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)
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:
Two papers out this week add to a pile of data addressing the role of neoantigens in tumor therapy. While these papers address tumor neoantigen “load” in the context of immune checkpoint therapy the results have implications for TIL therapeutics, TCR therapeutics and onco-vaccine development.
A really dramatic paper from diverse groups at the University of Pennsylvania and their collaborators, just published in Nature (link-1), explores the complex interplay of radiation therapy and anti-CTLA4 antibody therapy (ipilimumab, from BMS) in patients with stage IV metastatic melanoma (relapsed or previously untreated). In this Phase 1/2 clinical trial (NCT01497808) patients with multiple melanoma metastases received various doses of radiation therapy delivered to a single metastasis, termed the “index lesion”. They then received 4 doses of ipilimumab (3 mg/kg, i.v., once every 3 weeks) and non-irradiated lesions were evaluated within 2 months of the last dose.
Although the sample size reported is small (n=22) some interesting lessons emerged from the study. The response rate was low, and the progression free survival (PFS: 3.8 months) and overall survival (OS: 10.7 months) data bear this out. It appears that just shy of 40% of patients were still alive at ~30 months (see Figure 1c in the paper). It is too early to tell if there will be a “long-tail” effect going forward. In the original ipilimumab study a very small percentage of patients lived for a very long time, “pulling” the PFS and OS curves to the right. Regardless, most patients in this study did not respond and the questions posed in this paper are directed to the mechanisms of resistance.
The mouse B16-F10 melanoma model was used to model resistance. Mice with tumors were locally irradiated then treated with an anti-mouse-CTLA4 antibody, to mimic the clinical trial. Only 17% of the treated mice responded. Two predictors of response/non-response were elucidated: 1) the ratio of effector T cells (Teff) to regulatory T cells (Treg) and 2) a gene signature in the tumor cells that is dominated by the expression of PD-L1 and IFNgamma regulated genes. In short, if the melanoma cells are expressing PD-L1 and the tumor infiltrating lymphocyte (TIL) population is dominated by Tregs (which are PD-1+), then the radiation + anti-CTLA4 therapy failed.
To further subset TIL into active Teff versus non-responsive “exhausted” Teff, the authors used an expression profile of PD-1+/Eomes+ to identify exhausted Teff and PD-1+/Eomes+/Ki67+/GzmB+ for active Teff. Importantly, exhausted Teff could be reanimated upon treatment with PD-1 pathway antagonists: anti-PD-1 antibody or anti-PD-L1 antibody. This reanimation led to an improved CD8+ Teff/Treg ratio and led to tumor control in the majority of the mice (up to 80%) when the treatment consisted of irradiation plus anti-CTLA4 plus anti-PD-L1. Of note, radiation plus anti-PD-L1 did not achieve this effect; the triple therapy was required (see Figure 2d).
The striking conclusion is that upregulation of PD-L1 on tumor cells can subvert the effect of anti-CTLA4 antibody therapy, and this therefore qualifies as a mechanism of resistance.
What about the role of irradiation? In both the patients and the mouse model irradiation was local, not systemic. Further, this local irradiation was required to achieve complete responses in the mouse model. What is going on here? Irradiation was linked to a modest increase in TIL infiltration of melanoma tumors in the mouse model, but sequencing of the T cell receptors (TCR) revealed that there was an increase in the diversity of TCRs, meaning that more antigens were being recognized and responded to by TIL after irradiation. In this context then, anti-CTLA4 reduced the Treg population, anti-PD-L1 allowed CD8+ TIL expansion, and irradiation set the antigenic landscape for response.
Returning to the patients armed with this information from the mouse study, the authors find that low PD-L1 expression on the melanoma cells correlates with productive response to irradiation plus ipilimumab therapy, while PD-L1 high expressing tumors were associated with persistent T cell exhaustion. In addition, monitoring the state of the CD8+ T cell population (PD-1+/Eomes+ versus PD-1+/Eomes+/Ki67+/GzmB+) suggested that these phenotypes might be useful as peripheral blood biomarkers. The patient numbers are very small for this analysis however, which awaits further validation.
The conclusion: irradiation combined with ipilimumab plus anti-PD-L1 antibody therapy should be a productive therapeutic combination in PD-L1+ stage IV melanoma. Similar strategies may be beneficial in other solid tumor types. This is interesting news for companies developing anti-PD-L1 antibodies, including BMS-936559 (also from BMS), MPDL3280A (Roche/Genentech), MEDI4736 (AZN) and MSB0010718C (Merck Serono).
