Cancer immunotherapy/ Immunothérapie du cancer

Seminar organized by Antonio Coutinho from 8th to 13th of May 2023

Participants

Andrea ABLASSER (EPFL, Lausanne, Suisse), Rafi AHMED (Emory University, Atlanta, USA), Vassiliki BOUSSIOTIS (Harvard University, Boston, USA), Thiago CARVALHO (science writer, Germany), Kenji CHAMOTO (Kyoto University, Japan), Hongbo CHI (St. Jude’s Research Hospital, Memphis, USA), Antonio COUTINHO (Champalimaud Foundation, Oeiras, Portugal), Mikala EGEBLAD (Cold Spring Harbor Laboratory, USA), Sidonia FAGARASAN (Kyoto University, Japan), Alain FISCHER (Collège de France, Paris, France), Thomas GAJEWSKI (University of Chicago, USA), Silvio GUTKIND (University of California San Diego, USA), Honjo TASUKU (Kyoto University, Japan), Patrick MEHLEN (Lyon Cancer Research Center, France), Ton SCHUMACHER (Netherlands Cancer Institute, the Netherlands), Louis STAUDT (NCI/NIH, Bethesda, USA), Pamela OHASHI (Princess Margaret Cancer Center, New York, USA), Hans-Guido WENDEL (Memorial Sloan Kettering Cancer Center, New York, USA), Santiago ZELENAY (University of Manchester, England).

Résumé

L’immunothérapie anticancéreuse a révolutionné le traitement de certaines tumeurs solides métastatiques, à commencer par les premiers résultats définitifs d’essais cliniques sur le mélanome, publiés en 2011. Suite à l’effet perturbateur des thérapies bloquant CTLA4 et la voie PD1, le domaine de l’immunothérapie anticancéreuse s’est installé dans une phase de progrès progressifs. La plupart des patients et de nombreux types de tumeurs restent réfractaires à l’immunothérapie. À l’exception notable de l’introduction du traitement CART pour le lymphome au milieu des années 2010, les progrès ultérieurs ont été lents et le développement clinique s’est rapproché des succès antérieurs. Il existe actuellement environ 6 000 essais cliniques de médicaments ciblant la voie PD-1, la plupart sous forme de thérapies combinées avec des médicaments anticancéreux préexistants. De nouveaux agents immunomodulateurs ciblant d’autres molécules de point de contrôle telles que LAG3 font leur apparition en clinique. Mais les grands défis consistant à augmenter les taux de réponse des patients, à développer des marqueurs pronostiques fiables et, surtout, à réchauffer les tumeurs « froides » sur le plan immunologique demeurent. Un atelier aux Treilles, organisé par Tasuku Honjo et Antonio Coutinho, était prévu en 2020, mais n’a pu avoir lieu en raison de la pandémie de COVID-19. Trois ans plus tard, l’intérêt d’une discussion sur ce sujet persistait et « l’esprit » des Treilles semblait idéal pour revenir et discuter des mécanismes moléculaires et cellulaires de l’immunothérapie du cancer, de sorte que l’on puisse peut-être cerner les raisons de ses limites actuelles. et trouver de nouvelles opportunités. La réunion était parrainée par la Fondation Schlumberger pour l’éducation et la recherche, avec le soutien supplémentaire du Centre d’immunothérapie et d’immunobiologie du cancer (Université de Kyoto) et de la Fondation Champalimaud (Lisbonne). Les organisateurs ont annoncé leur intention de donner suite à d’autres ateliers sur l’immunothérapie du cancer, à Kyoto et à Lisbonne.

Summary

Cancer immunotherapy revolutionized the treatment of some metastatic solid tumors, starting with the first definitive clinical trial results with melanoma, published in 2011. Following the disruptive effect of therapies blocking CTLA4 and the PD1 pathway, the cancer immunotherapy field settled into a phase of incremental progress. Most patients and many tumor types remain refractory to immunotherapy. With the notable exception of the introduction of CART therapy for lymphoma in the mid 2010s, subsequent progress has been slow and clinical development has sailed close to the shoreline of earlier successes. There are currently about 6,000 clinical trials of drugs targeting the PD-1 pathway, most as combination therapies with pre-existing cancer drugs. New immunomodulatory agents targeting other checkpoint molecules such as LAG3 are trickling into the clinic. But the grand challenges of increasing patient response rates, developing reliable prognostic markers, and, especially, heating up immunologically “cold” tumors remain.

