminary study presented only in abstract form (17) reported a prevalence of CD in 96 children with AIH (3.4%) slightly lower than that obtained in our meta-analysis. CD-related AIH may be overlooked in true serum-negative cases (32) or in cases with autoantibodies still poorly examined in the clinical practice (eg, anti-liver cytosol 1 antibodies) (14,32). Cross-sectional studies conducted in adults with AIH revealed CD in 2% to 9.8% of patients (16,46) versus 6.3% in children of our meta-analysis. Conversely, the prevalence of AIH in adult patients with CD is similar to that observed in pediatric patients (47). As shown in Table 7, the prevalences of cryptogenic HTS and AIH in children with CD and vice-versa lay in the same order of magnitude (8,17,47).
In addition to AIH, liver damage may also be related to other autoimmune mechanism-based progressive liver diseases such as PBC and sclerosing cholangitis (32,48–50). The prevalence of PBC has been reported to be 3 times higher in adults with CD than in controls (19), but it is extremely rare in childhood, in which only 3 (nonceliac) cases have been reported: a 12-year-old girl with biopsy-proven anti-mitochondrial antibody–positive cirrhotic severe AIH and no histological features of PBC (38), and 2 children with anti-mitochondrial antibody–positive biopsy confirmed PBC, with histological stage II to IV (51) (references 51–56 are available online only at http://links.lww.com/MPG/A206). Two cases of AIH/cholangitis overlap syndrome and 2 cases of autoimmune isolated cholangitis have been reported by Caprai et al (32) in their series of patients with CD.
Prompt investigation of CD in patients with AIH can lead to early dietary treatment and probably prevent or attenuate other late complications of CD (33). Conversely, detection of AIH stigmata in an individual with CD may lead to timely immunosuppressive therapy to prevent liver disease deterioration. Fatty liver disease can be associated with CD, and an accurate evaluation of serological testing should therefore be done also in pediatric and adult patients with nonalcoholic fatty liver disease (NAFLD) (20,52). Patients with CD may be obese because jejunal abnormalities are not continuous but may have a patchy pattern (53), possibly sparing large surface areas. This phenomenon allows development of obesity and other obesity-related liver disease (NAFLD), especially in individuals with the epidemic unhealthy lifestyle. Moreover, both CD and NAFLD (54,55) belong to the group of leaky gut conditions, which may cause progression of NAFLD to nonalcoholic steatohepatitis. Possible hepatic damage caused by hepatitis B should be ruled out in HTS children with CD considered because of reported selective unresponsiveness to hepatitis B virus vaccination (56).
The main limitations of our meta-analysis are the poor number of data hitherto published for the pediatric age group, and the fact that some studies analyzing mixed-age populations could not be used. In addition, group sizes of the different pathological conditions here examined in spite of a statistical homogeneity were often clinically unequal. The strengths of our meta-analysis, which represents the first pediatric meta-analysis on this topic, include a comprehensive search strategy, independent eligibility assessment, and rigorous data abstraction.
In summary, our meta-analysis confirms that pediatric CD is frequently associated with a cryptogenic and reversible elevation of transaminases. Liver involvement in children with CD also includes severe forms of AIH, and, much more rarely, other autoimmune liver diseases such as autoimmune cholangitis and PSC. Awareness of this evidence may help clinicians to better orient their correct and timely diagnostic/therapeutic approach.
The authors are grateful to Jean Ann Gilder (Scientific Communication srl) for revising and editing the text.
1. Troncone R, Jabri B. Coeliac disease and gluten sensitivity. J Intern Med
2. Rewers M. Epidemiology of celiac disease: what are the prevalence, incidence, and progression of celiac disease? Gastroenterology
2005; 128 (suppl 1):S47–S51.
3. Pollock DJ. The liver in coeliac disease. Histopathology
4. Hagander B, Brandt L, Sjolund K, et al. Hepatic injury in adult coeliac disease. Lancet
5. Lindberg T, Berg NO, Borulf S, et al. Liver damage in coeliac disease or other food intolerance in childhood. Lancet
6. Bonamico M, Pitzalis G, Culasso F, et al. Hepatic damage in celiac disease in children. Minerva Pediatr
7. Panetta F, Nobili V, Sartorelli MR, et al. Celiac disease in pediatric patients with autoimmune hepatitis: etiology, diagnosis, and management. Paediatr Drugs
8. Sainsbury A, Sanders DS, Ford AC. Meta-analysis: coeliac disease and hypertransaminasaemia. Aliment Pharmacol Ther
9. Iorio R, Sepe A, Giannattasio A, et al. Hypertransaminasemia in childhood as a marker of genetic liver disorders. J Gastroenterol
10. Altuntaş B, Kansu A, Girgin N. Hepatic damage in gluten sensitive enteropathy. Acta Paediatr Jpn
11. Farre C, Esteve M, Curcoy A, et al. Hypertransaminasemia in pediatric celiac disease patients and its prevalence as a diagnostic clue. Am J Gastroenterol
12. Di Biase AR, Colecchia A, Scaioli E, et al. Autoimmune liver diseases in a paediatric population with coeliac disease—a 10-year single-centre experience. Aliment Pharmacol Ther
13. Bardella MT, Fraquelli M, Quatrini M, et al. Prevalence of hypertransaminasemia in adult celiac patients and effect of gluten-free diet. Hepatology
14. Vajro P, Fontanella A, Mayer M, et al. Elevated serum aminotransferase activity as an early manifestation of gluten-sensitive enteropathy. J Pediatr
15. Swarbrick ET, Fairclough PD, Campbell PJ, et al. Coeliac disease, chronic active hepatitis, and mesangiocapillary glomerulonephritis in the same patient. Lancet
16. Mirzaagha F, Azali SH, Islami F, et al. Coeliac disease in autoimmune liver disease: across-sectional study and a systematic review. Dig Liver Dis
17. Francavilla R, Castellaneta SP, Davis T, et al. Coeliac disease in children with autoimmune hepatitis. Dig Liver Dis
18. Sorensen HT, Thulstrup AM, Blomqvist P, et al. Risk of primary biliary liver cirrhosis in patients with coeliac disease: Danish and Swedish cohort data. Gut
19. Lawson A, West J, Aithal GP, et al. Autoimmune cholestatic liver disease in people with coeliac disease: a population-based study of their association. Aliment Pharmacol Ther
20. Franzese A, Iannucci MP, Valerio G, et al. Atypical celiac disease presenting as obesity-related liver dysfunction. J Pediatr Gastroenterol Nutr
21. Bardella MT, Valenti L, Pagliari C, et al. Searching for coeliac disease in patients with non-alcoholic fatty liver disease. Dig Liver Dis
22. Barta Z, Miltenyi Z, Toth L, et al. Hypokalemic myopathy in a patient with gluten-sensitive enteropathy and dermatitis herpetiformis Duhring: a case report. World J Gastroenterol
23. Stenhammar L, Klintberg B, Tevebring J, et al. Muscular dystrophy misdiagnosed as hepatic disease in a child with coeliac disease. Acta Paediatr
24. Schwarzer G. Package ‘meta’. Meta-analysis with R
, Version 2.1-4. 2012.
