March of 2011 marked an important day in the history of how we treat cancer. The reason for its importance? In March 2011, the FDA approved a little molecule called Ipilimumab, better known as Yervoy, for use as a first line therapy in metastatic melanoma patients. Now, the FDA approves around 20-30 new drugs a year, so to call an approval drug a historic moment might appear hyperbolic at first. However, Yervoy was different. Yervoy’s approval marked the dawn of the immuno-oncology age.

Traditionally, cancer of various stages has been treated through the use of radiation, surgical resection, chemotherapy, or targeted therapies. Each method has advantages and disadvantages.

Targeted therapies were the last therapeutic modality for which the word “cure” was whispered. They arrived on the oncology scene with the approval of Herceptin, a monoclonal antibody (biologic therapy) that targets Breast Cancer expressing the HER2 receptor in 1998. However, relapse due to the targeted therapies’ ability to modify the evolutionary fitness pressures of a cancer microenvironment, doomed them as a “cure”. In fact, improvement followed by relapse describes the most common pattern of therapeutic response seen in patients treated with targeted agents.

Yervoy is not the most potent immuno-oncology therapeutic on the market today. But it was the first, and one that marked the beginning of a new horizon for potential therapeutics – a strategic supernova that suddenly all biotechnology and pharmaceutical companies reoriented their strategies around. Suddenly the most effective way of treating cancer was to let the immune system to the hard work for us.

How Do I/O Therapies Work?

For those who worried that Yervoy miracle would not be replicated, the interim 7 years has provided relief. As of 2017, Immuno-oncology is here to stay. The effectiveness of the the therapeutic modality, which we will delve into detail about later, is clear.. However, to understand the I/O craze, we must discuss just exactly how these therapies work, and identify the factors make them effective therapeutic agents. All immunotherapies are built around the biological principle that the immune system, particularly the adaptive immune system, can actively recognize tumors as foreign agents and catalyze their destruction. Thus all the therapies make use of the antigen-Major Histocompatibility Complex (MHC) that a traditional immune response, e.g., to a virus or bacterium, would use. In particular, most immunotherapies the market and in development attempt to mobilize Cytotoxic T-cell populations (the “killer” T-cells) to target cancer, taking advantage of T-cell specificity , memory, and adaptability.

Firstly, T cells are specific killers. That is, T-cells use T-Cell Receptors (TCRs) on their cell-surfaces to recognize specific antigens (small digested peptides) presented on Major Histocompatibility Complex (MHC) I and MHCII complexes of foreign and host cells, which limits off-target responses and allows the T-cell to recognize cancers that do not present foreign cell-surface receptors. Additionally, T-cells have memory. Although the primary T-cell response to a foreign invader is quite devastating, the real potential – illustrated by the success of prophylactic vaccination in viral diseases – is that this primary response is usually followed by the production of memory T and B cells that can result in accelerated response if a previously seen antigen is recognized. This overcomes one of the chief problems with targeted therapies – relapse. A T-cell can remember any previously seen antigen, preventing a cancer from overcoming an attack by driving expression of previously dominant mutation. Finally, T-cells are adaptable. The highly combinatorial process by which TCRs are generated allows mature T-cells to recognize 10^15 potential antigens . Thus, while a cancer might evolve in response to a targeted therapy, with the initial targeted antigen dying out and allowing the cancer to recur, a T-cell response can theoretically adapt to a changing tumor by recognizing new antigens and responding accordingly.

Past this broad statement however, there is considerable variation in how each type of therapy achieves durable outcomes. However, 3 distinct categories can be identified: checkpoint inhibitors, cell-based therapies, and therapeutic vaccines.

Checkpoint Inhibitors

All of the approved therapies on the market today are part of a class of agents known as checkpoint inhibitors, initially developed by James P. Allison, the inventor of Yervoy. These molecules are antibodies (that is, biologic therapies, covered usually in an inpatient or outpatient setting and covered by the Medicare Part A and B programs) that are injected into patients and target (i.e. noncovalently bind to) specific cell-surface receptors on either the surface of T-cells or the tumor cells.

Currently, there are three broad types of checkpoint inhibitors on the market: Anti-CTLA-4 antibodies , Anti-PD1 antibodies, and anti-PD-L1 antibodies .

Anti-CTLA-4 antibodies, such as Yervoy, target the molecule CTLA-4 present on T-cell surfaces. Anti-PD-1 antibodies, such as Opdivo and Keytruda, target PD-1, another cell-surface receptor found on T-cells. Anti-PD-L1 antibodies on the other hand, attempt to affect the same core immunological pathway as anti-PD-1 antibodies, but target the PD-L1 ligand found not on T-cells, but the tumor cells themselves.

The Yervoy Playbook

Yervoy was discovered in the mid-1990s by James P. Allison, a current MD Anderson Cancer Center faculty member. He was studying the underlying biology of T-cell activation, trying to answer the following question – what molecules are involved when a T-cell becomes activated against an immunological target? Evidence showed that the simple recognition of a foreign antigen on an MHC molecule failed to trigger an inflammatory immune response towards to hostile agent. Rather, co-stimulatory agents, presented by Antigen Presenting Cells (APCs) were also needed to induce a full response. These co-stimulation signals were mediated by a binding complex of the B7 molecule on the APC and the CD28 molecule on the T-cell. These concurrent events – antigen recognition by the TCR and B7-CD28 costimulation by APCs – allowed T-cells to mount a full response against a threat by inducing cytokine production, cell-cycle progression, and production of anti-apoptotic factors that result in proliferation of T cells.