A workshop at Les Treilles, organized by Tasuku Honjo and Antonio Coutinho, had been planned for 2020, but could not be held due the COVID-19 pandemic. Three years later, intense interest in holding a discussion on this topic remained, and the “spirit” of Les Treilles seemed ideal to review and discuss the molecular and cellular mechanisms of cancer immunotherapy, such that we could perhaps pinpoint the reasons for its current limitations and find new opportunities. The meeting was sponsored by the Schlumberger Foundation for Education and Research, with additional support from the Center for Cancer Immunotherapy and Immunobiology (Kyoto University) and the Champalimaud Foundation (Lisbon). The organizers announced their intention to follow up with other workshops on cancer immunotherapy, in Kyoto and Lisbon. 

Report

Tasuku Honjo kicked off the meeting with a first-person account of the discovery of PD-1 and the development of anti PD-1 immunotherapy. The work began with a graduate student, Yasumasa Ishida, who wanted to decipher the molecular mechanisms of apoptosis. Ishida and Honjo identified Programmed Death 1 in a screen for genes involve in a model of thymocyte apoptosis. It soon became evident that PD1 does not play a role in apoptosis, but rather has immunoregulatory functions. The group shifted the study of PD1 towards the antipodes of autoimmunity and antitumor immunity. At the time, a separate immunoregulatory molecule, CTLA4, was shown by Allison and collaborators to put the brakes on T cell response to tumors. However, mice deficient for CTLA4 rapidly succumbed to a fulminant lymphoproliferative disease, indicating that blocking the pathway might be dangerous in the clinic. In contrast, the phenotype of mice genetically deficient for PD-1 was encouraging, because the autoimmunity was mild (and strain dependent), suggesting to Honjo and collaborators that therapeutic PD-1 blockade could be safer for patients. A decade after identifying PD1, the Honjo group identified the Honjo group and others discovered the PD1 ligands and signaling pathway. They showed that interfering with the PD1 pathway in mouse models promoted anti-tumor immunity. In 2004 the group, working with the company Medarex, produced a humanized version of the antibody. Clinical trials confirmed the efficacy of anti-PD-1 monoclonal antibodies in 2012. Anti-PD1 antibody, like anti-CTLA-4, produced durable responses in some patients; as Honjo had predicted, targeting PD1 in the clinic showed a better safety profile than blocking CTLA-4. At Les Treilles, Honjo also presented data on the potential for PD1 drugs in the treatment of autoimmune diseases: after several failed attempts, the group has finally produced PD-1 agonist antibodies.

Honjo proceeded to point out a key limitation of current animal models. The single most significant risk factor for cancer is ageing; most mice used in cancer research are at best young adults – the reasons are not mysterious, as Honjo put it, “to get aged mice is very expensive”. CD45RB levels increased in aged CD8 T cells, and he suggested that anti-PD1 therapy in combination with a CD45 inhibitor could synergistically enhance anti-tumor immunity in aged mice. The Honjo group showed that an age dependent decline in serum levels of the polyamine spermidine was linked to decreasing response rates to immunotherapy. Supplementing older mice with spermidine restored responsiveness. A few caveats emerged in the discussion: Louis Staudt pointed out that spermidine synthase is regulated by Myc and that inhibiting it prevents lymphoma in animal models; Tom Gajewsky cautioned that spermidine is metabolite that affects a wide range of cell types;Vassiliki Boussiotis said spermidine may promote survival of myeloid derived suppressor cells (MDSCs) in the tumor microenvironment. Sidonia Fagarasan replied that the net effect in vivo is to stimulate T cell immunity, perhaps because T cells are sensitive to lower spermidine levels than other cell types.

