Cancer Immunotherapy: The 2018 Dr. Paul Janssen Award Symposium

Cancer cells express and display novel antigens that the immune system can recognize and attack. This insight could not be exploited therapeutically, however, until researchers realized that the tumor microenvironment actively suppresses the immune system, as any successful pathogen would. Now, immune checkpoint blockades–antibodies targeting and neutralizing PD-1, PD-L1, and CTLA-4–have been used to release cytotoxic T cells from this immunosuppression and successfully treat patients with melanoma and leukemia.

This revolution in immunotherapy is due in large part to the work of Dr. James P. Allison, of the University of Texas MD Anderson Cancer Center, who has dedicated his career to better understanding the immune system and leveraging it to treat cancer. His lab developed ipilimumab, the CTLA-4 inhibitor that brought the concept of immunotherapy out of the lab and into patients’ lives as the first checkpoint inhibitor approved by the FDA in 2011. Immune checkpoint inhibition is now an essential part of cancer treatment, and has expanded to fight a broad range of tumors.

The symposium in his honor was devoted to emerging approaches in cancer immunotherapy. After James Allison’s keynote address on the past and future of cancer immunotherapy, Dana Pe'er, of  Memorial Sloan Kettering Cancer Center, described her fascinating work characterizing tumor immune ecosystems with mathematical and computer modeling methods in order to personalize immunotherapy. Her colleague Alexander Rudensky discussed his work on regulatory T (Treg) cells, a specialized lineage of cells that negatively regulate the inflammatory response.

Elizabeth Mittendorf, of Brigham and Women’s Hospital and the Dana-Farber Cancer Institute, gave her perspective as a surgeon on immunotherapy’s impact in breast cancer, where it has been investigated but hasn't had the same success seen in other tumor types. Padmanee Sharma, of the University of Texas MD Anderson Cancer Center,  talked about the center’s unique pre-surgical tissue based clinical trial platform, which helped identify the role of the ICOS/ICOSL pathway in anti-tumor immune responses induced by ipilimumab. And Matthew Mulvey, of Janssen Pharmaceutical Companies of Johnson & Johnson, discussed oncolytic viruses, a resurging facet of immuno-oncology that is now being used alongside checkpoint blockade.

The day ended with a panel discussion featuring all of the speakers and moderated by Dr. William Hait, the Global Head of Johnson & Johnson’s External Innovation. Topics included:  the potential role of cancer vaccines, the impact of hormone therapies on immunotherapy, and  potential drug combinations that have not yet been tried.

Highlights

• Immune checkpoint blockade was quite a paradigm shift in cancer therapy, because it doesn’t target tumor cells or use vaccines or cytokines to turn “on” immune responses.
• T cells are a perfect match for cancer because tumors are heterogeneous, which can lead to drug resistance and relapse, while T cells are specific, adaptable, and can be reactivated.
• The therapeutic mechanisms of anti-CTLA-4 (ipilimumab) and anti-PD-1 (nivolumab) are distinct, which may explain why using them in combination is so effective.
• The respective mechanisms of the two drugs remain the same in highly immunogenic and poorly immunogenic tumors and are each mediated by specific CD4 and CD8 T cell subtypes.

Immune Checkpoint Blockade in Cancer Therapy: New Insights, Opportunities, and Prospects for Cures

T cells are activated through the T cell receptor along with a costimulatory molecule, CD28. Cancer cells don’t have costimulatory molecules to engage CD28, so they remain invisible to the immune system until they die and their necrotic remains wind up on the surface of antigen presenting cells. Some tumors go even further in evading immune detection by also expressing CTLA-4, which shuts down T cell expansion. As a result, James P. Allison, hypothesized that disabling CTLA-4 with an antibody would be an effective cancer treatment. This "checkpoint blockade" cured all of the transplanted mouse tumors he threw it at, and the mice remained immune to the tumor for the rest of their lives. Nevertheless, as Allison explained, “it was hard to get people to believe we could cure cancer by ignoring the cancer.” A testament to his perseverance, in humans the anti-CTLA4 drug ipilimumab ultimately caused tumor regression in patients with metastatic melanoma.

