RESUMO
BACKGROUND: Current adult celiac disease diagnosis requires histological confirmation. However, pediatric guidelines have proposed biopsy-sparing algorithms. AIMS: To explore the applicability of the ESPGHAN criteria and assess the accuracy of serology in predicting disease in adults. METHODS: We evaluated 234 consecutive adults showing elevated anti-tTG titers, EMA-positivity, and genetic susceptibility. Patients underwent upper endoscopy with duodenal biopsy. We determined optimal anti-tTG cutoff levels using ROC curves. RESULTS: Mean anti-tTG levels were 71.1 ± 66.5 U/ml; mean normalized levels were 14.8 ± 14.1 × ULN (mean ± SD). Partial/total villous atrophy was present in 36%/55% of cases, respectively. Anti-tTG levels correlated with histology (r s = 0.397, p < 0.001). AUC was similar before and after normalization (0.803 vs 0.807). Applying the ESPGHAN criterion (≥10 × ULN), we calculated a 97.66% PPV. ROC curve analysis showed an optimal cutoff of ≥16 × ULN, with a PPV of 98.86%. Eleven different assays were used for anti-tTG titer determination: Two were prevalent, labeled A (n = 141) and B (n = 59). They performed differently regarding disease prediction (AUC = 0.689 vs 0.925, p < 0.01), showing distinct optimal cutoff values (14.3 × ULN vs 3.7 × ULN), even after standardization (-0.14 vs -1.2). CONCLUSION: In adult symptomatic patients showing EMA-positivity and genetic susceptibility, anti-tTG titers correlated with histology. ESPGHAN criteria performed similarly to previous studies. However, a calculated 16 × ULN cutoff showed an improved PPV. Among prevalent assays, PPV peaked differently both after normalization and standardization, indicating intrinsic differences in performance, thus preventing uniform prediction of disease in a real-life setting. Assay-specific optimal cutoffs seem possible, but would complicate diagnostic criteria. However, biopsy-sparing strategies in adults could prove useful in challenging patients.