However, unbridled stimulation of T-cells presents certain problems with regards to homeostasis. Continual cytokine production, T-cell differentiation, and proliferation can result in extreme damage to the body as a result of excess inflammation. Thus, genes exist to counteract the stimulatory effect of B7-CD28 binding, including those that encode a protein known as CTLA-4. CTLA-4, (please don’t ask me where the naming conventions come from in immunology) is a protein analog to CD28, similar in structure but inducing a depressive genetic program in T-cells. Thus, after initial co-stimulation, B7-CTLA4 binding would serve to compete with B7-CD28 binding and thus control T-cell responses – this is where the term “checkpoint” arises, as CTLA-4 appears to serve as a checkpoint in immune proliferation.

Yervoy is a humanized monoclonal antibody that blockades the CTLA-4 molecule on T-cells, attempting to ensure that T-cells continue down a proliferative developmental pathway and therefore improve the immune response to a foreign tumor.

While Yervoy has only one single agent indication and one combinatorial indication (with Opdivo), the fundamental model of checkpoint blockade has served to inspire a host of further research and developments. Indeed the 5 closest molecules to the clinic all follow the same pattern of checkpoint inhibition.

The key takeaways from the Yervoy success:

• T-cells are prevented from attacking tumors through tumor activation of suppressive developmental circuits in T-cells. Preventing this suppression is a route to therapeutic efficacy.
• Targeting molecules on the surface of the T-cell, as opposed to an inherent characteristic of the tumor, provides a methodology to allow therapeutic efficacy in multiple types of cancers, from different organs of origin or driven by differing genetic lesions. Checkpoint blockade then, is theoretically tumor agnostic .
• Checkpoint inhibition does not produce a therapeutic response or overall survival benefit in a majority of patients, but can produce durable responses in a subset of patients – the reasons for this are not yet clear.

Programmed Death-1 (PD-1): The Keystone Molecule for I/O.

While Yervoy can lay claim as the first major immunotherapy success to reach the clinic and the market, it cannot claim to be the most successful immunotherapy in oncology. This claim clearly lies with molecules targeting the PD-1/PD-L1 pathway. There are now 5 approved molecules targeting this pathway, with 24 of the 25 approved distinct uses (and all indications outside advanced melanoma) for these drugs coming from those that target the PD-1/PD-L1 regulatory axis.

Freeman et al 2000 was the first to identify PD-1 as a potential immune checkpoint, through elucidation of its ligand PD-L1. Activation of PD-1 by PD-L1 was shown to protect tumor cells by inducing T-cell apoptosis. PD-L1 (one of two ligands for PD-1)’s mechanism of action is not nearly as well characterized as CTLA-4’s. However targets against PD-1 and PD-L1 demonstrate greater and broader clinical efficacy than those targeting CTLA-4. It is known that PD-1, like CTLA-4 is expressed only after T-cell activation and co-stimulation. Mechanistic reasons for PD-1 effects include: downregulation of certain antiapoptotic molecules and proinflammatory cytokines, and effect the cell cycle, preventing progression through G1 phase.

Considerable thought has been given to the role of PD-1 and PD-L1 as predictive biomarkers of immunotherapy success, to little avail . We will give more thought to this question when discussing companion diagnostics. However it is sufficient for now to say that while increased expression of PD-L1 intuitively drives higher success rates in response to certain anti-PD-1 antibodies, there is no clear 1-1 mapping between PD-L1 expression and success of therapies targeting the PD-1/PD-L1 axis. Rather, success rates depend on the target antibody in question, the target companion diagnostic used, etc.

There are two additional concerns with PD-1 therapy. Firstly, it is unclear why a majority of patients do not see overall response rates rise (let alone considering the overall survival metric, for which an even fewer % see improvements). Secondly, it’s unclear what parameters govern the success of these therapies in different tumor types.

The concerns about the efficacy of PD-1 biomarkers as predictors of clinical success, as well as the lack of understanding of clinical responsiveness to immunotherapies in general demonstrate fundamental questions that remain to be answered around the question: “What governs the efficacy of immunotherapies?”, which additionally points to our lack of understanding of the underlying biology of tumor-immune interactions. Pace in the clinic due to initial successes with Keytruda, Opdivo, and Yervoy threaten to quickly overtake our biological understanding of their mechanisms of action. This problem is particularly compounded with the race to develop combination therapies as an alternative to immune single agent therapies, that is, combining immunotherapies with targeted therapies, other immune blockades, and traditional sources of therapy. One can quickly see that the combinatorial problems that arise when there is a lack of biological understanding – the sheer number of combination threatens to outpace the ability of clinicians to pick the best therapy for their patients.

