Part of the Article Series from Nature Reviews Drug Discovery, our paper hit the press today
Combination cancer immunotherapy and new immunomodulatory targets. Nature Reviews Drug Discovery 14, 561–584. 2015. doi:10.1038/nrd4591
by Kathleen Mahoney, Paul Rennert, Gordon Freeman.
a prepublication version is available here: nrd4591 (1)
Some thoughts to fill the space between AACR and ASCO (and the attendant frenzied biopharma/biotech IO deals).
Classical immune responses are composed of both innate and adaptive arms that coordinate to drive productive immunity, immunological expansion, persistence and resolution, and in some cases, immunological memory. The differences depend on the “quality” of the immune response, in the sense that the immunity is influenced by different cell types, cytokines, growth factors and other mediators, all of which utilize diverse intracellular signaling cascades to (usually) coordinate and control the immune response. Examples of dysregulated immune responses include autoimmunity, chronic inflammation, and ineffective immunity. The latter underlies the failure of the immune system to identify and destroy tumor cells.
Let’s look at an immune response as seen by an immunologist, in this case to a viral infection:
Of note are the wide variety of cell types involved, a requirement for MHC class I and II responses, the presence of antibodies, the potential role of the complement cascade, direct lysis by NK cells, and the potentially complex roles played by macrophages and other myeloid cells.
In the immune checkpoint field we have seen the impact of very specific signals on the ability of the T cell immune response to remain productive. Thus, the protein CTLA4 serves to blunt de novo responses to (in this case) tumor antigens, while the protein PD-1 serves to halt ongoing immune responses by restricting B cell expansion in the secondary lymphoid organs (spleen, lymph nodes and Peyer’s Patches) and by restricting T cell activity at the site of the immune response, thus, in the tumor itself. Approved and late stage drugs in the immune checkpoint space are those that target the CTLA4 and PD-1 pathways, as has been reviewed ad nauseum. Since CTLA4 and PD-1 block T cell-mediated immune responses at different stages it is not surprising that they have additive or synergistic activity when both are targeted. Immune checkpoint combinations have been extensively reviewed as well.
We’ll not review those subjects again today.
If we step back from those approved drugs and look at other pathways, it is helpful to look for hints that we can reset a productive immune response by reengaging the innate and adaptive immune systems, perhaps by targeting the diverse cell types and/or pathways alluded to above.
One source of productive intelligence comes from the immune checkpoint field itself, and its’ never-ending quest to uncover new pathways that control immune responses. Indeed, entire companies are built on the promise of yet to be appreciated signals that modify immunity: Compugen may be the best known of these. It is fair to say however that we remain unclear how best to use the portfolio of checkpoint modulators we already have in hand, so perhaps we can look for hints there to start.
New targets to sift through include the activating TNF receptor (TNFR) family proteins, notably 4-1BB, OX40, and GITR; also CD40, CD27, TNFRSF25, HVEM and others. As discussed in earlier posts this is a tricky field, and antibodies to these receptors have to be made just so, otherwise they will have the capacity to signal aberrantly either because the bind to the wrong epitope, or they mediate inappropriate Fc-receptor engagement (more on FcRs later). At Biogen we showed many years ago that “fiddling” with the properties of anti-TNFR antibodies can profoundly alter their activity, and using simplistic screens of “agonist” activity often led to drug development disaster. Other groups (Immunex, Amgen, Zymogenetics, etc) made very similar findings. Careful work is now being done in the labs of companies who have taken the time to learn such lessons, including Amgen and Roche/Genentech, but also BioNovion in Amsterdam (the step-child of Organon, the company the originally created pembrolizumab), Enumeral in Cambridge US, Pelican Therapeutics, and perhaps Celldex and GITR Inc (I’ve not studied their signaling data). Of note, GITR Inc has been quietly advancing it’s agonist anti-GITR antibody in Phase 1, having recently completed their 8th dose cohort without any signs of toxicity. Of course this won’t mean much unless they see efficacy, but that will come in the expansion cohort and in Phase 2 trials. GITR is a popular target, with a new program out of Wayne Marasco’s lab at the Dana Farber Cancer Institute licensed to Coronado and Tg Therapeutics. There are many more programs remaining in stealth for now.
