Explainable machine learning using perioperative serial laboratory results to predict postoperative mortality in patients with peritonitis-induced sepsis

Purpose@#Sepsis is one of the most common causes of death after surgery. Several conventional scoring systems have been developed to predict the outcome of sepsis; however, their predictive power is insufficient. The present study applies explainable machine-learning algorithms to improve the accuracy of predicting postoperative mortality in patients with sepsis caused by peritonitis. @*Methods@#We performed a retrospective analysis of data from demographic, clinical, and laboratory analyses, including the delta neutrophil index (DNI), WBC and neutrophil counts, and CRP level. Laboratory data were measured before surgery, 12–36 hours after surgery, and 60–84 hours after surgery. The primary study output was the probability of mortality.The areas under the receiver operating characteristic curves (AUCs) of several machine-learning algorithms using the Sequential Organ Failure Assessment (SOFA) and Simplified Acute Physiology Score (SAPS) 3 models were compared.‘SHapley Additive exPlanations’ values were used to indicate the direction of the relationship between a variable and mortality. @*Results@#The CatBoost model yielded the highest AUC (0.933) for mortality compared to SAPS3 and SOFA (0.860 and 0.867, respectively). Increased DNI on day 3, septic shock, use of norepinephrine therapy, and increased international normalized ratio on day 3 had the greatest impact on the model’s prediction of mortality. @*Conclusion@#Machine-learning algorithms increase the accuracy of predicting postoperative mortality in patients with sepsis caused by peritonitis.

Prognostic significance of computed tomography defined ascites in advanced gastric cancer / 대한외과학회지

PURPOSE: The aim of this study was to investigate the clinicopathologic features and prognosis in patients with computed tomography (CT) findings of ascites, with a focus on the correlation with peritoneal carcinomatosis. METHODS: This study included a total of 157 patients who underwent surgery for advanced gastric cancer from 2003 to 2008 at the Department of Surgery, Hallym University Kangdong Sacred Heart Hospital, Seoul, Korea, which were analyzed retrospectively. RESULTS: Fourteen patients (8.9%) presented ascites on their CT scan. Among them, 10 patients had peritoneal carcinomatosis, and showed significant difference with CT ascites positive group in the incidence of peritoneal carcinomatosis. The presence of CT ascites was significantly correlated with pathologic T stage, tumor size, histologic type, CT T and N stages, CT peritoneal nodularity and curability of surgery, statistically. The prognosis of CT ascites positive group was much poorer in the total advanced gastric cancer patients (P < 0.001), as well as in patients with pathologic T4 (P = 0.002). Also in patients without peritoneal carcinomatosis, CT ascites positive subgroup tended to have a worse prognosis than CT ascites negative subgroup (P = 0.086). Tumor size, CT T and N stages and the presence of CT peritoneal nodularity and ascites influenced the prognosis significantly; among which, if a tumor size larger than 5 cm, CT T4 stage and the presence of CT ascites were identified as independent prognostic factors. CONCLUSION: The presence of ascites was closely associated with peritoneal metastasis, and was the most significant independent prognostic factor in advanced gastric cancer in the present study.

Loss of Heterozygosity of E-Cadherin Gene and Protein Expression in Invasive Ductal Carcinomas / 한국유방암학회지

PURPOSE: The E-cadherin gene, located on chromosome 16q22, may play principal roles in cell adhesion with the loss of E-cadherin expression leading to a propensity for a great number of malignant properties. The loss of heterozygosity (LOH) on 16q22 has rarely been studied in invasive ductal carcinomas. Our objectives were to evaluate the LOH of E-cadherin and the protein expression in invasive ductal carcinomas and their correlation with various clinicopathological factors. METHODS: The LOH analysis was performed using polymerase chain reactions with three polymorphic microsatellite markers (D16S419, D16S3106 and D16S498) in 50 surgically resected tumors and their non-tumorous counterparts. The E-cadherin protein expression was studied using immunohistochemistry. RESULTS: The LOH and loss of protein expression were detected in 54% and 46% of the tumors, respectively. There was no LOH or protein loss detected in the non-tumor lesions. The LOH results were well correlated with the tumor size and lymph node metastasis. The protein loss results were well correlated with tumor histological grade. No correlation was found between LOH and protein loss. CONCLUSION: These results suggest that the LOH of E-cadherin may be associated with tumor metastasis and tumor progression and E-cadherin protein loss may be related with the dedifferentiation in some portions of invasive ductal carcinomas. We propose the LOH of E-cadherin and protein loss may contribute to tumor progression by independent mechanism.