The Future: Checkpoints and Combinations

Why Don’t Checkpoints Work for Everyone and Every Type of Cancer?

In addition to development of new checkpoints and new combinations, the second big area of research surrounding checkpoint immunotherapy surrounds the ability to further elucidate the mechanisms underlying effective response. This section summarizes research efforts in this area.

• –  The mutational burden theory. The prevailing theory in the oncology field today for explaining differential results to immunotherapies, both between patients, and between tumor types, surrounds the mutational burden of the tumor. How many mutations does the tumor have in it? This theory – that tumors with higher mutational burdens have higher objective response rates to checkpoint inhibitors – is derived from associative evidence from cohorts of patients that had higher objective responses to PD-1 checkpoint therapy. Nonsynonymous mutation load was correlated with improved objective response, durable clinical benefit, and progression-free survival. ( https://science.sciencemag.org/content/348/6230/124/tab-pdf , https://www.nejm.org/doi/full/10.1056/NEJMoa1406498, ). Additional evidence derives from a less quantitatively precise observation. Simply, the majority of approvals for checkpoint inhibitors have derived from cancer types with higher mutational burdens. However, what is also clear from these studies is that mutational burden alone is also not a sufficient predictor of response (let alone survival improvement) improvement in patients.
• –  Neoantigens, Inflammations, and Mechanics. To understand understand the differential response of patients to checkpoints, we first have to truly internalize what existing checkpoints do. They fundamentally modulate the entire immune system to respond to threats in a more aggressive way.
• They amplify the endogenous immune reaction (a.k.a. Preexisting immune reaction) to a tumor. This corresponds well to research ( https://www.jimmunol.org/content/196/1_Supplement/74.2  ) that indicates an inflammed tumor micro-environment is associated with improved response to checkpoint inhibitors. It is true that the “remove the brake” and “step on the gas” metaphors currently used to describe checkpoints accurately do so in a simplistic manner.. But checkpoint mechanisms thus far are 1.) Too broad and 2.) dependent on an initial immune reaction being present. After all, for releasing the brake to result in increased acceleration – there as to be an accelerator to press in the first place!
• In fact, new evidence from Steve Rosenberg at the NCI and Patrick Hwu at MDACC suggest that response to checkpoints can be completely replicated by set of clonally restricted (i.e. T-Cells responding to a set of specific antigens as opposed to the entire repertoire of T-cells) T-cells responding to a few number of neoantigens (antigens derived uniquely from tumor tissue and unique to tumor tissue). They have done this either bioinformatically or by isolating Tumor Infiltrating Lymphocytes (TILs) that do respond to tumors effectively and determining which antigens they responded to. That is, they’re answering the question: “What is specifically triggering the cells that DO get activated by the tumor”. The answer is that, in mouse models at least, one can recapitulate the response similar to that of our modern day checkpoint inhibitors with only a few activated T-cells. That’s not to say checkpoints aren’t necessary, but the reason for their lack of efficacy could be explained in part by certain types of cancers and certain types of tumors not containing the specific neoantigens and other preconditions that trigger an endogenous T-cell response. https://science.sciencemag.org/content/344/6184/641.long Work done by Steve Rosenberg represents the crowning achievement in this field. His lab demonstrated that a specific personalized neoantigen could be extracted and used to grow a set of T-cells that induced a durable response to a solid tumor – as close to the “holy grail” of personalized oncology that has been shown in the literature.The endogenous response theory of immunotherapy has additional support in some combination therapies discussed above. Combinations of chemotherapy and radiotherapy with checkpoint inhibitors could perform better than single-agents because the preceding lines of therapy injure cells, inducing more mutations, and a greater probability that an antigen will be created to induce an endogenous response. Aduro and Novartis are developing STING agonists that attempt to inflame the tumor microenvironment. And not two days ago , a paper in Nature demonstrated that a personal neoantigen vaccine could induce durable responses in 6/6 patients.

An Immuno-Oncology Timeline

April 2010: Seattle-based Dendreon wins FDA approval of sipuleucel-T (Provenge), a prostate cancer drug, the first treatment cleared to actively stimulate a patient’s own immune cells to fight cancer. The stock valuation soars to more than $6 billion for a while, helping other companies raise cash. The drug flopped in the marketplace amid reimbursement concerns, persistent questions about its efficacy, and tough competition from other prostate cancer drugs. Provenge however, is an oncolytic virus-based therapy, and currently one of two on the market along with Amgens’ Imlygic (laherparepvec).

March 2011: Bristol-Myers Squibb’s ipilimumab (Yervoy) won FDA approval. Melanoma patients on the targeted antibody drug had a median survival time of 10 months, compared with 6.4 months in the control group of a pivotal clinical trial. It was the culmination of years of work on the CTLA-4 target pathway. By making an antibody against the target, scientists thought they could “release the brakes” on the immune system. The drug didn’t work for everybody, and it had severe side effects. About 12.9 percent of patients suffered severe or fatal autoimmune reactions in the pivotal trial. One researcher said it was a ‘single, not a home run.’ Still, the drug was priced at $120,000 for a course of therapy.

