Low-dose radiotherapy enhances the efficacy of PD-L1 blockade and induces the abscopal effect

Abstract

Background Low-dose radiotherapy (RT) is a promising treatment likely to increase the efficacy of immunotherapy, including programmed cell death ligand 1 (PD-L1) blockade, in cancer therapy. Further exploration and optimization of such combinatorial strategies are required. Notably, the ability of low-dose RT to enhance the efficacy of immune-checkpoint inhibitors (ICI) in distant, unirradiated tumors is debated. Methods We used a stepwise preclinical approach in immunocompetent mice bearing different murine tumor models (MC38 or CT26), with one or two tumors per mouse. Mice received tumor-only irradiation consisting of either low-dose RT (2x0.5 Gy to 2x2 Gy) or high-dose RT (2x6 Gy to 2x8 Gy) combined with anti-PD-L1. Tumor growth rate and survival were compared across the different conditions. The immune microenvironments of both irradiated and distant unirradiated tumors were characterized using single-cell RNA sequencing. Results We first demonstrated that low-dose RT 2×2 Gy combined with anti-PD-L1 is as effective as high-dose RT 2×6 Gy in delaying the growth of irradiated tumors. Subsequently, we showed that low-dose RT to one tumor enhances the efficacy of anti-PD-L1 consolidation therapy in a distant, unirradiated tumor, thereby inducing an abscopal effect comparable to that observed with high-dose RT. Single-cell RNA sequencing analysis highlighted the polarization of tumor-associated macrophages (TAMs) within distant unirradiated tumors towards a pro-inflammatory phenotype following low-dose RT and anti-PD-L1. Depleting TAMs in distant unirradiated tumors using liposomal clodronate abrogated the abscopal effect driven by low-dose RT combined with anti-PD-L1. Conclusion Our findings demonstrate the ability of low-dose RT to increase the efficacy of ICI in a distant tumor, resulting in a significant abscopal effect, and highlight the critical role of TAMs in the underlying mechanism, as well as a potential immune crosstalk between TAMs and activated lymphoid cells. These data propose low-dose RT as a potential strategy to improve the efficacy of immunotherapy in patients with metastatic solid tumors receiving anti-PD-L1.

Keywords: Abscopal; Immunotherapy; Macrophages; Radiotherapy/radioimmunotherapy; Solid tumor.

Figure 1. Low-dose RT delivering two fractions of 0.5 Gy provides antitumor efficacy in a T cell-dependent fashion. (A) Schematic view of the experimental setting as described in the Methods section. Tumor volumes were measured three times per week. The experiments were repeated twice for C57BL/6 mice and athymic Nu^−/− mice. (B) Mean tumor volumes and (C) survival curves for each group of C57BL/6 mice treated as in the experimental conditions represented in A. Volumes are represented in mm³±SEM. **p 0.01 (two-way ANOVA); ***p<0.001 (log-rank test). (D–E) Clonogenic cell survival assay of MC38 in vitro after (D) single-dose irradiation at various doses or (E) two fractions of 0.5 Gy delivered with different time intervals. Normalized survival fractions are represented in mean±SEM with 10–12 replicates per condition. The experiments were repeated twice. (F) Mean tumor volumes and (G) survival curves for each group of athymic Nu^−/− mice treated as in the experimental conditions represented in A. ns, non-significant (two-way ANOVA/log-rank test). ANOVA, analysis of variance; D, day; h, hours; NIR, non-irradiated; RT, radiotherapy; s.c., subcutaneous; WT, wild type.