Kenji Chamoto further discussed the metabolic impact of spermidine, explaining that it binds to mitochondrial trifunctional protein (MTP) and acts as an allosteric modulator that enhances fatty acid oxidation (the spermidine metabolite spermine competes with spermidine and inhibits this effect). Mitochondrial respiration is reduced in older mice, and this contributes to CD8 T cell dysfunction. Spermidine increases T cell oxidative phosphorylation and fatty acid oxidation, and etomoxir blocks this effect. Ton Schumacher asked if spermidine affected T cells from younger individuals the same way; Chamoto replied that it did. A second approach to modulating mitochondrial function explored by the group was the use of bezafibrate, a small molecule drug used as a lipid lowering agent. Bezafibrate promotes fatty acid oxidation and acts a mitochondrial activator in T cells. The drug synergizes with anti-PDL1 treatment, promoting anti-tumor immunity; it also promotes CD8 cell survival by increasing Bcl2 expression via PPAR activation of Cpt1. Overall, the results point to fatty acid and mitochondrial metabolism as pathways that can be targeted to increase responsiveness to clinically available checkpoint inhibitors. Schumacher emphasized that this must be done cautiously, as other T cell types may also be functionally enhanced – he mentioned the possibility of promoting regulatory T cell mediated induction of suppressive tumor associated macrophage; Chamoto replied that so far, they have observed no increase in Foxp3 expressing cells or Arginase-1/MHCII positive cells.

Sidonia Fagarasan discussed an unexpected metabolic phenotype of PD-1 deficient mice, a drop in tryptophan and tyrosine levels that led to a deficiency in the neurotransmitters serotonin and dopamine in the brain. Knockout mice were anxious and showed increased fear responses (“they were very aggressive, the postdoc hated working with them, and so did I,” she said). Eventually all serum amino acid levels dropped, but the aromatic amino acids were the first to go. Ton Schumacher asked if this was also seen in animals treated with anti-PD1; yes, Fagarasan replied. Silvio Gutkind asked if they respond to antidepressants; they do. Fagarasan proposed that the specific amino acid deficiency is linked to dysbiosis in the gut flora caused by aberrant IgA responses in the PD1 deficient rodents; excessive T cell proliferation also directly depletes tryptophan and tyrosine in PD1 knockouts. Fagarasan et al. show that B cells themselves produce another neurotransmitter, γ-aminobutyric acid (GABA), which acts on monocytes to promote differentiation of immunosuppressive IL-10 producing macrophage that impair antitumor immunity. Interestingly, Fagarasan mentioned that Strandwitz et al. showed that the gut flora also regulates GABA levels, including in the central nervous system; consistent with this, Fagarasan finds the highest GABA levels in gut plasma cells. Overall, GABA levels in lymph nodes and spleen are the highest of any site outside the brain. Louis Staudt asked what the rate-limiting enzyme for GABA production is; “GAD1,” replied Fagarasan, and mice with B cell specific GAD1 deficiency show enhanced anti-tumor responses; less specifically, GABA receptor antagonists also enhance anti-tumor immunity. The geography of GABA in tumor immunity is not entirely clear – Fagarasan described the number of tumor-infiltrating IgA plasma cells as “pathetic” and proposed that the relevant interactions with monocytes/macrophage happens in the bone marrow. Alain Fischer asked if GABA levels are high in multiple myeloma; “That’s a good point. Let’s look,” replied Fagarasan. Tom Gajewksi mentioned that a glance at existing data on PDL1 responsiveness shows tumor cells making GABA; “MC38 cells and renal cancer cells don’t, but many other tumors do,” said Fagarasan (in his own presentation, Gajewski refers to stable colonies of patient-derived microbiota in his lab that show concordance between patient and mouse outcomes).

