Ken Garber

Nature Biotechnology, March 2016

On January 11, Innate Pharma in Marseilles, France announced a research collaboration and licensing deal with Sanofi, based in Paris, worth up to €400 ($451) million in milestone payments and royalties. The goal is to develop bispecific antibodies that engage natural killer (NK) cells with tumor targets. It’s the latest sign of a surging interest in NK cell-based cancer immunotherapies which, last July, saw NK cell company ConkWest (now NantKwest) in Cardiff-by-the-Sea, California, haul in $260 million in its initial public offering (IPO). A clutch of companies and institutions pushing for the next big thing in oncology is moving forward with NK cell-based immunotherapy (Table 1). Forty years after their discovery, NK cells “have come of age,” says Michael Caligiuri, director of the Ohio State University comprehensive cancer center in Columbus, Ohio. Much of the enthusiasm is spillover from the success of T cell checkpoint inhibitors, which have transformed the treatment of metastatic melanoma, lung cancer and kidney cancer. But NK cells belong to the innate immune system—the rough and ready arm of the immune response—and their peculiarities as cytotoxic lymphocytes offer certain advantages that drug developers are keen to exploit.

After decades of neglect, the resurgence in interest in NK cells “has been dramatic,” says Nicolai Wagtmann, Innate Pharma’s chief scientific officer. “In the last year it has been a total change.” Unlike T cells, NK cells do not require tumor antigen recognition or clonal expansion before killing tumor cells. So far, human trials (J. Clin. Oncol. 33, 15S, abstract 3065, 2015) have not shown  the severe autoimmune-like side effects seen with checkpoint inhibitors nor the cytokine release syndrome that often accompanies chimeric antigen receptor T cell (CAR-T) therapy (Nat. Biotechnol.32, 604, 2014). But doubts persist about the ability of NK cell-based therapies to shrink bulky solid tumors and to achieve long remissions as single agents.

NK cells are effective killers, but cytotoxicity function is tightly controlled by a wide range of receptors on the cell surface, some of which are inhibitory and others activating. The inhibitory receptors—the most important of which is is the killer immunoglobulin-like receptor (KIR)—recognize major histocompatibility complex (MHC) class I molecules present on almost all nucleated cells in the body, and prevent NK cell activation. This ‘self-tolerance’ mechanism prevents host cell killing.

To attack, NK cells must first become activated. In stark contrast to T cells, this takes place when NK cells encounter cells lacking MHC class I, for instance pathogens or tumor cells that often downregulate MHC class I expression. This peculiar activation pattern can tip the balance in favor of NK cell activation overturning the NK cells’ usual inhibited state.

A group in Perugia, Itay in 2002 first showed that NK cells could greatly enhance tumor killing (Science 295, 2097-2100, 2002). Patients with acute myeloid leukemia receiving KIR-MHC class I mismatched bone marrow transplants achieved 100% five-year relapse-free survival, compared to only 25% for matched patients. “That showed the world how powerful NK cells could be,” says Caligiuri, who also sits on Innate’s supervisory board.

Soon after, companies set out to design therapies to reproduce pharmacologically the dramatic Perugia transplantation results–and boost NK cells’ cytotoxic potential–for use in oncology. In 2007, Innate Pharma began trials of an anti-KIR monoclonal antibody (mAb), in hematological malignancies. The strategy aimed to block  KIRs to activate NK cells to attack tumors. In 2011 Bristol-Myers Squibb (BMS) in Princeton, New Jersey, licensed an improved anti-KIR antibody, lirilumab, from Innate. BMS is conducting six combination trials with lirilumab, mostly with other immunotherapies, in blood and solid tumors, whereas Innate has completed a single agent maintenance therapy trial in AML patients in their first remission. Results of this trial should be released in the second half of this year, and if positive could eventually lead to drug approval. “It will depend on the data, especially the survival data,” says Wagtmann.

Anti-KIR antibodies have shown good safety so far.  Although the antibodies block KIRs, receptors that normally tolerize NK cells to self antigens, these antibodies don’t unleash NK cells against normal tissue to cause autoimmunity. That’s because NK cell activation depends not only on the absence (or blockade) of inhibitory receptor ligands, but also on the presence of ligands that bind activating receptors on NK cells. “It’s all a matter of shifting the balance of signaling in the NK cell toward activation and away from inhibition,” says Kerry Campbell, an immunologist at the Fox Chase Cancer Center in Philadelphia. Tumors, due to the DNA damage response, express ‘stress ligands’, which bind to NK cell activating receptors (Nature 436, 1186-90, 2005). Healthy tissues don’t express these activating ligands,” says Wagtmann, so these tissues don’t present as targets to NK cells.