A second paper (link) bring our focus back to PD-1, in the context of non-small cell lung cancer (NSCLC). Using the anti-PD-1 antibody pembrolizumab (from Merck) a group from the Memorial Sloan-Kettering Cancer Center sought to determine correlates of response of NSCLC patients to anti-PD-1 therapy. Their findings again hone in on neoantigen load, as the best predictors of response were the non-synonymous mutational burden of tumors, including neoantigen burden and mutations in DNA repair pathways. What all this means is that mutations that change the amino acid sequence (thus, are non-synonymous) can produce neoantigens that can be recognized by CD8+ T cells; mutations in the DNA repair pathways increase the rate that such mutations go uncorrected by a cell.
The authors sequenced the exomes (expressed exons – these encode proteins) from tumors versus normal tissue, as a measure of non-synonymous mutational burden that could produce neoantigens. Patients were subsetted based on response: those with durable clinical benefit (DCB) and those with no durable benefit (NDB). High mutational burden was correlated with clinical efficacy: DCB patients averaged 302 such mutations, while NDB patients averaged 148; ORR, PFS and OS also tracked with mutational burden. In a validation cohorts the number of non-synonymous mutations was 244 (DCB) versus 125 (NDB).
Examination of the pattern of exome mutations across both cohorts was studied in an attempt to discern a pattern of response to pembrolizumab treatment. The mutational landscape was first refined using an algorithm that predicts neoepitopes that can be expressed in the context of each patients specific class I HLA repertoire – these are the molecules that bring antigens to cell surfaces and present them to T cells for recognition (I’m simplifying this process but that is the gist of it). The algorithm identified more potential neoepitopes in the DCB patient tumors than in the NDB cohort, more impressively, a dominant T cell epitope was identified in an individual patient using a high-throughput HLA multimer screen. At the start of therapy this T cell clone represented 0.005% of peripheral blood T cells, after therapy the population had risen 8-fold, to 0.04% of peripheral blood T cells. Note that most of this clone of T cell would be found in the tumor, not in circulation, so that 8-fold increase is impressive. The T cells were defined as activated CD8+ Teff cells by expression markers: CD45RA-/CCR7-/LAG3-. As in the first paper we discussed, it is useful that these markers of systemic response to immunotherapy treatment are being developed.
There is an interesting biology at work here. It is often noted that high mutational burden is associated with better outcome, for example to chemotherapy in ovarian cancer, and irrespective of therapy across different tumor types (link-2). This suggests that tumor neoepitopes are stimulating an ongoing immune response that is stifled by active immunosuppression, yet is still beneficial. Once unleashed by immune checkpoint blockade, the immune system can rapidly expand it’s efforts.
We recently reviewed the importance of neoantigens in anti-tumor therapy (link-3) although the focus then was on cellular therapeutics rather than on immune checkpoint modifiers such as anti-CTLA4 and anti-PD-1 or PD-L1 antibodies. We can mow add that our ability to track neoantigens and the immune response to neoantigens is opening new avenues for investigating immuno-oncology therapeutics and their efficacy.
I’m asked a lot about the onco-vaccine field, and if immune checkpoint inhibitors will be the key to unlocking the potential of this long-suffering therapeutic class. The answer is never simple, since we are often looking at thin patient data that can contain compelling hints of efficacy – those immunized late-stage patients who not only regressed but stay in remission, month after month and year after year. The problem for companies and investors is that such observational data can be very misleading, and the vaccine candidates most often go on to fail in later and larger clinical trials, sometimes spectacularly. These big failures burden the field with a high evidentiary bar.
Data have emerged that suggest several issues with most vaccines, and these issues are both distinct and related.
At the end of November Nature published two interesting papers that asked a very simple question: what immunogenic antigens are present in common mouse tumor models. Yadav et al from Genentech and Immatics Biotechnologies (link 1) used a genome-wide exome and transcriptome sequence analyses, mass spectrometry and structural modeling to identify immunogenic neo-antigens in the widely used MC-38 and TRAMP-C1 mouse syngeneic tumor models. These models are considered poorly immunogenic in wild-type syngeneic (C57Bl6) mice. The sequencing analysis was used to identify mutated proteins that were present at >20% allelic frequency. From the MC-38 model, 1290 expressed mutations were identified of which 170 were considered to be neo-epitopes, that is, modeling suggested they would be expressed by MHCI and sufficient residues would be solvent exposed to allow immunogenicity. Only 67 expressed mutations were found in the TRAMP-C1 model, and of these 6 were considered to be potential neo-epitopes. Of this total of 170 (MC-38) and 6 (TRAMP-C1) only 6 bound MHC1 by Mass Spec, with a predicted IC50 for MHCI < 500nM. Of these, 3 were actually immunogenic in vivo (using C57Bl/6 mice) and could protect wild-type mice from tumor challenge. The neo-epitopes were found in the proteins Dpagt1, Reps1 and Adpgk. Here is a schematic of the filtering scheme:
Working the other way, the authors confirmed the immunogenicity of neo-epitope peptides by analyzing tumor-infiltrating lymphocytes (TIL) and staining with peptide–MHCI dextramers to identify bound T cells. CD8+ T cells specific for Reps1, Adpgk and Dpagt1 were enriched in the tumor. Using the Adpgk neo-epitope, the TIL were further investigated and found to express PD-1 and TIM-3, inhibitory receptors associated with anergic or “exhausted” CD8+ T cells, showing that the murine immune system had indeed recognized and responded to the neo-epitopes, a response that was then actively immunosuppressed in the tumor microenvironment. Importantly, none of the identified neo-epitopes would qualify as tumor-antigens, that is, they are not specifically overexpressed at a sufficient level to qualify. The neo-antigen-specific TIL are also pretty rare, suggesting that use of tumor lysate as an immunogen to elicit an anti-tumor response may “miss” by failing to present enough of the right antigen to the immune system.