August 2011: Carl June, a researcher at the University of Pennsylvania, published work on chimeric antigen receptor-modified (CAR-T) T-cells in the New England Journal of Medicine. It told the story of a very sick patient with chronic lymphocytic leukemia who had a remarkable response to the experimental treatment, in which T-cells were reprogrammed in the lab to enhance their tumor-killing capability, and then re-infused into the patient. “It was unexpected that the very low dose of chimeric antigen receptor T cells that we infused would result in a clinically evident antitumor response,” June and his team wrote. The paper caught the attention of multiple media outlets, and the business development people at Novartis.

August 2012: Novartis obtained an exclusive worldwide license to chimeric antigen-receptor T-cell (CAR-T) technology from the University of Pennsylvania. The company and university jointly established a cell therapy facility on Penn’s campus. Novartis acquired exclusive rights to CART-19, which targets a protein called CD19 associated with B-cell malignancies. Remarkable responses were seen in an early clinical trial three patients with chronic lymphocytic leukemia, whose disease had worsened after extensive pre-treatment. Two of the patients were still in complete remission more than a year into the trial, and the third had a partial remission for more than seven months. There were risks: Patients were treated for tumor lysis syndrome, a potentially deadly metabolic condition that occurs when cancer cells are being killed so fast that the body’s metabolic and excretion systems can’t keep up.

March 2013: Still a private company known mostly for viral vector gene therapies at the time, Cambridge, Mass.-based Bluebird Bio announced a partnership with Celgene to develop gene therapy technology to modify a patient’s T cells as cancer immunotherapies. The deal included an undisclosed upfront payment and $225 million per product in option fees and milestones. Three months later, Bluebird went public.

June 2013: Bristol-Myers, Genentech/Roche, and Merck all competed for center stage at the American Society of Clinical Oncology with antibodies that act as “checkpoint inhibitors” against the PD-1/PD-L1 target axis. Merck’s antibody, at a high dose, significantly shrank tumors in more than half of melanoma patients whose disease worsened after prior therapy. The Genentech/Roche drug showed an effect against multiple malignancies, including lung cancer, melanoma, and kidney, colorectal, and gastric cancers. Bristol-Myers’ drug significantly shrank tumors in 41 percent of myeloma patients. “It was really the rapidity and the magnitude of the responses that was impressive,” Jedd Wolchok of the Memorial Sloan-Kettering Cancer Center told the New York Times. With three big drugmakers investing heavily in antibodies against the same pathway, and all showing results, immunotherapy hit the big time. Roger Perlmutter, the head of R&D at Merck, called the meeting a “very special moment” in oncology. Mario Snozl, an oncologist at Yale, memorably said, “If you look five years out, most of this meeting will be about immunotherapy.” His prediction sounded bold then, but turned out to be conservative.

September 2013: Long-term survival data showed some impressive responses among patients on Bristol-Myers Squibb’s ipilimumab (Yervoy). Stephen Hodi of the Dana-Farber Cancer Institute in Boston reported on a pooled clinical trial analysis that showed an overall survival curve that flattened over time. About 21 percent of Yervoy patients were alive at three years, 17 percent were still alive after seven years, and no deaths were reported after that, stretching out to 9.9 years.

December 2013: Seattle-based Juno Therapeutics raised a $120 million Series A financing, making it one of the biggest biotech startup financings ever. The company was founded by scientists at the Fred Hutchinson Cancer Research Center, Seattle Children’s Research Institute, and Memorial Sloan-Kettering Cancer Center. Juno also obtained intellectual property from St. Jude Children’s Hospital. Clinical data, which hadn’t yet been reported in peer-reviewed literature, drove the big investment: Juno researchers showed they were able to trigger a complete molecular response—with no trace of measurable cancer in the blood—for 15 of the first 17 patients (88 percent) with acute lymphocytic leukemia. The company was built on CAR-T technology, as well as high-affinity T-cell receptor technology, which were thought to help it reach molecular targets on the surface of cells, and inside cells. Juno went on to raise $310 million in private financing in its first 12 months from Arch Venture Partners, the Alaska Permanent Fund, Venrock, and Amazon founder Jeff Bezos.

December 2013: Science magazine named cancer immunotherapy its “breakthrough of the year” for 2013. The leading journal stated: “This year marks a turning point in cancer, as long-sought efforts to unleash the immune system against tumors are paying off—even if the future remains a question mark.”

April 2014: Two patients in a clinical trial at Memorial Sloan-Kettering Cancer Center, testing a CAR-T immunotherapy that had been licensed to Juno, died in a clinical trial. The study was put on clinical hold as researchers said they needed to modify the study protocol to better manage the new treatments, which can cause an excessive inflammatory reaction known as a “cytokine storm” or a dangerous metabolic problem known as tumor lysis syndrome, in which tumors are being killed so fast the body can’t break them down and excrete them fast enough. The hold was lifted and the study resumed within a couple weeks.

Many in the industry breathed a sigh of relief. “They handled that in a very upfront way,” Wilson said. “These are incredibly potent therapies that do have toxicities. The basic learning from that experience is that we’re learning how to manage those. That could have been a key event, but it hasn’t slowed Juno at all. Credit to them.”

