Secreted Factors Determine Resistance to Idelalisib in Marginal Zone Lymphoma Models of Resistance

Background. PI3Kδ is expressed in B-cells and has a central role in the B-cell receptor signaling in B-cell derived malignancies. Idelalisib was the first-in-class PI3Kδ inhibitors and several second-generation compounds are undergoing clinical investigation as single agents and in combinations. To identify modalities to overcome the resistance that develops to this class of agents, we have developed two idelalisib-resistant models derived from splenic marginal zone lymphoma (SMZL) cell lines.

Materials and Methods. Cells were kept under idelalisib (IC90) until acquisition of resistance (RES) or with no drug (parental, PAR). Stable resistance was confirmed by MTT assay after 2-weeks of drug-free culture. Multi-drug resistance phenotype was ruled out. Cells underwent transcriptome and miRNA profiling by RNA-Seq, whole exome sequencing (WES), lipidomics profiling, pharmacological screening (348 compounds), and FACS analysis. Cytokines and growth factor secretion was performed by ELISA.

Results. Two RES models were obtained from VL51 and Karpas1718 with 7-10 fold times higher IC50s than PAR counterparts. In both models, conditioned media from RES cells transferred the resistance in the PAR cells. While WES did not identify somatic mutations associated with resistance, RNA-Seq and lipidomics analyses showed that the two cell lines had developed resistance activating different modalities.

The VL51 RES model showed an enrichment in BCR-TLR-NFkB (TLR4, CD19, SYK), IL6-STAT3 (IL6, CD44), chemokines (CXCL10, CXCR4, CXCR3) and PDGFR (PDGFRA, PRKCE) signatures, paired with increased p-AKT and p-BTK levels, decreased cardiolipins and sphingomyelins levels, and increased levels of specific triacylglycerols and glycerophosphocholines. In particular, there was an over-expression of surface expression of PDGFRA and secretion of IL6 in the medium. Silencing of both IL6and PDGFRA by siRNAs reverted the resistance, while the silencing of the individual genes had only a partial effect. These data were paired with the acquired sensitivity to the PDGFR inhibitor masitinib, identified in the pharmacologic screening.

In the Karpas1718 model, we observed an increased p-AKT activity with an enrichment for B-cell activation signatures (RAG1, RAG2, TCL1A), proliferation (E2F2, MKI67), ERBB signaling (HBEGF, NRG2, ERRB4), increased levels of some triacylglycerols and repressed levels for specific glycerophosphocholines. HBEGF secretion was confirmed by ELISA. The addition of recombinant HBEGF to the medium induced resistance in the PAR cells. Combination with the pan ERBB inhibitor lapatinib was beneficial in the K1718 RES. Recombinant HBEGF also induced resistance to the BTK inhibitor ibrutinib in the PAR cells and in the mantle cell lymphoma SP-53 cell line.

Specific members of the let-7 family of miRNAs were repressed in the RES lines derived from both cell lines, indicating the involvement of miRNA deregulation in the mechanism of resistance. Indeed, let-7 members are known to directly target IL6-STAT3 and cytokine signaling cascade, as well PI3K-AKT network. In solid tumors, let-7 members are also expressed at low levels in tumors with constitutive active ERBB signaling, in accordance with the activation of ERBB pathway and p-AKT we observed in our Karpas1718model. Experiments with a LIN28B inhibitor are now on-going.

Finally, we validated the findings across a panel of 34 B-cell lymphoma cell lines, in which IL6, PDGFRA, HBEGF and LIN28 expression levels were negatively correlated with idelalisib sensitivity, while the latter was positively correlated with let-7 levels (P <0.05).

Conclusions. We developed two distinct models derived from MZL of secondary resistance to the PI3Kδ inhibitor idelalisib. We identified treatments that might overcome resistance to idelalisib and are worth of further investigations. The two models, driven by different biologic processes, will allow the evaluation of further alternative therapeutic approaches.