Peptic Ulcer Bleeding in Patients With or Without Cirrhosis
Peptic Ulcer Bleeding in Patients With or Without Cirrhosis
In this prospective study comparing PUB in two groups of patients with or without cirrhosis, we suggest that aetiology of PUB was probably different in cirrhotic patients; however, global outcome was not different.
Based on endoscopic screening, the prevalence of peptic ulcer in cirrhosis has been reported to be approximately 5–20%. In a recent prospective study, it has been shown that in 365 patients with cirrhosis who underwent upper endoscopy, 67 (18%) had peptic ulcer. The authors concluded that there was a lack of association between H. pylori infection and peptic ulcer disease in cirrhosis, underlying a probable different aetiology of peptic ulcer disease in cirrhosis. Furthermore, eradication of H. pylori in patients with cirrhosis does not effectively reduce the recurrence of peptic ulcers. To our knowledge, aetiology and prognosis of PUB have never been described in patients with cirrhosis. Our results are in accordance with these previous findings, as the large majority of patients with cirrhosis had negative H. pylori serology.
We reported a probable different aetiology, as PUB in group Cirr+ was found idiopathic in the majority of cases (79.3%) that is neither due to H. pylori nor to NSAIDs. Patients in Cirr+ group are significantly younger than in Cirr− group. However, this difference may not explain the difference of prevalence of H. pylori between Cirr+ and Cirr− group. Indeed, prevalence of H. pylori in the study area was about 40% at the age of 50 and was 50% at the age of 60 in a recent French cohort (JC Delchier, personal data). Use of NSAIDs may have been underestimated by some patients, especially patients with cirrhosis and encephalopathy. However, patients were asked by three different physicians and different names of NSAIDs to try to avoid this bias (local procedures). Moreover, all patients had a second questioning on the day of discharge. Physiopathology of PUB in cirrhosis is an unsolved question. Not surprisingly, we found stigmata of portal hypertension during initial endoscopy, suggesting that PUB could be related to portal hypertension in cirrhotic patients. As previously reported, the majority of patients with cirrhosis and peptic ulcer disease have stigmata of portal hypertension. Several other factors have been suggested in cirrhotic patients: hypergastrinaemia, impairment of gastric mucosal defense secondary to portal hypertension and a decrease in gastric prostaglandin E2 levels. However, those results have not been clearly re-evaluated. Other descriptions suggested a lower H. pylori colonisation in patients with severe gastropathy. Regarding general characteristics of patients, there was a difference concerning age and excessive alcohol consumption in the two groups; we found indeed that patients were younger in group Cirr+ than in group Cirr−.
Not surprisingly, alcohol consumption was more important in group Cirr+ than in group Cirr−, as all patients had cirrhosis in group 1. In our population of patients with cirrhosis, 85% have alcohol-related cirrhosis. However, among patients with cirrhosis, all patients with idiopathic PUB had alcoholic cirrhosis. Therefore, one can hypothesise that PUB in cirrhosis might be related in part to gastric/duodenal injuries related to alcohol. This phenomenon could be increased by portal hypertension, as in noncirrhotic patients, 75% of those with idiopathic ulcers did not have any alcohol consumption. Another explanation could be that ulcers are directly related to portal hypertension, as hepatic venous pressure gradient (HVPG) is higher in alcoholic cirrhotic patients.