June 2014: Combination therapy data on checkpoint inhibitors ipilimumab (Yervoy) and nivolumab (Opdivo) presented at ASCO showed a 79 percent 2-year survival rate in melanoma. More and more companies draw up plans for combo trials.

June 2014: Kite Pharma, the Santa Monica, Calif.-based company developing CAR-T technology from Steve Rosenberg’s lab at the National Cancer Institute, went public at $17. It climbed 50 percent on its first day.

June 2014: Shortly after its mega-bid for AstraZeneca fell through, Pfizer agreed to pay $80 million upfront to collaborate with Paris-based Cellectis, a developer of allogeneic CAR-T immunotherapies. These “off-the-shelf” CAR-Ts are designed so they don’t have to be custom manufactured, which ought to make them cheaper and easier to mass produce than autologous treatments based on a patient’s own cells.

September 2014: Merck wins FDA approval of pembrolizumab (Keytruda) for melanoma. The FDA finished its review almost two months ahead of schedule. It was the first antibody on the market against the PD-1 target, designed to work by inhibiting a cloaking mechanism tumors use to disguise themselves from the immune system. About one-fourth of melanoma patients had their tumors shrink in clinical trials. The price was set at $12,500 a month, or $150,000 a year. Tim Anderson, an analyst with Sanford Bernstein, predicted $3.5 billion in annual sales.

October 2014: Data published in the New England Journal of Medicine show that the Novartis/Penn CAR-T therapy produced complete responses for 27 of 30 pediatric and adult patients with relapsed/refractory acute lymphoblastic leukemia (90 percent). Sustained remissions were seen as long as two years, and median follow-up time was six months. The reported overall survival rate was 78 percent, in an otherwise very ill population.

November 2014: Pfizer, seeking to catch up to Merck, Bristol-Myers Squibb, and Genentech/Roche, pays $850 million upfront to Germany-based Merck KGaA, plus $2 billion in potential milestone payments, for the right to co-develop an experimental PD-L1 antibody. Pfizer maps out an ambitious clinical development plan with 20 clinical programs slated for 2015.

November 2014: Merck, along with researchers at UCLA’s Jonsson Comprehensive Cancer Center, the Fred Hutchinson Cancer Research Center and Adaptive Biotechnologies, reported in Nature that they were able predict which patients with melanoma were likely to respond to treatment with Merck’s Keytruda, based on whether they had enough tumor-infiltrating killer T-cells in the vicinity of the tumor. Patients with fewer tumor-infiltrating killer T-cells still responded to Keytruda, but the percentages were smaller. The findings offered hope for companies looking to stratify patients to boost the odds of success in clinical trials, and justify high drug prices to insurers.

December 2014: Developers of CAR-T immunotherapies jockey for attention at the American Society of Hematology meeting, a showcase for blood cancer treatments. Juno Therapeutics, setting the stage for its IPO, presented follow-up data to show that its treatment for acute lymphocytic leukemia was standing up to the test of time. Not to be outdone, Kite advisor Steve Rosenberg gave a talk titled “Curative Potential of Cell Transfer Therapy for Cancer.” The day after ASH, Kite completed a secondary stock offering at $54 a share—triple the IPO price from six months earlier.

December 2014: Juno netted $281 million in its initial public offering. Shares rocket 60 percent on the first day, driving the company’s market valuation beyond $4 billion. Another aspiring CAR-T company, Houston-based Bellicum Pharmaceuticals raises $140 million the same week before Christmas.

December 2014: Bristol-Myers Squibb won FDA approval for its anti-PD-1 antibody drug, nivolumab (Opdivo) for patients with metastatic melanoma who no longer are helped by other drugs. The price was set at $12,500 a month, or $150,000 a year, same as Merck’s pembrolizumab (Keytruda).

February 2015: Bristol-Myers Squibb agrees to pay $800 million upfront, and potentially $1.25 billion with milestones, to acquire Flexus Biosciences for its small-molecule IDO1 inhibitor in preclinical development. It’s a huge sum for a preclinical asset with its value still to be determined in combination trials with other immunotherapies.

March 2015: Bristol-Myers secures a crucial second FDA approval of nivolumab (Opdivo) for patients with non-small cell lung cancer. The approval came three months ahead of schedule. It was the first cancer immunotherapy for lung cancer, and a way for Bristol-Myers to wrestle back first-mover advantage back from Merck, which had taken that honor in melanoma. The lung cancer approval was important because it proved checkpoint inhibitors can work in cancers other than those considered “immune-related” like melanoma. “The robustness of the lung data helped break that mold,” said Rich Murray, CEO of Cambridge, Mass.-based Jounce Therapeutics, an immunotherapy developer.

March 2015: Cellectis raises $228 million in its IPO, raising about twice as much money as it had originally sought.

March 2015: Novartis pays $200 million upfront for a partnership with Berkeley, CA-based Aduro Biotech to develop its cyclic dinucleotide small-molecules that are made to modulate the Stimulator of Interferon Genes (STING) receptor. By using a different type of treatment, against a different biological pathway, Aduro said it hoped to trigger both a “broad innate and adaptive tumor-specific immune response.” The deal adds to the Novartis immunotherapy portfolio.