Figure 2. Low-dose RT in combination with anti-PD-L1 is as efficient as high-dose RT used alone. (A) Schematic view of the experimental setting according to the methods section. Tumor volumes were measured three times per week. The experiment was repeated twice. (B) Mean tumor volumes in each group of C57BL/6 mice treated as in the experimental conditions represented in A. Volumes are represented in mm³±SEM. ns, non-significant; *p<0.05; **p<0.01 (two-way ANOVA). (C) Individual growth profiles of s.c. MC38 tumors implanted in C57BL/6 mice and treated with tumor-only RT with or without anti-PD-L1 according to the experimental scheme in A. The individual growth profiles are represented separately for each RT regimen, with and without concurrent administration of anti-PD-L1. (D) Survival curves for each group of C57BL/6 mice treated as in the experimental conditions represented in A. ns, non-significant; *p<0.05; ***p<0.001 (log-rank test). (E) Synergy effect of the combination of the different low-dose RT regimens explored with anti-PD-L1. Synergy scores were calculated using the Bliss independence model according to the Methods section. ANOVA, analysis of variance; D, day; i.p., intraperitoneal; NIR, non-irradiated; PD-L1, programmed cell death ligand 1; RT, radiotherapy; s.c., subcutaneous; WT, wild type.

Figure 3. Low-dose RT in association with anti-PD-L1 provides a significant abscopal effect similar to that of high-dose RT. (A) Schematic view of the experimental setting according to the methods section. Tumors were measured two times per week. The experiment was repeated twice. (B–C) Mean volumes of (B) primary tumors and (C) secondary tumors in each group of C57BL/6 mice treated as in the experimental conditions represented in A. Volumes are represented in mm³±SEM. The curves for each group were truncated at the first time point when a sacrifice was performed, ensuring an equal number of mice in each group for calculating the average tumor volume. (D–E) Mean volumes at day 10 after randomization, corresponding to the day of the first sacrifice, for (D) primary tumors and (E) secondary tumors. Volumes are represented in mm³±SEM. ns, non-significant; *p<0.05; **p<0.01; ***p<0.001; ****p<0.0001 (one-way ANOVA). (F) Survival curves for each group of C57BL/6 mice treated as in the experimental conditions represented in A. The first of both tumors reaching 1200 mm³ or the appearance of suffering signs led to sacrifice. ns, non-significant; *p<0.05; ***p<0.001; ****p<0.0001 (log-rank test). ANOVA, analysis of variance; D, day; i.p., intraperitoneal; NIR, non-irradiated; PD-L1, programmed cell death ligand 1; RT, radiotherapy; s.c., subcutaneous; WT, wild type.

Figure 4. Low-dose RT enhances the activation of NK and CD8⁺ T cells within the microenvironment of irradiated tumors. (A) Experimental setting according to the scRNA-Seq protocol described in the methods section. In this figure, results of scRNA-Seq in the lymphoid cells from primary MC38 tumors are reported. (B) Single-cell transcriptomics of the lymphoid (CD45⁺, CD11b⁻) cells from primary MC38 tumors. Uniform Manifold Approximation and Projection plots representing a total of 6511 single cells (points) from all groups, colored by clusters, identified and annotated post hoc according to the Methods section and online supplemental figure S4A. (C) Heatmaps representing the expression of genes commonly used as activation (up) or inhibition (down) features in CD8⁺ T cells (cluster 1). The expression for each gene is represented in z-scores according to the methods section, for each condition without concurrent anti-PD-L1. (D) Heatmaps representing the expression of genes commonly used as activation (left) or inhibition (right) features in NK cells (cluster 5). The expression for each gene is represented in z-scores according to the methods section for each condition without concurrent anti-PD-L1. D, day; i.p., intraperitoneal; NIR, non-irradiated; NK, natural killer; PD-L1, programmed cell death ligand 1; RT, radiotherapy; scRNA-Seq, single-cell RNA sequencing; s.c., subcutaneous; Tregs, regulatory T cells; WT, wild type.

