# Predictive Value of tTG-IgA ≥ 10×ULN for Villous Atrophy in a Romanian Adult Cohort: The Modifying Role of Smoking

## TL;DR

This study shows that high tTG-IgA levels are a reliable indicator of intestinal damage in adult celiac disease, but lower levels may still require a biopsy for accurate diagnosis.

## Contribution

The study identifies smoking as a modifying factor in the diagnostic accuracy of tTG-IgA for celiac disease in adults.

## Key findings

- High tTG-IgA levels (≥10×ULN) are highly specific but not sensitive for villous atrophy in adults.
- Smoking status significantly influences the likelihood of villous atrophy in celiac disease.
- A diagnostic gap exists for patients with lower tTG-IgA levels, emphasizing the need for biopsy in such cases.

## Abstract

Introduction: Celiac disease (CD) is a prevalent autoimmune enteropathy that remains significantly underdiagnosed due to its multifaceted diagnostic pathway and diverse clinical presentations. While duodenal biopsy has historically served as the diagnostic gold standard, its clinical primacy has been challenged by the burden of invasive endoscopy and potential histological misinterpretation. Material and Methods: We conducted a prospective diagnostic accuracy study involving consecutive adult patients with suspected CD to evaluate the performance of serological markers against histological findings. Results: The study included 139 patients, with a female predominance of 105 (75.5%). Histological evaluation revealed Marsh 3a–c in 100 patients (71.9%), whereas Marsh 1–2 was observed in 39 patients (28.1%). Sixty-one patients (43.9%) presented with high-titer ≥10×ULN tTG-IgA levels, while 78 patients (56.1%) fell below this threshold. To determine the independent predictors of Marsh 3a–c, we performed a logistic regression analysis. In the univariate analysis, both tTG-IgA (OR: 1.880; 95% CI: 1.458–2.426; p < 0.001) and non-smoker status (OR: 1.865; 95% CI: 1.283–2.709; p = 0.002) were significantly associated with VA. After adjusting for confounding variables in the multivariate model, both factors remained highly significant (p < 0.001 and p = 0.014, respectively). The diagnostic performance of the ≥10×ULN tTG-IgA threshold for detecting VA was confirmed by AUROC of 0.737, CI 0.646–0.827, p < 0.001, with a sensitivity of 55.0% and a specificity of 84.6%, a PPV of 90.2% and a NPVof 42.3% (33/78), underscoring that while the ≥10×ULN tTG-IgA threshold is highly specific for atrophy, lower titers do not reliably exclude it in adults. Conclusions: In conclusion, our study demonstrates that the ≥10×ULN tTG-IgA threshold provides a reliable diagnostic surrogate for VA in adult CD. While high cut-off values minimize false positives, a diagnostic gap remains for patients with lower antibody levels or those influenced by modifiers such as smoking. The low sensitivity of the high threshold reinforces the continued necessity of duodenal biopsy for symptomatic patients with lower-range antibody elevations to avoid a significant diagnostic gap.