More worrisome are some of the legacy antibodies that made it into the clinic at pharma companies, as the mechanisms of action of some of these agonist antibodies are perhaps less well understood. But lets for the sake of argument assume that a correctly made anti-TNFR agonist antibody panel is at hand, where would we start, and why? One obvious issue we confront is that the functions of many of these receptors overlap, while the kinetics of their expression may differ. So I’d start by creating a product profile, and work backward from there.
An ideal TNFR target would complement the immune checkpoint inhibitors, an anti-CTLA4 antibody or a PD-1 pathway antagonist, and also broaden the immune response, because, as stated above, the immune system has multiple arms and systems, and we want the most productive response to the tumor that we can generate. While cogent arguments can be made for all of the targets mentioned, at the moment 4-1BB stands as a clear frontrunner for our attention.
4-1BB is an activating receptor for not only T cells but also NK cells, and in this regard the target provides us with an opportunity to recruit NK cells to the immune response. Of note, it has been demonstrated by Ron Levy and Holbrook Khort at Stanford that engagement of activating Fc receptors on NK cells upregulates 4-1BB expression on those cells. This gives us a hint of how to productively combine antibody therapy with anti-4-1BB agonism. Stanford is already conducting such trials. Furthermore we can look to the adjacent field of CAR T therapeutics and find that CAR T constructs containing 4-1BB signaling motifs (that will engage the relevant signaling pathway) confer upon those CAR T cells persistence, longevity and T cell memory – that jewel in the crown of anti-tumor immunity that can promise a cure. 4-1BB-containing CAR T constructs developed at the University of Pennsylvania by Carl June and colleagues are the backbone of the Novartis CAR T platform. It is a stretch to claim that the artificial CAR T construct will predict similar activity for an appropriately engineered anti-4-1BB agonist antibody, but it is suggestive enough to give us some hope that we may see the innate immune system (via NK cells) and an adaptive memory immune response (via activated T cells) both engaged in controlling a tumor. Pfizer and Bristol Myers Squibb have the most advanced anti-4-1BB agonist antibody programs; we’ll see if these are indeed best-in-class therapeutics as other programs advance.
Agonism of OX40, GITR, CD27, TNFRSF25 and HVEM will also activate T cells, and some careful work has been done by Taylor Schreiber at Pelican to rank order the impact of these receptors of CD8+ T cell memory (the kind we want to attack tumors). In these studies TNFRSF25 clearly is critical to support CD8 T cell recall responses, and may provide yet another means of inducing immune memory in the tumor setting. Similar claims have been made for OX40 and CD27. Jedd Wolchok and colleagues recently reviewed the field for Clinical Cancer Research if you wish to read further.
Looking again beyond T cells another very intriguing candidate TNFR is CD40. This activating receptor is expressed on B cells, dendritic cells, macrophages and other cell types involved in immune responses – it’s ligand (CD40L) is normally expressed on activated T cells. Roche/Genentech and Pfizer have clinical stage agonist anti-CD40 programs in their immuno-oncology portfolios. Agonist anti-CD40 antibodies would be expected to activated macrophages and dendritic cells, thus increasing the expression of MHC molecules, costimulatory proteins (e.g. B7-1 and B7-2) and adhesion proteins like VCAM-1 and ICAM-1 that facilitate cell:cell interactions and promote robust immune responses.
I mentioned above that interaction of antibodies with Fc receptors modulates immune cell activity. In the case of anti-CD40 antibodies, Pfizer and Roche have made IgG2 isotype antibodies, meaning they will have only weak interaction with FcRs and will not activate the complement cascade. Thus all of the activity of the antibody should be mediated by it’s binding to CD40. Two other agonist anti-CD40 antibodies in development are weaker agonists, although it is unclear why this is so; much remains to be learned regarding the ideal epitope(s) to target and the best possible FcR engagement on human cells. Robert Vonderheide and Martin Glennie tackled this subject in a nice review in Clinical Cancer Research in 2013 and Ross Stewart from Medimmune did likewise for the Journal of ImmunoTherapy of Cancer, so I won’t go on about it here except to say that it has been hypothesized that crosslinking via FcgRIIb mediates agonist activity (in the mouse). Vonderheide has also shown that anti-CD40 antibodies can synergize with chemotherapy, likely due to the stimulation of macrophages and dendritic cells in the presence of tumor antigens. Synergy with anti-CTLA4 has been demonstrated in preclinical models.
