Based On Hematological Indices Of PFR To Establish A Risk Predictive Model Of Myelosuppression Induced By Neoadjuvant Chemotherapy In Locally Advanced Nasopharyngeal Carcinoma

Objective:In order to build a predictive model of myelosuppression,clinical data were collected to identify risk factors of myelosuppression after neoadjuvant chemotherapy with nasopharyngeal carcinoma.The model can help us screen out high risk groups,so that clinicians have a basis for preventive treatment.Therefore,finding a balance between efficacy and toxic side effects can ensure the effectiveness of treatment,and reduce the occurrence of dangerous complications and the risk of death.Methods:·A total of 179 patients diagnosed with locally advanced nasopharyngeal carcinoma in The First Affiliated Hospital of Gannan Medical College from 2018 to 2022 were enrolled,including 92 in the experimental group and 87 in the control group,According to the WHO bone marrow suppression classification standard,Patients with myelosuppression of grade Ⅱ or above during induction chemotherapy were selected as the experimental group,and the grade Ⅰ or no myelosuppression were divided into the control group.The general data and clinical data indicators of the experimental group and the control group were compared.After univariate regression analysis,the risk factors which had significancep<0.05)were selected to be done by logistic regression analysis,so as to obtain independent risk factors for myelosuppression.The predictive model was established according to the regression analysis result,and the ROC-AUC curve was drawn to test the discrimination of the predictive model.Hosmer-Lemeshow was used to test the consistency of fit of the model.Another 50 patients with locally advanced nasopharyngeal carcinoma were selected to verify the predictive model according to the above grouping criteria.There were 26 patients with grade Ⅱ or above myelosuppression,and 24 patients with grade Ⅰ or no myelosuppression.Then Kappa test was used to verify the prediction model.Finally,the predictive model was visualized as a graph,and each risk factor was assigned according to the results of multi-factor analysis.The risk of myelosuppression was converted into a specific probability value,which made the model more convenient for clinical application.Results:1.The experimental group was compared with the the control group after induced chemotherapy,The decline of ALC and NLR with experimental group was statistically significant(p<0.05).The increase of PLR index was statistically significant(p<0.05).2.By single factor correlation analysis,the significant factors may lead to myelosuppression are Age(p<0.05),BMI(p<0.05),low level of WBC before chemotherapy(p<0.05),lymphocytes(p<0.05),alkaline phosphatase(p<0.05),fibrinogen(p<0.05),R-glutamine transferase(p<0.05),prealbumin fibrinogen ratio(p<0.u5),prealbumin alkaline phosphatase ratio(p<0.05).3.Muitivariate logistic regression analysis of myelosuppression showed that the independent risk factors of myelosuppression after induction chemotherapy were Age 134,95%CI:1.038,1.238,p=0.005),low level of WBC before chemotherapy(OR=0.577,95%CI:0.407,0.819,p=0.002),R-glutamine transferase(OR=1.3130,95%CI:1.061,1.205,p=0.000),prealbumin fibrinogen ratio(OR=0.957,95%CI:0.926,0.990,p=0.01).4.Based on the results of multivariate analysis,the prediction model of myelosuppression was establishes:Logit(p)=In[p/(1-p)]=0.125×[age]-0.549×[low level of WBC before chemotherapy]+0.123 ×[r-GT]-0.044×[prealbumin/fibrinogen]+4.810.the ROC-AUC curve was drawn to test the discrimination of the predictive model.Hosmer-Lemeshow was used to test the consistency of the mode.The sensitivity of the model was 92.3%,the specificity was 87.7%,the Yoden index was 0.80,and the AUC(0.956,95%CI:0.926,0.986).Hesmer-Lemeshow test(χ~2=5.467,p=0.707,p>0.05).5.To verify the prediction model,the specificity of this model is 96%and the sensitivity is 93%.The positive predictive value was 96.3%and the negative predictive value was 96%.Kappa test showed a strong consistency between the model and the actual test results(u test,p<0.05).Conclusion:This prediction model was verified to be reliable.Then by establishing a nomogram model and drawing a clinical decision graph,the probability of myelosuppression after chemotherapy for locally advanced nasopharyngeal carcinoma was calculated to provide important reference evidence for clinical decision.The predictive model makes clinical treatment more reasonable,and reduces the occurrence of serious complications and the risk of death of patients.