In 1984, Sarfeh et al.  proposed that portal hypertension would lead to specific gastric mucosal haemodynamics and morphological changes, manifesting as gastric mucosal hyperaemia, oedema, erosion and even haemorrhage under endoscopy. Therefore, "portal hypertensive gastritis" was proposed. In 1985, McCormack's team  described mucosal hyperaemia and mild inflammation in the fundus and body of the stomach. Histological biopsy showed that capillary dilation in the mucosal layer far exceeded mucosal inflammation, and the concept of "congestive gastropathy" was first proposed. In 1986, Sarfeh et al. formally proposed the term "portal hypertensive gastropathy", which is still used today. Endoscopic presentation of gastric vascular ectasia (GAVE) is a flat red spot that is located in the distal stomach (gastric antrum) and often arises in strips from the gastric antrum. It resembles a watermelon, the so-called "watermelon stomach". When it is difficult to distinguish from PHG, biopsy of gastric mucosal lesions should be considered for diagnosis. Histological manifestations of GAVE were telangiectasia, proliferation of lamina propria spindle cells, fibrin thrombus and fibrin deficiency .
PHG is a term of endoscopic diagnosis, in which endoscopic features include a typical snakeskin mosaic pattern, flat or bulging red marks or red spots resembling vascular ectasias. The histologic findings of PHG include mucosa and submucosa capillaries and small veins, no obvious inflammation and no microthrombosis. The most common location for PHG is the fundus and body of the stomach . Portal hypertension can promote short circuits in the gastric mucosal layer and submucosa, resulting in insufficient blood supply and oxygen to gastric mucosal cells. Portal hypertension can cause backflow obstruction and congestion of small blood vessels in the gastric mucosa, thus weakening the mucosal defence barrier . Liver hypofunction activates kinins and then limits vasoconstriction, leading to gastric mucosa repair disorders. Renin–angiotensin–aldosterone activation aggravates liver fibrosis and then aggravates congestion of gastric mucosa.
A total of 325 patients with cirrhosis were included in the present study, and 284 patients were diagnosed with PHG. The incidence of PHG in cirrhosis was 87.4% and included 247 patients (76.0%) with mild PHG and 37 patients (11.4%) with severe PHG. Abbasi et al.  found that the incidence of PHG in cirrhosis was 79.3%. The incidence of PHG in cirrhosis fluctuates from 20.0% to 98.0% due to inconsistencies in description, population differences, lack of uniform diagnostic and grading criteria and interobserver and intraobserver differences.
Most studies have shown no significant relationship between PHG and gender. However, Simbrunner's team  pointed out that the male ratio of patients with severe PHG was significantly higher than that of patients with mild PHG or patients without PHG, and gender was an independent risk factor for severe PHG. Meanwhile, this study found that male patients with cirrhosis were more likely to develop PHG than female patients. Animal studies have shown that oestrogen and progesterone therapy can reduce portal pressure, gastric mucosal blood flow, vessel number and relative vessel area. Currently, there is insufficient evidence to support gender-related endocrine or haemodynamic influences on PHG, and the relationship between the severity of PHG and gender needs further prospective study.
In the present study, haemoglobin, platelet count, albumin, prothrombin time and ascites were closely related to the occurrence of PHG. And we found that albumin was predictor of PHG. Anaemia is a common complication in patients with cirrhosis. The main reason is gastrointestinal bleeding, followed by splenomegaly and hypersplenism, which also cause increased destruction of RBCs and platelets. Studies have shown that patients with severe PHG have significantly reduced haemoglobin, and with the aggravation of PHG, moderate to severe anaemia is more common. Prothrombin time and albumin can measure liver reserve function. When liver function is damaged in cirrhosis, albumin and coagulation factor synthesis are decreased. Min et al.  retrospectively studied 232 patients with chronic liver disease, and multivariate analysis indicated that albumin, platelet count and spleen volume were independent risk factors for PHG.
Some studies  have indicated that among 24 patients with severe PHG, 18 (75.0%) were Child–Pugh C and six (25.0%) were Child–Pugh B, indicating that severe PHG was more common in Child–Pugh C patients. PHG was associated with the Child–Pugh grade of liver function. Some studies have indicated  that the occurrence of PHG is related to staged liver function, and the incidence of PHG is high in patients with Child–Pugh B and C cirrhosis, while the occurrence of PHG has nothing to do with Child–Pugh A. In contrast, El-Kalla et al.  pointed out that the occurrence and severity of PHG were independent of liver function. Mezawa et al.  pointed out that PHG was relieved after TIPS, but liver function did not change significantly before or two weeks after surgery, so it was speculated that PHG had no significant relationship with liver function.
The present study indicated that Child–Pugh liver function was correlated with PHG occurrence. It is speculated that the deterioration of liver function aggravates portal hypertension, shorting the arteries and veins of gastric mucosa and submucosa, resulting in insufficient blood supply and oxygen supply to gastric mucosa cells. Portal hypertension is prone to blocked reflux and congestion of small blood vessels in gastric mucosa, both of which will aggravate gastric mucosa injury. The deterioration of liver function resulted in activation of the kinin system, imbalance of vasoconstriction and vasodilation and impaired repair of gastric mucosal cell proliferation. Impaired liver function activates the renin–angiotensin–aldosterone system, which exacerbates liver fibrosis, liver function and portal hypertension, eventually leading to a vicious cycle.
Many studies have explored the relationship between PHG and portal pressure and varices, but no consensus report has been reached. Kumar et al.  retrospectively analysed 254 patients with cirrhosis. The average varices grade in PHG patients was grade 3 and that in patients without PHG was grade 2 and the probability of PHG in large varices was 2.83 times higher than that in small varices. Saleem et al.  showed that the occurrence and severity of PHG were independent of the degree of EV. However, Tiwari et al.  found no relationship between PHG and oesophageal and gastric varices.
The present study showed that the occurrence and severity of PHG were related to the severity of EV and GV. The more serious the varices are, the higher the incidence of PHG and the more serious the PHG. We found that the degree of EV and GV were predictors of PHG. Portal hypertension is the initiating factor of PHG, and varicose veins are important clinical manifestations of portal hypertension. When portal venous pressure is higher than gastric venous pressure, gastric venous return is blocked, resulting in mucosal and submucosal vascular dilation and intravascular congestion. Portal hypertension causes gastric mucosa and submucosal portal hypertension causes the formation of submucosal arteriovenous anastomosis branches, reduces gastric mucosa perfusion and causes gastric mucosa ischaemia and hypoxia injury. This abnormality, in turn, may lead to epithelial cell damage and establish an environment for excessive production of oxygen free radicals, nitric oxide, tumor necrosis factor-alpha, endothelin-1, prostaglandins and/or other factors leading to cell damage .
Overall, albumin and the degree of EV and GV, which were independently associated with PHG, were predictors of PHG. When albumin is reduced or varicose veins are serious in liver cirrhosis patients, more attention should be given to the occurrence and severity of PHG.
The limitation of this study was that it was a single-centre, retrospective study. However, the incidence and risk factors for PHG were systematically analysed. Further prospective studies are needed to clarify the relationship between PHG and gender, liver function, and varices to provide help for clinical diagnosis and treatment.