Long-term outcomes and bleeding risk factors in pancreatic portal hypertension: a retrospective cohort study.

backgroundPancreatic portal hypertension (PPH) is a distinct form of sinistral (left-sided) portal hypertension, primarily resulting from pancreatic diseases such as acute or chronic pancreatitis and pancreatic tumors. These conditions lead to splenic vein distortion, compression, and inflammatory thickening, which can progress to luminal obstruction and thrombosis. This thrombosis impairs venous return, elevates splenic vein pressure, and ultimately causes splenomegaly, hypersplenism, and left-sided portal hypertension. Current knowledge regarding the long-term prognosis of PPH is primarily derived from small-scale descriptive studies and case reports. Thus, large-scale, prospective cohort studies are needed to better characterize its natural history and outcomes.

objectiveThis retrospective cohort study aimed to determine the incidence of variceal bleeding (the primary complication) in patients with PPH and to identify its risk factors.

methodsA total of 207 patients diagnosed with PPH secondary to acute or chronic pancreatitis who were managed at Xijing Hospital between 2015 and 2023 were included. Demographic, clinical, laboratory, and imaging findings (e.g., thrombosis, stenosis, or occlusion of the splenic, portal, or superior mesenteric vein) were collected retrospectively. All patients were prospectively followed for the clinical endpoint of variceal bleeding. Cox proportional hazards regression analysis was performed to identify independent risk factors for variceal bleeding.

resultsThe crude incidence of bleeding was 11.6% (24/207). In contrast, Kaplan-Meier analysis revealed a cumulative incidence of approximately 30% at 3 years. On multivariate Cox regression analysis, smoking (hazard ratios [HR] 3.26, 95% confidence interval [CI]: 1.39–7.66, p = 0.007), chronic pancreatitis (HR 2.69, 95% CI: 1.04–6.91, p = 0.041), and diabetes mellitus (HR 2.83, 95% CI: 1.20–6.65, p = 0.017) were independent risk factors for bleeding. Conversely, a higher platelet count conferred a protective effect (HR 0.99 per 10^9/L increase, 95% CI: 0.98–1.00, p = 0.008). Of note, the clinical relevance of this per 10^9/L increase is likely limited. Among the intervention subgroups, bleeding occurred in 1 of 4 patients (25%) receiving endoscopic prophylaxis and in 1 of 15 patients (6.7%) on oral anticoagulation, whereas none of the 3 patients treated with non-selective beta-blockers (NSBBs) experienced bleeding. Formal statistical comparisons were not performed due to small and unequal sample sizes.

conclusionThis study suggests smoking, chronic pancreatitis, and diabetes mellitus as independent risk factors may be associated with variceal bleeding in pancreatitis-associated PPH, while a higher platelet count conferred a protective effect might confer protection. These findings provide preliminary insights for risk stratification. These findings enable risk stratification by highlighting modifiable factors and underscore the need for vigilant monitoring in high-risk patients, but require validation in prospective studies.