NantKwest's IPO was this week, and it is time for investors to take a closer look at the newest immunotherapy player on the market. The distinguishing feature for NantKwest is its focus on natural killer cells as cell therapy effectors, and more specifically, its focus on natural killer cell lines as a cell therapy effector. Their roadshow presentation has been posted on youtube: here.
NantKwest owns the NK-92 cell line, which is a natural killer cell line derived from a patient with leukemia. It is continuously growing like other tumor cell lines, and is dependent on IL-2 in the media for continued expansion. The NK-92 cell line appears to be frozen in time in a permanently activated state, with upregulated effector functions making it well suited for therapy. In order to transfer these cells into the clinic, they first need to be irradiated, which causes mutations in their DNA arresting further cell division. This procedure might sound complicated, but is fairly routine both in the laboratory (ex: irradiated MEF feeder cells) and in the clinic for a variety of applications.
|Image from NantKwest's website|
Since the NK-92 cells will be transient in the patient, repeated infusions are required. Furthermore, these repeated infusions may elicit an immune response to the NK-92 themselves. In the 2013 reported clinical trial, they reported 1 of 7 patients developed anti-HLA antibodies at 4 weeks post-infusion, and this patient might have actually developed those antibodies from unrelated transfusion during the treatment period. One thing noted in the academic papers is that the patients are already greatly immunosuppressed, with many having gone on various chemotherapy and bone marrow transplants before trying this therapy. They likely don't even have an efficient immune system in order to produce these antibodies. Ultimately, this is a question that requires close monitoring for the future, since the number of infusions required over a given span for treatment efficacy will be vital for NantKwest's success or failure.
Dr. Soon-Shiong outlined three different treatment paradigms in the roadshow video for investors. The first is the unalterated NK-92 cell line, which they refer to as aNK's. The second is a common derivative cell line expressing the high affinity CD16 receptor (haNK's). Note that NK-92 cells resemble the CD56 bright phenotype of NK cells and are CD16 receptor negative, contrasting with CD56 dim NK cells that are positive for CD16. This cell line is used in research to study the process of ADCC (antibody dependent cell-mediated cytotoxicity), both in the lab and in clinical testing for therapeutic monoclonal antibodies. This last fact is something Dr. Soon-Shiong emphasizes, since NantKwest intends to push the paradigm of co-administering NK-92 cells with therapeutic monoclonal antibodies (discussed more below).
|Image from NantKwest Website|
The third platform discussed is likely the most promising, which are CAR-directed NK-92 cells, which they call taNKs. The chimeric antigen receptors resemble the same one's Juno, Kite, and Novartis are pursuing as platform for T cells. The signaling would be the same, but with perhaps slightly different killing and cytokine release mechanisms due to differences between NK and T cell biology. For instance, a review by Dr. Hans Klingemann, discoverer of the NK-92 cell line, states that NK cells would not cause cytokine release syndrome, unlike T cell therapies, due to low IL-6 secretion upon CAR activation.
I will now discuss various specific elements of NantKwest's presentation and their strategies moving forward:
- The haNK platform is somewhat similar to Unum Therapeutics, wherein high affinity CD16 can bind any given anti-tumor antibody. Unum's platform engineers a high affinity CD16 ectodomain onto T cell signaling (CD28, Zeta, etc) endodomains for activation. Both Unum and now NantKwest highlight lots of partnering opportunities with current monoclonal antibody companies for their respective platforms. Presumably, Unum would argue that their T cells would be long-lived, allowing for continuous re-arming, while NantKwest is limited to a "hit-and-run" strategy of increasing the efficacy of the initial peak of monoclonal antibody concentration.
- The patients highlighted in the presentation with 15 and 18 infusions of aNK's over several months previously had a really harsh chemotherapy regimens, and seemingly don't have any immune system to react to the allogeneic cells. The patients that have tolerated multiple infusions therefore might not be truly representative of most cancer patients.
- The treated Hodgkin's lymphoma patient with complete remission (CR) patient for 8 years was speculated to achieve this status due to a stimulated host immune response, since the aNK therapy is only transient in nature. Dr. Soon-Shiong speculated that the aNK's triggered a host immune response via inducing the immune response against EBV infected the patients' tumor. This potential vaccination effect is very intriguing, since it would allow a transient NK therapy to turn into a long-term host T cell immune response. One time case reports are just that, though, as many therapies have rare astounding successes, so investors should taper enthusiasm until more data is in.
|Screen Capture from NantKwest Roadshow Presentation|
- Dr. Soon-Shiong showed some unpublished data in a mouse model that further investigates this hypothesis. The slide was brief, but appeared to be a syngeneic CD19+ lymphoma model with tumor inject into the right flank of the mouse. haNK's against CD19 were infused, and then this treatment was able to induce host mouse T cells to react and clear tumor compared to controls. This was one of the most potentially interesting phenomena I saw in the presentation, but again, this needs to be both rigorously evaluated in academic labs and in the clinic.