April 2015: Novartis agrees to pay Juno $12.25 million upfront, plus milestones and royalties, to settle an intellectual property dispute. That settles one of the lingering questions of IP ownership.

April 2015: Merck shows it is still in the game with Bristol-Myers Squibb with its PD-1 inhibitor for lung cancer. Merck said pembrolizumab (Keytruda) shrank tumors in about 45 percent of lung cancer patients who expressed high levels of the PD-L1 protein, compared with 16 percent who expressed lower levels of the drug’s target. Based on the data, the company said it filed an application for FDA approval for both squamous and non-squamous forms of lung cancer.

April 2015: Juno reports that 20 of 22 pediatric patients (91 percent) with acute lymphoblastic leukemia, who failed to respond to other treatment, had complete responses on its CD-19-directed CAR-T therapy. Results were presented at the American Association of Cancer Research.

April 2015: Celgene struck a deal to co-develop AstraZeneca’s anti-PD-L1 antibody, MEDI3746, across a range of blood cancers. Celgene paid $450 million upfront and agreed to pay the R&D costs through 2016, and cover three-fourths of the R&D costs thereafter.

April 2015: Berkeley, Calif.-based Aduro Biotech raises $119 million in its IPO. Its stock more than doubles the first day.

May 2015: U.K.-based Adaptimmune Therapeutics raises $191 million in an IPO to help advance its high-affinity engineered T-cells which can recognize targets inside cells, not just on the cell surface where many CAR-T cell therapies operate. The IPO gives Adaptimmune a market valuation of more than $1 billion from the start. Along with its partner GlaxoSmithKline, Adaptimmune is pushing ahead with an immunotherapy to fight cancers that have the NY-ESO antigen.

May 2015: The American Society of Clinical Oncology posts abstracts of clinical data presentations at its annual meeting in Chicago. Researchers post preliminary results from a wide variety of combinations, many of which include the CTLA-4 and PD-1/PD-L1 inhibitors.

These combinations include combining two I/O drugs together (e.g. Opdivo + Yervoy, Keytruda + Yervoy, etc.) nad  I/O + Chemotherapy (e.g. KEYNOTE-021 trial),   

January 2017: Neon Therapeutics raises  $70 Million Series B venture financing after a $55 Million Series A 15 months ago. Neon raises around the idea to create neoantigen vaccines that could be used in combination treatments with checkpoint inhibitors to increase overall response rates.

June 2017: The annual American Society of Clinical Oncology meeting is held.

Takeaways:

• Bluebird Bio amassed further evidence for its CAR-T program, bb2121, for relapsed and treatment-resistant multiple myeloma, aimed at the BCMA antigen, with a 4-1BB costimulatory motif and a CD3-zeta T cell activation domain. 89% objective response rate (16 of 18 patients), with 15 out of 15 with an objective response rate at active doses. Of those 15 on active doses, 73 percent of them were classified as very good partial responses, while the rest were Complete Responses.. Cytokine release syndrome was seen in 71 percent of patients, but no treatment-related deaths were noticed and no cerebral edemas of the sort that have been seen in CD19-directed CAR-Ts from Juno and Kite. JP Morgan analyst Cory Kasimov called Bluebird a “clear winner” from the weekend at ASCO.
• BMS announced data regarding Opdivo/Yervoy combos; Opdivo in combo with Incyte’s IDO1 inhibitor epacadostat; and Sprycel for chronic myeloid leukemia.
• Juno Therapeutics lead CD19-directed CAR-T program, which was shelved after a series of patients died from severe brain swelling (cerebral edemas). Juno advanced a new CD19-directed CAR-T program, JCAR017, with a different co-stimulatory domain (4-1BB) and a defined balanced between CD4 “helper” T cells that can dampen or regulate an immune response, along with the CD8 “killer” T cells that are needed to kill the tumor (but hopefully not kill the patient). At ASCO, Juno presented results from the Transcend study of JCAR017 in patients with relapsed and treatment-resistant non-Hodgkin’s lymphoma. Juno enrolled 71 patients in total in this safety study, but focused on 44 in the “core group” with diffuse large B-cell lymphoma, the indication on its radar for pivotal testing. In that core group, Juno reported overall responses for 38 of 44 (86 percent) of patients, and complete responses for 26 of 44 (59 percent). But after 3 months of follow-up, the overall response rate dipped to 21 of 32 (66 percent), and the complete response rate was 16 of 32 (50 percent). Most importantly, there were no patient deaths from cytokine release syndrome or neurotoxicity, although one 82-year-old patient died from diffuse alveolar damage, although that patient had a number of other problems that contributed to the situation.
• Kite Pharma released data on the ZUMA-3 trial in very sick patients with Acute Lymphoblastic Leukemia (ALL). There was 1 case of patient death due to cytokine release syndrome, with a majority experiencing Grade 3+ cytokine release syndrome side effects. But 8/11 patients had complete responses measured by minimal residual disease test. Kite’s market capitalization exiting ASCO was $5.1B.
• Merck released a lot data. Firstly long-term follow up data demonstrates that Keytruda outperforms Yervoy in a head-to-head study in advanced melanoma patients with a 30% relative improvement in overall survival. Secondly, responses in MSI-H and dMMR solid tumors held up long term.
• Novartis reported that its CTL-019 therapy generated a 45% overall response rate and 36% complete response rate after 3 months in 51 patients with relapsed diffuse large B-cell Lymphoma.