Anti-CTLA-4 “cured” melanoma—patients have survived for more than ten years on it—but only in about 20% of patients. One possible explanation is that there are other checkpoints that need to be targeted, and targeting multiple checkpoints at once would be a more effective therapy. Indeed, the inhibition of PD-1, another immune checkpoint, has been shown to be an effective cancer therapy though its mechanism is distinct. To better understand the mechanism of these pathways in the hope of translating to better therapies, and inform how to use combination therapies, Dr. Allison used mass cytometry of human melanoma cells and murine tumor models treated with either anti-CTLA-4 or anti-PD-1. This revealed that there are differences in how the two checkpoint inhibitors work, and these differences hold true across different tumor types.

Perhaps because the checkpoint inhibitors work differently, combining them is often more effective than using either in isolation. However, it’s important to note that each therapy can have opposite effects on particular subsets of cells. “We have to make sure we don’t give a combination of therapies that counteract each other,” Allison cautioned.  “We can learn something from every patient, even, or especially, if it looks like a therapy didn’t work.”

In an effort to get a better understanding of how immunotherapies work, Dr. Allison set out to examine the function of CTLA-4 in a normal, non-tumor context by generating mouse knockouts. T cells in these mice undergo “supernatural” stimulation; new, “alien” CD4 T cell phenotypes arise in the absence of CTLA-4 that do not develop in its presence. This raises the possibility that by manipulating these T cell differentiation pathways, anti-CTLA-4 may create new types of T cells.

Tumors contain many different T cell types. Dr. Allison is trying to determine which T cells in a tumor will be responsive to immunotherapy, and how that population might change after different treatments.

Highlights

• When looking at cellular data, the average dominates. This is a problem because although we can see a response, we can’t really determine the functionality of individual cells or molecules.
• Single cell resolution data has revealed that tonic signaling in CAR T cells can lead to exhaustion, and reduced ability to remodel signaling networks in response to stimulus. This analysis can be used to design more effective T cell therapies.
• Treg cells suppress the adaptive immune system’s inflammatory response.
• The X-linked transcription factor Foxp3 is specifically expressed in Treg cells and acts as a Treg lineage specification factor.

A Single Cell Lens in T-cell Behavior

Dana Pe’er uses single cell technologies, genomic datasets, and machine learning algorithms as “another set of tools to understand how T cells are functioning as we continue to engineer CAR T cells and develop immunotherapies.” She first described her discoveries about T cell signaling from single cell data of native T cells, and then explained how her methodology was applied to optimize CAR T cell design, identifying mechanisms and solutions for factors which reduce CAR T cell effectiveness.

Mass cytometry (CyTOF) can measure many proteins in millions of individual cells, but it is difficult to develop a computational algorithm that can handle the complexity and noise inherent in all of these measurements. Pe’er, whose background is in mathematics and computer science, has developed such an algorithm. Usually, a conglomeration of single cell CyTOF data is presented as a scatterplot colored by the density of the points. But scatterplots only convey the joint probability of two parameters–like how frequently two proteins co-occur. In contrast, Pe’er’s approach assays conditional probability, or how the state of one protein can vary based on the differing states of a second–even if the two proteins co-occur in one or very few cells. By monitoring how the relationships between pairs of proteins changes over time, Pe’er’s method can elucidate how signaling pathways in different cells respond to external stimuli. In this way, she was able to discern relationships between important proteins in T-cell signaling such as CD3, AKT, SLP-76, ERK, and RelA and visualize how the flow of signal was changed by T cell expansion and CAR expression.

Dr. Pe’er’s algorithm allows her to measure how two proteins vary as a function of each other in individual cells.

Pe’er’s colleagues and students used her methodology to see how CAR molecules work in individual cells. It had been assumed that the synthetic endodomains used in CAR T cells signal in the same manner as natural endodomains, and that artificially expanded CAR T cells have the same signaling dynamics as fresh T cells; but neither of these assumptions had been demonstrated experimentally. Using mass cytometry to measure phosphoproteins essential for TCR signaling in different T cell populations, Pe’er’s algorithm showed how tonic signaling arising from the CAR polarizes the signaling network and reduces its plasticity compared to native T cells.

Tonic signaling can lead to T cell exhaustion, and it is mediated by CD3ζ. αβT CARs need CD3ζ to replace signaling through the MHC-epitope restricted αβTCR, but Vδ2 γδTCRs are MHC independent, recognizing tumor cells with a high phosphoantigen burden. Since there is no need to replace the TCR signal, CD3ζ is dispensable for CARs expressed in Vd2 gdT cells. Pe’er’s work thus indicates that using γδT cells to make CAR T cells could circumvent the issue of CD3ζ mediated T cell exhaustion faced by αβ T cells, in addition to reducing the potential for on-target off tumor toxicity.