One of the more interesting CD40 agonist antibodies recently developed comes from Alligator Biosciences of Lund, Sweden. This antibody, ADC-1013, is beautifully characterized in their published work and various posters, including selection for picomolar affinity and activity at the low pH characteristic of the tumor microenvironment (see work by Thomas Tötterman, Peter Ellmark and colleagues). In conversation the Alligator scientists have stated that the antibody signals canonically, i.e. through the expected NF-kB signaling cascade. That would be a physiologic signal and a good sign indeed that the antibody was selected appropriately. Not surprisingly, this company is in discussion with biopharma/biotech companies about partnering the program.
Given the impact of various antibody/FcR engagement on the activity of antibodies, it is worth a quick mention that Roghanian et al have just published a paper in Cancer Cell showing that antibodies designed to block the inhibitory FcR, FcgRIIB, enhance the activity of depleting antibodies such as rituximab. Thus we again highlight the importance of this sometimes overlooked feature of antibody activity. Here is their graphical abstract:
The idea is that engagement of the inhibitory FcR reduces the effectiveness of the (in this case) depleting antibody.
Ok, moving on.
Not all signaling has to be canonical to be effective, and in the case of CD40 we see this when we again turn to CAR T cells. Just to be clear, T cells do not normally express CD40, and so it is somewhat unusual to see a CAR T construct containing CD3 (that’s normal) but also CD40. We might guess that there is a novel patent strategy at work here by Bellicum, the company that is developing the CAR construct. The stated goal of having a CD40 intracellular domain is precisely to recruit NF-kB, as we just discussed for 4-1BB. Furthermore, the Bellicum CAR T construct contains a signaling domain from MYD88, and signaling molecule downstream of innate immune receptors such as the TLRs that signal via IRAK1 and IRAK4 to trigger downstream signaling via NF-kB and other pathways.
Here is Bellicum’s cartoon:
If we look through Bellicum’s presentations (see their website) we see that they claim increased T cell proliferation, cytokine secretion, persistence, and the development of long-term memory T cells. That’s a long detour around 4-1BB but appears very effective.
The impact of innate immune signaling via typical TLR-triggered cascades brings us to the world of pattern-recognition receptors, and an area of research explored extensively by use of TLR agonists in tumor therapy. Perhaps the most notable recent entrant in this field is the protein STING. This pathway of innate immune response led to adaptive T cell responses in a manner dependent on type I interferons, which are innate immune system cytokines. STING signals through IRF3 and TBK1, not MYD88, so it is a parallel innate response pathway. Much of the work has come out of a multi-lab effort at the University of Chicago and has stimulated great interest in a therapeutic that might be induce T cell priming and also engage innate immunity. STING agonists have been identified by the University of Chicago, Aduro Biotech, Tekmira and others; the Aduro program is already partnered with Novartis. They published very interesting data on a STING agonist formulated as a vaccine in Science Translational Medicine on April 15th (2 weeks ago). Let’s remember however that we spent several decades waiting for TLR agonists to become useful, so integration of these novel pathways may take a bit of time.
This emerging mass of data suggest that the best combinations will not necessarily be those that combine T cell immune checkpoints (anti-CTLA4 + anti-PD-1 + anti-XYZ) but rather those that combine modulators of distinct arms of the immune system. Recent moves by biopharma to secure various mediators of innate immunity (see Innate Pharma’s recent deals) and mediators of the immunosuppressive tumor microenvironment (see the IDO deals and the interest in Halozyme’s enzymatic approach) suggest that biopharma and biotech strategists are thinking along the same lines.
Over 100 slides on immune checkpoint combination therapy, novel targets and drug development in immuno-oncology, created for a 3 hour workshop at ICI15 (link).
As always we work from indications to discovery and back again, keeping one eye on the rapid evolution of clinical practice in oncology and the other on novel targets and therapeutics.
on SlideShare now:
There is so much here to raise suspicion.
