Deep Profiling of Genetic Aberrations and Clonal Evolution in Follicular Lymphoma

Introduction:

Follicular lymphoma (FL) is an indolent malignancy, characterized by multiple relapses during the disease course. Annually around 2-3% of patients experience transformation to aggressive disease (tFL), most commonly to diffuse large B-cell lymphoma (DLBCL). Both transformation and progression of disease within 2 years (POD24) are associated with poor prognosis, yet the molecular events underlying these processes are not well understood. The existence of common progenitor cells (CPCs) has been inferred from genetic analyses of longitudinal biopsies. Improved characterization of genetic alterations associated with CPCs in cases with transformation and POD24 may improve our understanding of disease progression, and reveal molecular markers for high-risk disease.

Methods: We performed whole-exome sequencing of 97 serial tumor biopsies and matched normal samples purified from peripheral blood from 44 FL patients. An average sequencing coverage of 700X was achieved for both tumor and normal samples, which ensured high data quality and ability to detect SNVs and InDels using our bench-marked bioinformatics pipeline. SNP6.0 data was available for 93 sequenced tumors and used to infer allele-specific copy number alterations. Several computational tools were applied to identify potential cancer driver genes (MutSig2CV, 2020plus), mutational signatures (MutationalPatterns), and to study clonal evolution (PyClone, ClonEvol).

Results:

Twenty-two of the 44 FL patients experienced relapses without transformation (referred to as the nFL group), and 22 patients experienced transformation (the tFL group). Nineteen patients (including both groups) experienced POD24. Both transformation and POD24 were associated with inferior overall survival. The median non-synonymous mutational burden was 96 per biopsy (range 10-326). Pre-treatment FL biopsies from the tFL group had significantly higher mutational burden compared to the nFL group (p<0.02). The application of two different approaches for driver gene discovery resulted in a total of 55 as putative drivers. Sixteen of these genes were identified by both tools, including the known FL driver genes CREBBP, KMT2D, TNFRSF14 and BCL2, all being mutated in more than 40 % of cases, and the novel cancer driver genes CTSS and VPS39. By applying MutSig2CV to different categories of biopsies, we identified RRAGC, ATP6V1B2, and HNRNPU as being significantly mutated in pre-treatment tumors of the nFL group, whereas TNFRSF14 and EZH2 were significantly mutated in pre-treatment tumors of the tFL group. When comparing pretreatment and relapsed FL biopsies, we identified CTSS, ZNF493 and HLA-A as significantly mutated in relapsed FL, thus potentially linking these genes to FL relapse. Analysis of mutational signatures demonstrated the presence of the same signatures for longitudinal samples, and no contrasting difference was found between nFL and tFL biopsies. Finally, we constructed clonal phylogenetic trees for 31 patients with serial biopsies and observed that both non-transformed and transformed tumors can evolve directly from CPCs. Contrary to previous findings, we did not find a preference for divergent evolution from FL to tFL, as compared to FL to relapsed FL. There was no significant correlation between evolution pattern and time intervals in the nFL and tFL groups. Based on the mutational history, we confirmed that CREBBP and KMT2D were the most frequently mutated genes seen in CPCs and PCLO, ATP6V1B2 and LRRN3 were also identified as early mutated genes. TNFRSF14, TBL1XR1 and GNAI2 were often mutated later, but before clonal expansion.

Conclusions: We show that at diagnosis, the mutational burden in the tFL group is significantly higher than in the nFL group. By systematically applying driver discovery tools we have confirmed known driver genes and discovered novel genes that may be of importance for FL progression and transformation. Divergent evolution is a common feature both during relapse of FL and transformation to DLBCL, regardless of the time interval between the longitudinal biopsies. A better characterization of the CPCs may provide additional therapeutic opportunities towards a cure for FL patients.