We noted at the top that this paper came out of labs at Genentech and Immatics. An agreement between Roche/Genentech and Immatics will focus on the use of this technology. The two companies will develop new tumor-associated peptides (the neo-epitopes as cancer vaccine candidates, initially targeting gastric, prostate and non-small cell lung cancer. The most advanced candidate is IMA942, a peptide vaccine for the treatment of gastric cancer, in late preclinical development. Immatics CEO Paul Higham has publicly stated that a Phase I study of IMA942 with Roche’s PD-L1 inhibitor MPDL3280A in gastric cancer will be initiated soon. Immatics will also conduct research to identify neo-antigens for the additional indications. For those keeping score, Immatics received $17 million upfront, committed research funding, and potential milestones
The second work, led by Schreiber’s group at Wash U, used mouse models to ask a different but related question: what tumor antigens are recognized after immune checkpoint blockade with anti-PD-1 or anti-CTLA4 antibodies (link 2). The sarcoma lines d42m1-T3 and F244 were rejected in wild-type mice treated with either anti-PD-1 or anti-CTLA4 antibodies, in a CD4/CD8/IFNy/DC-dependent manner. As in the first paper, a filtering system built from diverse technologies was used to identify potential neo-epitopes. Mutations were identified by cDNA sequencing, translated to corresponding protein sequences, then tested against MHCI binding algorithms. Neo-epitopes were ranked by predicted median binding affinities and likelihood of productive immunoproteasome processing and antigen display. Using these methods, two MHCI restricted neo-epitopes were identified in Alg8 and Lama4.
As in the first paper, the authors then turned the system around, asking what neo-epitopes could be identified through analysis of TIL. Alg8 and Lama4 were found in tumor TIL and their frequency was increased by treatment with anti-PD-1 or anti-CTLA4. The neo-epitopes could successfully be used to induce anti-tumor immune responses. As in the first paper, these are not neo-epitopes that would qualify as tumor-antigens using the traditional criteria of selective and high expression.
So these are our two distinct but related issues with the current tumor vaccine landscape: those that have selected antigens have likely selected the wrong ones, while those that use lysates are likely too dilute.
Importantly, we can now compare these model systems to actual data from human patients treated with anti-CTLA4 antibody, as published recently in NEJM (link 3). The clinical group from Memorial Sloan Kettering Cancer Center obtained tumor tissue from melanoma patients treated with anti-CTLA4 antibodies (ipilimumab or tremelimumab). As in the mouse study, whole-exome sequencing was performed, somatic mutations identified and potential neo-antigens were characterized. Here is their schematic:
Baseline analyses showed that there was a significant difference in mutational load between patients with long-term clinical benefit and those with a minimal or no benefit, with higher mutational load associated with response. However the relationship was correlative, since some tumors with a high mutational burden were not responsive to anti-CLTA4 therapy. Using peptides predicted to bind to MHCI with a binding affinity ≤ 500 nM, the authors focused on mutated peptide sequences shared by multiple tumors and shared by patients having long-term clinical benefit. From this analysis a neo-epitope “signature” was derived, consisting of a distinct pattern of mutated peptide sequences. One of the peptide signatures identified matched an amino acid sequence in MART1, a known melanoma antigen. However the bulk of predicted neoantigens were tetrapeptide sequences shared across antigenic peptides, that is, they were encoded by diverse genes (you have to go into the huge supplemental data file to find this list, suffice to say it is very long). To make sense of this curious result the authors note that the some of the predicted sequences have high homology to viral and bacterial antigens, citing a CMV antigenic sequence as an example. They speculate, and here we quote from the paper: ” These data suggest that the neoepitopes in patients with strong clinical benefit from CTLA-4 blockade may resemble epitopes from pathogens that T cells are likely to recognize.”
