Lung cancer is the leading cause of cancer deaths, with approximately 20% of patients diagnosed with metastatic disease.[28] Although therapeutic strategies for lung cancer brain metastases have advanced, CNS spread still significantly affects survival and quality of life.[29] While immune checkpoint inhibitors show promise in some cases, most patients do not respond to these therapies, emphasizing the need for new treatments.[30] Challenges in cellular-based approaches for solid brain tumors include the consistent identification of targets.[31]
We used a fully immunocompetent murine model of lung cancer brain metastases and combined a chronic cranial window with repeated in vivo TPLSM to explore real-time dynamics of CAR T-cells at a single-cell level during combined administration of anti-PD1 and CAR T-cells. Our findings demonstrate the efficacy of EpCAM-directed CAR T-cells after intracerebral administration, resulting in a reduced tumor growth and prolonged survival. However, additional systemic anti-PD1 treatment did not increase the intratumoral persistence or the anti-tumor effects of CAR T-cells.
Locoregional injection of CAR T-cells into the surrounding brain tissue not only resulted in a noteworthy reduction in tumor growth but also achieved complete regression in selected cases. Consequently, mice receiving EpCAM-directed CAR T-cells showed significantly prolonged survival rates compared to control animals. These outcomes substantiate the findings established in our prior investigations using this model and are in line with similar observations in different tumor entities, including medulloblastoma or ependymoma[16],[32].
For T-cell suppression within the brain TME the PD-1/PD-L1 axis has been show to play a pivotal role.[33, 34] Antigen contact induces CAR T-cell effector function and production of IFN-γ. Next, IFN-γ binds to its receptor initiating the JAK/STAT signaling pathway, which regulates PD-L1 expression on brain tumor cells and tumor associated macrophages.[35] Accordingly, first clinical data indicates that anti-EGFRVIII-CAR T-cell infusion can paradoxically promote immunosuppressive tumor microenvironment via upregulating inhibitory immune checkpoint molecules in glioblastoma.[36] Interestingly, a phase I clinical trial investigating repeated peripheral infusions of anti-EGFRvIII CAR T cells in combination with pembrolizumab was not effective in glioblastoma.[37]
Several publications highlight the CNS penetrance and effectiveness of ICB antibodies in brain metastasis. Anti-PD1 treatment may reverse the immunosuppression within the TME and CNS tumors have been shown to respond to combined immune checkpoint blockade, resulting in elevated proportions of tumor-infiltrating lymphocytes (TILs).[27, 38, 39] Within the context of CAR T-cell treatment PD-1 suppression can be achieved through the co-administration of PD-1 targeting monoclonal antibodies or the PD-1 gene editing of CAR T-cells.[40]
In systemic tumor models, the additional value of PD-1 blockade to increase CAR T-cell efficacy has been debated. In a murine preclinical model for systemic melanoma, concurrent PD-1 blockade notably increased the persistence and efficacy of CAR T-cell treatment.[41] However, in another study using a immunocompetent murine model for systemic melanoma, PD-1 blockade primarily mediates its anti-tumor effect through endogenous T-cells and did not increase the anti-tumor effect of CAR T-cell treatment.[42] Furthermore, it has been demonstrated, that PD-1 silencing may impair the anti-tumor function of CAR T-cells by inhibiting proliferation activity in a murine model of systemic NSCLC.[43] Song et al. demonstrated that anti-EGFRvIII CAR T-cell therapy with PD-1 checkpoint blockade in a CNS tumor model using U87 glioma cells.[44] However, it's noteworthy that these experiments were conducted in immunodeficient mice, which may overlook the influence of endogenous T-cell immunity and an intact PD-1–PD-L1 signaling axis. In our study, we used autologous spleenocytes for CAR T-cell production and observe no significant difference in intratumoral densities of CAR T-cells and CD3+T-cells, nor in CAR T-cell persistence and survival following the co-administration of anti-PD-1 and CAR T-cells. Transduction with the retroviral CAR vector endows CAR T-cells with dual specificity via the CAR and the endogenous T-cell receptor (TCR). Although CAR T-cell-based therapies are recommended for the treatment of hematological malignancies, the effects of endogenous TCR signaling in CAR T-cell biology have not been well defined. Recent preclinical and clinical studies suggest that endogenous TCR signaling is not required for CAR T-cell effector function, whereas it could negatively affect proliferation and effector function[45, 46].