Many Faces and Functions of Regulatory T Cells

Treg cells are a specialized subset of cells that prevent the loss of tissue function in diverse biological and clinical settings. They act through negative regulation in trans, an unusual regulatory mechanism that can only be found in the adaptive immune system and the central nervous system. Alexander Rudensky and his lab identified the X-linked transcription factor Foxp3 as a Treg lineage specification factor; it is required for Treg cell differentiation and function and confers fitness and the suppressor function onto the cells. Mutations that ablate Treg cells precipitate fatal and rapid autoimmune disease in both mice and humans, underscoring the essential nature of their holding immune responses in check. IL-2 and TCR signaling are required to maintain FOXP3 expression, and thus the Treg cell population, throughout life.

Treg cells are more abundant in human breast carcinomas than in normal breast tissue or peripheral blood. Rudensky found that the transcriptomes of Treg cells in breast tumors are more similar to those in normal breast tissue than to those in peripheral blood, with some notable differences. Specifically, the chemokine receptor CCR8 is upregulated in tumor-resident Treg cells relative to normal tissue-resident cells. The Treg cells in breast tumors are decidedly suppressive; eliminating them deters growth of the primary breast tumor, and unexpectedly deters the growth of lung metastases as well. These data point to inhibition of Treg cells as a potential therapeutic mechanism.

Treg cells inhibit the anti-tumor immune response. When their inhibition is removed, tumors shrink.

Treg cells go through differentiation in the thymus, and in the periphery in placental mammals as well. Rudensky explained that this pathway for extrathymic differentiation of Treg cells probably arose in placental mammals as a way to enforce maternal-fetal tolerance. It occurs primarily in the colon and small intestine, where the Treg cells function to support the assembly of a functional and beneficial microbial community.

Highlights

• Breast cancer was ahead of the rest of oncology in targeted hormone therapy but is behind in immunotherapy.
• The immunotherapy platform at MD Anderson takes a “reverse translation” approach to linking clinical trials and lab work; rather than starting with a hypothesis and developing therapies for patients, they use what works in patients to generate hypotheses to test.
• Oncolytic viruses had previously been engineered to replicate only in cancer cells; from now on, they will be made to evade the immune system or will be given before immunotherapy so the immunotherapy doesn’t undermine their efficacy.

The Potential and Promise of Immunotherapy in Breast Cancer

“For a long time, people thought breast tumors wouldn’t be susceptible to immunotherapy; they’re not very immunologic, they don’t have a lot of T cells,” began Elizabeth Mittendorf. However, in her opinion, trastuzumab, the monoclonal antibody used against HER2/neu protein positive breast cancer, can be considered the first triumph of immunotherapy. In fact, about 30% of breast tumors have tumor infiltrating lymphocytes (TILs), which can be used to make prognoses.

Metastatic triple-negative breast cancers (mTNBCs) are the most likely to harbor TILs. Patients with this type of breast cancer have the lowest survival rate, and thus the highest need for new types of therapy. In the KEYNOTE 012 Phase Ib clinical trial, the PD-1 inhibitor pembrolizumab increased the survival of 18.5% of patients with mTNBC—which is akin to the 20% of patients it helped with melanoma, where it was initially described. Not surprisingly, pembrolizumab was much more effective against those breast tumors that were PD-1 positive. Other clinical studies using another PD-1 inhibitor, atezolizumab, have confirmed that there may be a role for checkpoint blockade in breast cancer, primarily as a first line of defense. But Mittendorf stressed that we need to delve more into its potential.

In order to learn more about patients who get these drugs and how they fare, TRIBUTE—a Translational Resource for Immuno-Biology to Understand Therapeutic Efficacy—was founded. Its stated goal is to establish a repository of clinical data, biospecimens, and imaging results from breast cancer patients treated with immunotherapeutic agents administered in the context of routine clinical care. It started by focusing on neoadjuvant therapy: giving therapeutic agents at the time of diagnosis, before surgery. The hope was that such an early administration would generate a complete response—a pathology report showing no disease.

