Targeting of colony-stimulating factor 1 receptor (CSF1R) in the CLL microenvironment yields antineoplastic activity in primary patient samples

Abstract

In many malignancies, the tumor microenvironment includes CSF1R-expressing supportive monocyte/macrophages that promote tumor cell survival. For chronic lymphocytic leukemia (CLL), these supportive monocyte/macrophages are known as nurse-like cells (NLCs), although the potential effectiveness of selective small-molecule inhibitors of CSF1R against CLL is understudied. Here, we demonstrate the preclinical activity of two inhibitors of CSF1R, GW-2580 and ARRY-382, in primary CLL patient samples. We observed at least 25% of CLL samples showed sub-micromolar sensitivity to CSF1R inhibitors. This sensitivity was observed in samples with varying genetic and clinical backgrounds, although higher white cell count and monocyte cell percentage was associated with increased sensitivity. Depleting CD14-expressing monocytes preferentially decreased viability in samples sensitive to CSF1R inhibitors, and treating samples with CSF1R inhibitors eliminated the presence of NLCs in long-term culture conditions. These results indicate that CSF1R small-molecule inhibitors target CD14-expressing monocytes in the CLL microenvironment, thereby depriving leukemia cells of extrinsic support signals. In addition, significant synergy was observed combining CSF1R inhibitors with idelalisib or ibrutinib, two current CLL therapies that disrupt tumor cell intrinsic B-cell receptor signaling. These findings support the concept of simultaneously targeting supportive NLCs and CLL cells and demonstrate the potential clinical utility of this combination.

Keywords: chronic lymphocytic leukemia; colony-stimulating factor 1 receptor; small-molecule inhibitors; tumor microenvironment; tumor-associated macrophages.

Figure 1. Ex vivo inhibitor screening reveals CSF1R sensitivity in CLL patient specimens

A. Mononuclear cells isolated from peripheral blood or bone marrow of CLL patients were added to 384-well plates containing dose-escalating concentrations of small-molecule inhibitors. Following incubation for 72 hours, the relative number of remaining viable mononuclear cells was evaluated by subjecting cells to a colorimetric cell viability assay. B. GW-2580 and ARRY-382 are highly specific small-molecule inhibitors of CSF1R (and not other class III receptor tyrosine kinases). C.-D. CLL primary patient specimens were exposed to C. GW-2580 and D. ARRY-382, as described in A., and dose-response curves for each specimen were included along with an average dose-response curve for all specimens. E.-F. Waterfall plot of the IC50 values for each patient specimen after exposure to E. GW-2580 and F. ARRY-382. The IC50 was calculated from the dose-response curve using a cubic logarithmic regression, and each specimen was positioned in order of increasing IC50.

Figure 2. No genetic or clinical characteristic readily co-segregate with sensitivity to CSF1R inhibition in CLL patient samples

The 197 CLL patient specimens that were evaluated by ex vivo inhibitor screening in Figure 1 were ordered by increasing AUC for GW-2580, which was calculated using a cubic logarithmic regression model. Various demographic, clinical, and genetic/cytogenetic characteristics of each patient were determined (the continuous variables are broken into quartiles) and each characteristic was evaluated for statistical significance (see Supplementary Figure 2 and Supplementary Tables 1-2).

Figure 3. CSF1R is not found on CD19+ CLL cells but instead expressed on a CD14+ myeloid subpopulation

A.-B. Mononuclear cells isolated from CLL patients were subjected to flow cytometry using antibodies specific for CSF1R and CD19, with CSF1R expression A. not observed in CD19+ lymphocytes (CLL cells) but B. observed in a subpopulation of CD14+ cells. C. Sensitivity to CSF1R inhibitors, as determined in Figures 1 and 2, was correlated with percentage of CD14-positive cells as determined in Figure 3B. Statistics were calculated using Spearman’s rank correlation.

Figure 4. Sensitivity of CLL cells to CD14+ depletion correlates with sensitivity to CSF1R inhibitors

A. CD14+ cells were depleted from patient specimens using magnetic cell-separation columns, and incubated in 384-well plates for 72 hours. The CD14+ depletion sensitivity ratio was calculated by comparing the relative remaining numbers of viable cells in depleted versus non-depleted conditions using a colorimetric assay. The ratio of cell viability readings in depleted to non-depleted cells at 72 hours was normalized to the same ratio at the start of the experiment to control for variance in cell input. B. The CD14+ depletion sensitivity ratio was generated for a panel of primary CLL patient samples as described in A.. This ratio was compared to GW-2580 and ARRY-382 AUCs. Statistics determined by Spearman’s rank correlation. C. Primary CLL patient samples cells were exposed to CSF1R inhibitors and were subjected to long-term culture conditions to produce nurse-like cells (NLCs). The number of NLCs was quantified using a hemocytometer. D. For one primary patient sample that did not produce NLCs, the addition of CSF1R inhibitors did not have a significant impact on cell viability compared to untreated control.

Figure 5. Synergy between ibrutinib or idelalisib and CSF1R inhibitors in majority of CLL patient specimens

A. Mononuclear cells from CLL patient specimens were cultured with dose gradients of single-agent CSF1R inhibitors, ibrutinib, or idelalisib, as well as equimolar ratio dose gradients of CSF1R inhibitors combined with ibrutinib or idelalisib. After a 72-hour incubation, relative numbers of remaining viable cells were assessed using a colorimetric cell viability assay, and synergy calculations were generated from the dose-response curves. B.-E. A hierarchically clustered heat map was generated showing the combination indices at increasing concentrations of inhibitors (rows) in CLL patient samples (columns) for B. GW-2580 with ibrutinib; C. ARRY-382 with ibrutinib; D. GW-2580 with idelalisib; and E. ARRY-382 with idelalisib. The single-agent sensitivity (IC50) to each drug used in the combination is included (depicted as a heat map) below the corresponding patient sample.