Genetic Determinants and Evolutionary History of Richter's Syndrome

Richter’s syndrome (RS) arising from chronic lymphocytic leukemia (CLL) is a striking example of an aggressive malignant histology that emerges from indolent cancer. A major barrier to disease control in CLL, RS is associated with poor clinical outcomes and limited survival. The genetic basis of RS is poorly understood and its relationship to antecedent CLL remains incompletely characterized. Notable challenges to the genomic study of RS includes those of sample acquisition, the distinction between true tumor events rather than sequence artifacts in archival fixed tissue, and the limitations of available computational techniques for deconvoluting admixtures of CLL and RS DNA within the same biopsy specimen.

To address these challenges and characterize the genetic profile of RS, we performed whole-exome sequencing (WES) on samples collected from 42 patients with RS of diffuse large B cell lymphoma (DLBCL) histology. For this genomic characterization, samples from 37 patients were analyzed as ‘trios’ (matched germline, CLL and RS tissue DNA) and those from 5 as ‘duos’ (matched CLL and RS DNA). CLL diagnosis preceeded RS diagnosis by a median of 60.6 months (range 0.1-234.5). The median number of prior CLL-directed therapies was 2 (range 0-10). 8 patients had no prior CLL-directed therapy, while 9 were exposed to novel agents. The median time from most proximal CLL sampling to RS was 2.7 months (range: 82.2 months pre- to 3.8 months post RS diagnosis). Critical analytic innovations applied to this dataset included addressing contamination of CLL DNA in the germline sample (through the tool DeTIN) and generating the ability to discriminate between clones arising from RS or from CLL, even while both histologies were commonly co-existing within originating biopsies (via the tool PhylogicNDT).

From this discovery cohort of 42 cases, 36 (86%) revealed RS and CLL to be clonally-related based on WES analysis, with a distinct RS clone emerging from an existing CLL subclone. Of the 6 (14%) cases determined to be clonally unrelated by WES, 4 had been previously examined by IGHV sequencing; only 1 of 4 was categorized as clonally unrelated, likely due to CLL and RS admixture. RS displayed mutational signatures reflecting aging (CpG), canonical AID, and non-canonical AID processes. Through deconvoluting clonal composition using PhylogicNDT in related sample trios (n=31), we established several notable differences compared to antecedent CLL. First, RS clones presented higher rates of additional mutations than the ancestral CLL clones from which they developed (2.47 vs. 0.86 Mut/Mb, p<0.0001). Second, the frequency of CLL-associated driver mutations in the RS clones was altered:TP53 mutation (n=21, 50%), NOTCH1 mutation (n=17, 40%) NFKBIE (n=5, 12%), EGR2 (n=6, 14%), XPO1 (n=5, 12%), and RPS15 (n=3, 7%). Third, the clonal transition to RS was marked by a diversity of additional driver mutations and particularly increased copy number variants (CNVs). RS biopsies had recurrent arm level events, including del(17p) in 44%, del(4p) in 27%, del(4q) in 27%, del(9p) in 22%, del(17q) in 22% and del(9q) in 20% (Q<0.05, GISTIC). The top focal alterations in RS biopsies, as identified by tool GISTIC included (Q value <= 0.05): amp(7q21.2) – CDK6, amp(8q24.3) – RECQL4, amp(13q31.3) – ERCC5, amp(19p13.3), amp(19q13.42), del(8p12), del(17p13.1) – TP53, del(13q14.3), del(7q35), and del(16q12.2). Finally, whole genome doubling was observed in 6 cases upon transition to RS.

To further investigate RS and CLL clonal evolution at high resolution, we performed single-cell RNA-sequencing (10x Genomics), on biopsies at the time of RS diagnosis in 5 individuals with clonally related transformation. Data analysis and clustering was conducted using Seurat(v3) with doublet removal (DoubletFinder). We identified CNV in single-cell transcriptomes that matched the WES copy-number profiles of individual subclones of the RS and CLL populations and thus linked subclones identified by WES to single-cell expression clusters and states, revealing CLL, RS and intermediate evolutionary cell states.

In conclusion, we identify that RS arises from CLL subclones through distinct mutational trajectories. Further molecular subclassification of RS, including genetic characterization of additional cases, and linking mutational data with clinical outcomes is ongoing and has potential to alter clinical classification and prognostication of RS.