- The Merkel cell carinoma target is interesting for their aNK technology. Dr. Soon-Shiong talked about the NK versus Merkel cell polyomavirus, but MCC's typically only express Large T antigen and are closer in phenotype to tumor cells. Regardless, it's an orphan indication that will help validate their model and get potentially an early FDA clearance. I'm curious about the efficiency of homing for aNK's to the tumor site, and how this effects the dose and number of infusions needed. Possible combinations with Abraxane were also mentioned, consistent with Dr. Soon-Shiong's pioneering of that therapy.
- A note on the clinical trials reported by NantKwest. Before Dr. Soon-Shiong got involved, the company was a small biotech and NK-92 testing was largely done in only a handful of academic centers among investigators there. Most academic researchers have been focused on T cells, so there haven't been many other reported trials for aNK's. With these largely academic trials, the scale was small and the tumors treated varied, since it was thought that aNK's could be active against multiple tumor types (which they are in vitro). There was no level of antigen specificity and targeting a specific cancer like there has been in the CAR T cell and CD19 / ALL story. Therefore, the Phase I data is very sporadic, lacks direction, and mostly reads like clinical case reports, rather than a vigorous clinical trial. With this disclaimer, only 2 responses out of 40+ patients in previous trials is not very inspiring, with the highlighted single Hodgkin's Lymphoma patient being from 8 years ago. One would hope more progress would have been made. With the large cash infusion from the IPO and scaling for Phase II trials, we will hopefully get a better sense of any therapeutic efficacy. On this note, the reported target dose for Phase II aNK trial is 3 x 10^9 cells, and anticipated number of infusions is 15 or greater.
- What the benefit of a large infusion of hanNK's with timing of current monoclonal antibody (ex: Hereceptin) infusions is uncertain, even though Dr. Soon-Shiong pushed this concept significantly. The connection between the CD16 affinity and MAb response in cancer patients is impressive, but the analogy to infusing haNK's isn't equivalent. NantKwest showed some mouse studies briefly that are unpublished with anti-CD123 that showed some evidence that it should work in vivo, but much more data is needed for clear evaluation. Concerning haNK's further, there is a lot of existing competition of mutations in the Fc domains that increase binding to CD16 receptors, to get around this low affinity problem. NantKwest might argue that their strategy using existing drugs that are already scaled compared to any new players, but this alternative strategy should be noted.
- NantKwest appears to be pushing the concept of a PDL1 CAR, which I don't think is being pursued by any of the T cell companies. I feel like the PDL1 taNK would be highly susceptible to being off-target, even if PDL1 is overexpressed on tumors. Immune cells express PDL1, and most anti-PDL1 antibodies are engineered to lack ADCC activity for safety reasons. The taNK's would be a powerful effector against PDL1, much more so than ADCC inducing monoclonal antibodes. The alternative view is that since the NK cells only around 2-3 days, toxicity might not matter since it will be transient and most cells will in theory go to the highly PDL1 positive tumor. I suspect that NantKwest is pushing this target to try to have the most broadly applicable antigen target for tumors as possible, allowing any approval to be quickly applied to other tumor types. It's a high risk, high reward strategy that differentiates them from some of the other CAR players.
- Dr. Soon-Shiong highlighted NantKwest's capabilities in neoantigen discovery (coming from technology of his other companies). I am not as sure how the sequencing fits into their platform. For whatever target they discover, they will still need to build a CAR or a TCR-like CAR into their NK cell platforms. Unless the targets from the deep sequencing of neoepitopes is applicable for multiple patients and not for discovering personalized mutations, I don't see how this integrates into their off the shelf strategy. Many academic, not to mention, industry players are continuously screening for new CAR targets, and the easiest route is likely translating these known targets to the clinic.
- Given the successes and notoriety of CAR-T therapies, the taNK appears to be NantKwest's most interesting approach, however a number of major questions remain. Will NK-92 CARs prove to be as good effector cells as T cells? Will the repeated infusions of NK cells make up for their lack of persistence relative to T cells? The targets NantKwest is pursuing for their taNKs include some that would potentially have severe off-tumor toxicities, such as HER2 and PDL1. One of the defining aspects of 2nd generation CAR-T therapies is their incredible potency against even very low levels of target antigen. This makes them very potent, but also quite dangerous if the target antigen is expressed even at low levels on essential off-tumor tissue. If taNKs are as potent as CAR-T cells, I would worry about the toxicities, even if taNK persistence is self-limiting. Otherwise, this might suggest NantKwest feels their taNKs are likely to be less potent either in magnitude of response or sensitivity to low levels of target antigen compared to CAR-T cells, and thus taNKs may be a less potent (but less toxic) adoptive cell therapy approach.
In all, while NantKwest's approaches are theoretically interesting, they'll need a lot more clinical validation to justify their record biotech IPO.