Gastric mucosal pit patterns to rule out premalignant conditions in an intermediate risk Western-Hispanic population:...

Endoscopic surveillance for early-detection of gastric cancer in Western populations relies on risk-stratification using OLGA/OLGIM staging. Validated endoscopic classifications to assess risk and guide biopsies are lacking. We aimed to validate the corpus Yagi AB and antral Pimentel classifications in an intermediate risk Western-Hispanic population.

We conducted a cross-sectional validation-study to assess the diagnostic-accuracy and interobserver agreement of both classifications. Adults undergoing endoscopy with NBI assessment and Sydney protocol biopsies were included. High risk status was defined as OLGA III-IV or OLGIM II-IV. For multivariate and diagnostic accuracy analyses, Yagi AB was dichotomized as B0-B1-B2-B3 (round/slightly elongated pits) vs A1-A2 (antralized pits), and Pimentel as Aa-Ab (circular/oval/round pits) vs B-C (abnormal patterns). Interobserver agreement between one-expert and five endoscopists was assessed using weighted-κ statistics on anonymized images.

Among 110 patients (mean age 59.1 years), 23.6% and 35.4% had high risk OLGA-OLGIM stages, respectively. Yagi AB patterns A1-A2 and Pimentel patterns B-C were independently associated with high risk OLGA/OLGIM (OR 24.4 and 40.4, both p<0.0001). In the image-based analysis (268 images), abnormal patterns showed sensitivities of 97-98% for moderate to severe chronic-atrophic-gastritis (CAG) or intestinal metaplasia (IM), whereas >97% of images with corpus patterns B0-B3 and antral patterns Aa-Ab corresponded to mild or non atrophic mucosa. Interobserver agreement was excellent for Yagi AB (weighted-κ=0.80-0.84) and moderate to substantial for Pimentel (weighted-κ=0.50-0.71).

In this Western-Hispanic cohort, the Yagi AB and Pimentel classifications showed appropriate interobserver agreement and allowed clinically significant CAG or IM to be reliably ruled out. Normal patterns may be used as a biopsy sparing gatekeeper within risk based surveillance strategies.