Whole-genome sequencing reveals the evolutionary trajectory of HBV-related hepatocellular carcinoma early recurrence

Abstract

Patients with hepatocellular carcinoma (HCC) have poor long-term survival following curative resection because of the high rate of tumor early recurrence. Little is known about the trajectory of genomic evolution from primary to early-recurrent HCC. In this study, we performed whole-genome sequencing (WGS) on 40 pairs of primary and early-recurrent hepatitis B virus (HBV)-related HCC tumors from patients who received curative resection, and from four patients whose primary and recurrent tumor were extensively sampled. We identified two recurrence patterns: de novo recurrence (18/40), which developed genetically independently of the primary tumor and carried different HCC drivers, and ancestral recurrence (22/40), which was clonally related to the primary tumor and progressed more rapidly than de novo recurrence. We found that the recurrence location was predictive of the recurrence pattern: distant recurrence tended to display the de novo pattern, whereas local recurrence tended to display the ancestral pattern. We then uncovered the evolutionary trajectories based on the subclonal architecture, driver-gene mutations, and mutational processes observed in the primary and recurrent tumors. Multi-region WGS demonstrated spatiotemporal heterogeneity and polyclonal, monophyletic dissemination in HCC ancestral recurrence. In addition, we identified recurrence-specific mutations and copy-number gains in BCL9, leading to WNT/β-catenin signaling activation and an immune-excluded tumor microenvironment, which suggests that BCL9 might serve as a new therapeutic target for recurrent HCC. Collectively, our results allow us to view with unprecedented clarity the genomic evolution during HBV-related HCC early recurrence, providing an important molecular foundation for enhanced understanding of HCC with implications for personalized therapy to improve patient survival.

Fig. 1

Genomic landscape of 40 pairs of primary and early-recurrent HCCs. a A concise diagram of the sequencing research in the present HCC genomic study. b The number of somatic mutations (SNVs + indels) in the whole genome across 40 pairs of primary and early-recurrent HCCs. c The number of structural variations (SVs) in 40 pairs of primary and early-recurrent HCCs, including insertions, inversions, tandem duplications, deletions, and inter- and intra-chromosomal translocations. d GISTIC analysis revealed the whole-genome distribution of copy-number alterations in paired primary (left panel) and early-recurrent (right panel) HCCs. GISTIC q-values (y-axis) for amplifications (upper, red) and deletions (lower, blue) are plotted across the genome (x-axis). e The genomic alteration of driver genes in 40 pairs of primary and early-recurrent HCCs, identified through whole-genome sequencing. Drivers present in ≥4 patients are shown. All mutations (SNVs + indels) were validated by Sanger sequencing

Fig. 2

Recurrence patterns during HCC early recurrence after curative resection. a Numbers of somatic mutations (SNVs + Indels) and frequency of shared mutations at the whole-genome level between primary tumors and recurrent tumors across 40 patients with HCC after force calling. b The proportion of shared mutations in patients with de novo recurrence and ancestral recurrence. c, d Clonality indices for the 40 cases of HCC analyzed in our study. Red dotted lines indicate the cut-off value to define clonal relatedness. e, f The proportion of shared mutations and its relationship to the recurrence location. g Comparison of recurrence time between patients with de novo recurrence and patients with ancestral recurrence. h Comparison of total somatic mutations (left panel), non-synonymous mutations (middle panel), and SVs (right panel) between primary tumors and recurrent tumors in patients with de novo recurrence or ancestral recurrence. i Comparison of whole-genome doubling (WGD), tumor cell differentiation, and tumor size between patients with de novo recurrence and patients with ancestral recurrence in primary or recurrent tumors. j Kaplan−Meier analysis revealed different overall survival rates between the two recurrence patterns after the second resection

Fig. 3

Phylogenetic trees and subclonal architectures underlying the evolutionary trajectory of tumor early recurrence in 22 HCCs with ancestral recurrence. a Each tree and corresponding subclonal architecture represents an individual patient. Trees were derived from genome-wide somatic mutations (SNVs + indels) based on subclonal architectures. The numbers of all somatic mutations per patient are labeled above the tree (the numbers of somatic mutations involved in constructing the phylogenetic trees are labeled in brackets). The asterisk indicates that trees and subclonal architectures were derived from somatic mutations in genic region. Line lengths reflect the proportion of clustered somatic mutations attributed to that clone or subclone. The whole tree is scaled to the maximum length of a tree that would be inferred from mutations identified in the primary tumor. In the subclonal architecture panel, the diameter of each oval with color is proportional to the estimated CCF, which reflects the proportion of cells in that sample that contain the mutations that constitute the same color of the relevant phylogenetic tree. b The relative fraction of patients in whom the indicated somatic alterations of drivers occupied the trunk clone, trunk subclone, or branch subclone of their respective phylogenetic trees. The numbers of patients with each of the indicated somatic alterations are labeled

Fig. 4

Phylogenetic trees and subclonal architectures underlying the evolutionary trajectory of tumor early recurrence in 18 HCCs with de novo recurrence. a Each tree and corresponding subclonal architecture represents an individual patient. Trees were derived from genome-wide somatic mutations (SNVs + indels) based on subclonal architectures. The numbers of all somatic mutations per patient are labeled above the tree (the numbers of somatic mutations involved in constructing the phylogenetic trees are labeled in brackets). The asterisk indicates that trees and subclonal architectures were derived from somatic mutations in genic region. Line lengths reflect the proportion of clustered somatic mutations attributed to that clone or subclone. The whole tree is scaled to the maximum length of a tree that would be inferred from mutations identified in the primary tumor. In the subclonal architecture panel, the diameter of each oval with color is proportional to the estimated CCF, which reflects the proportion of cells in that sample that contain the mutations that constitute the same color of the relevant phylogenetic tree. b The relative fraction of patients in whom the indicated somatic alterations of drivers occupied the clone or subclone of their respective phylogenetic trees. The numbers of patients with each of the indicated somatic alterations are labeled

