Construction Of Prognostic Classifier And Identification Of Immune Subtypes For Cervical Cancer Based On Machine Learning

Cervical cancer ranks as the fourth most common female malignancy and the fourth leading cause of cancer mortality in women worldwide,while low-and middle-income countries account for 90% of the deaths.Although the application of screening and human papillomavirus(HPV)vaccination provide effective prevention for cervical cancer,the imbalance of regional development leads to cervical cancer which will still be a serious health problem in the coming decades.For patients with stage I-III,15-61% of women will still experience metastatic disease within the first 2 years after completing treatment.Once the disease progresses,second-line and later treatment options are limited,and patients often have a poor prognosis.In recent years,the promising responses of immunotherapy based on inhibitors targeting cytotoxic T lymphocyte antigen 4(CTLA-4),programmed death receptor 1(PD-1)or its ligand(PD-L1)have brought revolutionary changes to the treatment of a variety of cancers.It has now become an important issue for cervical cancer.To date,the objective response rate(ORR)of immune checkpoint inhibitors(ICIs)for cervical cancer varies from 4% to 26.3%,with over 80% of responding patients obtaining long-lasting response(>6 months).Although several biomarkers have been applied to predict the clinical outcome of patients with cervical cancer,their sensitivity and/or specificity remain unsatisfactory.Therefore,it is extremely urgent to identify more valuable biomarkers for diagnosing and monitoring recurrence and evaluating prognosis.On the other hand,the major limitation of immunotherapy comes from the fact that it benefits only a minority of patients.In consideration of the economic burden and toxicity of ICIs,it is important to identify suitable patients who benefit from ICIs and combination therapy.In this dissertation,based on next generation sequencing(NGS)technology,a variety of bioinformatics technologies and machine learning algorithms were used to establish a classifier that can distinguish the risk of disease progression in cervical squamous cell carcinoma(CSCC).Based on this,a Normogram was constructed to predict the prognosis of patients with CSCC.Second,a classifier has been developed that recognizes the immune subtypes of CSCC to help screen candidate patients who may respond to ICIs.The main research contents are as follows:(1)After utilizing RNA sequencing(RNA-seq)data from 36 formalin-fixed and paraffin-embedded(FFPE)samples,the most significant modules were highlighted by the weighted gene co-expression network analysis(WGCNA).A candidate genes-based prognostic classifier was constructed by the least absolute shrinkage and selection operator(LASSO)and then validated in an independent validation set.Finally,based on the multivariate analysis,a nomogram including the FIGO stage,therapy outcome and risk score level was built to predict progression-free survival(PFS)probability.A mRNA-based signature was developed to classify patients into high-and low-risk groups with significantly different PFS and overall survival(OS)rate(training set: p <0.001 for PFS,p = 0.016 for OS;validation set: p = 0.002 for PFS,p = 0.028 for OS).The prognostic classifier was an independent and powerful prognostic biomarker for PFS in both cohorts(training set: HR = 0.13,95% CI: 0.05-0.33,p <0.001;validation set: HR =0.02,95% CI: 0.01-0.04,p <0.001).A nomogram that integrated the independent prognostic factors was constructed for clinical application.The calibration curve showed that the nomogram was able to predict 1-,3-,and 5-year PFS accurately,and it performed well in the external validation cohorts(concordance index: 0.828 and 0.864,respectively).(2)A real-world CSCC cohort of 36 CSCC samples were analyzed.We used a nonnegative matrix factorization(NMF)algorithm to separate different expression patterns of immune-related genes(IRGs).The immune characteristics,potential immune biomarkers,and somatic mutations were compared.Two independent data sets containing555 CSCC samples were used for validation.Two subtypes with different immunophenotypes were identified.Patients in sub1 showed favorable progression-free survival(PFS)and overall survival(OS)in the training and validation cohorts.The sub1 was remarkably related to increased immune cell abundance,more enriched immune activation pathways,and higher somatic mutation burden.Also,the sub1 group was more sensitive to ICIs,while patients in the sub2 group were more likely to fail to respond to ICIs but exhibited GPCR pathway activity.Finally,an 83-gene classifier was constructed for CSCC classification.This part of dissertation establishes a new classification to further understand the immunological diversity of CSCC,to assist in the selection of candidates for immunotherapy.In summary,starting from the two aspects of prognosis prediction and molecular subtype classification,which are closely related to clinical practice,this paper establishes a prognostic prediction system based on mRNAs and can be used as an independent prognostic predictor.Then,the Normogram of prognostic prediction is constructed which is expected to guide clinical practice.In addition,a new classification method of CSCC subtypes was established,which is helpful to further understand the immune diversity of CSCC,and has certain guiding significance for our basic research and clinical practice on CSCC in the future.