CICON18 Day 1 Update: T Cells and the Tumor Microenvironment

“This is not the meeting where you’re going to hear the readouts of late-stage clinical trials. Rather, it is about scientific discovery, immunological mechanisms, and basic biology—the essential grist that will enable the field to move forward in its quest to develop personalized immunotherapy that benefits more cancer patients,” Jill O’Donnell-Tormey, PhD, the chief executive officer and director of scientific and medical affairs at the Cancer Research Institute (CRI), said during her opening remarks at this year’s International Cancer Immunotherapy Conference (CICON18), co-hosted by the Cancer Research Institute (CRI), the Association for Cancer lmmunotherapy (CIMT), the European Academy of Tumor Immunology (EATI), and the American Association for Cancer Research (AACR).

Jill O'Donnell-Tormey, CEO and director of Scientific Affairs at Cancer Research Institute

Immune-based therapies, especially those targeting the PD-1/PD-L1 checkpoint pathway, have greatly improved our ability to treat several types of advanced cancers in the clinic over the past decade. Unfortunately, as O’Donnell-Tormey stated, further breakthroughs will be needed if we’re going to help all patients with cancer. To that end, Day one of CICON18 highlighted a variety of new approaches being explored, including many that are being combined with the PD-1/PD-L1 immunotherapies that have had significant success in the clinic.

SESSION 1: Regulating T Cells and Their Response to Cancer

Given the importance of the PD-1/PD-L1 pathway, it was fitting that the first speaker in the first session was Emory University’s Rafi Ahmed, PhD, who received the 2017 William B. Coley Award for his work that revealed the therapeutic potential of targeting this immune checkpoint. Ahmed first focused on a population of CD8+ T cells that express PD-1 and TCF and exhibit stem-like properties. In contrast to conventional T cells, these stem-like, self-renewing T cells that give rise to effector T cells don’t circulate throughout the body, but are instead largely restricted to lymphoid tissues, and while they express high levels of co-stimulatory molecules, they also have diminished expression of inhibitory receptors and effector molecules. After showing that these stem-like T cells had distinct gene expression profiles compared to the memory CD8+ T cells that arise after acute viral infection, Ahmed proposed that they represent a “specific adaptation of CD8+ T cells to chronic antigen stimulation [that] are crucial for maintaining T cell immunity under conditions of chronic stimulation.”

Rafi Ahmed, PhD, of Emory University

During the second half of his talk, Ahmed focused on combining PD-1 and IL-2 immunotherapy as well as how this approach impacts different types of T cells. Interestingly, he found that only the stem-like T cells appeared to reap the benefits of treatment-induced proliferation, which led to a 251-fold increase in T cells in circulation compared to a 2.5-fold increase after PD-1 treatment alone. This combination approach also led to increased expression of genes necessary for migration into tumors and other infected tissues. In fact, whereas PD-1 treatment by itself resulted in the generation of terminally differentiated effector T cells, the combination treatment produced effector T cells with unique transcriptional and epigenetic signatures that resembled those found in the setting of acute infection.

The second talk of the day, by Christian Ottensmeier, MD, PhD, of the University of Southampton (UK), looked at T cells in the clinical setting, specifically the tissue-resident memory (TRM) cells that are believed to be at the center of tumor control. After pointing out that TRM features—such as expression of the PD-1, LAG-3, and TIM-3 immune checkpoints as well as the co-stimulatory 4-1bb—are enriched in “hot,” immune-infiltrated tumors, Ottensmeier noted the prognostic value of the TRM phenotype. Interestingly, the presence of CD103+ antigen-presenting dendritic cells (DCs) predicted patient survival even better than the presence of “killer” T cells did. In tumors that contained these DCs, the clonality of the T cell repertoire was also expanded compared to those that didn’t. He went on to discuss global features of TRM cells, like expression of inhibitory molecules, that exist within both lung tumors and normal lung tissue, as well as TRM features that are unique to tumors. Using high-throughput single cell transcriptomics, he was even able to identify a novel TRM subset—characterized by the expression of TIM-3 and the absence of the IL-7 receptor—that was enriched in patients who responded to PD-1 immunotherapy.

