Erlotinib Increases Efficacy of 5-Azacitidine (AZA) by Inhibiting Drug Efflux Via ABC-Transporters

We and others showed that the TKI erlotinib (Erlo) has in vivo and in vitro efficacy in MDS and AML (Boehrer et al., Blood, 2008) and clinical studies defining more closely its therapeutic benefit are underway in MDS, including by our group (NCT00977548, NCT01085838). We tried to define in this preclinical study the potential interest of combining Erlo to hypomethylating agents, which have become reference treatments in MDS.

Erlo (10mM) was combined to AZA, (2mM) or decitabine (DAC, 2mM) and apoptosis over-time (24, 48, 72h) quantified by DioC₃(6)/PI staining in MDS- and AML-derived cell lines (MOLM-13, KG-1, HL-60) and ex vivo patient (pt) MDS and AML cells. To fully evaluate the extent of Erlo-induced sensitization, efficacy was compared to apoptosis induced by AraC (100nM), doxorubicine (Dox, 100nM), VP-16 (1mM) +/− Erlo. Single drug dosages had been chosen since they caused little to no apoptosis. Efflux via P-gp (by staining with DioC₂³), MRP (by calcein) and was quantified by FACS, specific biochemical efflux inhibitors (CSA, MK-571, KO143) served as controls, intracellular retention of mitoxantrone (MTZ) assessed overall functionality of ABC-transporters. Surface expression of P-gp was quantified by FACS.

Whereas co-incubation of Erlo with DAC did not increase apoptosis in any of the myeloid cell lines, combination of Erlo with AZA exhibited synergistic effects already observed at 24h (HL-60: Erlo: 3%, AZA: 10%, Erlo+ AZA: 30%), and increasing over-time (72h: Erlo: 6%, AZA: 15%, Erlo+ AZA: 40%). Noteworthy, the degree of apoptosis observed by combining Erlo+ AZA was comparable to cell death induced by 1mM VP-16, and largely exceeded apoptosis observed with 100nM Dox and 100nM AraC in the respective cell lines. To determine if Erlo also impacts on apoptosis in MDS- and AML-derived cells, we first screened 5 pt samples (with MDS or AML) for apoptosis observed under AZA and DAC, and demonstrated that, in all samples, AZA-induced apoptosis exceeded DAC-induced apoptosis (by up to 40% at 24h). Noteworthy, the “peak” of apoptosis in sensitive pt cells occurred at 24h and – under single agents - did not increase over-time. Co-incubation with Erlo and AZA was able to induce synergistic effects already at 24h (AML post MDS: Erlo: 10%, AZA: 33%, Erlo+ AZA: 55%) and apoptosis increased over-time. In addition, whereas malignant CD34- cells already showed a high rate of “spontaneous” apoptosis (that is in the absence of any drug, about 50% of apoptosis at 48h), CD34+ blasts not only exhibited a lower rate of spontaneous apoptosis (remaining below 15% at 72h), but also a considerable degree of chemoresistance (48h: no increase of apoptosis under AraC, Dox or VP-16). Testing if the combination of Erlo+ AZA also targets this resistant population, we showed that it is particularly efficient in CD34+ progenitors (inducing 80% cell death at 72h), exceeding efficiency of all other tested agents/combinations (AraC, Dox, VP-16, Dec+/−Erlo). Hypothesizing that this pattern of sensitization might be due to a decreased drug efflux via ABC-transporters (overexpressed on malignant myeloid progenitors), we tested this concept on cell lines and ex vivo pt cells. We found that erlotinib can inhibit drug efflux via P-gp (as evidenced by increased intracellular retention of DioC₂³) and MRP (increased retention of calcein). Notably, Erlo increased intracellular accumulation of MTZ in MDS- and AML-derived CD34- and CD34+ blasts, and exhibited - as single agent - a comparable degree of efflux inhibition as the combination of all three biochemical inhibitors. To more closely define the advantageous effect of Erlo+AZA in pt cells, we quantified drug-induced changes in P-gp expression separately in CD34+CD38dim and CD34dimCD38+ blasts. We found that Erlo down-regulates P-gp surface expression (decrease of 50% at 48h), and that this down-regulation is more pronounced in the immature CD34+CD38dim population than in the CD34dimCD38+ blasts.

Our study shows erlotinib's ability to reverse the drug-resistance phenotype in malignant blasts and its interesting combination with AZA. We also provide novel evidence that Erlo (combined to AZA) can target chemoresistant myeloid progenitors by diminishing functionality as well as surface expression of ABC-transporters. Those results suggest a potential clinical interest of combining erlotinib to AZA in MDS

Fenaux:CELGENE, JANSSEN CILAG, AMGEN, ROCHE, GSK, NOVARTIS, MERCK, CEPHALON: Consultancy.