Ton Schumacher presented a novel approach to a fundamental problem in immunology, particularly in tumor immunology: how to predict the specificity of the T cell receptor from DNA sequencing data. As Schumacher pointed out, immune checkpoint inhibition is only relevant in the context of what it is blocking and “we can read the TCR, but we can’t interpret it”. Schumacher and collaborators have developed an in silico method, termed STAPLER, that does not require prior knowledge of the target antigen, making the technique potentially useful not only for diagnostic purposes, but also for the discovery of new mechanisms of pathophysiology. Antonio Coutinho asked if the model accommodates degeneracy in TCR reactivity, and Schumacher replied that STAPLER does not limit cross-reactivity. Schumacher also showed data on the role of T cells and T cell derived cytokines in the tumor microenvironment. Are tissue resident memory T cells (TRMs) functional, or are they the immunological equivalent of scar tissue? Earlier data from the group showed that, in reproductive and skin infection models, TRMs are mobile cells, patrolling the tissue months after they were induced. They secreted high levels of IFN-γ, which promoted local immune responses even against unrelated pathogens. Schumacher et al. showed that in tumors, bystander cells respond to IFNγ and that the signal spreads over 600nm (about 40 cell diameters), leading to IFN-γ receptor dependent tumor cell death. Distant sensing of IFNγ also promoted PD-L1 expression. These long-range effects were not observed for TNF-alpha. “Some cytokines shout, others whisper,” said Schumacher, who argued that the field must incorporate an understanding of local cytokine secretion, turnover, and diffusion. Schumacher proposed that successful PD-1 blockade acts as a spark in the engine, allowing the IFNγ response to remodel the tumor microenvironment, leaving the target tumor cells with nowhere to hide. Later in the meeting, Tom Gajewsky recalled the work of Ayers et al. showing that, in patient samples, an IFNγ signature was a good prognostic marker of responsiveness to PD1 blockade, but a subset of patients with a T cell activation signature showed no response to treatment.

The success of immunotherapy is highly variable across not only patients, but also tumor types. There is currently little consensus on what makes a tumor a “hot” target for the immune system or a “cold”, immunologically inert tumor. Early bets on predictors like tumor mutational burden, PD1/PDL1 levels, or tumor infiltration did not pay off, and the uncertainty remains high. Santiago Zelenay presented work on a very early predictor of responsiveness to immunotherapy and new approaches to “heat up” cold tumors. Zelenay pointed out the paradoxical role of inflammation, which, depending on context, can promote tumor growth or favor anti-tumor immunity. Zelenay et al. had previously shown that tumor growth requires prostaglandin E2 (PGE2) and that inhibiting cyclooxygenase turns heats up cold tumors, making them attractive targets for the host immune system. The group found that PGE2’s early effects where mediated through natural killer cells, via EP2 and EP4 receptors. NK cell-derived IFNγ remodeled the tumor microenvironment, favoring oxidative phosphorylation over glycolysis in tumor macrophage, and drove potent cytotoxic T lymphocyte responses. Zelenay’s data on a broad panel of tumor cell lines showed that COX2 levels are an excellent predictor of tumor rejection in vitro. COX2 activation signatures were a reliable predictor of tumor rejection in vivo even for identical tumors in isogenic mice, where bilateral grafts allowed the group to correlate early inflammatory phenotypes with later tumor rejection or continued growth. Louis Staudt mentioned that inflammatory cell infiltration was already known to correlate with anti-tumor immunity; Zelenay replied that COX2 activation was more reliable – even when adjusted for inflammatory infiltration. Ton Schumacher asked what determines COX2 production by the tumor; Zelenay replied that this is currently unknown, with a long list of suspects. Because of the broad biological effects of COX2/PGE2, more specific targeting with antibody drug conjugates or bispecific monoclonal antibodies were suggested as possible clinical strategies.