Christian Ottensmeier, MD, PhD, of University of Southampton (UK)

Next, W. Nicholas Haining, B.Ch., B.M., of the Dana-Farber Cancer Institute, discussed the use of genetic screens to discover new targets for immunotherapy. In addition to TNF signaling and activation of NF-κB, Haining revealed the importance of a protein known as ADAR (Adenosine Deaminase that acts on RNA) that interferes with the sensing of, and subsequent immune responses against, double-stranded RNA. In a mouse model of melanoma, he showed that silencing the translation of ADAR sensitized tumors to immunotherapy, at least in part by reshaping the tumor microenvironment. In particular, anti-Adar1 sgRNA led to a significant decrease in M2 macrophages and a corresponding increase in both monocytes and CD8+ T cells. In response to IFN stimulation, these Adar-deficient tumor cells, which exhibited decreased editing of double-strand RNA, also expressed higher levels of IFNγ and IFNβ as well as decreased growth, even without immunotherapy, compared to controls. Furthermore, both PKR-induced growth inhibition and MDA5/MAVS-induced IFN production were shown to be sufficient to sensitize tumors to PD-1 immunotherapy. Notably, loss of Adar was also able to overcome immunotherapy resistance in β2M-deficient mouse tumor cells, which “still recruit effector T cell populations even in the absence of CD8+-mediated recognition of tumor cells.” Instead, in this case it appeared that CD4+ T cells and natural killer (NK) cells were responsible for the elimination of tumor cells.

W. Nicholas Haining, B.Ch., B.M., of the Dana-Farber Cancer Institute

The following speaker, Nicholas P. Restifo, MD, of the National Cancer Institute, highlighted another potential target—the elevated levels of potassium ions found within tumors—that could be used to improve the effectiveness of immunotherapy. In doing so, he also sought to address the “central paradox” of tumor-infiltrating T cells: “why are they dysfunctional and yet inherently capable of stem cell-like behavior?”

Nicholas P. Restifo, MD, of the National Cancer Institute

Essentially, when tumor cells die, in both mice and humans, they release their potassium ions into the tumor interstitial fluid, which results in hyperelevated levels that can inhibit—albeit reversibly—the effector function of T cells. Further investigation revealed that increased potassium concentrations can restrict glucose uptake and subsequently lead to autophagy in T cells, in addition to silencing expression of IFNγ. This caloric restriction was accompanied by an increase in mitochondrial oxidative phosphorylation, which depleted citrate and acetyl-CoA within T cells. Interestingly, whereas high intratumoral levels of potassium interfere with T cells’ immediate ability to attack cancer cells through epigenetic silencing of their effector program, this ion imbalance can actually benefit tumor-targeting T cells when it comes to making them more “metabolically fit” as well as increasing their persistence and potential anti-tumor activity. In fact, Restifo found that elevated potassium levels “preserve memory-like phenotype” and leads to increased expression of the same TCF protein that Ahmed found to be associated with stem-like T cell populations. This led Restifo to an innovative strategy that involved using hyperelevated potassium levels to “condition” T cells into a stem-like state prior to transferring them into melanoma-bearing mice, where they were then “liberated” and exhibited an enhanced ability to eliminate tumor cells.

Following Restifo, Roberta Zappasodi, PhD, of Memorial Sloan Kettering Cancer Center, showcased a mechanistic rationale for combining PD-1 blockade with a GITR-activating immunotherapy. First, she discussed how targeting GITR alone influenced immune activity and antitumor immunity in humans, showing that it primarily affects GITR+ regulatory T cells (Tregs) in the peripheral blood and effector Tregs in tumors in much the same way. Next, Zappasodi revealed the benefits of combining PD-1 and GITR immunotherapy in an immunotherapy-resistant model of mouse melanoma. When treated with anti-GITR alone at four days after injection of tumor cells, only 6/20 mice survived, and when either PD1 or GITR immunotherapy was administered occurred at seven days post-injection, which allowed the tumors to become established, no more than 1/20 mice survived. In contrast, however, the combination enabled the survival of 11/20 mice—even when they weren’t treated until seven days after injection. These results led to the launch of a phase I clinical trial in which this combination is being evaluated in patients with advanced solid tumors.

