Activating MAPK Pathway Mutations Mediate Primary Resistance to PI3K Inhibitors in Chronic Lymphocytic Leukemia (CLL)

Resistance to PI3K inhibitors including idelalisib remains poorly understood. The present study was designed to identify mechanisms underlying PI3K inhibitor resistance in CLL through WES, RNAseq and biochemical confirmation. We performed WES on matched tumor and germline samples from 28 relapsed CLL patients treated with PI3K inhibitors (idelalisib, 79% (22/28), pilaralisib 10.7% (3/28), voxtalisib 10.7% (3/28)). The median number of prior therapies was 4.5 and the adverse cytogenetic aberrations del(17p) and del(11q) were seen in 21% and 32% respectively. A median of two longitudinal tumor samples were sequenced per patient (N_samples= 68). An average of 23 ± 12 (range = 4-58) non-silent somatic mutations were identified in the initial sample. The non-silent mutation rate was 0.64 ± 0.32 per MB (range = 0.11-1.61) and typical of CLL. Significantly mutated genes (based on MutSig2CV) included SF3B1 36%, TP53 21%, DDX3X 14% and BRAF 11%. The cohort was sub-divided into responders (R) (N_cases= 18, N_samples= 47) and non-responders (NR) (N_cases= 10, N_samples= 21). Among the 10 NR, 7 had been treated with idelalisib and 3 with voxtalisib. Comparison of the mutational landscape between R and NR showed that, notably, mutations in BRAF(N=5), MAP2K1(N=2), KRAS(N=2),XPO1(N=1), PLEKHA1(N=1), INPPL1(N=1) and NXF1(N=1) were exclusively found in NR. Moreover, MAPK pathway genes (KRAS, BRAF and MAP2K1) were the only recurrent somatic mutations present exclusively in NR. We inferred the cancer cell fraction (ccf) for each somatic mutation using ABSOLUTE and found that the ccf values for MAP2K1 (p=0.01), XPO1 and PLEKHA1 mutations increased with time, indicating a positive selection pressure. Phylogic analysis of 6 R and 4 NR with at least 3 serial samples identified a median of 2 dynamic subclones (change of at least 10% between any two timepoints) and showed enrichment of the "Reactome_RAF_MAP_Kinase_Cascade" pathway in NR (p=3x10^-4, q=0.04). In addition, Gene Set Enrichment Analysis comparing RNAseq samples drawn prior to or after PI3K therapy in five of the patients from our cohort (1 primary NR and 4 progressors after response) demonstrated activation of the "Reactome_Prolonged_ERK_Activation_Events" pathway at the later time-points (p=2.3x10^-4, q= 0.03).

Given these findings, we further investigated the role of the ERK/MAPK pathway in idelalisib resistance. Western blot analysis of ERK1/2 phosphorylation in ex vivo idelalisib treated PBMCs from an initially responding patient who acquired resistance later showed that idelalisib failed to inhibit ERK1/2 phosphorylation only at the resistant timepoint. This persistent ERK1/2 activation was inhibited by combination treatment with the MAP2K1 inhibitor CI-1040. We then profiled pAKT and pERK relative to total, by western blot on ex vivo idelalisib-treated samples (5 R, 5 NR). Idelalisib treatment ex vivo inhibited AKT phosphorylation in both R and NR, but no reduction in ERK phosphorylation was seen in NR, in contrast to R.

Based on these findings we hypothesized that increased or persistent ERK activity confers resistance to PI3K inhibitors and tested the impact of MAP2K1 mutations on ERK activity. Among NR, we identified two different mutations in MAP2K1, E203K and Q56P (2/10 NR vs 0/18 R), both previously described as activating in solid tumors. In a separate cohort of multiply relapsed CLL patients, we found 4 cases of MAP2K1 F53L mutations, also previously reported in solid tumors. We studied the functional significance of these MAP2K1 mutations by generating stable MEC1 CLL cell lines expressing HA tagged E203K, F53L or Q56P MAP2K1 mutants. We found that all mutant cell lines had elevated basal ERK1/2 phosphorylation (E203K p<0.001; F53L p<0.001; Q56P p<0.001). Moreover, idelalisib did not inhibit IgM-induced ERK1/2 phosphorylation in the mutants when compared to the controls expressing wild type MAP2K1 (E203K p<0.001; F53L p<0.001; Q56P p<0.001). Preliminary results using the ERK1/2 inhibitor SCH772984 showed that the combination of SCH772984 along with idelalisib effectively decreased ERK1/2 phosphorylation in the mutant cell lines compared to idelalisib alone.

Taken together, these data implicate the ERK pathway in idelalisib resistance and suggest that ERK1/2 inhibitors either combined with idelalisib at therapy start or added to idelalisib at early progression might sensitize patients to PI3K delta therapy.