[Predictive value of serological tests in the diagnosis of celiac disease].Ann Ital Med Int. 2002 Apr-Jun; 17(2):102-7.AI
In the diagnostic work-up of celiac disease (CD) the simpler enzyme-linked immunosorbent assay (ELISA) for the identification of serum anti-transglutaminase (tTG) autoantibodies could substitute the immunofluorescence technique used for the detection of anti-endomysial antibodies (EmA). However, most of the studies on anti-tTG assay have considered pre-selected groups of patients and not consecutive subjects with suspected CD. The aim of this study was to compare the sensitivity, specificity and predictive value of anti-gliadin antibodies (AGA), EmA and two anti-tTG ELISAs, one based on guinea pig (gp)-tTG and the other on human (h)-tTG as antigens, in consecutive patients investigated for suspected CD. The study included 130 consecutive patients (age range 16-84 years), who underwent intestinal biopsy for suspected CD. They presented with one or more of the following symptoms: weight loss, anemia, chronic diarrhea, abdominal pain, dyspepsia, alternating bowel habits and constipation. At the time of admission in the study, an intestinal biopsy was performed and a serum sample was taken for immunoglobulin (Ig) G and IgA AGA, IgA EmA, anti-gp-tTG and anti-h-TG determination. Intestinal histology revealed that 15 patients had partial or total villous atrophy. In these patients the diagnosis of CD was confirmed at subsequent follow-up. The remaining 115 patients included in the study had an intestinal histology characterized by a normal villi/crypts ratio and were considered as controls. Serum EmA, anti-gp-tTG, and anti-h-tTG were positive in all the 15 CD subjects, whereas IgG and IgA AGA were positive in 10/15; in the control group, none were positive for serum EmA, but 11/115 (10%) were positive for anti-gp-tTG and 6/115 (5%) were positive for anti-h-tTG. The sensitivity was 100% for EmA, gp-tTG and h-tTG and 66% for IgA and IgG AGA. The specificity was 100% for EmA, 90% for anti-gp-tTG, 95% for anti-h-tTG, 74% for IgG AGA and 87% for IgA AGA. The negative predictive value was 100% for EmA, anti-h-tTG and anti-gp-tTG, 94% for IgG AGA and 95% for IgA AGA. The positive predictive value was 100% for EmA, 71% for anti-h-tTG (p = 0.03 vs EmA) and 58% for anti-gp-tTG (p = 0.003 vs EmA). Most of the patients who were false positive for anti-tTG had Crohn's disease or chronic liver disease. In conclusion, although both the anti-tTG ELISAs evaluated in the present study showed an optimum sensitivity, their low specificity determined positive predictive values which were significantly lower than those of EmA assay. Besides, the positive predictive value of gp-tTG was too low to warrant submitting a patient to intestinal biopsy for suspected CD.