Tom Gajewski stated the principle that “antigenicity and immunogenicity are two different things”. He recalled the correlation between immunogenic “hot” tumors and higher  type I interferon levels that in turn are correlated with T cell infiltration. Ton Schumacher remarked that the IFNγ and type I IFN gene signatures overlap significantly, to which Gajewski replied that this referred to type I IFN specific transcripts. Gajewski cited a list of culprits in interpatient variability in response to immunotherapy, including differences in the tumor itself, microbiome variation, and germline genetic differences, such as polymorphisms in PKC-delta. “We view spontaneous autoimmunity to be similar to spontaneous tumor resistance,” said Gajewski. Many factors beyond genetic and housing homogeneity lie behind the difference to mouse models that are highly susceptible to immunotherapy. Gajewski highlighted the common problem of tumor grafting protocols that introduce confounding factors like cell death, local inflammation, and vascularization that are not seen in patients with spontaneously arising tumors. Data from a genetically engineered mouse tumor model of BRAF-dependent melanoma showed that recruitment of Batf3 dependent dendritic cells into the tumor microenvironment was essential for the recruitment of effector CD8 cells. The response is triggered by the STING-type I interferon pathway. Gajewsky also shows that this step can be blocked by tumor-derived beta-catenin. Batf3 dendritic cells may also be required for the response to PD1/PDL1 blockade. Gajewski presented the interesting case of the “coldest tumor of all”, Wilms tumor, a rare pediatric kidney tumor, which he described as having “an almost negative immune infiltrate”.  In this case, low immunogenicity showed an association with increased expression of DNA repair genes by the tumor.

The canonical target for immune checkpoint therapy is the CD8 T cell – with a focus on reviving exhausted or dysfunctional cells, chronically stimulated killers that appear to be retired. Rafi Ahmed described the identification of a unique CD 8 T cell population that arises under conditions of chronic stimulation in a mouse model of viral (LCMV) infection, PD1+ stem-like cells expressing TCF1 and Bcl6. These cells do not express effector molecules and undergo self-renewal. In human cancer, they are found in tertiary lymphoid structures. The group then identified a transitory effector T cell type, derived from the stem-like cell. These cells, which express CX3CR1+, are an important target of checkpoint inhibitor therapy: transitory cells express PD1 and proliferate when exposed to anti-PD1 antibody. CD8 cells require CD28 to respond to PD1 targeted therapy. Data from peripheral T cells of lung cancer patients supports that these cells are the target of anti-PD1 therapy in humans as well as in mice. Ton Schumacher asked how strong the evidence is they are tumor reactive; “we have zero evidence in our studies,” replied Ahmed. Schumacher asked if these cells could be virus specific. “Absolutely,” was the reply. Ahmed went on to detail one case where anti-viral CD8 cells were the target of anti-PD1 therapy in human cancer. This was for HPV reactive cells in head and neck cancer, where a subset of PD1+ stem-like CD8 cells could be detected. Patients also had anti-HPV reactive B cells. The discussion pointed to pitfalls and promises of targeting stem like CD8s: they might by eliminated by radiotherapy; on the other hand, Louis Staudt asked if there was a way to expand them in a pool of tumor infiltrating lymphocytes, Ahmed replied that this might be possible using IL7. In a separate session, Hongbo Chi discussed targeting Ikaros to inhibit the transition from precursor to exhausted T cell, which sparked a lively discussion: Louis Staudt pointed out that lenalodomide degrades Ikaros family molecules; Tom Gajeweski cautioned that if you only enrich for a stem T cell population, it will never become a viable therapeutic option. Chi replied that it would have to be a transient treatment step.

In head and neck cancer, Silvio Gutkind focused on the EGFR, PI3K, PTEN, and mTOR pathways. This tumor type is accessible for serial biopsies, making it a good case study. Most patients with advanced disease treated with rapamycin followed by adjuvant therapy showed a tumor shrinkage response, without immune suppression. HER3 lies upstream of mTOR activation and inhibiting HER3 in combination with anti-PD1 is a promising approach. Interestingly, the largest share of responding patients were smokers (overall head and neck cancer has an intermediate mutation rate). Only about 20% of head and neck cancer patients respond to checkpoint inhibitors, with anti-CTLA4 showing particularly dismal results. In a mouse model, Gutkind tested the hypothesis that lymph node ablation in many patients may underlie the hyporesponsive phenotype. Mice with that underwent lymph node ablation lost the ability to respond to anti-CTLA4 and eradicate engrafted tumors. Effective immune responses were dependent on lymph node cDC1s. The results make a case for considering changes to clinical practice, which has yet to happen. As Gutkind put it, “the conclusion is that patients and physicians don’t read our papers.” There are about 800 G-protein coupled receptors (GPCRs) and they represent 30% of all known drug targets. Gutkind presented data from a screen of Gα_s-coupled GPCRs on exhausted CD8 in single cell RNA sequencing data from 19 different tumors and identified a set of candidates that could act synergistically with checkpoint inhibitor immunotherapy.

