Prognostic Factors Of Overall Survival In Cervical Cancer And Establishment Of Prediction Model

Background Cervical cancer is the most common gynecologic cancer worldwide.Radical hysterectomy and pelvic lymphadenectomy is one of the main therapy for operable cervical cancer patients and the survival is closely related to tumor stage.However,the predictive power of the FIGO staging system is not quite precise.So the key procedure for individualized prediction of survival is to investigate independent prognostic factors.A nomogram is a predictive statistical model that generates the continuous probability of a clinical event such as death or recurrence.It is often used for the estimation of prognosis of malignant tumor patients because of its accuracy and reliability and is useful to make treatment decisions and follow up planning.Objective To investigate the factors affecting the overall survival(OS)of patients with cervical cancer and develop a nomogram for predicting OS of patients with early stage cervical cancer.Methods In this retrospective study,the clinical,pathological,and hematogical data of 308 pathological diagnosed cervical cancer patients from January 2010 to December 2012 in Qilu Hospital of Shandong University were investigated.Clinical factors include FIGO stage,age,smoking history,medical comorbidities and surgery complications.Hematological indexes consist of preoperative white blood cell(WFBC),hemoglobin,platelet,monocyte,neutrophil,lymphocyte,lymphocyte/monocyte ratio(LMR),neutrophil/lymphocyte ratio(NLR),platelet/lymphocyte ratio(PLR)and serum albumin.Pathological factors include histological type,differentiation,depth of stromal invasion(DSI),tumor size,parametrial invasion,pelvic lymph node(PLN)status,lymph vascular space invasion(LVSI)and surgical margin status.Cox proportional hazard(CPIH)model were then used for survival analyses.Bootstrap cross-validation was used for the internal validation of CPH model.Besides,we identified 13,871 cases of stage ?A-?B operable cervical cancer in the SEER database,which was used as a testing set for the predictive model.In addition,calibration curves of the predictive model were drawn with the predicted 3-year survival probabilities against the actual survival status.Results1.clinico-pathologic and hematological characteristics of cervical cancer patients1.1 clinico-pathologic characteristics and follow-up data of all cervical cancer patients:A total of 29(9.4%)were in FIGO stage IA,122(39.6%)in stage IB1,50(16.2%)in stage IB2,75(24.4%)in stage IIA and the remaining 11(3.6%)were in stage IIB.A total of 259(84.1%)were squamous carcinoma,34(11.0%)were adenocarcinoma and 14(4.6%)were other histological type such as adenosquamous carcinoma,chorionic tubular adenocarcinoma,melanoma,etc.79 patients(25.6%)had positive PLN while 216 patients(70.1%)had negative PLN.The median follow-up time was 1926 days.Up to the last follow-up,60 of 308 patients died.1.2 The distribution of clinico-pathologic characteristics between death group and survival group of cervical cancer patients:The mean age of death group was 48.2111.2,and the mean age of survival group was 45.019.6(P=0.028).The distribution of FIGO stage,albumin,PLN status,tumor size,DSI and PI between these two groups remained statistically significant(P<0.05).And the distribution of smoking history,medical comorbidities,surgery complications,histological type,differentiation,and margin status had no statistical significance(P>0.05).1.3 The distribution of hematological indexes between death group and survival group of cervical cancer patients:Only serum albumin between these two groups remained statistically significant(P<0.001)and the others had no statistical significance.2.Prognostic factors for survival in cervical cancer patients2.1 Independent prognostic factors for survival in cervical cancer patients:By univariable analysis and subsequent multivariable analysis,we identified FIGO stage(HR=1.57,95%CI 1.046-2.350,p=0.030),monocyte(HR=2.67,95%CI 1.454-4.904,p=0.002),albumin(HR=0.50,95%CI 0.281-0.904,p=0.022),PLN status(HR=1.35,95%CI 1.202-1.526,p<0.001)and DSI(HR=1.70,95%CI 1.072-2.686,p=0.024)as independent prognostic factors for survival in cervical cancer patients.2.2 The validation of independent prognostic factors:ROC for death rate had been drawn using these independent risk factors and area under curves(AUC)was found to be 0.777.Also,Kaplan-Meier survival curves as well as Log-rank test(all of P<0.05)also lent support for the prognostic values of these predictors.3.Nomogram for prediction of 3-year and 5-year survival of cervical cancer patientsA nomogram model based on independent prognostic factors was built to predict 3-and 5-year OS for patients with ?A-?B cervical cancer.The c-index of this model for prediction of patient survival was found to be 0.745.Yet,only the clinical and pathological parameters of this nomogram had been external validated since the SEER database had not included the hematological parameters.C-index of nomogram in this external validation could also reach 0.718.Also,calibration curve which had been drawn for the training set showed satisfactory agreement between predicted and actual values.Conclusions1.We identified FIGO stage,preoperative monocyte,serum albumin,PLN status and DSI as a independent prognostic factors of OS for patients with cervical cancer.2.We developed a SEER database validated comprehensive nomogram that can provide prediction of 3-and 5-year OS for patients with cervical cancer individually.3.This nomogram could be useful for health care providers for counseling patients regarding treatment decisions and follow up planning.