Introduction The prediction of occurrence and form of a postoperative pancreatic fistula (POPF) could decrease the risk after pancreatic resections. We developed a score based on histomorphological criteria to predict POPF earlier on. In this study we test the scoring system to differentiate between patients developing grade B and C postoperative pancreatic fistula. Methods We analysed 92 patients who underwent pancreatic head resection and developed a grade B or C POPF postoperatively. There were no exclusion criteria. Pancreatic resection margins were analysed histologically and pancreatic duct size, fibrosis, fat content and signs of inflammation were transformed into a scoring system. Results Patients with a grade B fistula (n=48) were compared to patients with grade C fistula (n=44). Grade C group was characterized by higher mortality (0 vs 22, p<0.001), postpancreatectomy haemorrhage (3 vs 24, p<0.001) longer stays in ICU (2.6 vs 21.1, p<0.001) and in hospital (29.8 vs 44.6, p<0.001). The scoring system was unable to predict grade C fistula. We calculated receiver operator characteristics for all assessed characteristics, which revealed an area-under-the-curve of 0.561. The strongest predictors of POPF grade C were pancreatic fat and soft pancreatic tissue. A combination of only these two items led to a statistically significant difference (p<0.001) and an AUC of 0.771. Conclusion Establishing a score to differentiate between grade B and grade C POPF is a difficult task. Pancreatic fat seems to be the most relevant histomorphological feature to be used in any scoring system and it should be combined with other clinical features to enhance its precision and validity.
Read full abstract