Preoperative Prediction Of MVI For Early-stage Hepatocellular Carcinoma Based On Circulating Tumor DNA And High-throughput Sequencing

PART ?.Screening of MVI-related mutated signature based on high-throughput sequencing data from early-stage hepatocellular carcinoma tissue samplesOBJECTIVE:Microvascular invasion(MVI)is an important risk factor for early recurrence of hepatocellular carcinoma(HCC)after hepatectomy,and is closely related to the prognosis of patients.Previous studies have shown that multiple signal pathways may be involved in MVI of hepatocellular carcinoma.However,they mainly focused on the level of gene expression which is difficult to obtain before surgery.In this study,we used high-throughput sequencing technology to detect MVI-related mutated signature in genomic DNA and to assess the effect of genomic DNA mutation on MVI.METHODS:A total of 180 early-stage HCC patients from the Eastern Hepatobiliary Surgery Hospital from June 2013 to December 2014 were enrolled in this study.High-throughput sequencing combined with the MutSigCV(Mutation Significance Covariates)method were used to screen for MVI-related driver genes.Pathway enrichment analysis of the driving genes was performed using GO(Gene Ontology),KEGG(Kyoto Encyclopedia of Genes and Genomes)and Reactome databases.RESULTS:Through analysis of whole exon sequence(WES)data from 16 patients with early-stage liver cancer,133 driver genes from MVI group and 34 driver genes from non-MVI group were obtained.Pathway enrichment analysis indicated that extracellular matrix(ECM)signal pathway is a critical pathway in the HCC with MVI.Through targeted gene sequencing(TGS),57 genes and 123 high-frequency mutated genes were captured and sequenced in tissue samples from 164 patients.After MutSigCV analysis,we further obtained 24 genes with potential ability to distinguish the existence of MVI.Among them,ARID2,IGF2R,DENNDSA,PIK3CA,IL6ST and ABCG2 may be key genes affecting the mechanism of MVI.CONCLUSIONS:Through high-throughput sequencing data on early-stage HCC,we found that the mutated genes and related signal pathways associated with the progression of MVI,providing a theoretical basis for the subsequent detection of MVI mutated signature by liquid biopsy.The key ECM pathway in MVI suggests that tumor microenvironment changes may result in MVI progression and early metastasis of HCC.PART ?.Detection of MVI-related mutated signature in peripheral blood of early-stage hepatocellular carcinoma based on liquid biopsy and high-throughput sequencingOBJECTIVE:The diagnosis of MVI is mainly based on the histopathological examination of the resected specimens.Although preoperative prediction of MVI is important for surgical clinical decision-making,in addition to imaging and serological indicators,there is a lack of preoperative MVI-related features.In this study,we extracted peripheral blood free DNA(cfDNA)and performed high-throughput sequencing to detect MVI-related mutated signature.METHODS:This study collected peripheral blood samples from 214 patients undergoing hepatectomy in the Eastern Hepatobiliary Surgery Hospital from June 2016 to June 2017,and extracted cfDNA and leukocytes.After cfDNA cyclization and rolling circle amplification,library construction and sequencing for cfDNA were performed.The capture probes for cfDNA cover 42 genes,mainly derived from the ECM signaling pathway genes and 24 MVI-related driver genes of the "Part 1" study.Detection of somatic single nucleotide variation(SNV)and insertion deletion variation(InDel)was performed using VarScan v2 by sequencing the aligned white blood cells.RESULTS:Peripheral blood samples were collected from 214 patients with early-stage HCC,200 of which passed quality monitoring and used for subsequent sequencing analysis.The median concentration of cfDNA was 6.09 ng/mL,while the median concentration of circulating tumor DNA(ctDNA)was 62.54 ng/L,and the average sequencing depth per patient was over 8000X.A total of 3l774 germline variants and 702 somatic mutations were detected.The mutation detection rate of HCC was 78.8%,and 4 mutations were detected per patient on average.Significant mutated genes include tumor suppressor genes TP53,AXINI and RBI,WNT pathway oncogene CTNNBI,chromatin remodeling genes ARID1A and ARID2,cellular antioxidant defense gene NFE2L2,etc.Extracellular matrix signaling pathway genes,such as FBN3,COL6A3 and VCAN,were also found with a higher mutation detection rate.CONCLUSIONS:Through liquid biopsy combined with high-throughput sequencing technology,we were able to detect MVI-related mutated signature in patients with early-stage HCC preoperatively.The MVI-Selector probes designed by us have a higher mutation detection rate for an early-stage HCC,which has the potential to help liver cancer patients with preoperative screening for genetic mutations.PART ?.Preoperative prediction of MVI for early-stage hepatocellular carcinoma based on circulating tumor DNA and high-throughput sequencing technologyOBJECTIVE:Preoperative prediction of MVI has potential clinical significance in the surgical treatment of HCC patients.However,there is a lack of mature clinical application method with potential application prospects.In this part of the study,we establish a MVI predictive model for early-stage HCC by combining the liquid biopsy and high-throughput sequencing data and the clinically data.METHODS:Peripheral blood samples from 125 patients(training cohort)with early-stage HCC from June to December 2016 were collected for cfDNA sequencing.The MVI preoperative prediction model and prospectively validated by the data from 75 patients from March to June 2017(validation cohort).The liquid biopsy and sequencing methods are the same as the "TART ?" The endpoint of this study was the presence or absence of pathologically diagnosed MVI.Variables were screened by univarite Logistic regression and forward stepwise regression.Logistic regression and support vector machine(SVM)method were used to establish MVI predictive model.Decision curve analysis(decision curve analysis)compares the performance between models.The nomogram was constructed based on the results of multivariate logistic analysis.The best cutoff value of the nomogram score was determined by the receiver operating characteristic curve(ROC).The cutoff value was used to determine the risk of MVI,and the patients were divided into low-risk and high-risk groups.RESULTS:After screening,8 variables were used to construct MVI predictive model,including alpha-fetoprotein,PIVKA,imaging tunor diameter and tumor number,cfDNA concentration,TP53 mutation GAF value and two SNP from CDKNIA and VCAN.The models constructed by Logistic regression and SVM are similar in terms of MVI prediction performance.The multivariate analysis results based on the Logistics model were used to construct the nomogram.The risk of MVI was judged by using the cutoff value of 150-point value.The model predicted training and validation cohort with AUC of 0.850 and 0.788,respectively.CONCLUSIONS:Based on liquid biopsy technology and sequencing,it is found that peripheral blood circulating tumor DNA concentration,TP53 mutation,CDKN1A and VC AN SNP are risk factors for MVI.The MVI preoperative prediction model constructed by combining the above risk factors and clinical variables can accurately predict the risk of MVI preoperatively.