High-resolution Of Cell Type Deconvolution To Construct Rectal Cancer Radiotherapy Response Prediction Model

According to the report of the International Agency for Research on Cancer,there were19.3 million new cases of cancer worldwide in 2020,among which the incidence of colorectal cancer accounts for 10% ranking third.In China,about one-half of colorectal patients is rectal cancer,characterized as younger-onset and 15% of patients are locally advanced when discovered.For locally advanced rectal cancer,clinical guidelines recommend the use of preoperative chemoradiotherapy as an adjuvant method to reduce cancer stage,improve patient prognosis and promote the success of anus preservation in low rectal cancer.However,patients with rectal cancer have varying degrees of response to radiotherapy,with 20-40% not responding at all,and total mesorectal excision should be performed immediately in these patients.Therefore,the urgent problem to be solved is to process the data of rectal cancer,study the response mechanism of rectal cancer radiotherapy,and construct the response prediction model of rectal cancer radiotherapy.Mechanism discovery from gene expression data takes center stage in transcriptomics research.Since the analysis reliability of gene expression data will be affected by different data platforms,different formats,and different preprocessing batches,researchers hope to have a unified data processing platform and data processing method to reduce these effects and improve the accuracy of analysis results.Based on the R language,this study developed the gene expression data processing toolkit GEDPT,which aims to uniformly process the gene expression profiles of GEO and TCGA,including preprocessing,gene annotation,phenotype annotation,sample grouping,differential analysis,and result visualization,etc.After comparing the gene distribution,it was found that GEDPT uses the same preprocessing for multiple microarray raw data,which reduces the negative impact of the batch effect.Cell type deconvolution(CTD),a novel method to dissect cellular heterogeneity in the tumor microenvironment,can infer cell type-specific gene expression from bulk tissue gene expression data.This study revealed biomarkers associated with radiotherapy response in rectal cancer with the help of CTD high-resolution model results.And correlating them with existing cell states,tumor ecotypes,and ligand-receptor-mediated cellular communication to explore possible mechanisms by which rectal cancer patients have different degrees of response to radiotherapy.Furthermore,this study groundbreakingly used CTD high-resolution results as input,combined with random forest and neural network to construct a response prediction model for rectal cancer radiotherapy.The model improves the prediction accuracy(AUC=0.786)and provides cell type-specific differential genes that have a significant ability to distinguish responses.It provides a new perspective for the study of rectal cancer radiotherapy from the perspective of cell heterogeneity and is of great significance for promoting precise radiotherapy.Figure 38 table 6 reference 102...