Predicting Model Of Myelosuppression In Ovarian Cancer Patients Receiving Paclitaxel Plus Platinum Chemotherapy

Objectives:Ovarian cancer is the most common malignant tumors in women.Paclitaxel plus platinum is standard chemotherapy of ovarian cancer,but myelosuppression is the most popular toxicity.In order to guide the clinical practice,we tried to train a Logistic regression model to predict myelosuppression in ovarian cancer patients receiving paclitaxel plus platinum chemotherapy.Methods:Retrospective cohort study was conducted.A total of 803 patients who receiving paclitaxel plus platinum chemotherapy were recorded in the Third Affiliated Hospital of Kunming Medical University from May 2013 to May 2018.According to the criteria of WHO,two groups were divided:case group who suffered from ?-?degree of myelosuppression and control group who got 0-? degree myelosuppression after chemotherapy.Myelosuppression was defined by the lowest indicators of WBC,neutrophils,PLT,or HB.After univariate analysis,risk factors which P?0.25 were analyzed by multivariate Logistic regression to train the predicting model.The AUC was used to value the model,and bootstrap was used to validate the model.Results:predictive model:Myelosuppression probability(P)=ex/(1+ex)?Myelosuppression:x=4.885-0.539×(prealbumin)-0.712×(HB before chemotherapy)-0.775×(G-CSF)-0.966x(recurrence)+0.608x(previously degree of myelosuppression)0.745 ×(platinum RDI)-0.739×(cycles of chemotherapy)AUC=0.788(95%CI 0.755-0.820).If the threshold is set to 30.0%,the sensitivity is 79%and the specificity is 61.3%.??Leukopenia:x=2.636+0.591×(BSA)-0.512x(HB before chemotherapy)-0.977×(G-CSF)-0.889×(recurrence)+0.629×(previously degree of myelosuppression)-0.677×(cycles of chemotherapy)-0.984×(prealbumin)AUC=0.803(95%CI 0.769-0.837).If the threshold is set to 30.0%,the sensitivity is 70.8%and the specificity is 74.6%.?Neutropenia:x=2.368-0.89×(G-CSF)-0.993×(recurrence)+0.623×(previously degree of myelosuppression)-0.694×(cycles of chemotherapy)-0.693×(cholesterol)AUC=0.771(95%CI:0.734-0.808).If the threshold is set to 27.8%,the sensitivity is 75.1%and the specificity is 64.8%.?Thrombocytopenia:x=-2.130-0.733×(pre-chemotherapy PLT count)+0.359×(previously degree of myelosuppression)AUC=0.703(95%CI 0.624-0.782).If the threshold is set to 5.4%,the sensitivity is 68.2%and the specificity is 68.2%.?Hemoglobin suppression:x=2.625+1.010×(age)-1.626×(pre-chemotherapy RBC count)-0.941×(pre-chemotherapy PLT count)-1.312×(pre-chemotherapy HB)-0.741×(cycles of chemotherapy)AUC=0.864(95%Cl 0.810-0.918).If the threshold is set to 15.6%,the sensitivity is 67.3%and the specificity is 88.3%.Conclusions:Our Logistic regression bone marrow suppression predicting model are accuracy.The bone marrow suppression prediction model is recommended,and so are the leukopenia predietion model and the neutropenia Prediction model,but they should be verified by prospective study.