The unstated null hypothesis is that there is no relationship between the shared tetrapeptide sequences and clinical response and that the association between the two phenomena is due to a productive immune system response to anti-CTLA4 antibody therapy which has released the anti-tumor response as well as many otherwise quiescent immune responses, such as those to pathogenic viruses and bacteria (and to self antigens, as shown by the autoimmune toxicity associated with anti-CTLA4 antibody treatment). The in vitro response assay data sheds no real light on this, since these assays cannot distinguish anti-tumor responses from other immune responses. So we are left with an intriguing correlation, and a nagging sense that only a very few of the vast number of predicted neo-epitopes will actually trigger bona-fide anti-tumor T cells responses. Indeed the weakness of the paper is the reliance on predictive rather than experimental identification of productive peptide/MHC interaction. As we say in the mouse studies the majority of predicted interactions are not confirmed experimentally.
Regardless, the paper is a remarkable and important step forward, and shows us (as do the mouse studies) the level of investigation required to identify neo-antigens that might expand be used to expand patient TIL populations, as we have discussed in other posts. Returning to onco-vaccines, these three papers together show us that neo-antigen anonymity, rarity and variability from patient to patient are critical issues that will need to be addressed if we are to efficiently develop this therapeutic class.
Part 2 – The Border Wars.
One of the fascinating aspects of the toxicity of immune checkpoint therapeutics is that it is a lot of is triggered at the border between self and non-self, where non-self is everything that the immune system must encounter and sort through continuously. The sorting serves to identify pathogens and ignore non-pathogens among the myriad components of the microfauna and flora that inhabit these borders. The “sampling” of these ecosystems is continuous and highly reactive – one glass of unpurified water taken on foreign soil will teach you this lesson pretty quickly. When the immune system is unrestrained by blockade of CTLA4 and/or PD-1 it is not surprising that we see the breakdown of immune tolerance in these border zones.
There are three major surfaces where toxicity has been an issue: the skin, the gut mucosa, and the airspaces of the lung. Ipilimumab treatment can cause pretty intense inflammation of the skin, generally dismissed in the clinical trial literature as “rash”. In a pooled analysis of nearly 1500 patients enrolled in various ipilimumab clinical trials, 45% developed dermatological AEs considered drug related, and 2.6% (so 39 people) developed severe symptoms rating a grade 3-4 (where grade 5 is lethal) (see Tarhani, A. Scientifica 2013, Article ID 857519). A fair amount of the milder skin AEs can be ascribed to an anti-melan-A response, as this antigen is abundant in melanoma, the setting for the clinical development trials. In the Phase 3 registrational trials dermatologic AEs were reported in more than 40% of patients in the ipilimumab arms, and there were very severe AEs that cannot be ascribed to an anti-melan-A (i.e melanocyte) immune response. This is from Tarhani’s review of patients in the ipilimumab + gp100 (vaccine) and ipilimumab monotherapy arms having dermatological irAEs,
“of these, 2.1% and 1.5%, respectively, were grade 3 or higher … Severe, life threatening, or fatal immune-mediated dermatitis (Stevens- Johnson syndrome, toxic epidermal necrolysis, … full thickness dermal ulceration, or necrotic, bullous, or hemorrhagic manifestations; grade 3–5) occurred in 13 of 511 (2.5%) patients treated with ipilimumab. One patient (0.2%) died as a result of toxic epidermal necrolysis, and one additional patient required hospitalization for severe dermatitis… .”
That’s some rash. We note in passing that dermatologic AEs were see in a phase 2 trial of ipilimumab plus chemotherapy in non-small cell lung cancer (NSCLC) and so this is certainly not limited to the melanoma setting. PD-1 pathway antagonists also cause skin inflammation in both the melanoma and other settings, similarly suggesting that what we are seeing here are immune responses to antigenic stimulation that is normally immunologically inert. Nivolumab-induced dermatologic toxicity can be severe, but is less common than seen with ipilimumab therapy.
The issue of skin toxicity is well known clinically, and there are established treatment protocols requiring cessation of therapy and treatment with anti-inflammatories, usually steroids (i.e the REMS protocols). The gastrointestinal (GI, “gut”) AEs are also common, can arise suddenly, be resistant to therapy (corticosteroids, and rarely, anti-TNF antibody), and are of significant concern. Returning to the pooled analysis of ~1500 ipilimumab patients we see roughly half of the patients developing GI symptoms (this includes diarrhea). If we focus on grade 3/4 SAEs we have 10-12% of patients with GI disorders that include colitis, enterocolitis, intestinal perforations etc that can proceed to lethal septic complications. Of note, inflammatory infiltrates in the intestines include abundant T cells and neutrophils, showing that acute ongoing inflammation is occurring. GI toxicity is less common and less severe in nivolumab-treated patients, and this is true also of Merck’s anti-PD-1 antibody pembrolizumab and the anti-PD-L1 antibody MPDL3280A from Roche. Colitis is generally not a big issue, for example, GI SAEs are seen in less than 1% of nivolumab-treated patients. We might conclude here that other pathways are maintaining tolerance in the gut mucosa when the PD-1 pathway is blocked.
