Glioblastoma: the current state of biology and therapeutic strategies

Abstract

Over the past two decades, there have been advances in surgical technologies and chemoradiation strategies for glioblastoma, yet durable remissions are rarely seen. As the biological challenges and genetic basis of glioblastoma have become more understood, new therapeutic strategies may lead to more durable clinical responses and long-term remissions. We believe specialized academic centers that form meaningful corporate partnerships to complement basic science infrastructure and use adaptive clinical trial designs will achieve more rapid translation of innovative approaches to glioblastoma. Here we outline the core biological challenges to be overcome in the management of glioblastoma.

Glioblastoma (GBM) is a molecularly heterogeneous disease and other malignant astrocytomas share many of the phenotypic features noted during the course of this discussion. GBM often grows as a radiographically well-defined focus of disease but is also a microscopically diffusely infiltrative process, often penetrating into eloquent brain regions. As local control has improved with current treatment regimens, more diffusely infiltrative patterns of growth on recurrence are often seen, presenting major clinical hurdles. Determining the best therapeutic options, i.e., options that integrate both “focal” and “global/systemic” tumor targeting, remains a major clinical challenge. Here, we briefly highlight current challenges in GBM, with a focus on IDH wild-type tumors that now define GBM, and propose future therapeutic directions.

The following represent the “core biological challenges” to durable GBM clinical remission that need to be addressed with next generation therapeutics (Figure 1):

1. Infiltrative disease biology

2. Immunosuppressive tumor microenvironment (TME)

3. Tumor and TME heterogeneity

4. Biology of tumor initiation and recurrence

5. Central nervous system (CNS) penetrance of therapeutics:

Figure 1. Core biological challenges to durable GBM remission.

TME, tumor microenvironment; CNS, central nervous system.

Infiltrative disease biology

GBM patterns of infiltration are varied but can range from focal lesions with little infiltration to multi-focal disease that can be bi-hemispheric or involve both supra- and infratentorial structures, even at initial clinical presentation. Standard-of-care focal therapies, including MRI-based stereotactic neuro-navigation surgery and radiation therapy, are techniques that continue to evolve and that have resulted in finite incremental survival increases. However, local therapies are limited in the more diffuse cases. From the surgeon-clinician perspective, this is the major clinical barrier. Therapeutic approaches that limit migration and motility of GBM cells would have benefit by enhancing efficacy of focal therapies.

For more focal lesions, “maximal safe” surgical resection followed by adjuvant chemoradiation is accepted practice, while “supra-total” surgical resection has not yet been established in cases where it may be achievable. Radiation therapy planning is also evolving to incorporate AI and computational methods to predict recurrence regions and allow for the design of supra-total dose “boosts”. Advanced multiparametric MR imaging paradigms have been used to predict recurrence patterns, patient survival, tumor heterogeneity, and molecular features of GBM. The emerging fields of radiomics and radio-genomics that utilize AI-defined MR imaging modalities are now being more fully integrated into clinical practice and have the potential to significantly impact clinical outcomes (1).

Immunosuppressive tumor microenvironment (TME)

One of the major barriers to therapeutic efficacy for GBM is the immunosuppressive TME. Although the GBM TME is rich in macrophages, it is characterized by scant T cell infiltration, and high levels of T cell exhaustion in and expression of inhibitory receptors on those tumor-infiltrating lymphocytes (TILs) that are found, suggesting impaired immune anti-tumor activity (2). The contribution of peripheral and resident immune cells to both antigen presentation and the immune TME in the brain is still poorly defined. Recent work has shed light on the complex cell-to-cell communication dynamics in the GBM TME that allow for greater understanding of this major core challenge, and new methods are now available to dissect the GBM TME (3). We and others have employed a multi-omics approach to study the tumor cell-immune cell landscape. The GBM TME is composed of multiple cell types, including tumor cells, immune cells, neurons, and stromal cells and it seems as if the functional relationships within this TME network will need to be further defined for new therapeutics to achieve therapeutic efficacy.