Hongbo Chi stressed that the molecule-by-molecule approach to identifying targets in cancer therapy and immunotherapy is not enough – a systems approach is sorely needed. With correlations between transcripts and protein levels often revealing significant discrepancies, Chi and collaborators have also adopted a proteomics approach to understanding T cell activation. In parallel, the group has also employed a forward genetics approach, with a CRISPR screen identifying regnase-1 as an inhibitor of CD8 anti-tumor immunity.Ton Schumacher asked what regnase-1’s biological function was; “it controls T cell activation,” Chi replied – and explained that it is a ribonuclease which inhibits BATF. Louis Staudt pointed out that Shizuo Akira originally described regnase-1 as targeting as an IL6 regulator. Chi noted that Carl June and collaboratorshave shown that disrupting regnase-1 also enhances CART function. Alain Fischer cautioned that patients with loss of function mutations in regnase-1 show immune pathology. Chi also emphasized the need to understand metabolic control of dendritic cell function in the tumor microenvironment. At the metabolite level, Chi showed data on arginine, glutamine, and cysteine depletion from the tumor interstitial fluid (along with glucose), with glutamine depletion having significant effects on CD8 cell number and effector function. Glutamine supplementation synergized with the anti-tumor effects anti-PD1 treatment in a mouse model. Sidonia Fagarasan asked Chi if he traced what glutamine is used for in the context of tumor immunity; “it acts through DC1s,” which specifically require the SLC38A2 transporter for glutamine to exert its beneficial effects. Tom Gajewski pointed out that glutamine is crucial for T cells; Chi replied that T cells rely more on other transporters, but DCs are dependent on SLC38A2.

IL22 producing T cells (Tc22) also play a role in anti-tumoral immunity, and Pamela Ohashi et al. showed that Tc22 cells are induced by IL6 – with IL22 production by tumor infiltrating lymphocytes showing a correlation to tumor killing. In follow up work, group identified pantothenate, a metabolite required for coenzyme synthesis, enhanced oxidative phosphorylation, and Tc22 differentiation. In collaboration with Jennifer Wargo, Ohashi found that plasma pantothenic acid levels coincided with favorable responses to anti-PD1 therapy in melanoma patients. Kenji Chamoto asked if the culprit was pantothenate or coenzyme A; Ohashi replied that “this is a the-chicken-or-the-egg question,” but her bet would be coenzyme A. Ohashi also proposed a role for a unique CD56+CD3- innate lymphoid cell in antitumor immunity.

CD8 T cells hog the limelight in checkpoint inhibitor immunotherapy research, but PD-1 is also expressed on other cell types.  Vassiliki Boussiotis showed data on the effects of targeting PD-1 and PD-1 signaling in the myeloid lineage. PD1 is expressed on early myeloid progenitors and granulocyte/macrophage progenitors; PD1 can inhibit T cell activation by recruiting the SHP-2 phosphatase. However, T cell specific deletion of SHP-2 in Lck-Cre mice does not reduce antitumor immunity. Conversely, selective deletion of SHP-2 in LysM-Cre animals led to enhanced anti-tumor immunity – an effect that was not further boosted by treatment with anti-PD1 antibody. Mice with myeloid-specific deletion of either SHP-2 or PD1 were more efficient promoting antitumor immunity. These mice had enhanced neutrophil differentiation and tumor-associated macrophage that had improved antigen presenting function. Interestingly, cell transfer experiments also showed evidence of trained immunity in SHP-2 deficient neutrophils and monocytes harvested from tumor bearing hosts. Mikala Egeblad asked how trained immunity worked in this model and Boussiotis pointed out the enhanced antigen presentation and persistent epigenetic changes (imprinting). Boussiotis also presented data on a new reagent, an antibody that recognizes phosphorylated PD1, that allows a better understanding of where and when T cells are responding to PD1 stimulation.