Our results concerning prognosis of PUB are in accordance with those described in the literature, even if mortality is a little bit lower than described in other series; this may be related to the systematic treatment of Pub refractory to endoscopic and medical treatment with arterial embolisation. Noteworthy, as we included only patients admitted in our ICU, we were dealing with a majority of patients with high-risk PUB (78%). Moreover, another important finding is that global outcome is not different in the two groups of patients. This is an unexpected result, as usually global mortality of upper gastrointestinal bleeding is high in cirrhosis, in case of portal hypertension–related bleeding, and even in nonvariceal bleeding. All patients received high-dose PPI's since admission, were performed upper endoscopy within the 6 h after admission, and a double endoscopic treatment was always attempted. However, even if we tried to strictly follow the recommendations concerning dual endoscopic therapy, a single treatment was effectively applied in about 50% of cases. We changed our policy concerning salvage surgery, as many studies have now suggested that arterial embolisation could be an efficient and safe procedure for refractory PUB. This attitude could explain our good results concerning mortality in cirrhosis, as salvage surgery could always be avoided in our group of cirrhotic patients. Therefore, maximum nonsurgical treatment for PUB, including arterial embolisation, should always be applied to patients with cirrhosis. We last described that mean unit blood transfused was not different between the two groups. This result is in accordance with recent data concerning haemostasis in cirrhosis. When patients are assessed by global test such as thrombin-generation test, the results do not show hypocoagulability.
Our study has several pitfalls: the number of patients with cirrhosis is small. Even though, we could conclude that there was a difference of physiopathology. Second, we did not perform gastric biopsies to describe histological findings especially in cirrhotic patients. We do think that this could be a part of a further study. Last, we did not measure HVPG in our cohort of cirrhotic patients; the suggestion that ulcer bleeding may be related to portal hypertension is based on endoscopic findings, which are quite subjective, with a relatively inter-observatory variability. However, all the physicians in our ICU are experienced in PUB management and in portal hypertension–related gastrointestinal bleeding.
In conclusion, we found that PUB is neither explained by H. pylori nor NSAIDs in the majority of cases, when PUB occurred in patients with cirrhosis. PUB occurs almost only when alcohol is the aetiology of cirrhosis. Our study suggests that ulcer bleeding may be related to portal hypertension, as nearly all patients with cirrhosis and ulcer bleeding had endoscopic stigmata of portal hypertension. However, exact physiopathology should be further studied, with histology and data about HVPG. Even though, global outcome of PUB in a population of cirrhotic patients is comparable to general population, when maximal nonsurgical treatment is applied, including arterial embolisation.
Discussion
In this prospective study comparing PUB in two groups of patients with or without cirrhosis, we suggest that aetiology of PUB was probably different in cirrhotic patients; however, global outcome was not different.
Based on endoscopic screening, the prevalence of peptic ulcer in cirrhosis has been reported to be approximately 5–20%. In a recent prospective study, it has been shown that in 365 patients with cirrhosis who underwent upper endoscopy, 67 (18%) had peptic ulcer. The authors concluded that there was a lack of association between H. pylori infection and peptic ulcer disease in cirrhosis, underlying a probable different aetiology of peptic ulcer disease in cirrhosis. Furthermore, eradication of H. pylori in patients with cirrhosis does not effectively reduce the recurrence of peptic ulcers. To our knowledge, aetiology and prognosis of PUB have never been described in patients with cirrhosis. Our results are in accordance with these previous findings, as the large majority of patients with cirrhosis had negative H. pylori serology.
We reported a probable different aetiology, as PUB in group Cirr+ was found idiopathic in the majority of cases (79.3%) that is neither due to H. pylori nor to NSAIDs. Patients in Cirr+ group are significantly younger than in Cirr− group. However, this difference may not explain the difference of prevalence of H. pylori between Cirr+ and Cirr− group. Indeed, prevalence of H. pylori in the study area was about 40% at the age of 50 and was 50% at the age of 60 in a recent French cohort (JC Delchier, personal data). Use of NSAIDs may have been underestimated by some patients, especially patients with cirrhosis and encephalopathy. However, patients were asked by three different physicians and different names of NSAIDs to try to avoid this bias (local procedures). Moreover, all patients had a second questioning on the day of discharge. Physiopathology of PUB in cirrhosis is an unsolved question. Not surprisingly, we found stigmata of portal hypertension during initial endoscopy, suggesting that PUB could be related to portal hypertension in cirrhotic patients. As previously reported, the majority of patients with cirrhosis and peptic ulcer disease have stigmata of portal hypertension. Several other factors have been suggested in cirrhotic patients: hypergastrinaemia, impairment of gastric mucosal defense secondary to portal hypertension and a decrease in gastric prostaglandin E2 levels. However, those results have not been clearly re-evaluated. Other descriptions suggested a lower H. pylori colonisation in patients with severe gastropathy. Regarding general characteristics of patients, there was a difference concerning age and excessive alcohol consumption in the two groups; we found indeed that patients were younger in group Cirr+ than in group Cirr−.
