Immune Modulation with RANKL Blockade through Denosumab Treatment in Patients with Cancer

Abstract

Denosumab is a fully human mAb that binds receptor activator of NFκB ligand (RANKL). It is routinely administered to patients with cancer to reduce the incidence of new bone metastasis. RANK-RANKL interactions regulate bone turnover by controlling osteoclast recruitment, development, and activity. However, these interactions also can regulate immune cells including dendritic cells and medullary thymic epithelial cells. Inhibition of the latter results in reduced thymic negative selection of T cells and could enhance the generation of tumor-specific T cells. We examined whether administering denosumab could modify modulate circulating immune cells in patients with cancer. Blood was collected from 23 patients with prostate cancer and 3 patients with renal cell carcinoma, all of whom had advanced disease and were receiving denosumab, prior to and during denosumab treatment. Using high-dimensional mass cytometry, we found that denosumab treatment by itself induced modest effects on circulating immune cell frequency and activation. We also found minimal changes in the circulating T-cell repertoire and the frequency of new thymic emigrants with denosumab treatment. However, when we stratified patients by whether they were receiving chemotherapy and/or steroids, patients receiving these concomitant treatments showed significantly greater immune modulation, including an increase in the frequency of natural killer cells early and classical monocytes later. We also saw broad induction of CTLA-4 and TIM3 expression in circulating lymphocytes and some monocyte populations. These findings suggest that denosumab treatment by itself has modest immunomodulatory effects, but when combined with conventional cancer treatments, can lead to the induction of immunologic checkpoints. See related Spotlight by Nasrollahi and Davar, p. 383.

Figure 1.

Modulation of circulating myeloid cells with denosumab treatment. A, Schema outlining when patients received denosumab treatment and when blood samples were drawn. Blood samples were drawn prior to denosumab treatment. B, A UMAP plot of cells from all PBMC samples analyzed with Scaffold using data from CyTOF. Clusters correspond to canonical immune cells and are represented by distinct colors. C, Scaffold maps of the frequency of myeloid cell clusters are shown comparing pretreatment time point (TP) 1 (n = 16) with TP3 (n = 14). Black nodes represent landmark nodes of myeloid cell types identified by traditional manual cell gating. Remaining nodes represent unsupervised myeloid cell clusters created by scaffold analysis. Nodes are arranged and connected around each other, and the distance between connected nodes depends on cluster similarity. Cluster size is proportional to cellular abundance. The color of the nodes indicates statistically significant differences (q < 0.05; red = increase, blue = decrease, gray = not significant) between the two time points. There are no nodes with blue or red indicating that there are no myeloid populations that are significantly changed in frequency with treatment. D, Heat maps summarizing log₂ fold changes from statistical analysis of functional markers CD137, CD27, CD40, CD40L, CD71, CD95, CTLA-4, HLA-DR, ICOS, Ki-67, NKG2D, OX40, PD-1, PD-L1, PD-L2, TIGIT, and TIM3 when comparing pretreatment time point TP1 (n = 16) with TP3 (n = 14) and TP4 (n = 7). Each box in the heat map represents a myeloid cell cluster that was labeled according to the nearest landmark node they connect to on the scaffold map and ordered by cell count abundance. The pseudocolor represents clusters that showed a significant difference (q < 0.05) in the log₂ fold change, with red being significantly higher, while blue being significantly lower in group 2 of the two comparison groups.

Figure 2.

Modulation of circulating immune effectors with denosumab treatment. A, Scaffold map of significant changes in the frequency of lymphocyte cell subsets. Clusters are shown comparing pretreatment TP1 (n = 16) with TP4 (n = 7). Scaffold was used to analyze PBMC data obtained from CyTOF. B, Heat maps summarizing log₂ fold changes from statistical analysis of functional markers CD137, CD27, CD40, CD40L, CD71, CD95, CTLA-4, HLA-DR, ICOS, Ki-67, NKG2D, OX40, PD-1, PD-L1, PD-L2, TIGIT, and TIM3 when comparing pretreatment time point TP1 (n = 16) with TP3 (n = 14) and TP4 (n = 7). Each box in the heat map represents a lymphocyte cluster that was labeled according to the nearest landmark node they connect to on the scaffold map and ordered by cell count abundance. The pseudocolor represents clusters that showed a significant difference (q < 0.05) in the log₂ fold change, with red being significantly higher, while blue being significantly lower in group 2 of the two comparison groups.

Figure 3.

Changes in the peripheral T-cell repertoire with denosumab treatment. A, TCR β chain sequencing was performed on longitudinal samples from patients. Clonotypic frequencies of a representative patient are shown with pretreatment time point (TP1) on the horizontal axis and posttreatment time point (TP2) on the vertical axis. Clonality (B) and TCR convergent frequency (C) were calculated and compared across the time points using the Wilcoxon signed-rank test [TP1 (n = 25), TP2 (n = 14), TP3 (n = 15), TP4 (n = 12)]. In these box and whisker plots, the box covers the interquartile range, the horizontal line that splits the box is the median, and the whiskers capture the minimum and maximum.

Figure 4.

Effects of treatment on recent thymic emigrants. DNA isolated from PBMCs was assessed by qPCR for changes in copies of TRECs. A, Comparisons of TREC copies were done across time points for all patients (n = 14). Statistical analyses were calculated using the Wilcoxon signed-rank test (mean ± SEM). B, TREC copies were compared between patients who received chemotherapy, steroids, or both (combination, n = 4) to patients who did not receive chemotherapy or steroids (denosumab only, n = 10). Statistical analyses were calculated using the Wilcoxon signed-rank test between time points and the Wilcoxon rank-sum test between groups (mean ± SEM).

Figure 5.

Immune modulation in patients with or without chemotherapy (chemo)/steroids. A, Heat maps summarizing log₂ fold changes from statistical analysis of functional markers CD137, CD27, CD40, CD40L, CD71, CD95, CTLA-4, HLA-DR, ICOS, Ki-67, NKG2D, OX40, PD-1, PD-L1, PD-L2, TIGIT, and TIM3 when comparing pretreatment time point TP1 (chemotherapy/steroids, n = 5; Denosumab only, n = 11) with TP2 (chemotherapy/steroids, n = 6; denosumab only, n = 10) and TP3 (chemotherapy/steroids, n = 6; denosumab only, n = 8) in patients received chemotherapy, steroids, or both (top) and patients who did not receive chemotherapy or steroids (bottom). Each box in the heat map represents a lymphocyte cluster that was labeled according to the nearest landmark node they connect to on the scaffold map and ordered by cell count abundance. The pseudocolor represents clusters that showed a significant difference (q < 0.05) in the log₂ fold change, with red being significantly higher, while blue being significantly lower in group 2 of the two comparison groups. Heat maps were generated using statistical scaffold to analyze PBMC data obtained from CyTOF. TCR clonality (chemotherapy/steroids, n = 6; denosumab only, n = 18; B) and TCR relative clonality (chemotherapy/steroids, n = 8; denosumab only, n = 18; C) were calculated and compared between patients who did and did not receive chemotherapy or steroids across the time points using the Wilcoxon rank-sum test. In these box and whisker plots, the box covers the interquartile range, the horizontal line that splits the box is the median, and the whiskers capture the minimum and maximum.