Andrea Ablasser introduced STING-cGAS as a self/non-self discrimination pathway for nucleic acids that is involved in more than just responses to pathogens. With ageing, mitochondria leak DNA that can activate cGAS and cGAS deficient animals have a compromised cellular senescence response. cGAS activation triggers the production of senescence-associated secretory phenotype (SASP) factors, which can play important in tumorigenesis and metastasis. cGAS is normally found in the nucleus, tethered to nucleosomes. Interestingly, the Kaposi Sarcoma virus LANA peptide competes with cGAS for nucleosome tethering, blocking binding sites and preventing dimerization.

Nucleic acids were discussed not just as triggers of host innate immunity, but also as effector mechanisms, when Mikala Egeblad introduced the topic of neutrophil extracellular traps (NETs) – a DNA net secreted by the cell to trap extracellular threats. Her group showed that breast cancer cells can trigger NET secretion, which can in turn promote metastasis. The clouds of DNA were observed surrounding breast cancer cells in the mouse lung and digesting them with DNAse-coated nanoparticles reduced metastasis. NETs promote metastasis by remodeling the extracellular matrix. Tom Gajewski asked if there is cGAS in the NET; Ablasser replied that this not yet known. NETs are also implicated in post-chemotherapy resistance in breast cancer. In mice treated with cisplatin, NETs form around metastatic cells and DNAse treatment improves the response to chemotherapy. This is effect is mediated by NET associated integrins and metalloproteases that trap and activate TGF-beta. Gajewski asked if different neutrophils make different types of NETs; “probably,” replied Egeblad. There is also a potential role for NETs and neutrophils in primary tumors, where they can clog the vasculature, and NETs are observed in necrotic areas. The tumor may play an active role in this process, biasing bone marrow hematopoiesis towards the production of neutrophils that promote NET formation.

Patrick Mehlen presented data on netrin-1 in tumors. Originally described as a molecule involved in axonal guidance, netrin-1 is now known to also play roles in cell survival and angiogenesis. The Mehlen group showed in 2001 that signaling through the netrin-1 receptors, UNC5H and DCC, is important for neuronal survival. They then demonstrated that enforcing netrin-1 expression in the mouse gut led to the appearance of colorectal tumors, that high netrin-1 levels are a marker of poor outcomes in neuroblastoma patients, and generated a humanized monoclonal antibody that disrupted netrin-1 binding to UNC5H. This antibody, NP137, has progressed to clinical trials. The first-in-human phase I dose-escalation trial in hepatocellular carcinoma was powered for safety and pharmacokinetics analysis, and not for efficacy analysis. But preliminary analysisshows encouraging safety data and signs of an effect in patients – 40% showed stable disease, but only 2 showed target lesion reduction. Mikala Egeblad asked if any interesting correlations were observed in these two patients; unfortunately, no post-treatment biopsies were available. Mehlen considered that the results supported the scenario of using the monoclonal as part of combination therapies, and not as a single agent, as seems to be the case for combination with carboplatin-paclitaxel. In collaboration with Cedric Blanpain, Mehlen also found that the anti-netrin-1 inhibits the epithelial mesenchymal transition in squamous cell carcinoma. Tom Gajewski asked if there is still a tumor; “yes, but smaller and more drug-sensitive,” replied Mehlen. Mouse data indicates that anti-netrin-1 treatment may synergize with checkpoint inhibitor immunotherapy.