Mittendorf discussed three agents that are being explored in combination with immunotherapy: abemaciclib, PARP inhibitors, and HDAC inhibitors. Abemaciclib is a cell-cycle inhibitor that in mouse studies unexpectedly seemed to act through immune effects: including by stopping the proliferation of Treg cells and enhancing antigen presentation in tumor cells. Abemaciclib was further found to sensitize tumors to checkpoint blockade; and to date, at least two trials have indicated that it does in fact synergize with the PD-1 inhibitor pembrolizumab. PARP inhibitors have been effective against breast tumors with BRCA mutations—preclinical data suggested that this treatment, which inhibits DNA damage repair, would increase the development of mutations and therefore neoantigens and potentially sensitize to immunotherapy. Encouraging data suggest that in human trials immune checkpoint blockade in combination with PARP inhibitors is effective. Lastly, HDAC inhibitors have been found to recruit tumor associated macrophages in mouse models, and evidence suggest HDAC inhibitors in combination with an immunotherapy is effective. These examples highlight the importance of preclinical data in designing and rationalizing drug combinations.

The cell cycle inhibitor abemaciclib sensitizes the tumor environment to immunotherapy, synergizing with checkpoint blockade to shrink breast tumors.

Breast cancer is a “very different milieu for immunotherapy than melanoma or lung cancer because we have chemotherapies that work,” concluded Dr. Mittendorf. “It is hard to de-escalate therapies—to take away surgery and introduce immunotherapy—and to decide what level of toxicity can be introduced,” she continued.  “Breast cancer was ahead of the rest of oncology in targeted hormone therapy but is behind in immunotherapy.”

From the Clinic to the Lab: Investigating Response and Resistance Mechanisms to Immune Checkpoint Therapy

There are currently over 2,000 ongoing clinical trials for combination immunotherapies. “We need to be more rational” about them, said Padmanee Sharma. “We need to start from patients. We can’t just generate hypotheses; we must test them, in clinically relevant models.” To that end, the immunotherapy platform at the MD Anderson Cancer Center takes a “reverse translation” approach to linking clinical trials and lab work. Researchers collect pre- and post-treatment tissues from patients who receive a particular immunotherapeutic agent so they can compare data across different assays to determine what works; then they generate hypotheses to test in the lab.

Using this approach, Sharma’s lab observed that treating localized bladder cancer patients with the CTLA-4 inhibitor ipilimumab before surgery—surgery to resect the resect the tumor is the current standard of care—enriched T cell infiltration into tumors, compared to patients with surgery alone. Using gene expression analysis, they found that the ICOS (Inducible COStimulatory) pathway was highly upregulated in the infiltrating T cells, more so than any other pathway. This upregulation was confirmed in subsequent protein analysis, and was a surprising finding since ICOS had never before been shown to have antitumor activity, although it had been correlated with increased survival in metastatic melanoma. Based on these data, Sharma and her colleagues hypothesized that the ICOS pathway is necessary for the anti-tumor immune response in anti-CTLA-4 therapy. They confirmed this with experiments using ICOS knockout mice, where with anti-CTLA-4 treatment there was impaired anti-tumor responses compared to wild type mice. They further hypothesized that the ICOS pathway can be targeted as a combination therapy with anti-CTLA-4 or other immunotherapies. In mice they showed with combination therapy an increased anti-tumor response, not seen in ICOS knockout mice, suggesting this strategy may be effective in patients.

Treatment with CTLA-4 blockade induces the ICOS pathway.

Melanoma, bladder, and lung cancer, are tumor types thought to be “hot” tumors, with lots of antigens for immunotherapy to target. In contrast, “cold” tumors, such as breast and prostate cancer, have fewer antigens and infiltrating T cells, and it was thought that they would be therefore less responsive to immunotherapy. However, Sharma and colleagues hypothesized that since treatment with ipilimumab before surgery enriched T cell infiltration into tumors, that it may be used to treat these “cold” tumor types with lower mutational loads. Indeed, Sharma’s work has shown that the right combination of immunotherapies can make the prostate tumor environment more susceptible to treatment with checkpoint inhibition in some patients, in part by increasing the number of neoantigens expressed by the tumor.

However, immunotherapy also increases the immune checkpoints in the tumor that limit treatment response. PD-L1 expression is increased in prostate tumors treated with the anti-CTLA-4 drug ipilimumab, as is the expression of the novel immune checkpoint VISTA. Signaling through the interferon pathway—another mechanism the tumor uses to evade the immune response—ramps up in tumor cells, and epigenetic effects are seen as well. “All immune drugs generate a yin-yang response to control T cells,” said Sharma, “we must evaluate therapies in a longitudinal fashion” to see which combinations of therapies will ultimately work best for each cancer. This is the true value of the immunotherapy platform she heads at MD Anderson. Her access to patients and their tissue samples before and after treatment puts Sharma in a unique position to make just such longitudinal evaluations.

