Whole Exome and Transcriptome Sequencing in 1042 Cases Reveals Distinct Clinically Relevant Genetic Subgroups of Follicular Lymphoma

Follicular Lymphoma (FL) is the most common indolent lymphoma derived from light zone germinal center B cells and characterized by a t(14;18) translocation resulting in upregulation of BCL2 in over 80% of cases. This translocation alone is not sufficient for tumorogenesis, and must be combined with additional genetic mutations to transform B cells. FL is incurable and the disease course can be highly varied, with survival ranging from a few months to decades following diagnosis and treatment with standard chemoimmunotherapy. The heterogeneity of FL poses major challenges to identifying the association of genetic alterations and clinical outcome.

Current WHO guidelines recommend establishing grade for each FL case with grade 3 thought to be more aggressive than 1 and 2. The genetic basis and clinical implications of grade in FL are unclear. Recent sequencing studies have identified many genes found to be recurrently mutated in FL including KMT2D and CREBBP. However, the degree to which genetic alterations cooperate with each other or contribute to clinical outcome is unclear. Based on the observed mutational rates in follicular lymphoma, we estimated 900 cases were needed to comprehensively delineate the genetic alterations that underlie histologic grade and clinical outcome.

Accordingly, we enrolled a cohort of 1042 patients with newly diagnosed FL. All treated patients received rituximab-containing standard regimens. To go beyond the identification of gene-coding events, we developed a very large panel of 110 Mbp covering exonic (~40Mbp) and non-exonic regions (~70Mbp) of interest to enable a wide range of genomic analysis including mutation calling in both coding and non-coding regions, rearrangement detection, viral identification, and copy number analysis. In addition to the whole exome, we extended coverage to include introns, promoters, and untranslated regions of all known driver genes in cancer. We included the entirety of the immunoglobulin loci, T-cell receptor loci and CD3 loci to detect clonotypes and rearrangements. We also included lymphoma-relevant long non-coding RNAs, microRNAs, enhancers, and breakpoint-prone regions. For viral detection, we targeted the genomes of eight cancer-related viruses: Epstein-Barr virus, human papillomavirus, human immunodeficiency virus, hepatitis B, hepatitis C, Kaposi's sarcoma-associated herpesvirus, human T-lymphotropic virus, and Merkel cell polyomavirus. In addition, to enable high resolution identification of copy number variation (CNV) calls, the entire genome was tiled with probes spaced 10kb apart.

DNA and RNA were extracted from all tumors and their paired normal samples, prepared into DNA and RNA sequencing libraries and subjected to sequencing on the Illumina platform to a targeted coverage of 150X. Somatic events were identified and further filtered to identify driver events in both coding and non-coding regions.

FLs demonstrated a significant degree of genetic heterogeneity with over 100 genes mutated with a frequency of at least 2%. Nearly 100% of FL cases had a mutation in at least one chromatin-modifying gene. The most frequently mutated genes in follicular lymphoma were KMT2D, BCL2, IGLL5 and CREBBP. In addition, we identified frequent mutations in SPEN, BIRC6 and SETD2. To our knowledge, this is the first description of alterations in these genes in FL. Transcriptome analysis indicated a strong correlation between BIRC6 mutations and the previously described immune response 2 signature that is associated with a poor prognosis. We further performed unbiased clustering of genetic alterations in these FL cases. We identified a cluster that was specifically enriched in BCL6 and TP53 alterations and was strongly associated with grade 3 FLs which are predicted to have poorer outcomes with low intensity therapies. We further examined the genetic profiles of 1001 DLBCLs in comparison to this cohort of FLs. Our data indicate a continuum of highly overlapping genetic alterations with DLBCL displaying more complex patterns that included alterations in MYC, TP53 and CDKN2A (mainly copy number losses), indicating shared pathogenetic mechanisms underlying FL and DLBCL, particularly those germinal center B cell origin.