Roberta Zappasodi, PhD, of Memorial Sloan Kettering Cancer Center

Next, Jake S. O'Donnell, a doctoral student working with Michele Teng, PhD, and Mark J. Smyth, PhD, at the QIMR Berghofer Medical Research Institute (Australia), spoke about the benefits of neoadjuvant, or pre-surgical, immunotherapy. In a study using the 4T1.2 metastatic mouse cancer model, combined PD-1 and CD137 (4-1bb) immunotherapy conferred superior benefits when administered prior to surgery, compared to after. In more recent work that involved taking tumor-specific T cells from immunotherapy-treated mice and then adoptively transferring the T cells to different tumor-bearing mice, O’Donnell showed that the tumor-specific T cells generated in mice that received immunotherapy in the neoadjuvant setting were much better at controlling growth compared to the T cells that came from mice treated in the adjuvant setting. He also found that the extent of T cell expansion immediately after neoadjuvant immunotherapy correlated with the long-term survival of mice, and this association was observed in the context of four different immunotherapy combinations (PD-1 and CTLA-4; PD-1 and CD137; PD-1 and CD40; PD-1 and IL-2). O’Donnell’s team also identified a biomarker on proliferating T cells that correlated with long-term survival, and demonstrated that IL-2 could help rescue mice that lacked this biomarker and weren’t predicted to respond to the neoadjuvant immunotherapy.

The 2018 William B. Coley Lecture

Between the first and second sessions on Day 1, the crowd at CICON18 was treated to the 2018 William B. Coley Lecture, which was given by Padmanee Sharma, MD, PhD, of the University of Texas MD Anderson Cancer Center, who was recently named a recipient of this year’s William B. Coley Award from the Cancer Research Institute. Dr. Sharma detailed her efforts to bridge the clinic and the lab by investigating the mechanisms of response and resistance to checkpoint immunotherapy. Sharma first revealed that while CTLA-4 checkpoint immunotherapy expands CD4+ and CD8+ T cell populations within human melanoma, prostate, and bladder cancer, it did not deplete FOXP3+ regulatory T cells (Tregs) within these tumors; in fact, it increased Treg density. She then launched into a discussion of the importance of integrating laboratory and clinical research, and how the fruits of each process enable new avenues in the other. Whereas hypothesis testing is an efficient way to investigate the significance of a single variable at a time, and thus fits more with the reductionist nature of basic laboratory experiments, generating a hypothesis is more reasonably carried out as part of the discovery-driven approach that is applied in the clinic. She also stressed the need to re-think how clinical trials are designed in order to ensure that adequate samples are obtained for further analysis in the laboratory. The biomarker analyses that can be performed in pre-surgical or tissue-based trials can potentially reveal mechanistic insights that, if properly observed and understood, could be used to radically improve the effectiveness of immunotherapy approaches. According to Sharma, this need to properly integrate laboratory and clinical work “required that we developed a translational immunotherapy research program.” As of September 2018, that program is involved in more than 100 ongoing clinical trials that have enrolled more than 3,000 patients and collected and analyzed more than 6,000 blood samples, 3,000 tumor samples, and 40,000 tumor tissue slides, all in the hopes of identifying an important biological signal amid the noise.

Padmanee Sharma, MD, PhD, of the University of Texas MD Anderson Cancer Center

Next, Sharma discussed her pioneering efforts involving pre-surgical treatment with CTLA-4 immunotherapy in patients with localized bladder cancer. In addition to increased infiltration of T cells into tumors after treatment, the expression of a number of genes associated with T cell stimulation and activity were increased. Most strongly upregulated was the ICOS pathway. Beyond bladder cancer, Sharma demonstrated that pre-surgical treatment of prostate cancer patients with the combination of CTLA-4 immunotherapy and hormonal therapy was able to convert “cold” tumors into “hot” tumors characterized by high expression of the PD-1 and VISTA immune checkpoints. This led to preclinical work in which PD-1 was targeted along with CTLA-4 in a prostate cancer model as well as a trial in patients with metastatic, castration-resistant prostate cancer. Lastly, Sharma ended her talk by focusing on other resistance mechanisms. Two prime examples were: copy number alterations in the IFNγ pathway of melanoma cells that were associated with resistance to CTLA-4 immunotherapy, and an increased EZH2 expression in CD4+ T cells after CTLA-4 immunotherapy. After demonstrating the anti-tumor effects of EZH2 inhibition in mice, her team has now planned a clinical trial to test this approach in humans.

