The Tumor Microenvironment – A Big Tent

 We have talked repeatedly about the promise of immuno-oncology, and with good reason. Very recent data show that the landscape of cancer care is changing rapidly and dramatically for the better. We continue to see contributions from diverse therapeutic modalities: immune checkpoint modulation, novel antibodies, bispecifics, CAR T therapy, TCR therapy and others. Massive amounts of resources have poured into this space, and interesting new companies continue to launch in the Boston area: Surface, Unum, Potenza, Enumeral to name just a few.

The last decade has seen intense focus on the immune checkpoint field, and clinical development in that space is encompassing combination therapy as the defining principal to advance treatment (Mahoney et al. NRDD, submitted). While much of the effort is driven toward combining antagonists of T cell immune checkpoints (CTLA4, PD-1, TIM-3, etc) with T cell activators (4-1BB. OX40, CD27, etc), this approach may be self limiting due to the toxicity associated with hyper-activation of T cells (cf. CAR Ts and BiTES) alongside the limitation of targeting just one arm of the immunosuppressive armature deployed by tumors.

Further, we understand that affecting outright cures in more patients is a dramatic step change, and we are not there quite yet, outside of hematology perhaps. It is obvious (we think) that curing cancer will require taking down the infrastructure that supports tumor cell survival, proliferation, resistance and metastasis. For solid tumors and niche-homing leukemias/lymphomas this infrastructure is built on the foundation of tumor cell/stroma interaction, where we define the stroma as extracellular matrix and associated cells – tumor associated macrophages (TAM), myeloid-derived suppressor cells (MDSC), tumor associated fibroblasts, endothelium, other mesenchymal-lineage cells, etc. The composition may vary from indication to indication, with more or less complexity.

Let’s set the stage using three biology buckets:

 Screen Shot 2014-11-23 at 7.23.39 PM

Immuno-oncology focuses intently on tumor cells and tumor-expressed antigens, and on lymphocytes and the mechanisms by which they are shut down by tumor-mediated immunosuppression. Indeed, the current clinical immunotherapy data set is comprised primarily of measured T cell responses, in particular CD8+ T cell responses, with emerging respect for the role of NK cells (see, for example, the lirilumab post). We have not yet developed the readouts needed to accurately assess the role of microenvironmental cells and their soluble mediators but that’s not to say these elements are not be targeted.

It’s certainly true that some of the traditional approaches to cancer treatment impact components of immunity. Cytotoxic therapy (chemotherapy, irradiation) has the demonstrable effect of increasing antigen presentation by inducing tumor cell death and membrane disruption. Chemotherapeutics like cyclophosphamide have been shown in some tumor types to induce the cell surface expression of HLA, a molecule that is often specifically downregulated by tumor cells as a mechanism of avoiding recognition by CD8+ T cells and NK cells. Taxane based chemotherapeutics can induce dendritic cell (DC) activity and induce the local production of pro-inflammatory cytokines. Gemicitabine, until recently the standard chemotherapeutic for pancreatic cancer, induces DC maturation, induces antigen expression on tumor cells, and, of particular interest, reduces the MSDC “burden” within tumors. We stress this last activity as the accumulated data to date suggest that in nearly all tumor settings immunosuppression trumps immune activation, meaning that DC activation, epitope production and cytokine expression is for naught if overriding suppressive signals are not reduced.

Another interesting example is BRAF inhibition. Jennifer Wargo’s lab at MGH has published extensively on this subject. Critical observations from that group and others were recently summarized, and here we quote from their recent review:

“Oncogenic BRAF contributes to immune escape through … establishing an immunosuppressive tumor microenvironment. The administration of a BRAF inhibitor promotes clinical responses along with an increased expression of melanoma-differentiation antigens by malignant cells, an increased tumor infiltration by CD8+ T cells, and a decreased production of immunosuppressive cytokines such as interleukin (IL) -6, IL-8 and IL-1α as well as of the angiogenic mediator vascular endothelial growth factor (VEGF). This phenotype is reverted at time of disease progression. Importantly, the expression of immunomodulatory molecules on T cells (e.g., PD1) and on tumor cells (e.g., PDL1) is also increased within 14 d of BRAF-targeted therapy initiation.” (Cooper et al. 2013. OncoImmunology 2:5, e24320).

This review also makes the very interesting claim that targeting further downstream from BRAF, e.g. with MEK inhibitors, is less beneficial to the immune response because of a direct negative effect on T cells. These data were accumulated in studies of melanoma, but may be more broadly applicable. We’ll come back to BRAF inhibition in but suffice to say there are combination trials of immune checkpoint inhibition with BRAF and MEK inhibitors underway.

