Reliability in endoscopic diagnosis of portal hypertensive gastropathy.

Aim: To analyze reliability among endoscopists in diagnosing portal hypertensive gastropathy (PHG) and to determine which criteria from the most utilized classifications are the most suitable.

Methods: From January to July 2009, in an academic quaternary referral center at Santa Casa of São Paulo Endoscopy Service, Brazil, we performed this single-center prospective study. In this period, we included 100 patients, including 50 sequential patients who had portal hypertension of various etiologies; who were previously diagnosed based on clinical, laboratory and imaging exams; and who presented with esophageal varices. In addition, our study included 50 sequential patients who had dyspeptic symptoms and were referred for upper digestive endoscopy without portal hypertension. All subjects underwent upper digestive endoscopy, and the images of the exam were digitally recorded. Five endoscopists with more than 15 years of experience answered an electronic questionnaire, which included endoscopic criteria from the 3 most commonly used Portal Hypertensive Gastropathy classifications (McCormack, NIEC and Baveno) and the presence of elevated or flat antral erosive gastritis. All five endoscopists were blinded to the patients' clinical information, and all images of varices were deliberately excluded for the analysis.

Results: The three most common etiologies of portal hypertension were schistosomiasis (36%), alcoholic cirrhosis (20%) and viral cirrhosis (14%). Of the 50 patients with portal hypertension, 84% were Child A, 12% were Child B, 4% were Child C, 64% exhibited previous variceal bleeding and 66% were previously endoscopic treated. The endoscopic parameters, presence or absence of mosaic-like pattern, red point lesions and cherry-red spots were associated with high inter-observer reliability and high specificity for diagnosing Portal Hypertensive Gastropathy. Sensitivity, specificity and reliability for the diagnosis of PHG (%) were as follows: mosaic-like pattern (100; 92.21; High); fine pink speckling (56; 76.62; Unsatisfactory); superficial reddening (69.57; 66.23; Unsatisfactory); red-point lesions (47.83; 90.91; High); cherry-red spots (39.13; 96.10; High); isolated red marks (43.48; 88.31; High); and confluent red marks (21.74; 100; Unsatisfactory). Antral elevated erosive gastritis exhibited high reliability and high specificity with respect to the presence of portal hypertension (92%) and the diagnosis of portal hypertensive gastropathy (88.31%).

Conclusion: The most suitable endoscopic criteria for the diagnosis of PHG were mosaic-like pattern, red-point lesions and cherry-red spots with no subdivisions, which were associated with a high rate of inter-observer reliability.