SESSION 2: Tackling the Tumor Microenvironment

Following the previous session’s focus on T cells, the next session during Day 1 at CICON18 shined a light on the other non-cancer cells that make up the tumor microenvironment (TME). First up was Genentech’s Shannon J. Turley, PhD, the recipient of the 2017 Frederick W. Alt Award. Turley’s talk centered mostly on the role of TGFβ in the TME. In patients with metastatic bladder cancer who were treated with anti-PD-L1 immunotherapy, TGFβ and its receptor were significantly associated with lack of response and lower overall survival, especially in patients with immune-excluded tumors in which T cells reside in the stromal matrix architecture at the periphery of the tumor. In this stroma, “killer” CD8+ T cells interact with both carcinoma-associated fibroblasts (CAFs) and collagen fibers. Importantly, the signature of TGFβ-stimulated fibroblasts is also associated with non-responsiveness in the context of PD-L1 immunotherapy. Using a mouse tumor model that exhibits the immune-excluded phenotype, Turley showed that combining PD-L1 immunotherapy with a TGFβ-targeting treatment was capable of inducing tumor regression that corresponded with a significant increase in “killer” T cell gene expression and infiltration into tumors. These effects appear partly due to the inhibition of SMAD2 phosphorylation, which occurred after TGFβ treatment both alone and in combination with PD-L1 immunotherapy, particularly in the non-immune cells. While TGFβ treatment did decrease the expression of TGFβ response genes in fibroblasts, the combination therapy didn’t decrease the TGFβ response signatures in either T cells or macrophages. However, the combination therapy did decrease the cross-linking of mature collagen in tumors, which could partly explain the switch from the immune-excluded phenotype to the immune-inflamed phenotype characterized by increased infiltration by T cells.

Shannon J. Turley, PhD, of Genentech

The next speaker, University of Chicago’s Jeffrey A. Hubbell, PhD, sought to take advantage of the unique characteristics of tumor microenvironments in order to make current immunotherapies more effective. First, he showcased an approach in which checkpoint immunotherapy antibodies (targeting PD-L1 and CTLA-4) were conjugated with an extracellular matrix-binding peptide—PIGF-2123-144—that caused them to be retained locally within tumors after intratumoral injection. As a result, they caused far fewer systemic side effects, including a decreased incidence of type 1 diabetes, and even led to superior immune responses against breast tumors in mice compared to normal, unconjugated antibodies targeting those same immune checkpoints. Then, he unveiled another strategy to improve the effectiveness of immunotherapy, one that takes advantage of the “leakiness” of tumor blood vessels and their exposed collagen fibers. By conjugating antibodies with a collagen-binding domain (CBD), it enabled them to localize more effectively within tumors and was associated with decreased liver damage in mice. Compared to unconjugated antibodies, intravenuous injection of CBD-conjugated antibodies that targeted the PD-L1 and CTLA-4 immune checkpoints were able to slow the growth of a mouse melanoma model (B16F10) and increase the ratio of CD8+ effector T cells to Tregs within tumors in addition to restoring the cytotoxic capabilities of these T cells as evidence by increased expression of IL-2, TNF, and IFNγ. This approach, which also led to a decrease in myeloid-derived suppressor cells (MDSCs) within tumors, also proved effective in a mouse model of breast cancer (PyMT). Finally, he also demonstrated the improved characteristics of CBD-conjugated IL-2 protein and a CBD-conjugated version of the CCL4 chemokine. Unlike normal IL-2, the conjugated version caused neither the weight of the spleen not the water content of the lung to increase. CBD-IL-2 also mediated superior anti-tumor effects, slowing the growth of mouse melanoma and colorectal tumors compared to normal IL-2. Furthermore, treatment with a combination of CBD-PD-L1 and CBD-IL-2 suppressed tumor growth to a much greater extent than the combination of the two unmodified versions of these therapies. In the case of CCL4, which helps to attract DCs and T cells, CBD-CCL4  therapy, when used in combination with checkpoint immunotherapy, slowed tumor growth and improved survival in mice with melanoma. The combination of all three modified therapies—CBD-PD-L1, CBD-IL-2, and CBD-CCL4—also proved superior compared to the triple combination of the normal treatments. In this case, no tumor escape was evident in any of the mice treated with the combination of the three CBD-conjugated therapies.