A different picture emerges when we consider AEs in the lung. Pulmonary toxicity is rare in the context of ipilimumab monotherapy, with only scattered case reports in the literature (see Voskens et al for a review of rare ipilimumab-induced AEs: link). Anti-PD-1 pathway therapeutics, particularly nivolumab, are associated with pneumonitis, which is inflammation of the lung tissues. In the monotherapy setting, both nivolumab and pembrolizumab causes pneumonitis in 3-4% of patients – the condition is generally mild and treatable. Of note this AE rate is consistent across indications (e.g. melanoma, renal cell). The anti-PD-L1 antibodies (Roche’s MPDL3280A and Astra Zeneca’s MEDI4736) have not been associated with pneumonitis to date, perhaps reflecting a unique profile. The recent data from the anti-PD-L1 antibody MEDI4736 trial in NSCLC presented a tolerable profile. While response rate was low, significant numbers of patients remained on therapy with stable disease (ASCO 2014, Abstract #3002).
More worrisome is the pneumonitis rate and severity in combination therapy particularly in the NSCLC setting where diminished lung function is already a concern (smokers with lung cancer can’t breathe). When nivolumab was combined with platinum-based chemotherapy in NSCLC the SAE rate jumped to 45%, with notable findings of grade 3/4 pneumonitis (7%) and acute renal failure (5%) (ASCO 2014, Abstract #8113). Nivolumab plus erlotinib was not associated with pneumonitis (ASCO 2014, Abstract #8022) but response rates were low as well suggesting that these therapies were not particularly additive. The combination of nivolumab with ipilimumab was most worrisome, with grade 3/4 pneumonitis (6%) now seen along with grade 3/4 SAEs of skin (4%), GI (16%) and others (16%) (ASCO 2014, Abstract #8023). Most problematic is that 35% of patients discontinued, and between 3 to 5 patients died due to drug related SAEs including respiratory failure (caused by severe colitis), epidermal necrolysis (in a patient with multiple SAEs) and pulmonary hemorrhage (pneumonitis). As indicated above, the anti-PD-L1 antibody MEDI4736 may better suited for combination therapy. A combo trial in NSCLS with anti-CTLA4 mAb tremelimumab is enrolling, so we’ll wait and see.
It’s fair at this point to take a step back and say “so what?” These are close to terminal patients with deadly cancers usually highly refractory to treatment, and we cannot expect a free ride. The unmet need is acute and urgent, and these therapeutics offer potential cures and increase in life expectancy – as shown very clearly in last weeks early termination of the Phase 3 trial of nivolumab versus dacarbazine due to the obvious overall survival advantage offered by nivolumab (see John Carroll’s story in Fierce Biotech here: link)
The problem is that the response rates we are seeing are generally low, the discontinuation rates high, and for anti-CTLA4 and anti-PD-1 therapeutics there is no clear consensus regarding the use of biomarkers to select patients most likely to respond. Therefore the actual percent penetrance of therapy in the patient cohorts becomes quite low. For those relatively few patients who respond well the outcomes can be sustained and robust. It is critical however to get these response rates up. The blockbuster combination of nivolumab plus ipilimumab in metastatic melanoma gives us a sense of what is possible, if the drugs are tolerable. It is also critical to understand how and why immune therapy can make subsequent therapy intolerable, as we’ve seen in case reports, or conversely, how and why prior therapies can cause such problems for patients going onto an immune therapeutic (see that Voskens review mentioned above). We’ve seen some the issues that can bedevil combinations in metastatic melanoma (with vemurafenib) and in renal cell carcinoma clinical trials (pazopanib) When we look at all of the combination clinical trials underway with these agents we have to wonder what surprises lay in store.
Part 3 – The Fifth Column.
The fifth column refers to enemies lurking within the boundaries of the state, in this case the human body. These are a mixed collection of AEs that can be difficult to understand. While we are used to see liver and kidney inflammation in the setting of cancer therapy, it remains a bit mysterious that immune checkpoint therapy can cause severe inflammatory responses in these organs, the most notable is probably the induction of hepatitis in patients treated with ipilimumab. Even weirder (for me anyway) are the endocrinopathies, headlined by pituitary inflammation, seen with both CTLA4 and PD-1 directed immunotherapies. Primary thyroid inflammation is also seen although less frequently. These are of course autoimmune targets in this setting, but the triggers are obscure, as is also almost always true in autoimmune disease. Somewhat remarkable is the emergence of a sometimes fatal but normally very rare condition known as autoimmune hypophysitis or lymphocytic hypophysitis, which is inflammation of the pituitary gland. Hypophysitis is a unique toxicity of immune checkpoint inhibitors, and has been been seen in patients treated with ipilimumab, tremelimumab, and nivolumab. Because the pituitary sits in the middle of the limbic hypothalamic-pituitary-adrenal axis effects on the thymus and adrenal gland are also noted, with adrenal insufficiency being a severe and life-threatening complication. It must be stressed that the frequency of this AE is stunningly high, reaching 17% in some trials, as the disease has been described only very rarely, with a good deal less than 1000 cases ever known prior to the introduction of immune checkpoint therapeutics.