One potential limitation of anti-KIR therapy is that it might paradoxically prevent NK cell activation. KIR receptors may need to be positively engaged to ‘arm’ the NK cell (Trends Immunol. 32, 364-72, 2011). But continuous KIR blockade with lirilumab could prevent NK cell arming, leaving NK cells inert, anergic and unable to kill. For this reason Innate, in one arm of its single-agent AML trial, gave intermittent doses of lirilumab, in case KIR receptors need to be available, which the trial data should clarify. The clinical relevance of ‘arming’ is controversial, though, so Innate dosed continuously in another trial arm. The biotech is also in phase 2 testing, with London-based pharma AstraZeneca, a different antibody, monalizumab, which targets a different inhibitory receptor.

To spur NK cells into action without resorting to KIR inhibition, the Sanofi and Innate are collaborating to develop a bispecific mAb antibody to target NK cell activating receptors instead. One antigen-binding fragment (Fab) targets NKp46–an NK-specific activating receptor– and the other Fab targets a tumor antigen. This brings the activated NK cell in physical proximity with a cancer cell. Amgen in Thousand Oaks, California uses the same strategy, though with T cells, for its Blyncyto (blinatumomab) bispecific, approved in December 2014 for acute lymphoblastic leukemia. But there can be side effects resembling cytokine-release syndrome, causing severe respiratory symptoms and very high fevers. “We think it would be a lot better to stimulate NK cells with bispecifics” than T cells, says Wagtmann. “They don’t confer cytokine storms, even when given at very high doses to people.” (One possible reason is that NK cells don’t secrete the inflammatory cytokine interleukin-6.)

Wagtmann considers combining NK cell bispecifics with checkpoint inhibitors promising, because certain cytokines and chemokines released in an innate response drive T cells to infiltrate tumors. And patients usually respond to T cell checkpoint inhibitors only when such T cell infiltrates are present . Using an automobile driving analogy, Wagtmann explains that releasing the “brake” with a checkpoint inhibitor doesn’t do much without pressing the accelerator. Both only happen when T cells are in contact with tumors, which requires a prior innate immune response, which NK cells in theory can provide.

Affimed Therapeutics in Heidelberg, Germany, is pursuing a different bispecific antibody strategy that is already in the clinic. One-half of its AFM13 construct targets CD16 (FcγRIII), a receptor expressed on NK cells that binds the Fc (fragment crystallizable) arm of IgG antibodies. (The other half binds a tumor antigen.)  CD16 binding triggers NK cell killing of tumor cells by antibody-dependent cellular cytotoxicity (ADCC), an important NK cell effector mechanism. AFM13 is in phase 2.

The major reservation about all NK cell immunotherapeutics is antitumor efficacy. NK cells are mostly shortlived and tumors can inactivate them by secreting cytokines like IL-10  and TGFβ and by shedding soluble ligands that bind NK cell activating receptors, desensitizing the NK cell. (Science 348, 136-9, 2015.)

Blocking one KIR subtype may not be enough to activate an NK cell. A cocktail of antibodies may be necessary for optimal efficacy. (J. Immunol. 180, 6392-401, 2008.) And while NK cells are important early in the antitumor immune response, “evidence is mounting that they can become overwhelmed and cannot do it themselves, they cannot eliminate particularly a solid tumor,” says Campbell, who thinks T cells must also be mobilized. Caligiuri, however, cites the Perugia transplantation results as evidence for NK cells’ antitumor potency. Campbell agrees that NK cells “have a very good capacity to kill AML tumors,” but otherwise sees them mainly working against single cells, in antitumor surveillance, metastasis suppression or in minimal residual disease to prevent relapse.

That’s too pessimistic, argues NantKwest CSO Hans Klingemann, citing anecdotal results from his company’s clinical trials. Since 2002 NantKwest has treated roughly 50 cancer patients with cells from NK-92, an NK cell line originally derived from the blood of a patient with NK lymphoma, and has achieved some durable remissions of established tumors. (The cells are irradiated to eliminate malignancy risk.) The company existed on a shoestring, burning less than $18 million during its first decade, but that all changed last year, with the massive takings from it’s the IPO and a large infusion of cash from billionaire and serial biotech entrepreneur Patrick Soon-Shiong, who now owns close to 60% of the company.

Klingemann sees NK-92 cells as an affordable, off-the-shelf, easily renewable alternative to autologous NK cell transplants. Most patients, for unknown reasons, do not reject these cells. The cell line does not express KIRs, so the cells are primed to kill, and so far they’ve been safe and selective for tumors, including solid tumors. NantKwest is modifying them for potency by incorporating a tumor-targeting CAR, to be tried soon in glioblastoma, and also high-affinity CD16 to enhance mAb ADCC, with a Herceptin combination trial also to begin later this year. Klingemann considers NK cells better CAR vehicles than T cells, because unlike T cells they do not persist for months or years, with side effects such as B cell depletion. And NK cells kill tumors more ways, including the indirect stimulation of a T cell response via cytokine-activated dendritic cells. The downsides are that irradiated NK-92 cells only survive briefly, making frequent infusions necessary, and they are unproven in large randomized trials.

That’s true for all NK cell immunotherapies. But with companies now flush with cash, the necessary clinical trials will take place. “You’ve got to do the in vivo experiment in the patient,” says Caligiuri. “You just don’t know.”