Richter syndrome (RS), an aggressive lymphoma that develops in patients with chronic lymphocytic leukemia (CLL), is a striking example of histologic transformation. While recent advances have transformed the treatment landscape of CLL, RS remains associated with dismal overall survival and understanding of the genetic factors driving evolution of CLL to RS is limited. To decipher the genetics underlying this transformation, we performed an integrative analysis of whole exome (WES) and genome sequencing (WGS) data generated from matched RS and CLL samples from a discovery cohort of 52 patients with RS of diffuse large B cell lymphoma (DLBCL) histology followed by WES analysis of a validation cohort of 45 independent RS cases. Through computational deconvolution of CLL and RS clones, we constructed phylogenetic relationships and traced evolution of CLL to RS, confirming both clonal related (87%) and unrelated RS (13%). In addition to identifying recognized RS-risk genetic lesions, we discovered novel putative somatic driver genes (including IRF2BP2, SRSF1, B2M, DNMT3A, CCND3), copy number alterations (sCNAs) beyond CDKN2A/B loss and MYC gain (including del(15q15)[MGA], del(7q36.2)[EZH2, KMT2C, POT1], gain(1q23.1)[MCL1, NOTCH2] and arm level loss of 1p, 6q, 9p and 9q), whole genome duplication (WGD), and chromothripsis. Additional validation was obtained through re-analysis of 14 RS whole genomes (Klintman et al., Blood, 2021). Overall, our analysis highlighted alterations of NOTCH, DNA damage response and MAPK pathways as frequently preexisting in CLL. Alterations in pathways of MYC signaling, epigenetics, interferon/inflammatory signaling, cell cycle deregulation and immune evasion were newly occurring at transformation. To assess the degree of similarity between RS and DLBCL, non-negative matrix factorization (NMF) clustering was performed on the 97 RS WES samples and 304 previously characterized DLBCL samples (Chapuy et al., Nature Medicine, 2018) based on identified RS and known DLBCL drivers. The majority of RS (75 of 97) clustered separately from DLBCL, while 7 of 8 clonally unrelated RS cases clustered with DLBCL (P=6.75x10-6), with membership across the DLBCL molecular clusters. To further examine RS, NMF clustering was performed across the 97 RS samples, and 5 robust genomic subtypes were identified: RS-1: WGD; RS-2: Trisomy 12 with NOTCH1/SPEN mutations; RS-3: Mut-TP53 and Mut-NOTCH1; RS-4: Mut-EGR2 and Mut-SF3B1; and RS-5: TP53-altered RS without Mut-NOTCH1. Supervised and unsupervised transcriptome analysis of 36 of the 97 RS samples supported the identified genomic subtypes. The less-genome altered RS subtypes (RS-2, RS-4) had improved overall survival (3.3, 11.3, 5.0, 16.7 and 4.0 months for clonally related RS in subtypes 1-5, respectively; P = 0.0082). To investigate the stepwise evolution of CLL to RS at high resolution, single-cell RNA-sequencing was performed on diagnostic biopsy samples from 5 RS patients. We developed a novel tool, CNVSingle, to infer allele specific single-cell sCNAs that enabled identification of the single-cell clusters representing distinct CLL and RS genetic subclones as well as intermediate or transitional evolutionary states. Genetic changes often immediately preceded the transcriptional shifts towards RS and increasing chromothripsis was again observed as a hallmark of RS. RS cells displayed gene expression enriched in pathways of MYC activation and cell cycle, in line with similar analyses on paired RS and CLL bulk transcriptomes. Given the numerous RS-associated genomic features identified by our study, we assessed the feasibility of non-invasive detection of RS events in cell-free DNA (cfDNA) extracted from RS patient plasma by ultra-low pass-WGS sequencing of 46 plasma samples from 24 RS patients. Indeed, RS-specific sCNAs, chromothripsis, RS-specific driver mutations and whole genome doubling were detected in plasma cfDNA, distinct from circulating CLL. This included detection of RS in cfDNA of 6 of 8 patients (75%) at RS diagnosis and 2 of 7 (29%) patients within 10 months preceding RS diagnosis. In conclusion, we report molecular characterization of the largest RS cohort to date (111 patients), demonstrate that RS is genetically distinct from DLBCL and provide novel genetic and prognostic insights into this aggressive malignancy, with potential for clinical translation towards a more precise RS diagnosis.
Read full abstract