So we won’t dwell on this, as clinicians now know what to watch for, and treatment paradigms have been developed. As mentioned earlier, treatment generally involves initiation of steroids to control to autoimmune response, and cessation of immune checkpoint therapy.
Let’s return to the consideration of combination therapy, which I think we all agree is essential if we are really to expand use of immune therapeutics in the treatment of these difficult cancers. Great hope has been placed in the combination of CTLA4 and PD-1 targeting agents with “safe” immune checkpoint modulators, notably the IDO-inhibitor from Incyte. We have very little information to date, but it is notable that the dose limiting toxicity in the first combination trial of ipilimumab and INCB024360 from Incyte (INCY) was liver damage as measured by ATL elevation. It may be that merely piling on ways of disrupting Treg activity will not help with the toxicity profile; in fact, one might make the prediction that this approach will make things worse in some settings.
We’ve remarked in passing on the apparently mild safety profile of the anti-PD-L1 inhibitors compared to the PD-1 inhibitors. This makes some sense, as the ligands are expressed by the target tumor cells, and this may be the main sink for the injected antibody, i.e. antibody may not be evenly bio-distributed but rather predominantly localized to the tumors. The concordance of anti-PD-L1 antibody activity with tumor PD-L1 expression is consistent with a direct and localized effect. The fact that there is less consistent concordance of anti-PD-1 antibody activity with PD-1 expression by tumor-infiltrating T cells suggests less specificity in the induced immune response, and this may be why we see autoimmune toxicity in the nivolumab setting. As CTLA-4 is exclusively T cell expressed, the same seems to hold true for anti-CTLA4 antibody therapy. So combining these may not be the most ideal way forward.
We will discuss alternative approaches next time, but first there is some new data on novel immune checkpoint therapies to consider. These are the TNF receptor superfamily proteins that we discussed last month (link): 4-1BB, CD27, OX40 and GITR. There is admittedly very little data to date. Pfizer’s (PFE) anti-4-1BB antibody PF-05082566 reached a safe dose in Phase 1 without undue toxicity signals (ASCO 2014, Abstract #3007). Pfizer disclosed combination trials with rituximab in Non-Hodgkin Lymphoma (NHL) and pembrolizumab (anti-PD-1). The BMY antibody urelumab was tolerated through its’ dose escalation cohorts, and ex vivo analysis showed activation of CD8+ T cells and NK cells (ASCO 2014, Abstract #3017). The Celldex anti-CD27 mAb also has demonstrated safe dose escalation, although to date without signs of clinical activity (ASCO 2014, Abstracts #3024 and #3027). Celldex (CLDX) claims planned studies in combination with nivolumab, ipilimumab, and the targeted therapeutics darafenib and trametinib.
As we discussed in an earlier post, 4-1BB, CD27, OX40 and GITR are evolutionarily closely related receptors. Biomarker studies such as the one performed in the urelumab trial will be essential in understanding how these immune stimulatory pathways will differentiate clinically and which will be safe in combination settings. We’ve reviewed the biology of this superfamily recently (see these posts) so won’t do so again until we get some more clinical data.
Next we will introduce some novel targets in the TNF receptor superfamily, revisit some apoptotic pathway “influencers”, and will swing back around to PD-1 and PD-L1 in some other solid tumor settings (not necessarily in that order).
stay tuned.
Part 1 – wherein we introduce the issues
I think we are underestimating the impact of immune checkpoint mediated adverse events (AEs) and are too easily calmed by the notion that severe AEs (SAEs) can be managed or will reverse when drug is withdrawn. Indeed, we are beginning to see that the toxicity profiles of the anti-CTLA4 antibody ipilimumab and the anti-PD-1 antibody nivolumab will limit their use, at least in some settings. Anyone needing proof can look at Bristol-Myers Squibb (BMY) and their recent $50MM USD engagement with CytomX, a neat little Bay Area biotech that has developed “masking” technology for antibodies, called Probodies, that creates antibodies that are inert until they reach the tumor environment. Their website at http://www.cytomx.com/probodies.php has links to their Science Translational Medicine paper and also their 2014 AACR presentation.