To date, available studies reveal a complex interplay between GBM tumor cells and tumor-associated macrophages (TAMs), contributing to both mesenchymal cell-state change (4) and the immunosuppressive milieu (5). GBM TAMs tend to exhibit highly specialized, immunosuppressive, tumorigenic phenotypes. There also appears to be a pro-tumorigenic influence by the enriched myeloid compartment on neighboring GBM tumor cells. Recent data shows that GBMs acquire myeloid-affiliated transcriptional programs via epigenetic mechanisms to enhance immune invasion, similar to that observed in mesenchymal glioma stem cells (GSCs) (6). To add additional complexity to the GBM TME, networks between transformed astrocytes and neurons have been identified that may contribute to glioma-genesis and contribute to neuronal pathway dysfunction and neurologic deficits (7).

“Window of opportunity” trials have allowed us to learn that the immunosuppressive GBM TME is remarkably complex. GBM possess both a baseline immunosuppressive TME and an additional, adaptive wave of immunosuppression after CAR T cell therapy (8). Although precise kinetics need to be defined, this adaptive response seems to occur particularly after initial anti-tumor cytocidal activity.

Tumor and TME heterogeneity: Evolution during progression from de novo to recurrent GBM (rGBM)

Importantly, the tumor and immune cell compartments are a mosaic, and a pro-tumorigenic state can result from tumor-TME “cross-talk”. Recent preclinical data shows that immune cells in the TME enhance immune evasion by driving the loss of tumor suppressor genes or by enhancing oncogenic pathways (9). We and others have observed striking heterogeneity in oncogenic mutations in de novo and recurrent GBM, often with genetic profiles that are distinctly different. For example, across the EGFR locus, we have observed poly-allelic mutations, typically in the EGFR extracellular domain (ECD), presumably resulting in greater tendency for homo- and heterodimerization of erbB family proteins and enhanced oncogenic signaling. However, we have observed maintenance of wtEGFR amplification after both standard and EGFR-targeted treatments, which may have biologic significance. Mutations in other receptor tyrosine kinase (RTK) signaling systems are well described, and often co-expressed, further illustrating biologic complexity.

It is clear that in addition to molecular evolution of GBM during treatment, there is also an immune cell and TME evolution in GBM occurring with both standard and experimental treatment. In particular, we have observed enhanced inflammation in the GBM TME in recurrent GBM. Distinct spatial features of recurrent GBM, focused on CD3⁺ T cell densities found in perivascular regions, have been linked to a more effective anti-GBM response and better patient survival. Such differences were not appreciated using assays that did not preserve spatial relationships. Preclinical work looking at macrophage inhibition found recurrent tumors to be enriched in M2 macrophages. This was reinforced with longitudinal patient sample sequencing, which again found increases in the macrophage compartment, specific to the mesenchymal subtype GBM (10).

Biology of Tumor Initiation and Recurrent Disease

GBM long-term control is rare, and recurrence of disease is an inevitable biological reality. Molecular characterization of GBM subtypes has, to date, not resulted in a translational clinical breakthrough or defined a clear pathway to precision-based targeted therapy. Recent data has shown that recurrent GBMs do exhibit changes in both tumor and TME and acquire a more “stem-like” phenotype (11). These data merge with observations suggesting a stem cell origin to glial tumors (GSCs), although dynamic cell states and transcriptional programs may more accurately underlie GBM diversity and treatment response. Recent evidence proposes a role for a fundamental GSC-based neural wound response transcriptional program for GBM development (12). Preclinical data have previously demonstrated that GSC cells are more resistant to chemo- and radiation therapy. Treatment-resistant recurrent GBMs often exhibit a mesenchymal signature. More recent data has shed additional light on the biological basis of GBM recurrence, supporting the view that recurrent GBMs are a very different and distinct molecular and biological entity (13). An important challenge in the field is to develop therapies specific for recurrent GBM that are not based on observations in newly diagnosed tumors. Targeted, personalized therapeutic approaches to either de novo or recurrent tumors will offer the best route to disease control.

CNS penetrance and local environment

In addition to the observed inefficiency for small and large bulky molecules, including chemotherapeutics, tyrosine kinase inhibitors (TKIs), antibodies and antibody-drug conjugates to cross the blood brain barrier (BBB), the heterogenous nature of the GBM TME limits treatment efficacy. A high percentage of GBM tumors are notable for differential hypoxic gradients within the solid tumor core. Penetration and function of small molecules and cell therapies is limited in this environment. Moreover, T cell effector function is further limited in a hypoxic solid tumor environment. Regionally distinct oncogenic signaling also limits treatment efficacy across a tumor volume. Loco-regional therapies have been utilized for bulk flow of molecules across the interstitial space (convection) and more recently for the repeated administration of CAR T cells products into both the tumor cavity and the cerebrospinal fluid in recurrent GBM, with some suggestion that local delivery may be advantageous in some patient cohorts. Additional strategies to overcome the BBB include nanoparticle encapsulation, neurotropic virally mediated delivery, focused ultrasound, and receptor-mediated transcytosis.