Lion Biotechnologies was born ugly, from a merger with Genesis Biopharma, essentially forming a public company within the shell of what some people suspect was originally a pump and dump operation (link). This up-listing maneuver came at the cost of most of Lion’s equity, as Genesis shareholders held 83.6% of the combined company and Lion shareholders initially received 8.2%, with the promise of doubling that stake to 16.4% of the combined company based on achievement of certain milestones. The merger was completed in July of 2013 and the stock continued trading under the ticker symbol GNBP until September 2013, when Genesis announced a 1 for 100 reverse stock split that essentially took place immediately, locking shareholders in place. The merged company changed its name to Lion Biotechnologies and hatched the new listing symbol LBIO. Muddying the picture just a bit more, the company is run by Manish Singh, Ph.D., the former ImmunoCellular Therapeutics chief executive who resigned from that company in August 2012. One rumor circulating then was that he was sacked after suspicions were raised about ImmunoCellular’s Phase 1 data reporting and promotion. That company has stabilized since he departed, moving its’ oncology vaccine program into Phase 2.
LBIO has moved aggressively into the cellular immunotherapy space, licensing technology developed by Dr. Steve Rosenberg and colleagues at NCI (using the same CRADA model that Kite Pharma uses) and building collaborative relationships with MD Anderson and the Moffitt Cancer Research Institute. They pulled in an MD Anderson investigator, Laszlo Radvanyi Ph.D. as CSO and earlier this week appointed industry veteran Elma Hawkins Ph.D. as President and COO. These are likely all good moves toward establishing and building credibility, although the company remains dogged by bad PR, most recently being pulled (by subpoena) into an SEC investigation of Galena Biopharma, a seemingly unrelated company. Speculation about this “wide net” investigation by the SEC has focused on Dr. Singh and possible past involvement with an investor relations firm (link 2). This seems unlikely to have anything to do with LBIO itself.
The final piece of the puzzle is more transparent, which is that the shell merger/reverse split reboot was financed in large part via a private placement with Roth Capital. There is nothing wrong here, except that people tracking stocks in this space tire a bit of the relentless pumping that Roth does on behalf of LBIO, although it is of course their right and possibly their obligation to do so. One puff piece stated “we see 196% upside!” – and I’d have to comment, stealing a line here from Billy Bob Thornton, “that’s a pretty specific number”. I guess we could also wonder what if anything the original shareholders of Genesis and Lion have left of their equity.
Moving on.
LBIO’s CSO, Dr. Laszlo Radvanyi, spoke a few weeks ago at the Immunomodulatory Antibodies for Cancer Conference in Boston, part of the ImVacs package. It was an impressive talk, very upbeat, and contained some data and technology that was new to me. So I took a closer look.
The basis for the technology is a riff on something we discussed earlier, as presented by the NCI’s Rosenberg at ASCO (link 3). Tumor infiltrating lymphocytes (TILs) are found in large numbers in some solid tumor types, and can be isolated when the tumor is removed. The presence of TILs is correlated with improved survival at least in some indications. It has been known for quite some time that expanded TILs can be injected back into the patient where they, sometimes, effectively attack and eradicate the tumor. The problem with the technology is that, like all personalized cellular therapies (CAR, TCR, some types of tumor vaccines), it is cumbersome to perform. When logistics are such a challenge, you really want to see robust benefit from the treatment. This is what LBIO is suggesting it can deliver. Dr. Radvanyi walked us through a brief history of TIL technology, hitting the highpoints. His statement that standard TIL therapy has shown overwhelming and superior efficacy versus competing therapeutics for in melanoma was one I had not heard before, and I reserve judgment on this – after all if it was really that good everyone would be adopting this technology, and I really don’t see that happening.
What was really interesting though was the more experimental system that he introduced, and if you look at the LBIO website you’ll see it under “next-generation TIL” (link 4). In this system tumor fragments are made from biopsy samples and cultured with IL-2. This apparently works optimally because dendritic cells and monocytes persist in the culture for a week or so. They first activate and then sort the TILs using an agonist anti-4-1BB antibody to enrich for antigen specific activated T cells. This allows you to reduce the number of cells injected (a good thing). He did show dramatic enrichment of TILs that recognized known melanoma antigens, so at the very least this model system works. I think this also suggests the importance of the 4-1BB pathway, at least in this system. Notably, anti-OX-40 antibody failed to expand antigen-specific T cells (note we are talking specifically about CD8+ T cells here, in part because that’s all you’ll have left after a few weeks culture in IL-2).
