The Development And Evaluation Of Treatment Decisions And Individualized Prognosis Predictive Nomograms For Cervical Cancer

Cervical cancer is the most common malignant tumor of female reproductive tract.According to statistics in 2020,there are about 604,000 new cases of cervical cancer and 342,000 deaths worldwide every year[1].New hair and cervical cancer deaths in China respectively 110000 cases,59000 cases,respectively global cervical cancer morbidity and mortality by 18.2%and 17.3%of the total,seriously endanger the physical and mental health of women.Treatment for cervical cancer includes surgery,radiotherapy and chemotherapy.But common treatment does not take the place of individualized treatment,how to choose the most suitable for treatment of patients is still inconclusive.Cervical cancer radical surgery way including laparotomy and laparoscopic surgery,but in 2018 the New England journal of medicine published[2]a prospectie,randomized,controlled trial,hints of laparoscopic surgery for cervical cancer significantly worse prognosis than laparotomy.Based on the above research,NCCN[3]guidelines were revised,and recommended procedure as standard operation way of radical hysterectomy.However,the results[2]are not enough to conclude the safety of minimally invasive surgery for cervical cancer with tumor diameter less than 2cm.Therefore,is it possible to evaluate the oncology prognosis of patients undergoing laparoscopy and laparotomy respectively before surgery by comparing individual clinicopathological variables in patients with ⅠB1 cervical cancer,so as to serve as an auxiliary decision-making means for surgical approach selection?In addition,the corresponding treatment in patients with cervical cancer after end of oncology directly affect the follow-up of the consultation.After treatment,patients are expected to receive individualized follow-up recommendations as well as accurate assessment of their own oncology prognosis.Therefore,the formulation of cervical cancer treatment plan needs to take into account the specific situation of patients,in order to get the best treatment plan.The diagnosis and treatment of cervical cancer should not only follow the basic diagnosis and treatment standards,but also emphasize the individualization of treatment.So,can we find reliable methods to provide reference for doctors to make clinical decisions?Clinical prediction model for assessment of risk and benefit of quantitative tools,but for doctors,patients and health administrative personnel decision-making information more intuitive and rational.It is a common presentation method[4]to express complex mathematical formulas in simple and understandable graphs[5].Rosette has been used as a reliable tool to quantify risk by combining important clinical factors to quantify risk as accurately as possible,so as to accurately predict clinical outcomes[6-8].This research is divided into four parts:the frrst part,surgery and radiation and chemotherapy treatment prediction model is set up,respectively,by the same patients after underwent surgery and radiation and chemotherapy treatment to predict the outcome of oncology,so as to assist doctors and patients make individualized treatment strategies;In the second part,the prediction models of ⅠB1 stage open surgery and laparoscopic surgery were established respectively to predict the outcomes of the same patients receiving open surgery and laparoscopic surgery,so as to provide accurate and scientific reference for the selection of surgical approaches for patients with ⅠB1 stage cervical cancer.The third part,according to the postoperative pathologic suggest the existence of lymph node metastasis,respectively established stage ⅠB1-ⅡA2 and IIICp cervical cancer prognosis prediction model,the prognosis of cervical cancer patients with postoperative oncology for more accurate prediction,so as to assist clinical for the prognosis of patients with more accurate predictions,as well as individual follow-up plan.The first part:the multi-center cervical cancer data set is the selection of the best treatment(surgery/radiotherapy and chemotherapy)for cervical cancer patients and the construction and validation of the individual prognosis prediction modelObjective:In order to provide an effective reference for clinical decision making,the individual prognosis of cervical cancer patients undergoing surgery or radical chemoradiotherapy was predicted.Methods:Based on the Large database of Clinical diagnosis and treatment of cervical cancer in China,patients with ⅠB1-ⅡA2 and ⅢCr stage cervical cancer who received surgical treatment and patients with ⅠB1-ⅣA stage cervical cancer who received radical radiotherapy and chemotherapy were included in 47 units from 2004 to 2018.Overall survival rate in 5 years(overall survival,OS)and disease-free survival(diseases-free survival,DFS)to predict the end,the age,histological type,FIGO2018 stage and tumor diameter as prediction variables to construct the Cox proportional hazards regression model,Thus,the prognostic model of surgical treatment and the prognostic model of radical radiotherapy and chemotherapy were established.The area