Screening for celiac disease in patients with chronic liver disease

Dear Sir:We read with interest the paper of Kaukinen et al. reporting anelevated frequency of celiac disease (CD) (4.3%) in patients with previousliver transplantation due to severe hepatic dysfunction.1However, wewould like to report our experience of the serologic assays for CDdiagnosis in patients with chronic liver disease and comment on thescreening methods for CD used in that study. Between January andOctober 2001, we studied 96 consecutive subjects (65 male, 31 female,age range 18–87 years, median 42) with chronic hypertransaminasemiawho were attending for thefirst time the outpatients clinic for liverdisease at the Internal Medicine Division of the University Hospital ofPalermo. All patients underwent a complete work-up including routineliver function tests, serologic screening for viral hepatitis, presence ofserum autoantibodies, copper- and iron-related indexes,1-antitrypsinlevels, liver ultrasonography, and percutaneous liver biopsy. We foundchronic hepatitis in 70 cases, liver cirrhosis in 20 patients, nonalcoholicsteatohepatitis in 3 patients, and primary biliary cirrhosis in 3 patients.In all cases, serological screening for CD was also performed by deter-mining the values of total immunoglobulin (to exclude selective IgAdeficiency), IgA class anti-endomysial antibody (EmA) using monkey’sesophagus as the substrate, and IgA class anti-human transglutaminaseantibody (Anti-tTG). Both the serum EmA and anti-tTG assays wereperformed with commercially available assays as previously described andthe threshold value for serum anti-tTG was7%.2We found 3 of 96patients positive for serum anti-tTG, but only one of them was positivefor serum EmA. This last patient, an 18-year-old girl, had mild chronichepatitis due to hepatitis B virus; she had never reported gastrointestinalsymptoms or anemia and her body mass index was normal. She under-went the duodenal biopsy and the histology study, performed as previ-ously described,3which showed severe atrophy of the mucosa (villi/cryptsratio 0.5) with an elevated intraepithelial lymphocytes count, thus con-firming the CD diagnosis. The 2 other patients, positive for serumanti-tTG but negative for EmA, had mild chronic hepatitis and livercirrhosis respectively, both due to hepatitis C virus infection. These 2patients accepted to undergo the intestinal biopsy but their intestinalhistology was normal (villi/crypts ratio3.5, normal intraepitheliallymphocytes count). Furthermore, both these patients were negative forthe presence of the HLA-DQ2 and HLA-DQ8 antigens, which charac-terize CD patients. The 2 patients with false positive anti-tTG results didnot differ from all the other subjects included in the study with regard toliver histology grading and staging, severity of liver function impair-ment, serum-globulin, and etiology of the liver disease. Furthermore,both were negative for serum auto-antibodies. In total, we consequentlyfound 1% frequency of CD and 2% of false positive anti-tTG results forCD diagnosis in our group of patients with chronic liver disease.Kaukinen et al. reported a much higher frequency of CD (4.3%)very probably linked to the high frequency of autoimmune liverdisease (primary biliary cirrhosis, primary sclerosing cholangitis, au-toimmune hepatitis) in their series.1However, although our dataindicated that the frequency of CD was lower in consecutive patientswith chronic liver disease, not including exclusively subjects withend-stage liver disease, we think that CD screening should be per-formed in any case as Kaukinen et al. clearly demonstrated thatdietary treatment in CD patients may prevent the progression tohepatic failure.1In regard to the serologic screening for CD, in theFinnish study there was a 2-fold higher frequency of false positiveanti-tTG results (8 of 185 cases, 4.3%) than in our study. This wasprobably in part due to the use of an anti-tTG ELISA based on tTGfrom guinea pig liver as the antigen4; in fact, we have previouslydemonstrated that in patients with chronic liver disease this systemgives a higher number of false positive results than the anti-tTGELISA based on human recombinant tTG as the antigen and shouldnot be used as a screening tool for CD.5However, our present dataunderlined that in patients with chronic liver diseases, false positiveanti-tTG results can also be observed using the anti-tTG assay basedon human antigen. Consequently, we suggest that when EmA assay doesnot confirm the positivity of the anti-tTG assay, determination of theHLA haplotype should be the subsequent step and only DQ2 or DQ8positive subjects should undergo intestinal biopsy for CD diagnosis.