Future GBM Treatment:

Recent data outlined above converge to justify developing future approaches to a) modulate stem/mesenchymal GBM cell state; b) target TAMs; c) target unique features of rGBM; and d) optimize immuno-oncology combinations. As more of the complex interplay between GBM and immune cells is defined for specific molecular GBM cohorts, personalized approaches will be more widely integrated into treatment plans.

Advanced diagnostics and novel biomarkers will allow for more accurate disease detection and more precise therapeutic selection. Liquid biopsy using circulating cell-free and tumor DNA may be used to define tumor burden and response to treatment. MRI strategies will continue to improve detection of infiltrative tumor driven by oncogenic pathway activation [e.g., EGFR variant III (EGFRvIII)] and can also stratify higher risk patients. Adaptive clinical trial designs (e.g., GBM AGILE) will allow for faster evaluation of multiple trial agents. We advocate smaller clinical trial cohorts with extensive tissue analysis to inform multiple trials.

Durable responses to targeted agents, e.g., TKIs, have been limited in both de novo and recurrent GBM. At Penn, our bias continues to be immuno-oncology approaches based on expertise and infrastructure at our institution, along with scientific rationale. Our early CAR T cell therapy experience has included de novo and recurrent GBM patients in small cohorts and we have demonstrated safety, CNS penetrance and antigen reduction in GBM tumors with systemic delivery with single antigen targeting (EGFRvIII) CAR T cells.

Immune checkpoint blockade (ICB), as a monotherapy or in combination, has been largely ineffective in GBM disease control, likely due to poor CNS penetrance of full-length antibodies and a lack of TILs to stimulate anti-tumor activity. However, a recent evaluation of patients demonstrating stable disease with anti-PD-1 therapy revealed elevated pERK levels correlated with patient response, suggestive of a potential biomarker for ICB therapy in a cohort of patients (14). There has been some suggestion of efficacy with both mono- and polyvalent vaccine approaches, with a suggestion that neoadjuvant administration may be more beneficial. Limited patients have responded to monovalent CAR T cell approaches, which is noteworthy given the number of lines of treatment such patients had previously received along with the anatomical distribution of recurrent disease. Cell therapy approaches have the advantage of providing effector T cells to the “cold” local GBM TME to yield a cytocidal effect that may be more durable than the cytostatic effect seen with molecularly targeted agents. There is much to be learned with cell therapy trials, including, in part, the optimal valency of antigen targeting, effective combinations that enhance T cell effector function and persistence and the most effective use of allogeneic versus autologous cell products. Loco-regional delivery may be required to complement systemic administration in various GBM cohorts.

From biobank to bioclinic: GBM treatment in the future.

Defining T cell functional status in both de novo and rGBM is likely to be necessary as our early data indicate elevated, but variable, levels of T cell exhaustion in the blood and tumor tissue of GBM patients. Our early efforts involve plasmapheresis of all new GBM patients for our clinical trial programs. In addition to characterization of matched blood, GBM tumor tissue and TME using multi-omics platforms, we believe that routine generation of glioblastoma organoids (GBOs) will allow for real-time clinical testing of chemotherapeutic agents, TKIs, adoptive cell therapies and ICBs. These patient-derived tumor cultures can be tested quickly and maintain over 98% genetic fidelity of GBM-TME heterogeneity. Biomarkers that confer sensitivity and resistance to therapies can be identified quickly and inform adaptive trial designs. Effective drug combinations can be identified particularly in rGBM while patients are undergoing real-time treatment. GBOs will also give clues to the biology of the complex tumor-TME multicellular network. High-fidelity models will allow for more efficacious integration of personalized therapies into clinical practice. Importantly, academic-corporate alliances will be more critical to support rapid testing of novel combinations. Collectively, these efforts will allow for a more seamless workflow between bench and bedside practices that will increase the chances of durable remission in GBM.