July 2017:

• An FDA expert advisory panel unanimously (10-0) reported to recommend approval of the Novartis CTL-019 product. It was recommended for children and young adults aged 3 to 25 who have recurring forms of the rare blood cancer B-cell acute lymphoblastic leukemia (ALL).

The I/O Landscape as of 2017

In the 6 years since Yervoy was approved, 5 additional immuno-oncology therapeutics have been approved. They are:

• Opdivo (Nivolumab) – Manufactured by BMS
• Keytruda (Pembrolizumab)- Manufactured by Merck.
• Bavencio (Avelumab) – Manufactured by Pfizer.
• Imfinzi – Manufactured by AstraZeneca
• Tencentriq (Atezolizumab)  – Manufactured by Genentech/Roche (both are co-commercializing the molecule)

However, the main 3 immuno-oncology therapeutics are Opdivo, Keytruda, and Yervoy. Each other drug was approved in 2016 or 2017, limiting the knowledge we know about their performance post-approval.

*Note: The main difference between Opdivo and Keytruda are in the constructions of the antibodies themselves , information which is generally kept proprietary. However, both target the same pathway and company beliefs about each molecules’ relative effectiveness can be expressed as a function of the trial design and companion diagnostics used (See: “Companion Diagnostics” section).

Currently, these 6 therapies are approved for 170+ unique ICD-10 indications and 27 unique use cases across 7 therapeutic categories: Bladder Cancer, Head and Neck Cancer, Kidney Cancer, Lung Cancer, Lymphoma, Melanoma, and Merkel Cell Cancer, including one “organ-independent” indication*.

* Keytruda was approved in 2017 for for adult and pediatric patients with unresectable or metastatic, microsatellite instability-high (MSI-H) or mismatch repair deficient (dMMR) solid tumors that have progressed following prior treatment), marking the first organ-independent oncology approval.

The 27 Approval Use Cases and their associated approval clinical trials are As Follows:

As the chart above demonstrates, a majority of immuno-oncology drugs are not first-line therapies. Rather they are used for late-stage cancers (metastatic neoplasms, which are generally Stage III or IV cancers on the TNM scale, see here for more information: https://en.wikipedia.org/wiki/TNM_staging_system ) often after multiple rounds of prior therapy including resection, radiation, chemotherapy, use of targeted therapies. Thus far, the risk of I/O therapies limit them to drugs of last resort.

An example of a current physician workflow for I/O in NSCLC:

Companion Diagnostics

Also relevant to the current immuno-oncology landscape are “companion diagnostics”. According to the FDA, a companion diagnostic is defined as, “medical device, often an in vitro device, which provides information that is essential for the safe and effective use of a corresponding drug”. In plain english, when a drug’s’ clinical efficacy is mediated through a particular molecular pathway – one that can be tested for – its worth it to do so. The presence of available, relatively-cheap, sequencing, histology-based, and microarray diagnostics have additionally made it possible to reduce the doubt involved in choosing a therapy. These companion diagnostics are controlled by FDA approval pathways analogous to those used of medical device approvals, and prior to the onset of immuno-therapeutics, were primarily used in associated with targeted therapies. This makes a certain amount of intuitive sense – why give an anti-HER2 antibody therapy to a patient whose breast cancer does not contain appreciable amounts of HER2 receptor in the tissue? However, two things are changing with regards to this usage of companion diagnostics:

Firstly, precision oncology is generally becoming an accepted principle in FDA approvals . Generally, thus far precision medicine has rightly been considered to be an unrealized idea. But in 2017 thus far , the FDA had approved 13 oncology-related treatments that either are explicit therapy-diagnostic pairs, as with Illumina’s RAS sequencing test for use of Vectibix, or are drugs that are conditioned on molecular results of FDA-approved diagnostics.

Secondly, immuno-oncology itself is becoming precise. The current wave of approved therapies in I/O only involve 2 different immunological pathways, that involving the cell-surface receptors CTLA-4 (Yervoy) and PD-1/PD-L1 (the rest). Thus it becomes tenable to understand the the quantified presence of those molecules in a patient’s’ tumor through the use of histological evidence. In the case of Keytruda, the clinical trials design have explicitly ensured that a positive result from a companion diagnostic (22C3 PharmDx IHC assay) is needed to progress with therapy. But even with the other

approved therapies, companion diagnostics have been developed (e.g. 28-8 PharmDx assay for Opdivo) as a guide for physicians to use in prescribing the therapy, even if not explicitly required to treat. The current practice of

pharmaceutical companies is to independently develop an anti-PD-L1 IHC diagnostic assay for their therapies.