Jeffrey A. Hubbell, PhD, of University of Chicago

The Icahn School of Medicine at Mount Sinai’s Miriam Merad, MD, PhD, who was recently named a recipient of the 2018 William B. Coley Award, spoke next and focused on the work that earned her the award—the myeloid cell lineages that play important roles in the tumor microenvironment. Unfortunately, little is known about their diversity in tumors and their distinct contributions to T cell-mediated immune responses. By mapping these cells and their genetic transcripts, she strives to uncover crucial information about their origins, identities, and behaviors, which she hopes will shed light on their roles in the development of diseases such as cancer as well as enable the prioritization of potential targets for immunotherapy. Focusing on early-stage lung cancer, she used unbiased analysis of the immune cells within tumors using ScRNA sequencing to get a better grasp of the complexity of the myeloid cell populations within tumors. In addition to working to identify markers that provide insight into the distinct phenotypes and activities of myeloid clusters, she also discussed her efforts to explore the impact of these myeloid cells in the clinic through a phase I “window of opportunity” trial involving patients with lung and liver cancer who were treated with neoadjuvant PD-1 immunotherapy. Through analysis of analogous mouse models of lung cancer, she was able to identify mice correlates of the myeloid cell clusters found in humans. Now, she’s utilizing this knowledge to map the fate of blood-derived macrophages that infiltrate lung tumors, and found that embryonic, tissue-resident macrophages—which do not result from adult hematopoiesis—are the first cells to interact with tumor cells prior to their redistribution around tumor lesions. These tissue-resident macrophages were found to support cancer in several ways: through early induction of Tregs; by reducing T cell infiltration into tumors; and through promotion of epithelial-to-mesenchymal transition and early dissemination to distant tissues. After identifying the human equivalent of these tissue-resident macrophages, she’s now using protein barcoding in conjunction with CRISPR screens to further characterize the myeloid compartment within tumors at the single cell level.

Miriam Merad, MD, PhD, of Icahn School of Medicine at Mount Sinai

Natural killer T (NKT) cells represent another population that can interact with tumors and influence their progression or elimination, as Leonid S. Metelitsa, MD, PhD, of the Baylor College of Medicine, pointed out in the following talk. These cells share properties of both NK cells and T cells (hence their name) and are able to recognize glycolipids presented by the CD1d surface receptor, and their presence within tumors has been associated with improved outcomes in patients with stage 4 neuroblastoma. As a result, Metelitsa and his colleagues sought to enhance the tumor-fighting capabilities of these cells by equipping them with chimeric antigen receptors (CARs) targeting the GD2 protein that’s often overexpressed in neuroblastoma. These NKT cells were capable of targeting both GD2-expressing tumor cells as well as CD1d-expressing M2 macrophages that can support and protect tumors. Compared to CAR T cells, these CAR NKT cells trafficked to tumors more effectively and were associated with significantly less graft-versus-host-disease (GvHD) and other off-target side effects. To further enhance their effectiveness in cellular immunotherapy approaches, both the endodomain of the co-stimulatory CD28 receptor as well as IL-15 were incorporated into the CAR construct, enabling these engineered NKT cells to expand after transfer as well as to accumulate and persist within metastatic lesions. Additionally, a method for expanding these CAR NKT cells under cGMP conditions was developed, enabling their use in a phase I trial for children with neuroblastoma. Thus far, one patient has been treated, and there was evidence that these CAR NKT cells expanded in the peripheral blood of this patient and also infiltrated into the tumor and bone marrow. While this trial utilizes patient-derived, or autologous, NKT cells, the lack of therapy-related toxicity in preclinical models suggests that donor-derived NKTs could feasibly be safe and effective in the clinic too.

