Clinical Characteristics And Survival Model Of Adult Acute Myeloid Leukemia

Objectives:1.Identify the epidemiological statistical indicators of Acute Myeloid Leukemia(AML)such as incidence and survival rate based on Surveillance,Epidemiology,and End Results(SEER)database.2.Analyze AML cases in SEER database by Kaplan-Meier survival analysis,Cox proportional risk model,competitive risk model,nomogram model and other methods,explore the prognostic factors that affect overall survival(OS)and cancer-specific survival(CSS)of AML patients,establish statistical models and verify them.3.Through big data analysis based on SEER,we can provide reference for diagnosis,treatment,prevention and research of AML,understanding the clinical features of AML,eventually assist clinical decision-making.Methods:1.The patient data of 18 registration centers in the United States were retrieved from SEER database according to the acceptance criteria,and the data were cleaned,converted and encoded to form the included population data set of this study.2.The age-adjusted morbidity,survival rate and other epidemiological statistical indicators were calculated and derived by SEER*Stat software.3.After setting the random number seed,all data sets were randomly allocated to training cohort(n=41477)and validation cohort(n=17776)at a ratio of 7:3,which were used for the construction and verification of Cox proportional risk and nomogram prediction model.4.Kaplan-Meier survival analysis and Kaplan-Meier survival curve were made on the OS of AML patients with all data sets.Log-rank test was used to compare each curve group to explore the influence of different factors on OS.5.Using the training cohort,single-factor and multi-factor Cox proportional risk analysis of the OS of AML patients was carried out and a nomogram model was constructed.The validation cohort was used to evaluate and verify the model.To evaluate the model by calculating the Consistency Index,Net Reclassification Improvement,Integrated Discrimination Improvement,at the same time,the working curve of the subjects was drawn,followed by the calculation of area under the curve.Finally,a calibration curve was drawn to evaluate the performance of the nomogram model.Clinical decision curve analysis was used to evaluate the clinical validity of the model.6.Fine-Gray competitive risk analysis was carried out and cumulative risk curve was drawn with all data sets,and the proportional risk model of CSS of AML patients was established,to identify the survival analysis indicators of patients who died of AML and other causes.Results:1.A total of 64,436 AML cases were initially retrieved from the SEER database,and 59,253 AML cases were selected and included for further analysis according to the inclusion and exclusion criteria,which were randomly allocated to training cohort(n = 41,477)and validation cohort(n= 17,776).2.From 2000 to 2018,the age adjusted incidence rates(AAIR)of AML patients increased from 3.9/100,000(95% CI 3.8/100,000-4.0/100,000)in 2000 to 4.2/100,000(95% CI 4.0/100,000)in 2018.The 1-year,3-year and 5-year OS of AML patients were 46.81%(95%CI 46.27%-47.36 %),30.94%(95%CI 30.42%-31.47%),27.65%(95%CI 27.12%-28.17%.3.Kaplan-Meier survival analysis showed that the median survival time of all AML patients in this cohort was 7 months(95%CI 7-7 months).The patients were grouped according to age,sex,race,year of diagnosis,ICD-O-3 histological type,income level and place of residence.The results of Log-Rank test showed significant difference.4.Multivariate Cox proportional hazard model showed that age,sex,year of diagnosis,median family income,histological classification of ICD-O-3 and residence were independent prognostic factors of OS in patients with AML.There was no statistical difference in OS among different races.5.Multivariate Fine-Gray competitive risk model showed that age,sex,year of diagnosis,median family income,histological classification of ICD-O-3 and residence were independent prognostic factors of CSS in patients with AML.There was no statistical difference in OS among different races.Conclusion:1.In this study,we established a multi-factor Cox proportional hazard model and a line chart model for predicting OS in patients with AML.It has been proved that the line chart model can better predict the prognosis of patients with AML(OS).This model can be used to evaluate the prognosis of patients and assist clinical diagnosis and treatment.2.In this study,a multi-factor Fine-Gray competitive risk model was established to predict CSS in patients with AML.It has been proved that Fine-Gray competitive risk can better predict the prognosis of AML patients(CSS).This model can be used to evaluate the prognosis of patients and assist clinical diagnosis and treatment.3.The results of multivariate Cox proportional hazard model and multivariate Fine-Gray competitive risk model showed that age,sex,year of diagnosis,median household income,histological classification of ICDO-3 and residence were independent prognostic factors for the prognosis of AML patients(OS and CSS).