Another interesting class of traditional inhibitors with described impact on anti-tumor immune responses are the growth factor targeting agents, notably VEGFR inhibitors. For example sunitinib, a VEGFR inhibitor and a standard therapy for patients with metastatic renal cell carcinoma (RCC) blocks growth factor signaling in tumor and vascular cells to disrupt tumor-induced angiogenesis. Sunitinib also reportedly reduces the accumulation of myeloid-derived suppressor cells (MDSC) and the number of T regulatory (suppressive) T cells within tumors. Diverse types of VEGF pathway inhibitors are now in clinical trials with immune checkpoint inhibitors like nivolumab (anti-PD-1, BMY).

A few days ago I pooled readers for their favorite “tumor microenvironment” targets. The response was interesting for it’s diversity of approaches and companies. Here are some very good ones:

Screen Shot 2014-11-23 at 5.10.32 PM

So we have targets, a company (AstraZeneca), a link to a Science TM paper – all in 144 characters. Also included was the a slide from an AZN presentation about the paper.

Screen Shot 2014-11-23 at 5.16.27 PM

The subject of the paper is a specific tumor type, rhabdomyosarcoma (RMS), the most common childhood soft tissue sarcoma (Highfill et al. 2014. Sci Transl Med 6: 237). In a mouse model, RMS tumors were notable for selective recruitment and/or expansion of CXCR2+ MDSC, which mediate local immunosuppression, while inhibiting or blocking inflammatory CD11b+ macrophages. CXCR2 is a chemokine receptor, and chemokines and their receptors are cell attraction gradients widely used in normal and diseased tissue to “pull” cells into the appropriate environment at the right time. CXCR2 gene-deficiency, or anti-CXCR2 monoclonal antibody therapy, enhanced an anti-PD1 antibody-induced anti-tumor immune response and anti-tumor efficacy, informing the design of the clinical trial mentioned in the slide above. AZD5069 has been profiled extensively in inflammatory diseases, and has quite a good tolerability profile. The GlaxoSmithKline (GSK) inhibitor reparixin blocks binding of the chemokine CXCL8 to both it’s receptors, CXCR1 and CXCR2. GSK has positioned this therapeutic as a method of targeting cancer stem cells, which are inherently mobile and rely of chemokine receptors to exit the primary tumor and find suitable host sites for metastasis. Reparixin has been tested in a variety of chronic inflammatory and tissue injury models like acute lung injury and COPD. More recently, a variety of tumor types are being targeted with this drug in clinical trials. The use of riparixin with chemotherapy is being pursued in breast cancer and pancreatic cancer.

Interest in these targets has also grown with the appreciation that chemokine inhibitors can be used in the context of chemotherapy to “flush” tumor niches like the bone marrow and lymph nodes by disrupting chemokine gradients. A good example of this type of gradient is SDF-1 and CXCR4, a target picked out by Dan Marks aka @Festivus159:

Screen Shot 2014-11-23 at 6.36.37 PM

We’ll come back to that other target (CSF-1R) later. CXCL12 is the official designation for SDF01, mentioned above as the ligand for CXCR4. The approach is validated by the use of plerixafor (AMD3100, Genzyme), a small molecule CXCR4 inhibitor, and ulocupumab (aka BMS-936564/MDX-1338, from BMS), a fully human anti-CXCR4 antibody. These therapeutics are used to treat hematological malignancies, especially in the context of bone marrow invasion by the tumor cells. Plerixafor is approved as Mozobiltm). The bone marrow is an exquisite example of a tumor-supportive niche as it is highly vascularized, rich in adhesion pathways that can be hijacked to induce pro-survival signaling, and readily remodeled to accommodate tumor-specific stromal interactions.

CXCR4 is an interesting target, and there are now multiple companies pursuing diverse therapeutics targeting this receptor. A quick search of the American Society of Hematology meeting abstracts for this year reveal a new antibody from Pfizer and a peptidic antagonist from Eli Lilly in clinical development, with other compounds coming along behind.

A lesser-appreciated role for CXCR4 is articulated in the preclinical and translational medicine literature, in which the critical role of CXCR4 in glioblastoma, ovarian cancer, renal cell carcinoma and other solid tumors in detailed. In this setting the role of CXCR4 as a critical regulator of immune suppression within the tumor microenvironment has been revealed. Specifically, CXCR4 inhibition has been shown to block recruitment of regulatory T cells, block recruitment and retention of MDSC and a newly appreciated stromal suppressor cell, the FAP+ tumor-associated fibroblast, thereby reducing immune suppression. Reviews from Doug Fearon and colleagues cover this biology in detail (e.g. Fearon 2014. Immunol Res2;187).

In part 2 we will look at ideas sent in by @JSWatercooler, @PaulyDeSantis, @WilliamGerber1, @zDonShimoda, @csr1223, @mcbio316, @AZBiomarkers and others.