25. Nobili V, Reale A, Alisi A, et al. Elevated serum ALT in children presenting to the emergency unit: relationship with NAFLD. Dig Liver Dis
26. Tommasini A, Not T, Kiren V, et al. Mass screening for coeliac disease using antihuman transglutaminase antibody assay. Arch Dis Child
27. Poupon R. Chronic autoimmune hepatitis. http://www.orpha.net
. Updated March 2007. Accessed December 7, 2012.
28. Woynarowski M, Woźniak M, Socha J. Transition of adolescents with autoimmune hepatitis from pediatric to adult care. Exp Clin Hep
29. Maggiore G, Sciveres M. Autoimmune hepatitis: a childhood disease. Curr Pediatr Rev
30. Arslan N, Büyükgebiz B, Oztürk Y, et al. The prevalence of liver function abnormalities in pediatric celiac disease patients and its relation with intestinal biopsy findings. Acta Gastroenterol Belg
32. Caprai S, Vajro P, Ventura A, et al. Autoimmune liver disease associated with celiac disease in childhood: a multicenter study. Clin Gastroenterol Hepatol
33. Ventura A, Magazzù G, Greco L. Duration of exposure to gluten and risk for autoimmune disorders in patients with celiac disease. SIGEP Study Group for Autoimmune Disorders in Celiac Disease. Gastroenterology
34. Diamanti A, Basso MS, Pietrobattista A, et al. Prevalence of celiac disease in children with autoimmune hepatitis. Dig Liver Dis
35. El-Shabrawi M, El-Karaksy H, Mohsen N, et al. Celiac disease in children and adolescents with autoimmune hepatitis: a single-centre experience. J Trop Pediatr
36. Ludvigsson JF, Elfström P, Broomé U, et al. Celiac disease and risk of liver disease: a general population-based study. Clin Gastroenterol Hepatol
37. Lacaille F, Canioni D, Bernard O, et al. Celiac disease, inflammatory colitis and sclerosing cholangitis in a girl with Turner's syndrome: a case of digestive autoimmunity. J Pediatr Gastroenterol Nutr
38. Gregorio GV, Portmann B, Mowat AP, et al. A 12-year-old girl with antimitochondrial antibody-positive autoimmune hepatitis. J Hepatol
39. Invernizzi P, Alessio MG, Smyk DS, et al. Autoimmune hepatitis type 2 associated with an unexpected and transient presence of primary biliary cirrhosis-specific antimitochondrial antibodies: a case study and review of the literature. BMC Gastroenterol
40. Maggiore G, De Giacomo C, Scotta MS, et al. Celiac disease presenting as chronic hepatitis in a girl. J Pediatr Gastroenterol Nutr
41. Drago S, El Asmar R, Di Pierro M, et al. Gliadin, zonulin and gut permeability: effects on celiac and nonceliac intestinal mucosa and intestinal cell lines. Scand J Gastroenterol
42. Maggiore G, Caprai S. Liver involvement in celiac disease. Indian J Pediatr
43. Maggiore G, Caprai S. The liver in celiac disease. J Pediatr Gastroenterol Nutr
44. Leonardi S, Bottaro G, Patané R, et al. Hypetransaminasemia as the first symptom in infant celiac disease. J Pediatr Gastroenterol Nutr
45. Kooy-Winkelaar Y, van Lummel M, Moustakas AK, et al. Gluten-specific T cells cross-react between HLA-DQ8 and the HLA-DQ2α/DQ8β transdimer. J Immunol
46. Volta U, Rodrigo L, Granito A, et al. Celiac disease in autoimmune cholestatic liver disorders. Am J Gastroenterol
47. Jacobsen MB, Fausa O, Elgjo K, et al. Hepatic lesions in adult coeliac disease. Scand J Gastroenterol
48. Kingham JG, Parker DR. The association between primary biliary cirrhosis and coeliac disease: a study of relative prevalences. Gut
49. Duggan JM, Duggan AE. Systematic review: the liver in celiac disease. Aliment Pharmacol Ther
50. Bridoux-Henno L, Dabadie A, Briard D, et al. A case of celiac disease presenting with autoimmune hepatitis and erythroblastopenia. J Pediatr Gastroenterol Nutr