It seems a nice simple system and worth watching. There were other bells and whistles (transduction techniques) that we’ll skip for now. Other technology LBIO is funding includes the use of the anti-CTLA4 antibody ipilimumab in conjunction with TIL therapy (to turn off active immune suppression in the tumor microenvironment). That is being done at Moffitt in a Phase 1 expansion cohort. Be interesting to see what else the company has in mind.
Can LBIO – using this new TIL technology – achieve clinical and commercial success?
stay tuned…
One of the puzzles in thinking about the available costimulatory receptors on T cells (and NK cells) is the unsettling number of them. Sticking just to the TNF receptor superfamily (TNFRSF) we have OX40 (discussed earlier), 4-1BB, GITR, CD27, and also the TNF receptors themselves (1 and 2), the lymphotoxin beta-receptor, HVEM, and TNFRSF25. There may be some I’ve forgotten. As noted in part 1 OX40, GITR, 4-1BB and CD27 are evolutionary cousins, as are their cognate ligands. Why did the immune system evolve such a complexity of T cell costimulators?
The answer is not entirely clear although the expression patterns and kinetics of expression suggest some rationale for understanding the number of different receptors. Also, as it’s understood that all the TNF receptors signal via NF-kB, Jun and p38, we might see these receptors either compete (for signaling proteins) or cooperate. All of the available genetic and pharmacologic data suggest they cooperate or even synergize, thereby powering the T cell response when needed. Since T cell responses (and immune responses generally) are so dangerous when dysregulated, the multiplicity of on and off switches presumably allows for redundancy of control.
As we said previously, OX40 comes on slow and easy, starting about 12 hours after TCR stimulation, and riding along for up to 96 hours. This is in vitro, cell culture data … so lets recognize that in vivo, in response to the chaotic presentation of antigen, the population of T cells is likely to be turning over, proliferating … so it’s unlikely we will see a finely tuned kinetic response in the real world, as regards the population of responding cells. Nonetheless we can focus on a single T cell, just the one. And we’ve guessed it will be expressing OX40 say from day 2 to day 5 after activation. Lets ignore the fact that activated effector T cells are likely dividing more rapidly than every 5 days, and just ask the simple question – of the other TNFSF receptors, what else is expressed and when, and on what T cell and other cell types?
Cue 4-1BB.
4-1BB is also expressed following T cell activation, and is expressed on other cell types also. 4-1BB expression come up much more quickly after T cell activation, within a few hours, and wanes after several days. This receptor is critical to supporting activated T cell proliferation, differentiation and survival. Expression on T cells may be coincident with OX40, but the consequences of engaging the receptor with an agonist antibody are different. 4-1BB preferentially supports the proliferation and survival of CD8+ T cells, although at least in some settings the activity of CD4+ T cells is also stimulated through 4-1BB. Much of the anti-tumor activity of 4-1BB agonist antibodies in preclinical studies can be traced to stimulation of NK cells although this depends of the tumor type and model used. Less well understood is the role on 4-1BB on other cells types. This receptor is also found on DCs, macrophages, granulocytes, and Tregs. Expression has also been described on vascular endothelium and on some tumor cells. The role of 4-1BB is confusing, with various studies showing expansion of the Treg subset and others suggesting that 4-1BB dampens Treg responses, perhaps via direct effects on DCs. The 4-1BB gene-knockout mouse shows aspects of autoimmune disorders (at least in the mouse strains tested), suggesting a role for 4-1BB in maintaining immune homeostasis following activation. 4-1BB knockout mice have trouble handling tumor challenge, and at least some spontaneously develop B cell lymphomas as they age. It is all a bit complicated.