under receiver operating characteristic curve(AUC),calibration curve and risk stratification system were used to conduct internal and external verification of the performance of the model in the training set and verification set respectively.Results:(1)Prognostic prediction model of radical radiotherapy and chemotherapy:5-year OS prediction model of radical radiotherapy and chemotherapy(hereinafter referred to as"OS model")and 5-year DFS prediction model of radical radiotherapy and chemotherapy(hereinafter referred to as "DFS model")were established in the included patients according to the different predicted endpoint.OS model showed better prediction accuracy than FIGO stage in the training set(AUC:OS model vs.FIGO stage=59.3%vs.56.4%).Verify the result of concentration have been similarly(AUC:concurrent chemoradiation OS model vs.FIGO staging 62.3%vs.54.5%);Similarly,the DFS model showed better prediction accuracy than FIGO stage in the training set(AUC:DFS model vs.FIGO stage=58.6%vs.56.3%).Verify the result of concentration have been similarly(AUC:chemoradiotherapy DFS model vs.FIGO staging=56.2%vs.53.6%).The calibration curve also shows that the model has good calibration degree.Based on the risk stratification of nomogram system can be accurately distinguish the prognosis of patients with different oncology.(2)Prognosis prediction model of surgical treatment:5-year OS prediction model of surgical treatment(hereinafter referred to as the "Surgical OS model")and 5-year DFS prediction model of surgical treatment(hereinafter referred to as the "surgical DFS model")were established for the included patients according to different predicted endpoints.Surgery OS model centered in training shows more FIGO staging a better prediction accuracy(AUC:OS operation model vs.FIGO staging=70.4%vs.57.9%),and verify the result of concentration have been similarly(AUC:OS operation model vs.FIGO staging 68.7%vs.64.1%);Similarly,the surgical DFS model showed better predictive accuracy than FIGO staging in the training set(AUC:Surgical DFS model vs.FIGO staging=71.6%vs.64.4%).Similar results were obtained in the validation set(AUC:surgical DFS model vs.FIGO staging=66.1%vs.62.8%).The calibration curve also shows that the model has good calibration degree.The risk stratification system based on the lirogram can accurately distinguish patients with different oncology outcomes.Conclusion:In this study,the prediction model of cervical cancer after surgical treatment was established based on IB1-ⅡA2 and ⅢCr stage cervical cancer patients who received surgical treatment,and the prediction model of cervical cancer after radical radiotherapy and chemotherapy was established based on ⅠB1-ⅣA stage cervical cancer patients who received radical radiotherapy and chemotherapy.The results showed that the prediction efficiency of the model was better than that of FIGO2018.ⅠB1-ⅡA2 and ⅢCr stage cervical cancer patients can be predicted by the two prediction models after surgery and radiotherapy and chemotherapy respectively,so as to assist clinicians to select the optimal treatment plan for patients by comparing the predicted outcomes.The prediction model of radical radiotherapy and chemotherapy can be used to predict the oncology outcome of patients with stageⅡB-ⅣA cervical cancer after receiving radical radiotherapy and chemotherapy,so as to help doctors make follow-up treatment and follow-up plans for patients.The second part:the selection of the best surgical method for cervical cancer patients and the establishment and evaluation of the individual prognosis prediction modelObjective:In order to provide reference for the choice of surgical approach for patients,the oncology prognosis of stage ⅠB1 cervical cancer patients who received open or laparoscopic surgery was predicted.Methods:In order to provide reference for the choice of surgical approach for patients,the oncology prognosis of FIGO IB stage 1 cervical cancer patients who received open or laparoscopic surgery was predicted.Results:(1)Prognosis prediction model of ⅠB1 open surgery:The 5-year OS prediction model of ⅠB1 open surgery(hereinafter referred to as the "OPEN OS model")and the 5-year DFS prediction model of ⅠB1 open surgery(hereinafter referred to as the "open DFS model")were established among the included patients with predicted ⅠB1 open surgery treatment according to different predicted endpoints.In the training set and validation set,IB1 period of open surgery for 5 years OS prediction model of the area under the ROC curve were 70.1%and 75.9%respectively,ⅠB1 period of open surgery for 5 years DFS prediction model of the area under the ROC curve were 78.7%and 75.8%respectively.In addition,the risk stratification systems established based on the above rolograms could accurately distinguish the risk subgroups with different prognoses.