The factors mediating GvL resistance following allogeneic stem cell transplant (SCT) in lymphoid malignancies remain incompletely characterized. Because cell-intrinsic features shape chemotherapeutic relapse, we hypothesized that they also shape GvL outcomes by influencing evolutionary trajectories of CLL relapse after reduced intensity conditioning SCT (RIC).
We identified 9 heavily pre-treated patients (pts) (range: 1-5 therapies, median: 3) with various times to CLL relapse after RIC (range: 83-1825 days), of which 8 had at least partial responses before relapse. To define evolutionary trajectories, we generated paired whole-exome and RNA sequencing data from purified CLL cells pre/post-RIC, using MuTect2 and ABSOLUTE algorithms to identify somatic alterations (SAs) and corresponding cancer cell fractions (CCFs). 5 pts had clonal SAs in TP53 and/or SF3B1 pre-SCT, and no single SA was specific to post-RIC. Furthermore, we found no SAs nor altered expression of HLA class I/II or b2M in either baseline or post-RIC samples. However, we found 6 relapse pairs to exhibit complex branched evolution involving CCF shifts of at least 0.2 in subclonal and clonal SAs whereas 3 pairs showed genomic stability. Clonal evolution was associated with longer time to relapse (Wilcoxon, p=0.02; median 798 versus 304 days) as well as complete response (p=0.05), suggesting that GvL immune escape may be facilitated by clonal evolution.
To determine the phenotypic consequences of clonal evolution, we examined single cell transcriptomes using scRNAseq from paired pre/post-RIC CLL cells from 2 pts with early (304, 442 days; "ERs") and 2 pts with late (1801, 1825 days; "LRs") relapses after RIC. Using the inDrop platform, we profiled a median of 3560 CLL cells/pt (range: 2254-5278). Clustering using Seurat revealed marked transcriptional stability after RIC in ERs whereas dramatic shifts in gene expression programs were observed in LRs. Single cell trajectory analysis using Monocle identified ordered biological processes through which LRs, but not ERs, progressed. Branched expression analysis revealed multiple patient specific pathways defining LRs, including within chromatin regulators (EBF1, BANK1), oncogenic pathways (AFF3, DENND4A) and ribosomal biosynthesis (EEF1G, NACA). Thus, genetic evolution in LRs results in distinct phenotypic consequences.
To directly link SAs with transcriptional outcomes, we interrogated scRNAseq data for known SAs identified by WES. In one LR, loss of a CLL cancer driver (RPS15mut) was observed in two of three post-RIC transcriptional clusters, either through deletion of chr.19p (where RPS15 resides) or reversion to the wildtype allele (implying loss of heterozygosity). In addition, genomic and transcriptional loss of HLA genes were detectable in pre-RIC clusters that failed to expand at relapse in both LRs, suggesting that pre-existing HLA loss does not provide a selective advantage for CLL relapse after RIC, consistent with our bulk analyses. These data highlight how scRNAseq can delineate genetic selection pressures within subpopulations of a single patient.
To investigate whether epigenetic dysregulation underlies these genetic changes, we measured locally disordered methylation (LDM), a known epigenetic mechanism of CLL genetic variability. Genome-wide methylome profiles revealed increases in LDM in LRs compared to ERs for various genomic regions (Kruskal-Wallis (KW), p<0.05 for promoters, genes, distal regulatory modules); no increases in LDM were observed in an independent cohort of late CLL relapse after chemotherapy alone (n=7; time between samples: 496-1511 days). Moreover, we controlled for time between samples by calculating the rate of change in LDM and still found significant differences only during LR after RIC (versus ER or late relapse after chemotherapy; KW, p<10-13). Finally, genes with increased LDM were enriched for multiple stem cell gene sets (q<0.01), implicating a common stem-like state in LRs.
Altogether, these data highlight important features of GvL resistance in CLL: 1) GvL selective pressure, shown by LRs, can shape evolutionary trajectories through genotypic alterations that directly exert phenotypic consequences; 2) alterations in HLA genes have less influence in CLL than in myeloid malignancies; and 3) GvL immune editing may select for epigenetic variability that facilitates evasion through stem-like states.
Disclosures
Brown: Octapharma: Consultancy; Novartis: Consultancy; Loxo: Consultancy, Research Funding; Kite: Consultancy, Research Funding; Janssen: Honoraria; Invectys: Other: other; Gilead: Consultancy, Research Funding; Genentech/Roche: Consultancy; Dynamo Therapeutics: Consultancy; Catapult Therapeutics: Consultancy; BeiGene: Consultancy; AstraZeneca: Consultancy; Acerta Pharma: Consultancy; Morphosys: Other: Data safety monitoring boards ; Sun Pharmaceuticals, Inc: Research Funding; Sun: Research Funding; Verastem: Consultancy, Research Funding; TG Therapeutics: Consultancy; Teva: Honoraria; Sunesis: Consultancy; Pharmacyclics: Consultancy; Pfizer: Consultancy. Getz:MuTect, ABSOLTUE, MutSig and POLYSOLVER: Patents & Royalties: MuTect, ABSOLTUE, MutSig and POLYSOLVER; IBM: Research Funding; Pharmacyclics: Research Funding. Ho:Jazz Pharmaceuticals: Consultancy; Jazz Pharmaceuticals: Research Funding; Omeros Corporation: Membership on an entity's Board of Directors or advisory committees. Neuberg:Celgene: Research Funding; Pharmacyclics: Research Funding; Madrigal Pharmaceuticals: Equity Ownership. Soiffer:Gilead, Mana therapeutic, Cugene, Jazz: Consultancy; Jazz: Consultancy; Kiadis: Other: supervisory board; Mana therapeutic: Consultancy; Cugene: Consultancy; Juno, kiadis: Membership on an entity's Board of Directors or advisory committees, Other: DSMB. Ritz:TScan Therapeutics: Consultancy; Equillium: Research Funding; Merck: Research Funding; Kite Pharma: Research Funding; Aleta Biotherapeutics: Consultancy; Celgene: Consultancy; Avrobio: Consultancy; LifeVault Bio: Consultancy; Draper Labs: Consultancy; Talaris Therapeutics: Consultancy. Wu:Neon Therapeutics: Other: Member, Advisory Board; Pharmacyclics: Research Funding.
Longitudinal single-cell transcriptomics reveals distinct patterns of recurrence in acute myeloid leukemia