Regardless, there are two antibodies that can provide some early clinical data. Bristol-Myers Squibb (BMY) started development of the BMS-663513 antibody quite early, before the immunotherapy wave had really gotten started. BMS-663513 is a specific anti-4-1BB agonist antibody, isotype IgG4. The antibody ran into toxicity issues in a phase 2 trial of metastatic melanoma in 2011, leading to a halt of three trials with that antibody. As the toxicity was correlated with dose, BMY has restarted the clinical campaign with BMS-663513, now called urelumab to establish a safe and efficacious dose. A monotherapy trial is being run in patients with advanced/metastatic solid tumors or with relapsed/refractory Non-Hodgkin Lymphoma (NHL). A second trial in NHL is being run in combination with rituximab treatment. A third trial in advanced/metastatic colorectal and head and neck cancers is being run in combination with cetuximab (anti-EGFR). Pfizer’s (PFE) PF-05082566 is an agonist anti-4-1BB antibody (IgG2 isotype). One trial is listed at clinicaltrials.gov, a 3×3 dose escalation phase 1 trial run as monotherapy in patients with advanced cancers, and as combination therapy with rituximab in NHL.
At the June ASCO meeting there will be updates on both the BMY and PFE programs. BMY has a presentation focused on mechanism of action and biomarker analyses. Abstract #3017 outlines the goal of monitoring the immune status of 4 patients prior to and during the phase 1 study of urelumab (BMS-663513: clinical trial NCT01471210). The antibody was given every 3 weeks and the analysis presents results through 3 cycles. PBMCs were isolated from whole blood, and stimulated for 4 hours with PMA/ionomycin to activate lymphocytes. There was an increase in CD8 T cells up to 41% and NK cells up to 62%. CD4 T cells decreased by as much as 23% and regulatory CD4 T cells decreased by as much as 18% comparing the 3rd cycle to baseline. The results are consistent with a preferential impact on CD8+ T cells and NK cells. The level of the cytokines GM-CSF and IFNgamma were increased.
The PFE study (abstract #3007) describes very early data from clinical trial NCT01307267. Patients received PF-05082566 IV every 4 weeks (one cycle) with an 8 week period for assessment of dose-limiting toxicity (DLT) and radiographic analysis of tumor burden (RECIST 1.1). 27 patients were up to 0.3 mg/kg, the highest dose reported in the abstract. The majority of patients had either colorectal cancer (n=11) or Merkel cell carcinoma (n=6), the rest were a collection of solid tumor patients and 2 patients with B cell lymphoma. 25/27 patients completed the DLT assessment period (first cycle). No DLT was established but only 7 patients remain on therapy. All discontinuations from treatment were due to disease progression. A best overall response of stable disease was observed in 6 patients. No duration data is supplied.
These two early trials suggest that safe dose levels can be achieved, that a mechanism of action can be confirmed (urelumab: expansion of CD8+ T cells and NK cells), and that some clinical response can be observed (PF-05082566: stable disease). That’s a pretty good picture coming out of Phase 1. As preclinical data suggest that 4-1BB is most effective in various combination formats, it will be interesting to monitor advances in the rituximab and cetuximab co-therapy arms of these trials. Several potent combinations arising in the preclinical literature include 4-1BB with immune checkpoint inhibition (CTLA4 or the PD-1 pathway) and in combination with agonist OX40 antibody therapy.
Lets get back to the Treg cells, whose function is to suppress immune responses, primarily those of CD4+ T cells. These express 4-1BB constitutively, although it’s not clear how or if they are responding to treatment with agonist anti-4-1BB antibodies. Let’s turn to a different pathway known to have a profound effect on Tregs, the Glucocorticoid-Induced TNFR Related gene (GITR). GITR was first identified as a regulatory T-cell marker and was shown to play a critical role in breaking T cell tolerance by direct suppression of Treg activity. The preclinical evaluation of GITR produced some very striking data, including in combination settings in which anti-GITR antibodies essentially synergized with other immune checkpoint therapeutics to eliminate established tumors. Such combinations have included PD-1 blockade and CTLA4 blockade. GITR agonism is also synergistic with chemotherapy in preclinical models. Clinical development of GITR antibodies has been slow. A program initiated at TolerX and reborn at GITR, Inc., is recruiting for a phase 1 trial in advanced melanoma and other solid tumors. The antibody is TRX518 (NCT01239134). Merck is advancing a phase 1 study with the anti-GITR antibody MK-4166, although this trial (NCT02132754) is not yet recruiting patients. Review articles have mentioned an ongoing clinical program at MSKCC – this is one of the three sites enrolling patients in the TRX518 trial.
