| Purpose:Intratumoral heterogeneity is an important reason that affects diagnostic accuracy and leads to drug resistance,recurrence and metastasis in tumor therapy.The purpose of this study was to investigate the degree and regional distribution of intratumoral heterogeneity in esophageal squamous cell carcinoma genome,and to explore the significance and contributing factors of genomic heterogeneity.Materials and Methods:Five patients with untreated esophageal squamous cell carcinoma diagnosed by endoscopic biopsy were included.4-5 separated samples from surgically resected tumor,together with normal esophageal epithelium and peripheral blood mononuclear cells were collected and storage in liquid nitrogen.DNA was extracted to build libraries after strict quality control,and whole-exome sequencing was performed.1.Heterogeneity and phylogenetic analysis:After sample screening and data quality control,somatic mutations in various regions of the tumor were called.Variation of mutation distribution in different regions of tumor after filtering and annotation was assessed.The relationship of different tumor regions in molecular evolution was analyzed by constructing phylogenetic trees;To extrapolate and generate a list of driving genes in the five patients,records on cancer public database COSIMIC and common significant mutated genes(SMGs)reported in previous ESCC genome studies were considered together.The location of each driver gene in individual’s evolutionary tree was analysed and its potential biological role in tumor evolution was inferred according to the graph theory of evolutionary tree.The evolutionary model of the five ESCCs was discussed considering the function of driver mutations.2.Mutation pattern analysis:According to the location of mutations in the phylogenetic tree,they can be divided into two categories:trunk and branch mutations.Mutation types for trunk and branch variations(base substitution types)were extracted to assess the composition and difference of mutation spectrum.By converting all mutation types to a frequency array in the form of 96 nucleotides,the dominant mutation signature(single base replacement mode)is inferred by software according to the reference data of the COSMIC database.The composition and differences of SBS in trunk and branch variation were analyzed,and their potential roles in ESCC heterogeneity and evolution were explored too.3.Subclone analysis:The data of mutation allele frequency(VAF),tumor ploidy and tumor purity(purity)were input into software to calculate and analyze the subclone composition of individual tumors and their evolutionary relationship,and the subclone evolution trees were generated.The subclone distribution in different regions of the tumor was examined to analyze the relationship between subclone composition and the spatial/anatomical location.The earliest and the latest subclones in the evolution tree were manually selected,and the spatial distribution differences of the early and late subclones were analyzed.Genes specific to early and late subclones were extracted,the differences of mutation profiles between the two were analyzed,and the differential genes were functionally enriched to explore the biological mechanism behind the differences.Results:1.Of the five tumors,heterogeneous mutations not in all regions averaged 60.2%,including 27.6%in only individual tumor regions.Of the inferred driving variants,37.5%(12/32)is shared by all of the tumor regions,while 31.3%(10/32)are only present in a single region within the tumor.In evolution trees built for individual tumors,TP53 and NOTCH1 mutations are positioned on the trunk in four patients,while ERCC3 mutations are on the branch of two patients,suggesting that the above mutations may be early and advanced events in tumor development,respectively.There was no significant difference in the number distribution of oncogenes and tumor suppressor genes in trunk and branch.Mutations unique to different regions of the tumor have different biological functions,manifesting a pattern of branched evolution.2.Six mutation types of all the patient differ significantly in trunk and branch variants(p values<0.05,Fisher’s exact test).The proportion of C>G substitution,associated with APOBEC enzyme abnormalities,was lower in branch than in the trunk variants in 4 patients.Mutation pattern analysis found that,APOBEC related patterns SBS2、SBS13 and DNA damage repair disorder related patterns SBS15 are the main mutation patterns.The proportion of SBS15 was higher in all patient in branching variants than the truncal ones,but the difference is not statistical significant(p=0.29,Wilcoxon test).3.The average number of subclones in five patients was 6.2.Tumor samples adjacent to the sampled anatomical location have a more similar subclonal composition;Early clones are more widely distributed,occupying 3.4 areas(range 3～5)in average.Late clones in three patients were located only in a single tumor region,showing a pattern called spatial limitations.No differences in the number and distribution of subclones were observed between the central,peripheral,and upper and lower group samples.The proportions of APOBEC-related C>G substitution as well as proportions of SBS2、SBS13 signatures in late clones are both higher than early clones(p values<0.05,Fisher’s exact test and Wilcoxon test).The GO enrichment analysis of genes unique to late clones indicated ion channel binding correlation(adjusted p value=0.041).Conclusions:There is significant intratumoral genomic heterogeneity in the five ESCCs,and multi-regional sampling may be more accurate in characterizing operable targets and making therapeutic decisions.The distribution of tumor heterogeneity corresponds to the spatial position at the subclonal level,which provides theoretical support for the optimization of local treatment schemes.Different mutation patterns in branch evolution and mode of tumor evolution may be the source of intratumoral genomic heterogeneity of the five ESCCs. |
PDF Full Text Request