Characteristics of gastrointestinal hemorrhage associated with pancreatic cancer: A retrospective review of 246 cases.

While gastrointestinal (GI) hemorrhage is common in the general population, few studies have evaluated large numbers of GI hemorrhage patients with pancreatic cancer. The clinical features and potential risk factors of GI hemorrhage with pancreatic cancer was investigated in the present study and the effect of GI hemorrhage on survival rate was examined. Patients enrolled in the present study had pathologically proven pancreatic cancer, and received treatment between August 2006 and 2012. Their medical records were retrospectively reviewed. The data for the present study were obtained from a review of 246 patients with pancreatic cancer (average age, 63.4±10.92 years; 190 male cases, 56 female cases). In addition, 73 cases had stage I-II, 173 had stage III-IV, and only 67 cases (27.2%) were candidates for curative pancreatectomy. Among them, 32 cases (13.0%) were clinically diagnosed with GI hemorrhage. A total of 24 cases were male patients and the other 8 cases were female, the cases of hemorrhage history and alcoholism were 2 and 29 cases, respectively. The major initial clinical symptoms of GI hemorrhage included 18 patients with melena or blood stool (56.25%), 9 with haematemesis (28.13%), 3 with abdominal distention (9.37%) and 2 with stomach ache (6.25%). The independent risk factor for GI hemorrhage was tumor initial stage of IV. A continuous increase in carbohydrate antigen 19-9 (CA19-9) may be a warning of GI hemorrhage, particularly when it is >1,000 U/ml. The most frequent method of hemostasis was combination therapy (n=12, 37.5%). Only 3 cases (9.3%) of these 32 GI hemorrhage patients were blood stanched and only 10 patients (31.2%) received gastroscopy. The time from GI hemorrhage to fatality is extremely short (median 30 days, range from 1 h to 65 days), and the median overall survival time of the patients with GI hemorrhage was 9.0 months (range, 2.0-16.0 months) and was significantly shorter than that of patients without GI hemorrhage [14.5 months (range, 0.5-48.0 months)]. In conclusion, although GI hemorrhage was not common in patients with pancreatic cancer, it is critical. GI hemorrhage was controlled with endoscopic hemostasis. Clinicians should fully assess the risk factors of GI hemorrhage (such as alcohol, smoking, past hemorrhage history, initial stage, tumor location and CA19-9 level at diagnosis of pancreatic cancer) when the pancreatic cancer patients were on admission, particularly for patients of the late stage, preventive measures should be investigated to reduce suffering.