Chronic Lymphocytic Leukemia
EHA 2016. Abstract LB2070. Whole exome sequencing of 278 chronic lymphocytic leukemia patients from the CLL8 trial uncovers driver mutations with clinical impact and their evolutionary history in cancer pathogenesis
Eugen Tausch et al.
Results: This resulted in detection of 55 putative CLL driver events (44 sSNVs, 11 sCNVs), many of which have not been previously associated with CLL. At least one potential driver event was found in 91.1% of samples, with 65.4% harboring 2 or more. These putative drivers appeared to affect 8 major cellular processes, including the MAP-ERK pathway (KRAS, NRAS, BRAF, MAP2K mutations; affecting 5.6% of cases), and the MYC pathway (MGA, FUBP1, PTPN1, IRF4 mutations; 8% of cases).
For the majority of samples, two driver events were found within the same sample, one at clonal and the other at subclonal frequency. Such cases enabled us to infer a temporal sequence of acquisition and built a network based on significantly recurrent temporally directed events. Thus we found distinct points of origin restricted to few events such as del13q and tri12, with early convergence to del11q and subsequent divergence in diverse late-occurring driver events including mutated TP53 and KRAS.
We assessed the clinical impact of driver events in 278 patients from CLL8 in the context of front line therapy. Mutated TP53 and SF3B1 cases had significantly shorter PFS and OS. XPO1 (HR 2.02, P =0.02), BRAF (HR 2.05, P =0.05) and EGR2 (HR 2.14, P =0.05) mutations were associated with reduced OS and trend towards shorter PFS. ATM and BIRC3 mutations had no significant impact on outcome. Both the presence of a pre-treatment subclonal driver (66.5% of subjects) and clonal driver (84.2%) were associated with a significantly shorter PFS (HR 1.64, p=0.003 and HR 1.78 p=0.009) with only subclonal drivers achieving significance in the FCR arm.
To directly observe clonal evolution following FC/FCR, we compared the cancer cell fractions (CCFs) of the putative driver alterations in 59 of 278 samples for which a matched relapse samples was available. We observed clonal shifts in 57 of 59 cases of which 21 showed linear and 36 branched evolution with the latter significantly associated with FCR treatment. Neither type of evolution nor presence of the relapsed clone at baseline (n=18) had an impact on outcome. Notably events inferred as early events by our temporal network were stably clonal over time. Late events demonstrated either increasing (i.e. in TP53) or shifting (i.e. in SF3B1, ATM) CCFs after FC/FCR suggesting evolution in relation to therapy (figure). In addition, this temporal map suggests that the copy number loss in 11q precedes point mutations in ATM or BIRC3 in cases with biallelic loss. By contrast, TP53 and 17p show a concordant rise of both events suggesting that biallelic inactivation of TP53 is typically required for fitness advantage.
Conclusion: WES of a large, uniform cohort revealed several novel putative drivers in CLL, enabled their categorization into 8 major functional pathways and investigation of their prognostic impact within the CLL8 trial cohort. Furthermore we developed a network model of CLL evolution with earlier initiating and late accelerating events.