To the extent that the driving mechanism of action (MOA) of GITR stimulation is shown to be downregulation of Treg activity, this pathway should be a good candidate for combination therapy with 4-1BB, OX40 or CD27 agonists (see below) as well as with the CTLA4 and PD-1 pathway antagonists. If the MOA in human cancer patients is different or more complex than proposed, different combinations may be more or less attractive.
One last receptor – CD27.
The costimulatory molecule CD27 is constitutively expressed on most effector T cells, memory B cells, and an NK cell subset. So its expression may also overlap with those of the other receptors. CD27 appears to be important for sustained T cell effector function and also the development of T cell memory. CD27 is a marker of memory T cells, conversely, it is low or absent on Tregs. More broadly, CD27 supports germinal center formation that drives B cell maturation and the differentiation of plasma cells that produce high affinity antibodies, and is also important in driving the cytolytic activity of some NK cells.
Although there is a large preclinical literature on CD27 and its ligand CD70, there are few antibodies in clinical development. There are several historical explanations for this I think. CD70 is expressed at high levels on certain tumor types, particularly renal cell carcinoma (RCC). Much effort has gone into the development of cell-depleting antibodies targeting CD70. This expression pattern also called into question the relevance of CD27 in controlling tumor growth, as the ligand would be expected to stimulate immune responses. We now know that RCC and other solid tumors expresses high levels of PD-L1, and likely disables immune responses via this pathway. Not surprisingly then, one of the ongoing clinical efforts is a combination trial of nivolumab, the anti-PD-1 antibody from BMY with CDX-1127, an anti-CD27 antibody from Celldex (CLDX) in a collaboration announced by the 2 companies last week. In the meantime we have 2 abstracts from CLDX at ASCO in June to look forward to. In a 3×3 phase 1 dose escalation study of B cell lymphoma patients the drug was well tolerated with weekly IV dosing, and there were signs of clinical response, including a durable complete response in one patient with advanced refractory disease (abstract #3024; clinical trial NCT01460134). In the same trial, solid tumor patients were treated in a dose escalation phase and then an expansion phase (RCC and melanoma). The drug was well tolerated, there were preliminary signs of clinical response, and measureable activation of the immune system (Abstract #3027). With the potential to support memory T cell differentiation, CD27 may provide an important additional signal to drive long term tumor control. We’ll have to wait and see.
So, we have 4 receptors with overlapping activities and we have multiple antibodies in various stages of development. There will be plenty to learn about these targets and their roles in the future of combination immunotherapy. One of the most promising paths forward is the analysis of immune checkpoint and costimulatory proteins on tumor infiltrating lymphocytes (TILs). It seems very likely that the makeup of the tumor cell defense against the immune system will diverge between tumor types, and perhaps between patients with the same tumor types, or even with the same patient tumor at different times, or in different metastatic locations. Profiling TILs, and perhaps sentinel lymph nodes, for the expression patterns of lymphocytes and antigen presenting cells is likely to help guide combination therapy.
We’ll come back to that. And we’ve not forgotten those NK cells either.
stay tuned.
In Part 1 we introduced the evolutionarily related TNF receptor superfamily receptors OX40, 4-1BB, CD27 and GITR (link). All of there receptors are being targeted for cancer therapy due to their ability to stimulate CD4+ and/or CD8+ T cells while potentially shutting down T regulatory cells (Tregs).
The receptor OX40 (CD134) is expressed primarily on CD8+ T-cells, NK cells, NKT cells, and neutrophils. OX40L is expressed on dendritic cells (DCs), B-cells, and macrophages, and at sites of inflammation (e.g. activated endothelium). The expression pattern suggests a critical role for OX40L/OX40 is supporting immune responses. Additional information on the role of this pathway comes from the analyses of expression kinetics upon immune cell activation. OX40 is transiently upregulated on activated CD4+ and CD8+ T cells after engagement of the T cell receptor (TCR). The window of expression is from ~12 hours through ~96 hours, after which the receptor is downregulated. This receptor therefore functions to support the survival and expansion of already activated effector T cells. Importantly, OX40 agonism also reactivates the memory T cell population.