Figure 5. The combination of low-dose RT and concurrent anti-PD-L1 increases the number and cytotoxicity of CD8⁺ T cells in irradiated tumors. (A) Schematic view (left) of the same experimental setting as in figure 4. In this figure, results of scRNA-Seq in the lymphoid cells from primary MC38 tumors are reported. The clustering of lymphoid cell subtypes (right) is the same as in figure 4B. (B) Individual UMAP plots of the lymphoid cells from primary tumors treated as indicated. The annotation of clusters is the same as represented in the reference UMAP plot in A, with~700 representative single cells plotted for each condition. (C–D) Heatmaps representing the expression of the same (G) activating and (H) inhibiting genes in CD8⁺ T cells than those reported in figure 4C, comparing here the different conditions with concurrent anti-PD-L1. (E–F) Differential gene expression of (I) Nkg7 and (J) Gzmk in CD8⁺ T cells from the conditions with concurrent anti-PD-L1. *p<0.05; **p<0.01; ****p<0.0001 (Wilcoxon test with Bonferroni correction). D, day; i.p., intraperitoneal; NK, natural killer; PD-L1, programmed cell death ligand 1; RT, radiotherapy; scRNA-Seq, single-cell RNA sequencing; s.c., subcutaneous; Tregs, regulatory T cells; UMAP, Uniform Manifold Approximation and Projection; WT, wild type.

Figure 6. (complementary to online supplemental figures S6 and S7). ScRNA-Seq analysis of the myeloid infiltrates in secondary MC38 tumors. (A) Identification of the cell subtypes corresponding to the different clusters following the clustering of myeloid cells according to the Methods section. The different clusters correspond to those represented in the top right corner and in online supplemental figure S7B. (B) Volcano plots representing the genes upregulated or downregulated in macrophages (clusters 3, 4, 6, 7, 10, 11 and 12 pooled) between each of the irradiated conditions with concurrent anti-PD-L1 and anti-PD-L1 alone. Among these genes, we highlighted those upregulated (in red) or downregulated (in blue) in the lists by Coates et al or Foster et al exploring M1-like and M2-like macrophages respectively. See the Methods section for further information. (C) Individual expression of important genes for the identification of the different clusters represented in the Uniform Manifold Approximation and Projection plots of 30,707 single cells (points) in the top right corner. NK, natural killer; NPC, neural progenitor cell; OPC, oligodendrocyte precursor cell; PD-L1, programmed cell death ligand 1; aNSC, active neural stem cell; qNSC, quiescent neural stem cell; scRNA-Seq, single-cell RNA sequencing.

Figure 7. Secondary tumor-associated macrophages activated following low-dose RT to the primary tumor and concurrent anti-PD-L1 are crucial for the abscopal effect mediated by low-dose RT. (A) Differential expression of Arg1 in the entire tumor-associated macrophage (TAM) population (corresponding to the sum of clusters 3, 4, 6, 7, 10, 11, and 12 in online supplemental figure S7B) between the indicated conditions. ****p<0.0001 (Wilcoxon test with Bonferroni correction). (B) Single cell transcriptomics of secondary MC38 tumor TAMs overlapping (2×2 Gy+anti-PD-L1) versus (2×8 Gy+anti-PD-L1) conditions. UMAP plots of 5798 single cells (points) colored by treatment group. (C) Experimental setting of macrophage depletion assay in C57BL/6 mice bearing two MC38 tumors, as described in the Methods section. Tumor volumes were measured three times per week. The experiment was repeated twice. (D–E) Mean tumor volumes of (D) primary tumors and (E) secondary tumors for each group of C57BL/6 mice treated as in the experimental conditions represented in C. The volumes are represented in mm³±SEM. ns, non-significant; *p<0.05; **p<0.01; ***p<0.001 (two-way ANOVA). (F) Survival curves for each group of C57BL/6 mice treated as in the experimental conditions represented in C. ns, non-significant; *p<0.05 (log-rank test). Arg1, Arginase 1; ANOVA, analysis of variance; i.p., intraperitoneal; i.t., intratumoral; Lip-clod, liposomal clodronate; NIR, non-irradiated; PBS, phosphate-buffered saline; PD-L1, programmed cell death ligand 1; RT, radiotherapy; s.c., subcutaneous; UMAP, Uniform Manifold Approximation and Projection.