The bone marrow stromal microenvironment comprises “niches” that promote survival and quiescence of LSC and, together with molecular features intrinsic to LSC or bulk leukemic cells, may contribute to therapeutic resistance. AML patients treated with FLT3 inhibitors have had better clinical responses in the peripheral blood than in the bone marrow, suggesting that the marrow niche may represent an important resistance mechanism in this setting. We have shown that FLT3 is a valid therapeutic target in MLL-rearranged (MLL-R) ALL, and FLT3 TKIs are being tested in newly diagnosed infants with MLL-R ALL. We hypothesized that: 1) marrow stroma may provide a protective niche for MLL-R ALL LSCs, and 2) disrupting the LSC-stroma interaction may render MLL-R ALL LSCs more vulnerable to FLT3 TKI.

N=28 primary childhood ALL samples (9 infant MLL-R, 21 other) were split into 2 aliquots, one cultured in medium only, one in medium + normal human bone marrow feeder layers. Survival of viable cells (defined as CD19+, annexin−) was quantified as: (# of viable cells at baseline)/(# at 48 hrs). Surviving fraction in medium was similar for MLL-R group (56%) vs. others (42%, p=0.23). On stroma, surviving fraction was greater for MLL-R group (106%) vs. others (75%, p=0.03). The effect of stroma co-culture on FLT3 TKI (lestaurtinib)-mediated 48 hr MTT cytotoxicity was determined on 3 MLL-R and 3 other samples. Stroma was protective for the MLL-R samples at all 6 dose levels (5 – 100 nM). At 50 nM, e.g., the mean OD (% control) was 25% (medium) vs. 47% (stroma, p=0.03). A protective effect was not seen for the other samples. Thus, stromal interactions selectively promote the survival of MLL-R ALL cells in absence and presence of FLT3 TKI.

The relative contribution of physical interaction vs. production of soluble factors by stromal cells was addressed using N=3 MLL-R samples, each cultured in 3 conditions: in physical contact with stroma (S); with leukemia cells and stroma separated by a permeable Transwell membrane (T); and medium only (M). WST-1 and annexin binding assays were done @ 4, 24 and 48 hrs −/+ 5–100 nM TKI. Mean OD (% control) in untreated wells in S/T/M conditions were: 257%/172%/97% @ 24 hr, and 321%/229%/95% @ 48 hrs. Annexin+ fraction in S/T/M conditions (untreated) was: 12%/22%/21% @ 48 hrs. Mean OD (% control) in 50 nM TKI in S/T/M conditions were: 172%/86%/88% @ 24 hr, and 163%/110%/71% @ 48 hr. Thus: S conditions enhanced proliferation, reduced apoptosis, and abrogated TKI-mediated cytotoxicity. These effects in T conditions were generally less striking, and absent in the case of apoptois. While soluble factors are able to mediate some effects of stroma on MLL-R ALL cells, physical interaction is required for maximal effect.

The effects of the S/T/M conditions on the LSC (as opposed to bulk leukemia cells) were then assessed: 48 hr S/T/M ex vivo culture (−/+ 20 nM TKI) was followed by in vivo xenografting in NOD/SCID mice (tail vein injection, 6 cohorts, N=6 per cohort). Marrow engraftment of ALL cells was assessed after 10 weeks (reflecting LSC activity). Mean engraftment (% human ALL cells in marrow) with S/T/M conditions was 96%/74%/34% (no TKI) and 87%/44%/7% (20 nM TKI) (p<0.05 for all comparisons except S @ 0 nM vs. S @ 20 nM), showing that S is superior to T, and T superior to M, in promoting survival of LSC ex vivo, an effect that is particularly pronounced under the pressure of FLT3 TKI.

The combination of G-CSF and the CXCR4 antagonist AMD3100 potently mobilize CD34+ stem progenitors. We tested this combination as a therapeutic strategy to disrupt the stroma/LSC interaction and improve the efficacy of FLT3-targeted therapy in MLL-R ALL. Cohorts of 10 NOD/SCID mice were injected with N=2 primary MLL-R ALL cells (5 mice per sample), allowed to engraft for 2 weeks, then treated with 1 of 4 regimens: vehicle control (C), G-CSF-AMD3100 (G-A), lestaurtinib (L), or G-CSF-AMD3100 + lestaurtinib (G−A + L). At week 10, mice were sacrificed and assessed for marrow engraftment of ALL cells. Mean engraftment for C/G−A/L/G−A+L was 94%/90%/69%/23% (p<0.05 for all comparisons except C vs. G−A), showing that the G-CSF/AMD3100 combination potently enhance the efficacy of FLT3 inhibitors against MLL-R ALL LSCs.

Thus, stromal-leukemia interactions are protective for MLL-R ALL bulk cells and LSCs and represent an important mechanism of resistance to FLT3-targeted therapy. Disruption of these interactions with stem cell mobilizing agents may be able to overcome therapeutic resistance and improve the efficacy of FLT3 TKI.

Disclosures: No relevant conflicts of interest to declare.

Author notes

Corresponding author