Targeting CEA in metastatic triple negative breast cancer with image-guided radiation followed by Fab-mediated chimeric antigen receptor (CAR) T-cell therapy

Abstract

Introduction: Although CAR-T cell therapy has limited efficacy against solid tumors, it has been hypothesized that prior treatment with Image-Guided Radiation Therapy (IGRT) would increase CAR-T cell tumor infiltration, leading to improved antigen specific expansion of CAR-T cells.

Methods: To test this hypothesis in a metastatic triple negative breast cancer (TNBC) model, we engineered two anti-CEA single-chain Fab (scFab) CAR-T cells with signaling domains from CD28zeta and 4-1BBzeta, and tested them in vitro and in vivo.

Results: The anti-CEA scFab CAR-T cells generated from three different human donors demonstrated robust in vitro expression, expansion, and lysis of only CEA-positive TNBC cells, with the CD28z-CAR-T cells showing the highest cytotoxicity. IFN-γ and granzyme B release assays revealed significantly higher IFN-γ production at a 4:1 effector-to-target (E:T) ratio in CD28z-CAR-T cells compared to 4-1BBz-CAR-T cells. Treatment of CEA-positive TNBC MDA-MB231 xenografts in the mammary fat pads of NSG mice, that produced spontaneous lung metastases over time, resulted in significant tumor growth reduction compared to either therapy alone (p<0.01). Immunohistochemical (IHC) analysis revealed that only combined IGRT and CAR-T therapy resulted in the elimination of lung metastases.

Discussion: These findings demonstrate that the combination of IGRT and anti-CEA scFab CAR-T therapy induces a strong antitumor response, effectively targeting both the primary tumor and distant metastatic lesions in the lungs, thus demonstrating that IGRT enhances CAR-T cell infiltration, persistence, and overall efficacy within both primary and metastatic lesions.

Keywords: carcinoembryonic antigen; cell immunotherapy; chimeric antigen receptor T cells; image-guided radiation therapy; triple-negative breast cancer.

Figure 1

Schematics of humanized anti-CEA (M5A) scFab-CAR-T. (A) 4-1BB and (B) CD28 CAR constructs. Two constructs were tested, with cytoplasmic domain corresponding to second generation CAR-Ts. (C, D) The two scFab CAR-T constructs have distinct signaling domains; CD28-CD3zeta and 4-1BB-CDzeta configuration, both co-expressing hCD19t to allow identification of expressed CARs.

Figure 2

scFab-CAR-T production and expression on T cells. (A) Transduction efficiency was assessed by flow cytometry for CD19t expression of transduced T cells following enrichment with anti-CD19 beads. (B) Percentage expression of scFab-CAR-T cell production across T cells from three different donors. (C) Growth curves showing the expansion of scFab-CAR-T from three donors’ post-transduction.

Figure 3

In Vitro scFab-CAR T-cell cytotoxic assays; (A) Cytotoxicity against CEA-negative MDA-MB231 parental cells. (B) Cytotoxicity against CEA-positive MDA-MB231 cells. Both versions of scFab CAR-T cells exhibit antigen-specific lysis of target cells. (C, D) Analysis of IFN-γ and granzyme B production by mock and two versions of scFab CAR-T cells against CEA-positive MDA-MB231 cells. (E) Expression of the CD107a degranulation marker in both CAR-T cells when co-incubated with CEA-positive and negative MDA-MB231 target cells. (Statistical analysis was performed using two-way ANOVA and Sidak’s multiple comparison test *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, ns - not significant).

Figure 4

In Vivo therapeutic Efficacy of Anti-CEA scFab-CD28-CAR-T in combination with 10 Gy of IGRT. (A) Experimental design: orthotopic MDA-MB-231CEA-Luc positive tumors implanted in NSG mice were treated with 10 Gy IGRT on day 10, followed by an intravenous injection of 1 × 10⁶ anti-CEA scFab CAR-T cells on day 11. Tumor size was monitored with calipers until reaching 1500 mm³ (for controls and CAR T groups). (B) MDA-MB-231CEA-Luc tumor growth curves (n=4-5 mice per group). The combination therapy of 10 Gy IGRT and scFab-CAR-T cells was statistically significant (p<0.01). (C) Tumor weight measurements after euthanasia on day 48 showed a significant difference (p<0.0001) in tumor weight between the control and combined treatment groups. (D) Tumor analysis for T cell subpopulations (CD4/CD8) using flow cytometry and their corresponding expression of exhaustion markers in CD4 (E) and CD8 (F) subpopulations. (Statistical analysis was performed using two-way ANOVA and Sidak’s multiple comparison test *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001).

Figure 5

Comparative immunohistochemistry analysis of infiltrating anti-CEA CAR-T cells in MDA-MB231CEA-Luciferase Tumors with or without IGRT. NSG mice with orthotopic MDA-MB-231CEA-Luciferase-positive tumors were analyzed as (A) Tumors treated with CAR-T cells only and (B) Tumors receiving combination treatment of 10Gy IGRT and anti-CEA CAR-T cells. Tumors were monitored until they reached terminal size or until day 48, and then analyzed using immunohistochemistry. Anti-CD3 staining indicates infiltrating anti-CEA CAR-T cells, that were significantly increased in the combination therapy group. (C) Quantification of infiltrating CD3-positive T cells in MDA-MB-231CEA-Luciferase tumors (2 regions per tumor, 5 tumors each group) showed a significant increase in the combination treatment group compared to the anti-CEA CAR-T cell-only group (p < 0.01). (Statistical analysis was performed using student t-test **p < 0.01).

Figure 6

Immunohistochemistry Staining of MDA-MB231CEA-Luc Tumor Metastasis to the Lungs. At termination, one lung lobe from each mouse was collected and stained to detect Luc+ tumor cells (A-D). (E) Quantification of Luc+ areas (average of 5 areas for each lobe) revealed a significant difference between control, single CAR-T treatment, and combination treatment groups (p < 0.001). Each group consisted of 4-5 mice. (Statistical analysis was performed using one-way ANOVA and Sidak’s multiple comparison test *p < 0.05, ***p < 0.001).

Figure 7

In vivo analysis of CAR-T cell tracking kinetics. (A) the treatment schedule schematic, (B) representative bioluminescence scans for CAR-T cell luciferase activity, and (C) the graphical quantification of luciferase in photons per second across the treatment groups. The untreated control group included 2 mice, while the other groups consisted of 3-5 mice each. (Statistical analysis was performed using two-way ANOVA and Sidak’s multiple comparison test **p < 0.01, ns - not significant).