Stewart Charmaine A, Sanyal Arun J
Division of Gastroenterology, University of Pennsylvania Medical Center, Philadelphia, Pennsylvania, USA.
Am J Gastroenterol. 2003 Aug;98(8):1758-65. doi: 10.1111/j.1572-0241.2003.07595.x.
The specific aim of this study was to independently evaluate the reproducibility and validity of the Baveno portal hypertensive gastropathy (PHG) grading system.
This is a prospective study of 100 consecutive patients with cirrhosis and portal gastropathy. An endoscopy was performed at entry on all subjects and on 70 patients during follow-up (median 15 months). Interobserver reproducibility was assessed by the kappa statistic. Patients with PHG-related bleeding were compared with those without bleeding. The relationship of PHG-related bleeding to the PHG score was assessed by a receiver operating characteristic curve and by multiple logistic regression analysis.
Interobserver agreement about the presence of the mosaic pattern and red marks was high (kappa >0.75). However, the agreement about the extent of the lesion was poorer (kappa = 0.4-0.45). There was a stepwise increase in PHG-related bleeding risk with increasing PHG scores. Multiple logistic regression confirmed that the PHG score independently correlated with PHG-associated bleeding (OR = 2.5, 90% CI = 1.4-4.6, p < 0.009). During follow-up, red marks developed de novo in one of 70 patients. The severity of red marks worsened in five of 25 patients, whereas they disappeared in eight of 62 individuals. Nodular lesions in the antrum were found in five subjects. The risks of recurrent bleeding during follow-up were related to severe PHG scores (>4), presence of gastric antral vascular ectasia, and nodular lesions in the antrum.
The PHG scoring system is reproducible and accurately reflects the risks of PHG-related bleeding in patients with cirrhosis.
本研究的具体目的是独立评估巴韦诺门静脉高压性胃病（PHG）分级系统的可重复性和有效性。
这是一项对100例连续性肝硬化和门静脉性胃病患者的前瞻性研究。所有受试者在入组时均接受内镜检查，70例患者在随访期间（中位时间15个月）也接受了内镜检查。通过kappa统计量评估观察者间的可重复性。将发生PHG相关出血的患者与未出血的患者进行比较。通过受试者工作特征曲线和多因素逻辑回归分析评估PHG相关出血与PHG评分的关系。
观察者间关于马赛克样改变和红色征的一致性较高（kappa>0.75）。然而，关于病变范围的一致性较差（kappa=0.4 - 0.45）。随着PHG评分的增加，PHG相关出血风险呈逐步上升趋势。多因素逻辑回归证实，PHG评分与PHG相关出血独立相关（OR = 2.5，90%CI = 1.4 - 4.6，p < 0.009）。随访期间，70例患者中有1例新发红色征。25例患者中有5例红色征严重程度加重，而62例患者中有8例红色征消失。5例受试者胃窦部发现结节样病变。随访期间再出血风险与严重PHG评分（＞4）、胃窦血管扩张的存在以及胃窦部结节样病变有关。
PHG评分系统具有可重复性，能准确反映肝硬化患者PHG相关出血的风险。
