The Significance Of White Blood Cells Count,C-reactive Protein And Procalcitonin In Predicting Anastomotic Leak After Radical Gastrectomy For Gastric Cancer

Objective:Anastomotic leak(AL)is a noteworthy postoperative complications after radical gastrectomy for gastric cancer,and it’s difficult to predict it early.The purpose of this study is to explore the significance of white blood cells count(WBC),C-reactive protein(CRP)and procalcitonin(PCT)in predicting AL,and to provide a nomogram prediction model for predicting AL.So we can take measures to intervene in the occurrence of AL as soon as possible.Methods:This study is a retrospective case-control study.A total of 136 patients who were treated gastric cancer by the same surgery team in the General Surgery Center of the Affiliated Hospital of Qingdao University were included from September 2016 to September 2019.Patients were divided into non-anastomotic leak group(NAL group),including 119 cases,and anastomotic leak group(AL group),including 17 cases.Their basic clinical data,surgical data and pathological results were analyzed.TheX~2 test was used to do univariate analysis,and the Mann-Whitney U nonparametric test was used to compare the continuous variables of non-normal distribution between the two groups.The receiver operating characteristic(ROC)curve is drawn to obtain the sensitivity,specificity,the area under the receiver operating characteristic curve(AUC)and 95%confidence interval.Use Med Calc statistical software to compare ROC curves and get a better predictive index.The R software is used to construct a Nomogram prediction model,and the samples are included to calculate the total score of the Nomogram.The total score was subjected to ROC curve analysis and calibration line verification.Results:The difference between NAL group and AL group in the patient’s age,gender,BMI,preoperative hemoglobin level,history of hypertension,history of diabetes,surgical methods,resection range,reconstruction method,tumor diameter,p N staging,p T staging,Lauren classification and pathological diagnosis were not statistically significant,while the preoperative serum albumin levels,smoker,and p TNM staging were statistically significant.Comparing the data of POD1,3,5 WBC,CRP and PCT between the NAL group and the AL group,the difference between the groups of POD3,5WBC,POD1,3,5CRP and POD3,5PCT were statistically significant.The ROC curves were drawn,and the comparison of ROC curves show that the predictive effects of POD5,WBC,CRP and PCT on AL were significantly better than those of other time test indicators.The cut-off values are 11.56*10^9/L,113.00 mg/L and 0.841 ng/ML.The corresponding sensitivity and specificity are 0.941 and 0.899,0.941 and 0.933,and 0.706and 0.966.The corresponding AUC are 0.949,0.973 and 0867,and the 95%confidence intervals of AUC are(0.913～0.985),(0.941～1.000)and(0.774～0.960).The cut-off values of POD5 WBC,CRP and PCT were used as the classification threshold,and the classification results were included in the multi-factor logistic regression for variable screening.The screening results show that the three indicators were independent influencing factors of AL.Then construct the Nomogram predicting model,and the total score obtained from the Nomogram was subjected to ROC curve analysis and calibration line verification.When the results of WBC,CRP and PCT in the Nomogram are greater than or equal to their respective cut-off values,the scores are 100,92.075 and 74.167.From the ROC curve,the cut-off value of the total score is 133.121,the corresponding sensitivity is 0.941,the specificity is 0.975,AUC is 0.988 and the 95%confidence interval is(0.974-1.000),and the predictive accuracy is better than the single index test.The P-value of this model is less than 0.001,showing a good distinction degree.The inspection level of the calibration line isα=0.05,and the absolute error between the prediction and the actual occurrence of AL is 0.032.In practical applications,if the total score of Nomogram is greater than or equal to 133.121,it is considered that the probability of AL is high.Conclusion:POD5 WBC,CRP and PCT have the greatest predictive value for AL and their cut-off values are 11.56*10^9/L,113.00mg/L and 0.841ng/ML.The cut-off value of the total score of Nomogram prediction model from the ROC curve is 133.121.Using the established Nomogram prediction model can predict the occurrence of AL with a better degree of discrimination.The predictive accuracy of this model is better than the single index test.This Nomogram can be helped to decide discharge patients early or delaying.