Oncolytic Viruses: Past Present Future

Matthew Mulvey began by giving an overview of the history on oncolytic viruses. As early as 1904, doctors noticed that when patients with blood cancers contracted an infection, their cancers went into spontaneous remission. When cell culture practices were established in the 1950s, clinical trials of wild-type viruses were undertaken to take advantage of the phenomenon; some remission was observed, but there were significant adverse effects. When chemotherapy was developed, it was much more efficacious and easier to control than viruses, so research and development into oncolytic viruses slowed considerably.

Onyx-015 was the first recombinant oncolytic virus, generated around the turn of the millennium. Its E1A gene is deleted, so it can only replicate in cancer cells that lack p53. It was designed to replicate until it killed off all of the cancer cells it infected. The patient’s viral response was never considered. It worked, but was abandoned for business reasons, though it was approved for use in China in 2005.

In 1999, an abscopal effect was first observed with the oncolytic virus G207, a highly attenuated HSV1; it killed uninjected tumor cells in mice by eliciting an anti-tumor CD8 effector T-cell response, when injected into a different tumor in the mice. This proof of concept paved the way for T-Vec, an attenuated HSV1 virus encoding human GM-CSF. T-Vec eliminated melanoma metastases and became the first FDA-approved oncolytic virus in 2015. T-Vec monotherapy had limited systemic efficacy, though, and only helped patients with Stage IIIB/C, IVM1a cancers. But T-Vec in combination with pembrolizumab was a different story. The oncolytic virotherapy promotes intratumoral T cell infiltration, and thus improves immunotherapy with anti-PD-1. A 62% response was seen, even though many patients had cold tumor microenvironments—they lacked the PD-1 that pembrolizumab blocks. Now, oncolytic viruses are designed with a “friendly” paradigm: to work in tandem with immunotherapy and the immune system.

“Oncolytic viruses are many drugs in one,” said Dr. Mulvey. “As such, they can hit three different steps in the cancer immunity cycle.” They can release cancer antigens by killing infected tumor cells; they can promote cancer neoantigen presentation by attracting dendritic cells into tumors; and they can prime and activate anti-tumor T-cells by inducing production of Type I Interferon and TLR/STING agonists (e.g. dsRNA/DNA) in the tumor microenvironment. The viruses also act to induce dendritic cell maturation, and can be engineered to express a customized suite of cytokines and antibodies.

Now that T-Vec has been approved as a first-in-class drug product—the first virus that can resist innate (but not adaptive) immunity—challenges remain in optimizing both its delivery and efficacy. This led to the development of T-Stealth at Benevir, a new oncolytic virus platform, which can evade adaptive immunity—anti-viral T cells—in addition to innate immunity elicited by interferons. Its replication mechanism has been hobbled, though; it can still replicate only in cancer cells, limiting off-target effects. To address this and other challenges, the next wave of oncolytic viruses should: evade the immune system; encode multiple transgenes to reverse complex immuno-suppressive tumor microenvironments; and be affordable so that intravenous dosing is commercially viable. Thus, the creation of “antigen agnostic” tumor viruses is an important next step; these viruses will act almost like an in situ vaccine, allowing the immune system to infiltrate tumors and fight them.

What are the fundamental mechanisms underlying checkpoint inhibitor induced tumor rejection?

What aspects of tumor intrinsic properties and immune response determine sensitivity to immune checkpoint blockade?

What makes CD8 cells responsive, or nonresponsive, to immune checkpoint blockade?

What is the impact of the tumor microenvironment on T cells’ infiltration into tumors and their recognition of cancer cells?

Are there biomarkers that might indicate which patients will respond to which (combinations of) therapies, and which patients won’t be able to tolerate those therapies?

Are there more cancer immune cycle pathways that can be targeted to improve clinical outcomes?

Large virus scaffolds, like HSV and vaccinia, are required to accommodate all of the new genetic material that makes oncolytic viruses. Currently, the only known way to culture them is to grow them in slowly rolling culture bottles. Can this process be affordably scaled up to generate enough virus to be clinically meaningful?