Clinical trials using a mouse anti-OX40 agonist antibody were spurred on by impressive results in preclinical tumor models, both as single agent therapy (in immunogenic tumors) and in combination with chemotherapy and irradiation. Various other combination therapies have since been tested preclinically, and that work is continuing. Early clinical trials with anti-OX40 agonist mAb show that the therapeutic is well tolerated, and have provided evidence of immune system stimulation and modest signs of clinical activity.
The latest clinical update I can find is from the SITC meeting in November 2013 (link). In a poster entitled “Phase I/II clinical trial of anti-OX40, radiation and cyclophosphamide in patients with prostate cancer: immunological analysis”, Andrew Weinberg’s group from the Providence Cancer Center, Portland, OR report out as follows. The Phase 1/2 trial was run in chemotherapy and hormone therapy-resistant metastatic prostate cancer patients (mCRPC). The therapeutic anti-OX40 antibody was given in three doses, on top of radiation therapy and escalating doses of cyclophosphamide. Expansion of CD4+ T cells, CD8+ T cells and NK cells was observed. This observation supports the hypothesis that OX40 agonism promotes the proliferation and survival of activated T cells, the proposed mechanism of action. In mice OX40 signaling also supports the differentiation of activated T cells into memory T cells that are required for long term protection. There is no corresponding human data yet. There was no expansion of the Treg subset (FoxP3+) except at the highest dose of cyclophosphamide. There were transient changes in circulating PSA, and 5/9 patients, metastatic lesions were stable during the course of the study. Importantly, given that we know that T cells must first be activated in order to express OX40, monotherapy may not be the best setting for testing anti-OX40 agents.
This is a promising start that gives us a few things to think about: 1) PSA was not a useful biomarker, at least in this patient cohort, although larger sample sizes may help sort this out; 2) the immune response was measured in the peripheral blood, not from TILs recovered from the tumors – be nice to see that data; 3) there were no DLTs and the therapeutic appeared safe.
Given that some of the most provocative data being generated around this pathway is in the setting of combination immunotherapy, those trials can be expected to start over the next few years. A combination trial of anti-OX40 plus ipilimumab (anti-CTLA4) in metastatic melanoma was recently withdrawn. It was to be run by the Ludwig Institute for Cancer Research, as sponsored by AgonOx. The rights to the program have gone to Medimmune/AstraZeneca and this may simply reflect a desire to proceed with a better anti-OX40 antibody (the current one being a mouse antibody). If anyone knows the back story here please share in the comments.
OX40 agonist antibodies are a hot commodity. Some recent deals involving OX40 assets are shown below:
Importantly, all of these deals involve human or humanized antibodies. This should reduce the anti-drug responses seen in the early Phase 1/2 trials that limited dosing.
No doubt there are other deals in the works. The level of activity is driven by the early clinical observations, the apparently favorable toxicity profile, and key advances in the preclinical literature. Several examples of interesting preclinical results include the data demonstrating that anti-OX40 antibody can deplete Tregs directly via engagement of Fc-receptors. The degree of depletion in this mouse model was correlated with the intensity of expression of OX40, with the highest expression found on intra-tumoral Tregs. This is an interesting finding although the translational relevance remains unclear as human Tregs only OX40 when activated through the TCR, and even then expression levels are lower than in mouse. Further, other studies have shown that anti-OX40 antibodies can activate Tregs in mouse tumor models, so there is certainly some complexity here. An attractive hypothesis is that the degree of depletion/activation of Tregs depends not only on the degree of OX40 expression but also the biophysical properties of the antibody (potency, valency, composition of the Fc-binding domain).
Also of critical interest (as mentioned above) will be the utility of anti-OX40 in combination therapies. Preclinical data suggest that OX40 agonism is synergistic with CTLA4 blockade (ipilimumab), PD-1 blockade, 4-1BB agonism, IL-2 cytokine treatment, and targeted small molecule therapeutic drugs. This is rich landscape and we anticipate an explosion of clinical data as fully human anti-OX40 antibodies begin moving through clinical trials.