Another potential mechanism contributing to the limited efficacy of the combinatorial approach is the complex composition of the tumor microenvironment (TME) within the brain which frequently harbors fewer proliferating immune cells compared to primary tumors and other metastatic sites. Additionally, T-cells in brain metastases exhibit elevated expression levels of immune checkpoint proteins compared to those in other sites, while macrophages in the brain are more prone to expressing an immune-suppressing M2 gene signature. These factors collectively contribute to impeding the effectiveness of CAR T-cells in CNS tumors.[47–49]
By utilizing repetitive in vivo TPLSM we are capable of elucidating intratumoral CAR T-cell dynamics from early stages of tumor formation until late timepoints, when large tumors have formed. After intracerebral injection of CAR T-cells, we initially observed higher intratumoral densities of EpCAM-directed CAR T-cells compared to undirected CAR T-cells. In general, CAR T-cells recognize surface antigens independently from MHC restriction. Based on the intracranial administration, early contact of EpCAM-directed CAR T-cells with EpCAM-transuced LL/2 tumor cells may lead to receptor-antigen-interaction inducing activation, proliferation and the development of a cytotoxic phenotype.[50, 51] Interestingly, intratumoral CAR T-cell density and proliferation diminished during the observation period indicating insufficient CAR T-cell persistence within the tumor. Consequently, a decreasing amount of EpCAM-directed CAR T-cells was paralleled by tumor growth. Consistent with our data, several preclinical and clinical studies in other solid brain tumors observe decreasing CAR T-cell numbers and T-cell exhaustion even when a sufficient T-cell infiltration has been achieved.[52, 53] Immunologically, large tumor burden requires persistent CAR T-cell function upon repeated antigen stimulation in an immunosuppressive environment to eventually achieve tumor eradication. However, chronic antigenic stimulation by the tumor results in endogenous T-cell exhaustion characterized by loss of lytic function and cytokine secretion with simultaneous expression of inhibitory receptors like PD1/PD-L1[54, 55]. Consequently, we sought to elucidate the impact of concomitant anti-PD1 treatment on CAR T-cell migration and effector function. Surprisingly, we do not observe any differences in CAR T-cell migration to and persistence within the tumor after anti-PD1 treatment. In line with that, no survival differences could be observed between animals receiving ICB and the isotype control antibodies, respectively. In general, anti-PD-1 antibodies mainly function by disrupting the interaction between PD-1 on T-cells and PD-L1 on tumor cells. Paucity of PD-L1 on tumor cells is a well-defined factor associated with resistance to anti-PD-1 antibody treatment while high expression usually indicate better response rates.[56–58] However, the PD-L1 expression varies among patients and between different tumor entities.[59] Furthermore, the upregulation of alternative immune checkpoints or the activation of alternative signaling pathways within tumor cells may contribute to resistance. For instance, tumor cells may exploit pathways other than the PD-1-PD-L1 axis to evade immune surveillance. Liu et al. engineered CAR T-cells by modifying PD-1, incorporating the extracellular and transmembrane domains of PD-1 with the intracellular signaling domain of CD28. This adaptation facilitated the transformation of inhibitory signals within the TME into activating signals.[60] The resultant 'switch-receptor' CAR T-cells exhibited enhanced efficacy in tumor control compared to the concurrent administration of anti-PD-1 with CAR T-cells. The conversion of multiple inhibitory signals within the TME to stimulatory signals holds significant potential for improving anti-tumor cytotoxicity. This is particularly noteworthy given the elevated expression of checkpoints, including PD-1, LAG-3, TIM-3, and TIGIT, along with their ligands, in solid brain tumors.[61] Converting the ubiquitous inhibitory signals into stimulatory signals can thereby greatly improve CAR T-cell infiltration and persistence and has to be investigated in further studies.
Although CAR T-cell therapy shows promising results in B cell malignancies, CNS affection is a common exclusion criterion in clinical trials mainly driven by fear of neurotoxicity. Additionally, most CARs targeting solid tumors use antigens shared by normal tissues, carrying the risk of on-target off-tumor toxicity. The additional use of ICB theoretically increases efficacy while also increasing the risk of toxicity. Such side effects most frequently comprise neurological symptoms, epileptic seizures, systemic immune reactions like the cytokine release syndrome (CRS), organ dysfunction and death[62, 63]. Amongst others, the CAR construct in our model is constituted by an scFv capable of recognizing murine EpCAM in most epithelial tissues. As a pan-epithelial marker, EpCAM is homogenously expressed on the surface of healthy alveolar tissue[64]. Due to shared expression on the surface of tumor cells and healthy tissue, the risk of on-target-off-tumor reactions is significantly increased in the context auf EpCAM-directed CAR T-cells and anti-PD1 treatment[65]. Notably, we did not observe any clinically relevant side-effects in our fully immunocompetent mouse model. Nevertheless, it remains to be mentioned that especially due to the small sample size and the translational nature of our experimental set-up we cannot fully predict on on-target/off-tumor reactions of our combinatorial approach.
In conclusion, we demonstrated that locally injected CAR T-cells adjacent to the tumor lead to intratumoral accumulation and reduced tumor growth translating into a survival benefit of EpCAM/GFPCAR T-cell treated mice. Even though additional anti-PD1 treatment was safe and well-tolerated, it does not elicit unconstrained proliferation or intratumoral persistence.