Leonid S. Metelitsa, MD, PhD, of Baylor College of Medicine

The following speaker, Dmitry I. Gabrilovich, MD, PhD, of The Wistar Institute, explored the two distinct types of neutrophils found within the tumors of lung cancer patients, and revealed how their presence and activity depended on the stage of a tumor. One type, referred to as polymorphonuclear-MDSCs (PMN-MDSCs) are associated with late-stage tumors and characterized by strong inflammation, production of reactive oxygen species, and NF-κB activity. These PMN-MDSCs are highly immunosuppressive and appear to promote the growth of both primary and metastatic tumors. In contrast, the other type of neutrophil is similar to PMN-MDSCs (PMN-MDSC-like cells), associated with early-stage tumors, and characterized by increased glycolysis and oxidative phosphorylation. These PMN-MDSC-like cells aren’t very immunosuppressive, but are capable of potent spontaneous migration and appear to promote tumor seeding and the formation of the initial metastatic niche.

Dmitry I. Gabrilovich, MD, PhD, of The Wistar Institute

Next, Wei-Wu Tom Chen, MD, of the National Taiwan University Hospital, discussed his efforts to classify soft tissue sarcomas (STS) based on their immune contexture. Through analysis of more than 500 cases involving the three most common histologies—leiomyosarcoma (LMS), dedifferentiated liposarcoma (DDLPS), and undifferentiated pleomorphic sarcoma (UPS)—he showed that these sarcoma immune clusters (SIC) were associated with distinct phenotypes and differential patient survival. The C1 and C2 clusters, which comprised roughly 35% of cases, were “hot” tumors characterized by strong immune cell infiltration, including T cells and myeloid cells, and were associated with the longest overall survival. The A1 and A2 clusters, which comprised roughly 50% of cases, were “cold” immune deserts, and leiomyosarcoma predominated in the A1 cluster, which had slightly fewer immune cell signatures compared to the A2 cluster. Finally, the B cluster, which comprised roughly 15% of cases, was characterized by high infiltration of endothelial cells and intermediate infiltration of neutrophils, with all other immune cells registering low signatures. Importantly, the C1 and C2 clusters also exhibited increased expression of CXCL13, which is important for the formation of tertiary lymphoid structures. Consequently, TLS formation was most highly associated with the C1 and C2 clusters, with the highest frequency in C2—all tumors that had TLS were either in the C1 or C2 cluster, and all of the cases in the C2 cluster were TLS-positive. Given that the presence of TLS is associated with increased infiltration of immune cells and good prognosis in many cancer types, it makes sense that these C1 and C2 clusters represented the cases with the longer average overall survival.

Wei-Wu Tom Chen, MD, of the National Taiwan University Hospital

Finally, the University of California, San Francisco’s Kevin C. Barry, PhD, wrapped up day one at CICON18 with a discussion of the cross-presenting dendritic cells (DCs) in the tumor microenvironment that can stimulate T cells. High levels of these stimulatory DCs (SDCs) are able to predict higher overall survival in melanoma, and their presence also correlates with responsiveness to PD-1 immunotherapy. However, in order to tap into their full immunotherapeutic potential, Barry sought to determine what stimulates these SDCs, and found that SDC levels correlated with the concentration of the FLT3LG within human melanoma tumors. After finding an association between the levels of SDCs and NK cells, which form close and stable contacts with SDCs in tumors, he then confirmed in humans that NK cell levels correlate with levels of intratumoral FLT3LG as well as SDC levels. Furthermore, he observed that higher NK cell levels were associated with increased overall survival and increased responsiveness to PD-1 immunotherapy. Levels of NK cells and SDCs within tumors were shown to be the best correlate of responses to PD-1 immunotherapy, providing a significant incentive to identify novel therapeutic agents that can enhance the activity of NK cells, and subsequently promote SDC differentiation and survival.

Kevin C. Barry, PhD, of University of California, San Francisco

That’s it for day one of CICON18. Check back later for our recap of the highlights from day two!

All photos by Arthur N. Brodksy for the Cancer Research Institute.