Mutational Profile and Clonal Evolution of Relapsed/Refractory Diffuse Large B-Cell Lymphoma

Abstract

Primary refractory/relapsed diffuse large B-cell lymphoma (rrDLBCL) is an unresolved issue for DLBCL treatment and new treatments to overcome resistance is required. To explore the genetic mechanisms underlying treatment resistance in rrDLBCL and to identify candidate genes, we performed targeted deep sequencing of 430 lymphoma-related genes from 58 patients diagnosed with rrDLBCL. Genetic alterations found between the initial biopsy and biopsy at recurrence or refractory disease were investigated. The genes most frequently altered (> 20%) were (in decreasing order of frequency) CDKN2A, PIM1, CD79B, TP53, MYD88, MYC, BTG2, BTG1, CDKN2B, DTX1, CD58, ETV6, and IRF4. Genes mutation of which in pretreatment sample were associated with poor overall survival included NOTCH1, FGFR2, BCL7A, BCL10, SPEN and TP53 (P < 0.05). FGFR2, BCL2, BCL6, BCL10, and TP53 were associated with poor progression-free survival (P < 0.05). Most mutations were truncal and were maintained in both the initial biopsy and post-treatment biopsy with high dynamics of subclones. Immune-evasion genes showed increased overall mutation frequency (CD58, B2M) and variant allele fraction (CD58), and decreased copy number (B2M, CD70) at the post-treatment biopsy. Using the established mutational profiles and integrative analysis of mutational evolution, we identified information about candidate genes that may be useful for the development of future treatment strategies.

Keywords: chemotherapy resistance; immune evasion; next-generation sequencing (NGS); prognostic marker; refractory diffuse large B-cell lymphoma; relapsed diffuse large B-cell lymphoma; tumor evolution.

Figure 1

Workflow of patient selection. Total of 2,604 patients were diagnosed with DLBCL. Of the 2,217 patients with informed consent to review of medical records, 828 patients have refractory or relapsed disease. DNA extraction was performed using samples from 116 patients with rrDLBCL, but samples from 58 patients failed to pass the quality control. DLBCL, diffuse large B-cell lymphoma; rr, refractory/relapsed; NGS, next-generation sequencing.

Figure 2

The details of acquired samples. The pretreatment biopsy samples were obtained from 29 patients (18 refractory DLBCL and 11 relapsed DLBCL). Among them, samples obtained from 18 patients (12 refractory DLBCL and 6 relapsed DLBCL) were paired samples and other samples from 11 patients (6 refractory DLBCL and 5 relapsed DLBCL) were unpaired pretreatment samples. The post-treatment biopsy samples were obtained from 47 patients (28 refractory DLBCL and 19 relapsed DLBCL). Excluding paired samples from 18 patients, samples from 29 patients (16 refractory DLBCL and 13 relapsed DLBCL) were unpaired post-treatment samples.

Figure 3

Mutational profiles of post-treatment tumor samples of relapsed/refractory diffuse large B-cell lymphoma (rrDLBCL) and pretreatment tumor samples of cured DLBCL. rrDLBCL patients include 28 refractory, 14 early recurred, and 5 late recurred patients. Genes mutated in > 3 samples and pathways having > 1 mutated genes are shown. The pathways and genes are ordered by the frequency of mutation. Copy numbers > 6 are marked as amplifications and copy numbers < 0.8 are marked as deletions. COO, cell of origin; ABC, activated B-cell-like subtype; GCB, germinal center B-cell-like subtype.

Figure 4

Mutational frequency of genes in post-treatment tumor samples from patients with relapsed/refractory diffuse large B-cell lymphoma (rrDLBCL). (A) The mutational frequency ABC-type rrDLBCL including 19 refractory and 15 recurred patients. (B) The mutational frequency of GCB-type rrDLBCL including 9 refractory and 4 recurred patients.

Figure 5

Differences in the frequency of mutated pathways between post-treatment tumor samples of the ABC-type rrDLBCL and GCB-type rrDLBCL. Mutations of genes are aggregated for each pathway and the differences between ABC-type rrDLBCL and GCB-type rrDLBCL are shown for (A) refractory disease and (B) recurred disease.

Figure 6

Associations between survival probability and mutations in pretreatment tumor samples. Hazards ratios (HRs) are shown separately for pretreatment samples of rrDLBCL and all DLBCL including cured disease. HR and P values were corrected by the COO and IPI score. (A, B) Genes with P values < 0.05 in the univariate analysis for rrDLCL are included. (C, D) Pathways with > 9 mutated rrDLBCL samples are included.

Figure 7

Tumor mutational burden (TMB). (A) TMB of pretreatment tumor samples. (B) Change in TMB between pretreatment and post-treatment paired tumor samples. (C) The number of SNV/indels that were not present in pretreatment tumor samples but appeared for the first time in post-treatment tumor samples. (D) The number of SNV/indels that were present in pretreatment tumor samples but disappeared in post-treatment tumor samples.

Figure 8

Mutational profiles of 18 paired pretreatment and post-treatment rrDLBCL samples. Genes mutated in > 3 cases are shown. The genes are ordered by the frequency of mutation. Unlike the Figure 3, copy numbers > 4 are marked as amplifications. COO, cell of origin; ABC, activated B-cell-like subtype; GCB, germinal center B-cell-like subtype.

Figure 9

Change in alterations between pretreatment and post-treatment paired tumor samples. Genes altered in > 1 patient are included. P values were calculated using the Wilcoxon signed-rank test. Numbers of samples with alterations are shown in parentheses. (A) Change in variant allele fraction (VAF) of each gene and percentages of mutation SNV/indels that appeared for the first time in post-treatment tumor samples (newly appeared), disappeared in post-treatment tumor samples (disappeared), or were present in both pretreatment and post-treatment tumor samples (remaining). (B) Change in the log2 copy number of genes with amplification. (C) Change in the log2 copy number of genes with deletion.

Figure 10

Molecular classification of pretreatment tumor samples according to the classification of Chapuy et al. (9). Among the 71 features used for classifications 51 features with > 2 alterations are shown. Features were ordered first by the cluster in which the feature contributed the most and second by the degree of contribution.

Figure 11

Differences in survival according to the classification of Chapuy et al. (9). Kaplan-Meier curves shows the difference in survival rate according to each class for (A) progression free survival and (B) overall survival. C1 and C4 were reported to have good prognosis and C3 and C5 were reported to have poor prognosis. This trend was also seen in our data on (C) progression free survival and (D) overall survival.

Figure 12

Molecular classification according to the classification of Schmitz et al. (8). (A) Proportion of assigned classes. (B) Difference in progression free survival among the ABC-type DLBCLs. (C) Difference in overall survival among the ABC-type DLBCLs. (D) Difference in progression free survival among the GCB-type DLBCLs. (E) Difference in overall survival among the GCB-type DLBCLs.