Abstract

Introduction: The landscape of gene mutations in CLL prior to therapy is well-characterized. Comparatively less is known about gene mutations and their frequency in CLL patients that have relapsed after potent chemo-immunotherapy. Further, despite knowledge of subclonal TP53 mutations that enrich and likely drive CLL relapse in a fraction of cases, a comprehensive profile of gene mutations and their variant allele frequencies (VAFs) and clonal dynamics before and after chemo-immunotherapy in CLL is lacking.

Methods: We have procured paired pre-treatment and post-treatment samples from 53 CLL cases that had relapsed after chemo-immunotherapy and purified CLL CD19+ cells and CD3+ T-cells to purity with FACS. DNA from relapsed CLL was subjected to exome capture and whole exome sequencing (WES) at a mean coverage of 72-fold (range 52-102) and sequence data analyzed using three variant callers: MuTect v.1.1.4, Strelka v.1.0.13, and VarScan2 v.2.3.7. Somatically acquired gene mutations occurring in 2 or more rCLL cases were confirmed by Sanger sequencing in relapsed CLL samples and also re-sequenced in pre-treatment samples. Genes with mutation frequencies ≥5% in rCLL underwent custom gene panel-based deep coverage re-sequencing in paired pre-treatment and post-treatment samples. Analysis of deep re-sequencing data was done using the Broad GATK HaplotypeCaller v3.3.0 in parallel with VarScan2. Selected low-level variants were measured using droplet digital PCR (ddPCR) that was adapted to detection of VAFs as low as 1/10,000.

Results: In CLL relapsed from potent chemo-immunotherapy, we detected mutated TP53, NOTCH1, SF3B1, XPO1, BIRC3, MYD88, NXF1, POT1, CACNA1E, CHD2, EGR2, FAM50A, FAT3, FBXW7, MGA, SAMHD1 and ZMYM3 with frequencies ≥5%. An additional 64 genes were mutated in 2/53 rCLL cases each. We performed ultra-deep panel-based re-sequencing of the 17 genes with frequencies ≥5% in 53 paired diagnosis and relapse samples, complementing selected variants with ddPCR validation to determine VAFs.

TP53 mutations constituted the most frequently enriched gene at relapse (7/53=13%) and the VAFs of all TP53 mutations substantially increased at relapse often from very minor subclones at diagnosis. Importantly, none of the clonal TP53 mutations in rCLL appeared directly induced by chemotherapy, but instead all were selected from pre-existing subclones. Similarly, subclonal mutations in SAMHD1 substantially enriched in four cases at relapse (4/53=8%) suggesting a role in resistance to chemotherapy.

The majority of NOTCH1 mutations (8/13) were already fully clonal at diagnosis without further enrichment at relapse. Three (3/13) subclonal NOTCH1 mutations substantially enriched at relapse, while two (2/13) clonal NOTCH1 mutations substantially decreased. The VAFs for SF3B1 mutations similarly demonstrated three patterns: i) clonal that remained clonal (4/10), ii) clonal that substantial declined and became subclonal at relapse (4/10), and, iii) subclonal that enriched but remained subclonal (2/10) at relapse. Of the 13 remaining genes, most demonstrated no consistent enrichment or depletion or remained subclonal at relapse.

Of biological interest, the genes FBXW7, MYD88, NOTCH1, NXF1, ZMYM3, XPO1, SF3B1 and POT1, were often already fully clonal in the pre-treatment samples, suggesting an early role in CLL pathogenesis rather than a later role in the development of CLL relapse.

Conclusion: In this large WES study focused on gene mutations in relapsed CLL paired with analysis of subclone dynamics using deep panel re-sequencing and ddPCR, we identify the genes TP53 and likely SAMHD1 as drivers of CLL relapse in 20% of cases. Multiple other genes previously implicated as CLL drivers did not consistently enrich at relapse. Further, a subset of the mutated genes was often already fully clonal pre-treatment; these genes likely serve an important role early in CLL pathogenesis that is independent of therapy. The majority of relapsed CLL in this cohort were not associated with the recurrent clonal emergence of known CLL driver mutations and based on the gene mutations frequencies reported here, much larger rCLL cohorts would need analysis to confirm possible additional low frequency gene drivers of rCLL.

Disclosures

Malek:Gilead Sciences: Equity Ownership; Abbvie: Equity Ownership; Janssen Pharmaceuticals: Research Funding.

Author notes

*

Asterisk with author names denotes non-ASH members.