(2)Prognosis prediction model of ⅠB1 laparoscopic surgery:The 5-year OS prediction model of ⅠB1 laparoscopic surgery treatment(hereinafter referred to as the"laparoscopic OS model")and the 5-year DFS prediction model of IB 1 laparoscopic surgery treatment(hereinafter referred to as the "laparoscopic DFS model")were established in the included IB1 patients according to different predicted endpoints.In the training set and validation set,the AUC of laparoscopic OS model was 63.7%and 64.5%,and that of laparoscopic DFS model was 64.5%and 64.0%,respectively.In addition,the risk stratification systems established based on the above rolograms could accurately distinguish the risk subgroups with different prognoses.Conclusion:In this study,ⅠB1 open surgery model and ⅠB1 laparoscopic surgery model were successfully established respectively based on ⅠB1 stage cervical cancer patients who received open surgery or laparoscopic surgery,and the verification results indicated that the model had good efficacy.Combined with the prediction model of radiotherapy and chemotherapy in the first part of this project,the prediction results of oncology outcomes of ⅠB1 stage cervical cancer patients after radical radiotherapy and chemotherapy,laparotomy and laparoscopic surgery can be obtained at the same time,so as to obtain the best treatment mode for this patient.The third part the construction and validation of the postoperative individualized prognostic prediction model for cervical cancer patients based on the multi-center cervical cancer data setObjective:Combined with postoperative pathological factors,the oncology prognosis of patients with cervical cancer after surgical treatment was more accurately predicted.Methods:Based on the Large database of Clinical diagnosis and treatment of cervical cancer in China,cases of ⅠB1-ⅡA2 and ⅢCp cervical cancer undergoing surgical treatment in 47 institutions from 2004 to 2018 were included.The 5-year OS and DFS were used as the predicted endpoint.Age,histological type,FIGO stage(2018),tumor diameter,depth of cervical interstitial invasion,LVSI(lymphatic interstitial invasion),paracytal metastasis and vaginal margin were used as the predictive variables to construct Cox proportional risk regression models,so as to establish ⅠB1-ⅡA2 stage cervical cancer prognosis prediction model and ⅢCp stage prognosis prediction model.The area under receiver operating characteristic curve(AUC),calibration curve and risk stratification system were used to conduct internal and external verification of the performance of the model in the training set and verification set respectively.Results:(1)ⅠB1-ⅡA2 prognosis prediction model:The 5-year OS prediction model forⅠB1-ⅡA2 cervical cancer(hereinafter referred to as"ⅠB1-ⅡA2 OS model")and the 5-year DFS prediction model for ⅠB1-ⅡA2 cervical cancer(hereinafter referred to as"IB1-IIA2 stage")were established in the included cases of cervical cancer according to different predicted endpoints DFS model ").ⅠB1 ⅡA2 period OS model centered in training shows more FIGO staging a better prediction accuracy(AUC:ⅠB1-phaseⅡA2 OS model vs.FIGO staging=73.2%vs.55.4%).Similar results were obtained in the validation set(AUC:IB1-IIA2 OS model vs.FIGO stage 79.3%vs.60.2%).Similarly,ⅠB1-ⅡA2 5-year DFS prediction model showed better prediction accuracy than FIGO stage in the training set(AUC:ⅠB1-ⅡA2 DFS model vs.FIGO stage=70.0%vs.55.1%).Similar results were obtained in the validation set(AUC:IB1-IIA2 STAGE DFS model vs.FIGO stage=71.7%vs.60.4%).The calibration curve also shows that the model has good calibration degree.The risk stratification system based on the lirogram can accurately distinguish patients with different oncology outcomes.(2)ⅢCp prognosis prediction model:The 5-year OS prediction model of CERVICAL cancer at ⅢCp stage(hereinafter referred to as "ⅢCp OS model")and the 5-year DFS prediction model of cervical cancer at ⅢCp stage(hereinafter referred to as "ⅢCp DFS model")were established in the included cases of cervical cancer according to different predicted endpoints.ⅢCp OS model showed better prediction accuracy than TNM stage in the training set(AUC:ⅢCp OS model vs.TNM stage=66.9%vs.54.5%).Similar results were obtained in the validation set(AUC:ⅢCp OS model vs.TNM57.5%vs.52.9%).Similarly,DFS model at ⅢCp stage showed better prediction accuracy than TNM stage in the training set(AUC:DFS model at ⅢCp stage vs.TNM stage=67.2%vs.54.8%).Similar results were obtained in the validation set(AUC:ⅢCp DFS model vs.TNM staging=65.5%vs.51.3%).The calibration curve also shows that the model has good calibration degree.The risk stratification system based on the lirogram can accurately distinguish patients with different oncology outcomes.Conclusion:In this study,prognostic models of ⅠB1-ⅡA2 and ⅢCp cervical cancer were successfully established based on the cases of ⅠB1-ⅡA2 and ⅢCp cervical cancer who underwent surgical treatment.In the prediction model of ⅠB1-ⅡA2 stage cervical cancer,the prediction efficiency of the model was better than that of FIGO2018 stage.In ⅢCp period prognosis prediction model at the same time,the model prediction is better than TNM staging.By combining the clinicopathological variables of patients,it can provide more accurate and individualized prediction of postoperative prognosis of patients with cervical cancer,and provide effective reference for clinicians to develop personalized follow-up plans and follow-up treatment for patients.