Yervoy however, does not have an associated companion diagnostic. The ostensible reason for this is not that a companion diagnostic, should one be developed, would not be useful (still, only a minority of patients have durable responses under Yervoy), but rather because the targeted molecule CTLA-4 is a cell-surface receptor on the T-cells ( therefore a minority of the tumor microenvironment, and thus hard to detect with a histologically based test) as opposed to PD-L1, a ligand present on tumor cells , and thus more easily detectable in bulk using histological tests. Currently, it is not known whether there are predictive markers for the success of Yervoy. However, the presence of hybrid-capture genetic sequencing methods could potentially make a sequencing based companion diagnostic possible for Yervoy.

The case study for the use companion diagnostics surrounds Merck’s Keytruda. Keytruda has a strict cut off on PD-L1 % required by the FDA to prescribe it for NSCLC indications – 50%. This is comparatively high compared to Opdivo ( PD-L1 % > 1%). However, in the KEYNOTE-001 clinical trial, patients meeting this cutoff and experienced a 58.3%, median progression-free survival of 12.5 months, and 24-month overall survival of 60.6%, with minimal outcomes benefits seen at any lower level of PD-L1 expression. While this use of a companion diagnostic as a required element of treatment as opposed to its complementary use as in Opdivo, might seem disadvantageous, it does yield benefits. As an example, the 2016 Phase III CheckMate-02 trial, Opdivo’s more liberally defined criteria for inclusion (PD-L1 expression > 5%) failed as a first-line therapy, compared to Keytruda’s success as a first-line therapy for patients with a stricter PD-L1 expression.

However, the use of comparatively blunt companion diagnostics makes the direct effect of these cutoffs unclear. While overall survival does improve with strict PD-L1 requirements, these cutoffs are insufficient to generate response in a majority of patients, indicating there is more to learn about the presence of certain molecules in controlling response to immuno-oncology therapeutics. Cellular, spatial, genetic, and temporal heterogeneity all contribute to the poor prediction accuracy of PD-1 as a clinical biomarker.

https://www.nejm.org/doi/full/10.1056/NEJMoa1606774#t=article

Additionally, it’s unclear what the net effect of these restrictions on companion usage have been on sales. Overall, Opdivo has outsold Keytruda by a significant margin. Opdivo’s 6 year sales in the Medicare Part B population total is $58,944,617,340, while Keytruda’s 6 year sales total is just $5,323,826,186. However, Opdivo’s first approval was in September of 2014, while Keytruda has only ben on market since late 2015. Additionally, although both Keytruda and Opdivo are approved for 9 indications as of the end of 2016, the differences in the underlying patient populations for each of these indications makes it difficult which drug has a larger total-addressable market. CareSet’s Medicare Claims data only contains proof of Medicare-claims for J9271 (Keytruda) and J9299 (Opdivo) for 2016, making it difficult to track utilization trends over time, and thus identify the sales ramps and growth tendencies of these drugs (prior to 2016, these drugs were given generic biologic J-codes, which undermines the quality of the data to track trends over time). As additional data becomes available, an answer could emerge regarding the differential effects of companion diagnostics on sales.

A Comprehensive Overview of the I/O Pipeline

However, despite the only 6 FDA-approved therapies, Clinicaltrials.gov demonstrates how full immuno-oncology pipelines are, and why they truly represent a “strategic supernova” for pharmaceutical companies.

As of 2017, Clinicaltrials.gov noted 1,184 clinical trials from PhI – PhIV using only approved immune-checkpoint inhibitors. Keytruda alone is an agent in 543 clinical trials. This too discounts the true number of immuno-oncology related clinical trials because it does not include “other” types of immuno-oncology treatments, including molecules that attack different targets, but in a similar mechanism to existing drugs, as well as completely differing approaches to immunotherapy in cancer, such as cell-based therapies and vaccines.

As mentioned in prior sections, the next set of checkpoint inhibitors that will come to market fall into two categories, checkpoint combinations, as well as new checkpoint inhibitors.

Implications for Pharmaceutical Industry Strategy

As I mentioned previously, the I/O landscape will likely define many biopharma strategic initiatives moving forward. It is a so called, ‘strategic supernova’.

References:

https://www.targetedonc.com/publications/targeted-therapy-news/2017/february-2017/mutational-load-is-only-one-piece -of-the-immunotherapy-puzzle

https://www.merck.com/product/usa/pi_circulars/k/keytruda/keytruda_pi.pdf https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4993154/ https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3892798/ https://www.nejm.org/doi/full/10.1056/NEJMoa1003466#t=abstract

https://www.nejm.org/doi/full/10.1056/NEJMoa1606774#t=article

https://lifescivc.com/2015/04/its-the-antigens-stupid/

https://lifescivc.com/2016/06/io-strategic-supernova-cancer-today/

https://science.sciencemag.org/content/344/6184/641.long

https://pharmaintelligence.informa.com/resources/key-topics/immuno-oncology?utm_source=twitter&utm_medium=soci al&utm_campaign=Immuno-Oncology%2520Combinations

https://www.targetedonc.com/publications/targeted-therapies-cancer/2016/april-2016/next-generation-diagnostics-for-i mmuno-oncology?p=2

https://science.sciencemag.org/content/344/6184/641.long