So for their fifty million upfront, plus R&D support, BMY gets to apply the Probody technology to ipilimumab, and three other targets. Since BMY also owns nivolumab, we might make the reasoned guess that this is a second target, although that has not been disclosed and remains unclear. Another interesting target for masking is urelumab, the agonist anti-CD137 (aka 4-1BB) antibody. And it’s clear that there are other targets for which antibodies could benefit from being masked.
Why “mask” immune checkpoint antibodies? The issue is that these antibodies can induce immune responses that are off-target, that is, not directed at the tumor. Some off-target AEs induced by different agents are similar, some are quite unique and others appear only in combination settings. Lumping these all together as “immune-related” AEs (irAE) may be convenient but teaches us nothing about the underlying mechanisms.
So lets have a look at some general mechanisms.
Recent posts have concentrated on ipilimumab and the PD-1 pathway antagonists (see here) and have detailed some of the AE issues arising. Other toxicities have been seen with the TNFR agonists, such as 4-1BB and related receptors, whose biology and use are reviewed in another recent post (link).
Breaking irAEs down by class is helpful. The irAEs are most commonly associated with barrier tissues: skin, gut mucosa, the lung airways and the eyes. Less common are irAEs triggered in the context of “sterile” inflammation, that is, inflammation directed toward tissues and organs within the body and walled off from the outside environments. This class includes the endocrinopathies and abnormalities in liver, kidney and other organs. The third class captures central nervous system (CNS) toxicities. Some irAEs can appear with monotherapy, some are worse in the context of dual immune checkpoint blockade and some are associated with immunotherapy in the context of combination therapy with other types of drugs.
The cause of irAEs is very likely to be due to the breakdown of immune tolerance to commensal flora antigens and/or to self-derived antigens. We can look at the biology of the pathways targeted by ipilimumab and the PD-1 inhibitors for some clues to how this arises. The ipilimumab target CTLA4 normally acts as a shut-off switch for primary immune responses, blocking the activity of the closely related T-cell costimulatory receptor CD28, normally triggered by the B7 proteins, B7-1 and B7-2. The nivolumab target PD-1 acts as a similar stop signal but is associated with shutting off activated T cells by binding to its’ ligands PD-L1 and PD-L2. It is worth noting that these two families of receptors and ligands are closely related – CD28, CTLA4 and PD-1 are evolutionarily related receptors and B7-1, B7-2, PD-L1 and PD-L2 are closely related ligands. One consequence of this relatedness is seen in the ability of B7-1 to bind to and activated PD-1. Of interest, the evolutionary tales suggests a primordial single pathway that has diverged and specialized to control different aspects of the adaptive immune response. This is very similar to the situation we described for the closely related TNFR family members 4-1BB, OX-40, CD27 and GITR, and their ligands 4-1BBL, OX-40L, CD70 and GITRL.
The irAE story begins with ipilimumab, as does the immune checkpoint field. There are several mechanisms by which CTLA4 blockade might induce irAEs. The first is by lowering the threshold for activation of T cells, thereby allowing these cells to respond to antigens they would normally ignore. CTLA4 is expressed exclusively on T cells and regulates T cell activation. The second, related, mechanism is the breakdown of tolerance to antigens. On the surfaces of the organism antigens to which were have become tolerant (microorganisms, commensal flora) are routinely presented by APCs (antigen-presenting cells including dendritic cells and monocytes/macrophages) in the draining lymph nodes, Peyer’s patches, spleen and other lymphoid organs. Normally, these antigens are ignored. Furthermore, any aberrant T cells response to tolerant antigens is blocked by the action of T regulatory cells (Tregs). These Tregs “see” tolerant antigens and secrete immune suppressive factors to keep nearby T cells from becoming activated (T-effectors) and moving out into the environment to respond wherever antigen is encountered. It has been demonstrated that ipilimumab derails Treg function as well as pumping up T-effector cell function. As a consequence, in the presence of ipilimumab, T-effectors overwhelm any tolerance mechanisms and triggers immune responses to cells and tissues presenting antigens that are normally ignored. The setting for PD-1 pathway-mediated toxicity is similar although the details are a little different. PD-1 inhibition is mediated by expression of PD-L1 and PD-L2 on diverse cells types in the periphery (i.e. outside the lymphoid organs). PD1 is highly expressed on Treg cells. Peripheral tolerance is maintained by this system, and blockade of the system disrupts Treg function directly, and licenses T-effector cells for inappropriate responses to antigens normally ignored. The irAEs seen in the context of PD-1 blockade are generally considered less common than those seen in the context of CTLA4 blockade, suggesting that compensatory mechanisms of tolerance maintenance may be operative in the periphery.
