Multi-dimensional Analysis Of Hepatocellular Carcinoma Prognosis Based On Cell-free DNA Copy Number Variation

Background and Objective Hepatocellular carcinoma(HCC)is one of the most common malignant tumors in the world and one of the main causes of death.Radical surgery and radiofrequency ablation are the main treatment methods for patients with early HCC.However,due to the lack of effective markers to predict recurrence and prognosis,recurrence and metastasis of liver cancer are still important reasons for poor prognosis of HCC.Alpha fetoprotein(AFP)is the most commonly used serum marker for clinical monitoring of HCC,but its sensitivity is relatively low.Therefore,it is urgent to develop new biomarkers for HCC prognosis that are easy to detect.Plasma free DNA(cf-DNA)is a short fragment of double-stranded DNA(160-180 bp in length),which contains tumor genome information such as methylation,point mutation and copy number variation(CNV),and can be used for tumor diagnosis and prognosis judgment.In this study,the whole genome sequencing(WGS)of the plasma cf-DNA of 117 HCC patients was carried out to analyze the relationship between the whole genome,chromosome arm and Bin(1M bp region)of the three dimensions of cf-DNA CNV and the clinical characteristics of HCC patients.Further analyze the relationship of different dimensions of cf-DNA CNV on the total survival(OS)and relapse-free survival(RFS)of HCC patients,explore the predictive value of cf-DNA CNV as a new blood biomarker for the prognosis of HCC patients,and screen potential functional genes that affect the prognosis of hepatocellular carcinoma.MethodsFrom May 2016 to July 2017,a total of 117 patients with hepatocellular carcinoma(HCC)were included.All patients received radical treatment of liver cancer and reached a tumorfree state after surgery.The clinical data of patients including age,sex,tumor size,AJCC stage,serum AFP level,microvascular invasion,BCLC stage,CTP stage,recurrence,recurrence time,death and survival time were collected.This study was approved by the school ethics committee,and the ethics number was KY20183331-1.Collect 5m L of whole blood from 117 patients with liver cancer before operation,and use QIAamp circulating nucleic acid kit(Qiagen)to extract plasma cf-DNA;Another 19 paired liver cancer tissue samples were taken to extract tissue DNA.Qubit3.0(Thermo)and Agilent 2100 biological analyzer(Agilent)were used to evaluate the content and quality of DNA samples,and the WGS library of tissue DNA and plasma cf-DNA samples was constructed by the NEBNext Ultra II DNA library preparation kit of Illumina(NEB).The QPLOT algorithm is used for quality control and genome-level statistical analysis,and the ichor CNA algorithm is used for genome-wide CNV analysis,and the tumor score(TFx)is calculated.This study borrows the definition of standard deviation,calculates the mean of the square sum of the CNV(log2 ratio)and the difference of 0 at each Bin level(that is,divide by the total number of Bins),and then takes its positive square root,and innovatively defines the stability score(S-score).The regression analysis of all Bins in the whole genome was carried out by using single factor Cox to screen the Bins with significant prognosis,and then the Bin-score of liver cancer prognosis based on the level of Bin was constructed by using the least absolute contraction and selection operator(LASSO)regression and multivariate Cox regression.The effects of CNV indicators of all dimensions on total survival and relapse-free survival were evaluated by Kaplan-Meier’s Log-Rank test,and the predictive efficacy of each indicator in predicting 1 year and 3 year survival was evaluated by ROC curve.The genes corresponding to the three core Bins regions were annotated by the human genome reference sequence,and the potential regulatory genes affecting the prognosis of liver cancer were screened by single-factor Cox.ResultsThe whole genome sequencing results of plasma cf-DNA samples from 117 patients with liver cancer in this study showed that the average depth was 4.01 ×(2.78-7.86),Q30 quality is 92.4%;The average depth is 29.51 ×(18.42-49.23),Q30 quality is 92.5%,showing good sequencing quality.This study innovatively defined the CNV stability score(S-score),which can effectively reflect the degree of discrete variation of cf-DNA CNV at the whole genome level on the normal diploid baseline.There was a significant positive correlation between the S-score and the previously reported CNV indicators "TFx and P-score"(r=0.767 and r=0.706).The CNV analysis of chromosome arm dimension showed that the first four most common arms of CNV increased in 117 patients with HBV-HCC were 20 p,8q,1q and 20 q,respectively;The most common first four arms of CNV reduction are 17 p,4q,19 p and 16 q,respectively.The consistency between cf-DNA and tissue DNA shows that the CNV frequency of cfDNA and tissue DNA in the center’s liver cancer cohort is highly consistent at the Bin level and in the dimension of chromosome arm,and has a similar change pattern with the CNV in the TCGA liver cancer cohort,suggesting that cf-DNA has the possibility of replacing tissue DNA and becoming a new biomarker.The results of chi-square test of clinical characteristics of liver cancer and CNV indicators at the whole genome level showed that the proportion of patients with liver cancer with higher TFx,P-score and S-score in the middle and late stage(phase III)and tumor diameter ≥ 5cm were significantly higher than those with higher and lower indicators(P<0.05);At the same time,the risk of recurrence and death in patients with liver cancer with higher TFx,P-score and S-score were also significantly higher than those with lower indicators(P<0.05);There was no significant difference in age,sex,AFP level and CTP grading between the two groups(P>0.05).In this study,1406 candidate Bins were identified from 2475 Bins by single-factor Cox regression analysis,which were significantly related to the prognosis of liver cancer;Three core Bins(B1021 at Chr6 q12;B1318 at Chr8 q13.2;B2173 at Chr17 p13.2)were further screened based on the LASSO regression model and multivariate Cox regression to establish the Bin-score.The results showed that the patients with high Bin-score had later BCLC stage,larger tumor and higher risk of recurrence and death.Bin-score is a biomarker of liver cancer prognosis independent of sex,age,BCLC stage,MVI and AFP(OS: HR=1.09,P<0.001;RFS: HR=2.16,P=0.004).Using the gene expression and survival data in the TCGA hepatoma cohort,two prognostic-related genes CYB5D2 and MED11,as well as two potential prognostic trend genes RNF167 and GLTPD2,were screened by single-factor Cox.Compared with normal liver tissue,CYB5D2,MED11,RNF167 and GLTPD2 genes were significantly lower expression in liver cancer tissue,and their lower expression was associated with later TNM stage and worse prognosis of patients.ConclusionThe CNV stability score(S-score)can effectively reflect the degree of discrete variation of the whole genome dimension cf-DNA CNV on the basis of normal diploid.The plasma cfDNA and tissue DNA of patients with liver cancer are highly consistent with CNV in Bins and chromosome arm dimensions.Plasma cf-DNA is a reliable alternative marker for tissue DNA detection.The decrease of CNV in chromosome arm dimensions 4q,17 p and 19 p and the increase of CNV in chromosome arm dimensions 8q and 1q are independent prognostic factors for OS and RFS of hepatocellular carcinoma.The S-score and Bin-score constructed in this study are significantly related to the BCLC stage and tumor size of hepatocellular carcinoma,which can effectively predict the prognosis of patients with hepatocellular carcinoma.The decreased expression of potential functional genes such as CYB5D2,MED11,RNF167 and GLTPD2 screened in this study suggests a later TNM stage and a worse prognosis of patients.