Fig. 5

Mutational spectrum and signatures in primary and early-recurrent HCC. a Six substitution patterns sorted by the total mutation number in primary tumors and recurrent tumors from all 40 patients (left panel), 18 patients with de novo recurrence (middle panel), and 22 patients with ancestral recurrence (right panel). b Major mutational signatures extracted from paired primary and recurrent HCC samples. c Dot plots showing the distribution of mutational signatures to each patient in the trunk clone, the trunk subclone, and primary and recurrence branch subclones from patients with ancestral recurrence, based on our classification of clonal and subclonal mutation clusters. d Comparison of major mutational signatures across trunk clone, trunk subclone, and primary and recurrence branch subclones in 22 patients with ancestral recurrence. Left panel: mutational signatures with elevated proportions; middle panel: mutational signatures with decreased proportions; right panel: mutational signatures with variable proportions. e Dot plots showing the distribution of mutational signature to each patient in the clone and subclone of the primary or recurrent tumor, respectively, from the patients with de novo recurrence, based on our classification of clonal and subclonal mutation clusters. f Comparison of major mutational signatures across clone and subclone in primary tumors or recurrent tumors, respectively, from patients with de novo recurrence. Left panel: mutational signatures with elevated proportions; middle panel: mutational signatures with decreased proportions; right panel: mutational signatures with maintained proportions

Fig. 6

Spatiotemporal heterogeneity and polyclonal dissemination in four patients with ancestral recurrence revealed by multi-region WGS. a Phylogenetic trees constructed from four patients with ancestral recurrence (total of 30 tumor samples). Trees were derived from somatic mutations (SNVs + indels) in genic region based on subclonal architectures. The numbers of all genome-wide somatic mutations per patient are labeled above the tree (the numbers of somatic mutations in genic region involved in constructing the phylogenetic trees are labeled in brackets). Lengths of lines are proportional to the number of mutations in each cluster. Quantification of tumor spatial and temporal heterogeneity of each patient is on the right of the corresponding phylogenetic tree. b Fishplot showing the tumor clonal evolution process from primary tumor to early-recurrent tumor. c Oval plots showing the subclonal structure of tumor samples from each patient. The diameter of each oval is proportional to the corresponding CCF value. Subclonal structures are illustrated by the nested ovals. d Clonal evolutionary history. The dashed box shows the predicted historical states during the tumor ancestral recurrence. Black arrows denote clonal evolution during primary tumor progression. Red arrows denote clonal evolution from the primary tumor to recurrent tumor. Arrows are labeled the acquisition of new subclones with orthogonal lines. Each colored line corresponds to a subclone

Fig. 7

Clinical significance of BCL9 genomic alteration in 40 pairs of primary and recurrent HCCs. a Distribution of somatic mutations in BCL9 identified in the recurrent HCC samples. b Calculated BCL9 copy numbers identified in 40 pairs of primary and recurrent HCCs, ***P < 0.001. c Representative BCL9 staining and its density statistics in peritumor tissues, primary tumor tissues with wild-type BCL9, and recurrent tumor tissues with mutant BCL9 (G14E, S278N, and D349V) and BCL9 amplification; scale bars = 50 μm, *P < 0.05, **P < 0.01. d Kaplan−Meier survival analysis showing overall survival rates after the second resection based on BCL9 mutation and copy-number variation in recurrent tumors. e BCL9 copy numbers revealed by qPCR in one normal liver cell line (L0-2) and six HCC cell lines. f BCL9 mRNA and protein levels examined by qRT-PCR and western blot in one normal liver cell line and six HCC cell lines

Fig. 8

Oncogenic role of BCL9 in HCC. a BCL9 expression examined by western blot in stably transfected cells. b Proliferation of HCCLM3 cells after BCL9 knockdown and of HepG2 cells expressing wild-type or mutant BCL9 compared with that of controls, **P < 0.01, ***P < 0.001. c Colony formation and invasion of HCCLM3 cells after BCL9 knockdown and of HepG2 cells expressing wild-type or mutant BCL9 compared with that of controls. The bar graphs illustrate the quantification of the assay results, *P < 0.05, **P < 0.01, ***P < 0.001. d Representative bioluminescence images of mouse liver tumors and pulmonary metastasis. The color scale bar depicts the photon flux emitted from the mice, *P < 0.05, **P < 0.01, ***P < 0.001. e Western blot showed the expression of β-catenin in HCCLM3 cells after BCL9 knockdown and of HepG2 cells expressing wild-type or mutant BCL9. f Results of dual-luciferase assays; reporter activity was normalized to Renilla luciferase activity, *P < 0.05, ***P < 0.001. g Immunofluorescence staining showing subcellular β-catenin localization in the indicated cells. h Representative BCL9 and CD8 staining in peritumor tissues and tumor samples from patients with HCC. The color scale bar depicts the CD8-positive cell number in all 40 matched sets of peritumor, primary tumor, and recurrent tumor tissues (left panel) and in 40 recurrent tumor tissues (right panel); scale bars = 50 μm, *P < 0.05, **P < 0.01

Fig. 9

Schematic figure illustrating the recurrence pattern and evolutionary trajectory of HBV-related hepatocellular carcinoma early recurrence