Not surprisingly, alcohol consumption was more important in group Cirr+ than in group Cirr−, as all patients had cirrhosis in group 1. In our population of patients with cirrhosis, 85% have alcohol-related cirrhosis. However, among patients with cirrhosis, all patients with idiopathic PUB had alcoholic cirrhosis. Therefore, one can hypothesise that PUB in cirrhosis might be related in part to gastric/duodenal injuries related to alcohol. This phenomenon could be increased by portal hypertension, as in noncirrhotic patients, 75% of those with idiopathic ulcers did not have any alcohol consumption. Another explanation could be that ulcers are directly related to portal hypertension, as hepatic venous pressure gradient (HVPG) is higher in alcoholic cirrhotic patients.
Our results concerning prognosis of PUB are in accordance with those described in the literature, even if mortality is a little bit lower than described in other series; this may be related to the systematic treatment of Pub refractory to endoscopic and medical treatment with arterial embolisation. Noteworthy, as we included only patients admitted in our ICU, we were dealing with a majority of patients with high-risk PUB (78%). Moreover, another important finding is that global outcome is not different in the two groups of patients. This is an unexpected result, as usually global mortality of upper gastrointestinal bleeding is high in cirrhosis, in case of portal hypertension–related bleeding, and even in nonvariceal bleeding. All patients received high-dose PPI's since admission, were performed upper endoscopy within the 6 h after admission, and a double endoscopic treatment was always attempted. However, even if we tried to strictly follow the recommendations concerning dual endoscopic therapy, a single treatment was effectively applied in about 50% of cases. We changed our policy concerning salvage surgery, as many studies have now suggested that arterial embolisation could be an efficient and safe procedure for refractory PUB. This attitude could explain our good results concerning mortality in cirrhosis, as salvage surgery could always be avoided in our group of cirrhotic patients. Therefore, maximum nonsurgical treatment for PUB, including arterial embolisation, should always be applied to patients with cirrhosis. We last described that mean unit blood transfused was not different between the two groups. This result is in accordance with recent data concerning haemostasis in cirrhosis. When patients are assessed by global test such as thrombin-generation test, the results do not show hypocoagulability.
Our study has several pitfalls: the number of patients with cirrhosis is small. Even though, we could conclude that there was a difference of physiopathology. Second, we did not perform gastric biopsies to describe histological findings especially in cirrhotic patients. We do think that this could be a part of a further study. Last, we did not measure HVPG in our cohort of cirrhotic patients; the suggestion that ulcer bleeding may be related to portal hypertension is based on endoscopic findings, which are quite subjective, with a relatively inter-observatory variability. However, all the physicians in our ICU are experienced in PUB management and in portal hypertension–related gastrointestinal bleeding.
In conclusion, we found that PUB is neither explained by H. pylori nor NSAIDs in the majority of cases, when PUB occurred in patients with cirrhosis. PUB occurs almost only when alcohol is the aetiology of cirrhosis. Our study suggests that ulcer bleeding may be related to portal hypertension, as nearly all patients with cirrhosis and ulcer bleeding had endoscopic stigmata of portal hypertension. However, exact physiopathology should be further studied, with histology and data about HVPG. Even though, global outcome of PUB in a population of cirrhotic patients is comparable to general population, when maximal nonsurgical treatment is applied, including arterial embolisation.
Source...