BackgroundEsophageal variceal diameter (EVD) is a crucial factor in determining the risk of esophageal variceal bleeding, which is associated with an increased portal pressure gradient (PPG). However, research into the relationship between EVD and PPG has been limited, primarily because the assessment of EVD depends on visual estimation during endoscopy. Recently, we developed an artificial intelligence (AI)-based method to accurately detect EVD. In this study, we aim to investigate the correlation between EVD and PPG, with the goal of evaluating EVD as a potential non-invasive indicator of PPG.MethodsThis study included both retrospective and prospective data from 128 patients diagnosed with portal hypertension and gastroesophageal varices, gathered from two medical institutions. Clinical data including PPG, biochemical markers, and routine blood tests were collected. In the retrospective phase, EVD was evaluated using an AI-based virtual ruler. In the prospective phase, PPG was measured using radiological intervention methods, and EVD was measured during endoscopy with the aid of AI.ResultsA positive correlation between PPG and EVD was identified (r = 0.521, P < 0.001), which was further supported by multivariate linear regression analysis (b = 6.521, t = 6.872, P < 0.001). When patients were stratified into two groups based on PPG levels (27 patients with PPG < 20 mmHg and 101 patients with PPG ≥ 20 mmHg), a significant difference in EVD was observed between the groups (OR = 29.275, 95% CI 5.590–153.304, P < 0.001), with larger EVD in the higher PPG group. These findings suggest that EVD may serve as a predictor of adverse events associated with elevated PPG levels. In addition, receiver operating characteristic (ROC) curve analysis showed that EVD had an accuracy of 0.814 in diagnosing PPG function (standard error 0.048, 95% CI 0.720-0.908; P < 0.001), indicating that PPG levels are likely to exceed 20 mmHg when the variceal diameter is greater than 1.1 cm.ConclusionEVD demonstrated a positive correlation with PPG and could potentially be used as a predictive marker for assessing PPG levels. These findings provide novel insights for the non-invasive evaluation of PPG in clinical practice.
Read full abstract