Geriatric Nutritional Risk Index Less Than 92 Is a Predictor for Late Postpancreatectomy Hemorrhage Following Pancreatoduodenectomy: A Retrospective Cohort Study

SIMPLE SUMMARY: The definite risk factor of postpancreatectomy hemorrhage (PPH) is still unknown in spite of a lethal complication of pancreatoduodenectomy (PD). In this study, we evaluated whether GNRI is a reliable marker for PPH following PD. The present study retrospectively evaluated 121 patients treated with PD at Ageo Central General Hospital in Japan. Ten patients had developed PPH. Among them, the patients were divided into bleeding group (n = 10) and non-bleeding group (n = 111). The bleeding group had significantly low geriatric nutritional risk index (GNRI) values compared to those in the non-bleeding group (p = 0.001). The cut-off value of GNRI was determined by 92 accounting for a sensitivity 80.0%, specificity 82.9% using receiver operating characteristic curve analysis. A GNRI of <92 was statistically identified as an independently risk factor of PPH risk following PD. ABSTRACT: Postpancreatectomy hemorrhage (PPH) is the most lethal complication of pancreatoduodenectomy (PD). The main risk factor for PPH is the development of a postoperative pancreatic fistula (POPF). Recent evidence shows that the geriatric nutritional risk index (GNRI) may be predictive indicator for POPF. In this study, we aimed to evaluate whether GNRI is a reliable predictive marker for PPH following PD. The present study retrospectively evaluated 121 patients treated with PD at Ageo Central General Hospital in Japan between January 2015 and March 2020. We investigated the potential of age, gender, body mass index, serum albumin, American Society of Anesthesiologists classification (ASA), diabetes mellitus and smoking status, time taken for the operation, estimated blood loss, and postoperative complications (POPF, bile leak, and surgical site infections) to predict the risk of PPH following PD using univariate and multivariate analyses. Ten patients had developed PPH with an incidence of 8.3%. Among them, the patients were divided into bleeding group (n = 10) and non-bleeding group (n = 111). The bleeding group had significantly lower GNRI values than those in the non-bleeding group (p = 0.001). We determined that the cut-off value of GNRI was 92 accounting for a sensitivity 80.0%, specificity 82.9%, and likelihood ratio of 4.6 using receiver operating characteristic curve analysis. A GNRI of <92 was statistically associated with PPH in both univariate (p < 0.001) and multivariate analysis (p = 0.01). Therefore, we could identify that a GNRI < 92 was an independently potential predictor of PPH risk following PD. We should alert surgeons if patients have low level GNRI before PD.