Ageing is, of course, not the only significant risk factor for cancer. Alain Fischer spoke about a particularly vulnerable group, patients with primary immune deficiencies (PID). There are about 500 described primary immunodeficiencies, with common variable immunodeficiency (CVID) the most prevalent. Overall, PID patients show an age-dependent increase in cancer incidence, evenly split between hematological tumors and non-hematological tumors. A decrease in immunity to cancer-causing viruses accounts for some of this effect, with Epstein-Barr Virus (EBV) being a case in point. EBV has a tropism for CD21+ B cells and is linked to 200,000 new cancer cases each year. Patients with a PID affecting NK cells or T cells are at risk for uncontrolled EBV infections and consequent B cell malignancies. Perturbations in the CD70-CD27 checkpoint in T cells are specifically linked to a susceptibility to EBV and B cell malignancy; PID patients with mutations in linked pathways show similar phenotypes. Fischer and Sylvain Latour have recently summarized the data on T cell signaling-defect PIDs that predispose to EBV linked malignancies. Silvio Gutkind mentioned that Kaposi’s Sarcoma is also linked to PIDs in some populations; Fischer replied that described PIDs affecting Kaposi’s Sarcoma include OX40 deficiency and Wiskott-Aldrich Syndrome.

Louis Staudt focused on cancers of the immune system, on activated B cell (ABC) and germinal center (GC) B cell tumors. ABC tumors are the most problematic and depend on chronic antigenic stimulation (by self-antigens); B cell receptors show greater clustering on the cell surface in ABC tumor cells than in GC tumors. ABC tumors have a 38% cure rate when treated with ibrutinib, lower that the GC tumor response rate. ABC tumors had a high incidence of BCR and MyD88 mutations, and most of these patients responded to treatment. However, even some patients without BCR pathway mutations responded to ibrutinib. Staudt went on to examine a series of genetic vulnerabilities in B cell malignancies that play a role in treatment outcomes with different drug regimens, with a focus on diffuse large B cell lymphoma (DLBCL). He presented a strong case for combination regimens matched to specific tumor subtypes – not personalized medicine, but still a more discerning strategy than either throwing the kitchen sink at tumors or relying on single genetic markers. As Staudt put it, focus on rationally designed combination therapy targeting multiple vulnerabilities at once is “reinventing the wheel. It’s HAART. It’s tuberculosis treatment”. While not opposed in principle, Ton Schumacher brought up the practical problem that biopsies sampled only a small fraction of the tumor, and hidden tumor heterogeneity might compromise the approach in some patients.

Hans Guido Wendel continued along the lines of targeting multiple tumor vulnerabilities, taking as his guiding principle the dictum by Dutch football legend and amateur philosopher Johan Cruyff that “every advantage has a disadvantage”.  Thus, the approach of the Wendel group is to ask “what is the cost of malignancy?” Wendel shows that Nrf2 (the stress response gene can function as an oncogene) activity in tumors is regulated through de-glycation by the enzyme fructoseamine-3-kinase (FN3K), because the glycated form of Nrf2 is unstable and does not effectively bind target genes. This dependency on FN3K creates a metabolic vulnerability in tumors that rely on Nrf2 which can be exploited therapeutically. In another example, 45% of follicular lymphomas have mutations in HVEM, a cell surface receptor on B cells. Loss of HVEM leads to B cell activation and proliferation and the appearance of tumor supportive niches. HVEM levels show an inverse relation to BTLA in tumors, pointing to complementary possibilities for targeting. Wendel also showed data on the vulnerability created by eIF4A dependency of myc translation.

In the closing discussion, Antonio Coutinho recalled the goals of the meeting, adding that we have all learned many new things and paid attention to usually separate research topics, but had progressed little in solving the major problem of the rarity of patients responding to checkpoint immunotherapy. As pointed out by several participants, heterogeneity in responsiveness to immunotherapy is also observed in mouse models, even if genetically homogeneous, in the same age group and with common housing! He suggested that this is likely to reflect the complexity of the system’s dynamics with irreversible “branching points”. In a general note, Coutinho concluded that, while we know much about the tumors, we remain ignorant of many aspects of the relationship between tumors and the host, similarly to several other complex biological systems (e.g., autoimmune disease). He thanked all participants and supporting organizations, as well as all the personnel at Les Treilles who contributed to the outstanding environment we had enjoyed. 

Report written by Thiago Carvalho.

 

---

---