Predictive Value Of Radiogenomics-identified Genes On CTL Cells Infiltration In Renal Cell Carcinoma

Purpose During the past several years,checkpoint blockades such as PD-1/PD-L1 inhibitor turn on a new paradigm breakthrough in immunotherapy for renal cell carcinoma due to the promising OS benefits.However,the response rate was only about 25% according to a recent clinical trial based on the treatment of nivolumab,which reminds researchers of the great importance to explore effective biomarkers to predict the response of checkpoint blockades.It has been proved PDL-1 expression is correlated with the presence of cytotoxic T lymphocyte cells(CTL,also called CD8+ T lymphocyte cell)infiltration in tumor microenvironment,and searching a predictive biomarker of CTL cells infiltration maybe a potential prediction marker to identify real patients who would benefit from immunotherapy.Radiogenomics,a new field of study method that identifying genomic data through imaging features,showed great potential in the prediction of molecular subtype and drug response.Based on big data resource database,our research intends to analyze the quantified CT imaging features of renal cell carcinoma patients in order to establish different radiomics phenotypes,and then identify differential expression genes of each subtype to analyze the associations between the expression level of candidate genes and CTL cells infiltration.The overall purpose of the present study is to evaluate potential value of radiogenomics in identifying biomarkers for clinical response of checkpoint blockades in patients with renal cell carcinoma.Materials and methods CT images of qualified patients with renal cell carcinoma were downloaded from The Cancer Imaging Archive,an open-source TCGA's companion database funded by National Cancer Institute with images data of cancer and the genomics and clinical information of the corresponding patients were obtained from The Cancer Genome Atlas through the same TCGA-ID.Maximum axial level segmentation was performed by Image J to obtain regions of interest(ROI).Radiomics characteristics such as gray value histogram statistics,textures,sharpness of lesion boundaries,metrics of compactness were extracted from ROI and quantitative radiomic data acquired which also need to be normalized for further analysis.Next,unsupervised k-means clustering analysis manipulated by R ggplot2 package was used for discovery on the development cohort to classify imaging subtypes after the optimal number of clustering was determined by R Nb Clust package.To identify the Differential gene expression,we used R limma package,and genes with P<0.05 and |log FC|>1.5 was defined as differential expression genes which are also regard as candidate genes to be analyzed in Tumor Immune Estimation Resource to investigate the potential correlation between the specific genes and CTL cells infiltration.Results A total of 88 patients with renal cell carcinoma were identified in our study and was grouped as Cluster1 and Cluster2 by R language ggplot2 due to a result of the optimal number k=2 and all the patients TCGA genomic and clinical data in the above two groups were summarized.A total of 14636 expression genes were included in the RNA sequence data after elimination of sample expression data with a signal value of 0 and an mean<1 in all 20532 genes.There were 19 genes identified due to the radiomic subtype,of which 7 genes(QRFPR,SLC16A12,SLC5A1,FAM196 B,TRHDE,FOSB,PKHD1)are down-regulating while the other 12(IGFN1,RPS28,PPP2R2 C,PRSS50,EEF1A2,COL7A1,DES,ADAMTS14,PANX2,PLXNB3,CORO6,APOD)showed up-regulating.It was found that a total of 11 gene expression levels were associated with CTL cells infiltration,of which SLC5A1,FAM196 B,QRFPR,SLC16A12 are positively correlated while CORO6,PANX2,IGFN1,PRSS50,COL7A1,EEF1A2,RPS28 exhibit negative correlation according to the TIMER database analysis.Among all genes above,CORO6 showed the most significant negative association(cor=-0.2188,P<0.0001)and SLC5A1(cor=0.1951,P<0.0001)showed the strongest positive association with CTL cells infiltration.Conclusion The differential gene expression of renal carcinoma can be effectively identified based on radiogenomics and the candidate genes showed associations with CTL cells infiltration,which indicates the predictive value of radiogenomics on CTL cells infiltration.Those positive correlated genes are expect to be biomarkers to identify checkpoint blockades beneficial patients while those negative genes are potential predictors of ineffective populations and may be a potential mechanism for the resistance of immune checkpoint inhibitors.In conclusion,our radiogenomic method offers a promising noninvasive approach to identify candidate genes and correlate images with CTL infiltrating which showed significant prospect of further exploration and validation.