Introduction
Autoimmune gastritis (AIG) targets parietal cells and intrinsic factor, causing progressive corpus/fundus atrophy [1]. Global prevalence is 3.8% [2]. Advanced disease leads to iron/B12 malabsorption and elevated neoplasia risk: type 1 gastric neuroendocrine tumors (10%–37%) [3,4] and adenocarcinoma (2.8%–9.8%) [5,6]. Despite this, AIG remains underdiagnosed due to insidious onset [7]. Diagnostic gold standard is endoscopy with biopsy [7,8], but this is invasive, subject to sampling bias, and overlaps with H. pylori-associated gastritis, with interobserver variability [9]. Objective models integrating serological and endoscopic markers are urgently needed. Parietal cell antibodies (PCA) offer high sensitivity but limited specificity (false-positives from H. pylori, other autoimmune diseases, or advanced age) [10,11]. Conversely, progressive parietal cell loss can cause PCA seroconversion to negative, underdiagnosing severe AIG [12]. Serological markers reflecting gastric function—pepsinogen I, PGI/II ratio, and gastrin-17—are useful for assessing corpus atrophy [13], though their optimal combination for AIG diagnosis remains unclear.
Aims & Methods
This study aimed to determine whether integrating serum Gastrin-17 (G-17) and specific endoscopic features with PCA could overcome these diagnostic gaps to establish a highly accurate dual-model diagnostic strategy.
This single-center retrospective cohort study enrolled 792 patients with complete serological and endoscopic histopathological data. Using histopathology as the reference standard, patients were classified into AIG and non-AIG group. We analyzed clinical characteristics, endoscopic findings, and serological markers including PCA, intrinsic factor antibody (IFA), G-17, and pepsinogen.The optimal predictors were selected via Least Absolute Shrinkage and Selection Operator (LASSO) regression, and a diagnostic nomogram model was constructed using multivariable logistic regression. Model performance was evaluated using receiver operating characteristic (ROC) curves, 10-fold cross-validation, calibration curves, and decision curve analysis (DCA).
Results
Among 792 patients, 635 (80.2%) were diagnosed with AIG and 157 (19.8%) served as non-AIG controls. G-17 levels were markedly elevated in AIG patients (4.7 vs. 0.28 fold the upper limit of normal [ULN], P < 0.001). Notably, G-17 levels remained normal in PCA-positive patients without AIG, but were elevated in PCA-negative patients with AIG (median 2.9 fold ULN, P<0.001). LASSO regression retained only PCA status and log₁₀-transformed G-17 as the strongest independent predictors. The serological model incorporating these two markers demonstrated excellent discriminatory ability, with an AUC of 0.951, sensitivity of 87.2%, and specificity of 93.7%. Furthermore, integrating endoscopic features (corpus-restricted atrophy and adherent thick mucus) significantly improved the model's performance, yielding an AUC of 0.983. The models achieved a statistically significant improvement in diagnostic performance and showed robust stability and favorable clinical net benefit in internal validation.
Conclusion
Serum G-17 remains consistently elevated even in PCA-negative AIG patients, effectively addressing the diagnostic gap left by PCA alone and resolving challenges related to false-positive results. The combined serological nomogram (PCA and G-17) serves as an effective non-invasive diagnostic tool. Furthermore, the addition of endoscopic features yields excellent diagnostic accuracy in patients eligible for endoscopic evaluation.
References
1. Lenti MV, et al. Nat Rev Dis Primer 2020;6:56.
2. Lahner E, et al. Dig Liver Dis 2026;58:38-50.
3. Jove A, et al. Am J Gastroenterol 2025;120:1140-43.
4. Panozzo MP, et al. J Transl Autoimmun 2025;10:100273.
5. Zhang T, Tang X. QJM 2025;118:203-47.
6. Zhang H, et al. Scand J Gastroenterol 2017;52:150-6.
7. Shah SC, et al. Gastroenterology 2021;161:1325-1332.e7.
8. Rugge M, et al. Gut 2024;73:407-41.
9. Song Y, et al. J Gastroenterol Hepatol 2026;Feb 24.
10. Sugimoto M, et al. J Clin Gastroenterol 2025;May 5.
11. Conti L, et al. Dig Liver Dis 2020;52:1310-4.
12. Nishizawa T, et al. Scand J Gastroenterol 2022;57:143-8.
13. Kishikawa H, et al. Sci Rep 2022;12:4202.
Disclosure
The authors have no conflicts of interest to disclose.
