Are radiation and response biomarkers the missing elements for efficacious immunotherapy for glioma patients?

See the article by Ene et al in this issue, pp. 639–651.

The abscopal effect is the phenomenon of tumor shrinkage beyond the immediate vicinity of localized treatment. This effect would be particularly beneficial in widely metastatic disease and in diffusely infiltrating gliomas. The mechanism of action is presumably immunologically mediated by the release of antigens that prime an immune response. There are multiple case reports of the abscopal effect from radiotherapy alone.¹ Emerging data from clinical trials in solid cancers demonstrate that a significant number of patients have radiographic evidence of the abscopal effect when various types of immunotherapy are used in combination with radiation therapy,^2,3 with well over 20 clinical trials now under way in a variety of oncological indications. The study by Ene et al provides preclinical data that abscopal responses can also be induced against gliomas,⁴ which have been particularly refractory to immunotherapy, using the combination of irradiation and anti–programmed cell death ligand 1 (PD-L1) blockade. To prove this, the investigators engineered an innovative bihemispheric fluorescent murine glioma model using the immune-competent replication-competent avian sarcoma-leukosis /nestin leukosis virus receptor (RCAS/Ntv-a) system. This murine model can be used to simulate both low- and high-grade gliomas and various genetic backgrounds, define the optimal immune-modulatory radiation dose, prioritize lead immune-modulatory agents, and clarify the relative doses and schedules of each treatment to inform clinical trial design, thereby providing the scientific community with a valuable resource.

The lack of correlation between immune-therapeutic responses in preclinical murine glioma models and human subjects has impeded meaningful research on the immunology of glioblastoma and optimization of immunotherapy. Clonotypic models such as GL261 express high levels of PD-L1 and have a high tumor mutational burden. Mice engineered to spontaneously develop glioblastoma in situ with the RCAS/Ntv-a system require little or no genetic “second hits” beyond the driver oncogenes (eg, platelet derived growth factor [PDGF]). As such, this model probably lacks immune system–targetable mutations, which are present at a low to moderate frequency in human glioblastoma and which the investigators have compensated for with the introduction of epidermal growth factor receptor variant III (EGFRvIII) expression. Clearly, there still remains an unmet need for preclinical models that more closely recapitulate the immune biology of human gliomas, but the Ntv-a model systems can certainly be used to validate immune-therapeutic approaches^5–7 and to ascertain the induction of therapeutic mechanisms such as the abscopal effect, as described.

Ene et al⁴ mechanistically show that the combination of irradiation and anti–PD-L1 enhances both Iba1+ (ionized calcium binding adaptor molecule 1) macrophage and CD3 T-cell infiltration in the PDGF+EGFRvIII+ glioma microenvironment. The investigators indicated that anti–PD-L1 leads to direct phagocytic activity by macrophages that is extracellular signal-regulated kinase (ERK) dependent, with the abscopal effect being mediated by T cells. A key question arises as to whether these glioma-infiltrating macrophages are directly interacting with the T cells. If this is the initial antigen presentation event occurring in the tumor microenvironment, then the T cells may be capable of effector responses⁸ and may not yet be exhausted, although this was not directly evaluated. In addition to the commonly cited mechanism of antigen release being induced by radiation, the combination of radiation with immune-modulatory agents, especially those that control tumor-mediated mechanisms of immune suppression of the antigen-presenting cell (APC), may induce direct antigen presentation and T-cell activation in the glioma microenvironment, which has previously been considered to occur exclusively in the peripheral lymph nodes. The radiation may be inducing an influx of APCs to the glioma microenvironment, whereas the anti–PD-L1 may be preventing M2 polarization and/or impairment of antigen presentation, thereby allowing activation and effector function of the T cell. This is a mechanism that has not yet been described as a mode of activity for radiation and immunotherapy for any type of malignancy.

Anti–programmed cell death (PD) 1 has been shown to enhance the phagocytic activity of tumor-infiltrating macrophages⁹ but did not trigger a survival advantage in the preclinical murine study, even in the presence of an immunogenic antigen such as EGFRvIII. This indicates that various types of immune checkpoint inhibitors may have different immune-modulatory effects on macrophages such as activation through phosphorylated ERK, antigen presentation, and/or M2 to M1 polarization capabilities, and represents another area for future investigation.

It is very likely that the genetic and baseline immune phenotype in the glioma influences responses to various immune therapeutics, including the susceptibility to the abscopal effect. As such, based on the limited preclinical modeling performed with Ntv-a mice thus far, additional modeling with various common genetic backgrounds such as PTEN and p53 alterations should be evaluated. Ene et al showed that ERK activation in macrophages is induced by PD-L1 blockade and leads to a tumor-phagocytic phenotype.⁴ This is consistent with the established role of mitogen activated protein kinase (MAPK) signaling in the modulation of macrophage development and phenotype. The results presented by this study provide mechanistic insight into how PD-L1 blockade might elicit antitumoral immunity in this context. We recently reported enrichment of activating mutations of MAPK regulators such as BRAF and PTPN11 (protein tyrosine phosphatase non-receptor type 11), which are upstream of ERK, among recurrent glioblastomas that respond to PD1 blockade.¹⁰ Moreover, in lung cancer, activating mutations of KRAS are associated with robust tumor infiltration by macrophages.¹¹ Thus, several lines of evidence implicate MAPK signaling in the modulation of the tumor microenvironment, and specifically tumor-infiltrating macrophages. These observations suggest that MAPK might influence the response to immune checkpoint blockade in glioblastoma. Yet, what remains unclear is how MAPK activation in tumor cells relates to this signaling activation in macrophages.

The lack of significant therapeutic and abscopal effect in mice with only PDGF-induced gliomas indicates that some degree of tumor immunogenicity is required for the abscopal effect to be triggered. Notably, the expression of EGFRvIII is present in only about 30% of glioblastomas,¹² and high tumor mutational burden is present in only 3.5% of glioblastomas.¹³ As such, the cut-points for “immunogenicity” will need to be defined to enrich for those patients who respond to combinatorial strategies, such as the subjects enrolled in NCT03174197. Finally, high PD-L1 expression is detected in 38% of glioblastoma patients¹⁴ in both tumor cells and the myeloid stroma, including macrophages. It is unknown how PD-L1 expression in various cells may or may not influence responses to anti–PD-L1 and radiation therapy. Thus, future directions need to focus on which of these biomarkers, or combination thereof, need to be considered for patient stratification purposes.