Genetic deletion studies in mice can shed some light on these mechanisms. The CTLA4 gene-deficient (aka “knockout”) mouse dies within five weeks after birth from systemic autoimmunity. I’ve seen these mice several times – they have lymph nodes the size of large peas – and are half the size of wildtype littermates due to multiorgan autoimmune inflammation that is ultimately lethal. The phenotype of the PD-1 knockout mouse is less severe. These mice develop specific autoimmune diseases depending on the background strain of mouse use to derive the knockout. Autoimmune cardiomyopathy and lupus-like nephritis have been described. The PD-L1 knockout mouse has a very mild phenotype unless crossed back on strains of mice prone to autoimmunity, as was shown for lupus nephritis. The PD-L2 knockout mouse also developed renal disease although the mechanism differed from that of the PD-L1 knockout. Both manifestations required active challenge to induce disease, i.e. autoimmunity was not spontaneous.
So in this case the mouse genetic studies have previewed the human experience to the extent that CTLA4 blockade induces more toxicity than PD-1 blockade. As we’ll soon see, the combination of these immune checkpoint blockades can be more problematic that either given alone.
We’ll return with a toxicity by indication breakdown of irAEs induced by CTLA4 antagonists, PD-1 and PD-L1 antagonists and combinations of these agents with each other and other types of therapeutics. Then we’ll swing back to the TNF receptor superfamily for a look at the known and expected issues to be encountered there. We’ll finish with a few suggestions on novel ways forward – triggering T-effector immune responses to tumor without upending the systemic Treg environment. And as we do all that we’ll highlight a few key companies to watch.
stay tuned.
Other PD-1 pathway therapeutics in advanced melanoma therapy.
Yesterday we focused on nivolumab, particularly in combination with ipilimumab, for the treatment of advanced melanoma. There are competing PD-1 pathway inhibitors that have now reported out substantial trial data. See part 1 for a list of PD-1 pathway therapeutics in development. Much attention has gone to Merck’s pembrolizumab, formally called MK-3475. The activity of pembrolizumab in melanoma is very similar to that of nivolumab, so it’s worth taking a closer look at the characteristics of the antibodies. Included here is pidilizumab, another anti-PD-1 antibody, developed by CureTech.
Attributes of note include the different sources of the antibodies (fully human vs humanized murine antibody), different isotypes (IgG4 vs IgG1) and affinities ranging more than 200-fold from sub-100pM to 20nM. However, this is a small number of antibodies and it will be hard to discern how each of these attributes contributes to efficacy. Pembrolizumab closely resembles nivolumab except that the affinity for PD-1 is as much as 10 fold better. At the doses given it is difficult to know if this makes any difference, as drug levels may be saturating. We’d have to dig out target occupancy data from the trials to figure this out, but let’s look at the pembrolizumab results first, as it will become clear that this antibody has similar efficacy as nivolumab. How these therapeutics are being developed is different, as we’ll see.
The pembrolizumab (“pembro”) data reported at ASCO are from a huge Phase 1 clinical trial in advanced melanoma. Importantly, Merck made the strategic decision to stratify patients by prior exposure to the anti-CTLA4 antibody ipilimumab (“ipi”), from Bristol-Myers Squibb. This gave the company a jump on the field, allowing them to pursue FDA approval first for ipi-refractory patients. Due in part to the toxicity associated with ipi therapy, there are a lot of these patients. First, though, a brief look at the data, which has been widely reported. The data are compared to published data for nivolumab (“nivo”) treatment of ipi-naive advanced melanoma patients (http://jco.ascopubs.org/content/32/10/1020.long). A guide to the clinical abbreviations is included in part 1.
If we focus on the ipi-naive ORR and 1 year survival data I think we have to conclude that these drugs are pretty comparable, and we’ll wait for additional data before trying too hard to differentiate these. That data will have to come from longer duration of ongoing trials and various combination studies. It is clear from the monotherapy data is that for advanced melanoma patients, anti-PD-1 therapeutics offer a chance at extended benefit. If we look more closely however,we see that in the nivo trial referenced above, half of the responding patients stopped therapy for reasons other than disease progression, most likely dropping off study due to AEs. It is true that 3/4s of the nivo patients stopping therapy maintained a response, some for extended periods. In the pembro study, the SAE rate was 12% but only 4% of patients discontinued therapy as a result of AEs, so that’s good. The catch is that in order to move ORR higher than 40%, combination therapy may be needed. As we saw with the ipi/nivo combo, this comes with much higher toxicity and drop-out rates. Of course the hope is that moving to earlier line therapy will boost response rates with the same or less toxicity and that data will come with time. As an aside, the question of ORR is the reason we have basically ignored the anti-PD-1 antibody pidilizumab, which had a 5-6% ORR. The 1 year OS was similar to the other anti-PD-1 therapeutics, but with such a low ORR it’s hard to believe this therapeutic from Curetech will gain much traction.
Anti-PD-L1 antibodies constitute the second class of therapeutics targeting the PD-1 pathway. These are in early clinical development in multiple tumor types, and will be addressed later. PD-L1 is also important in the context of predicting response to therapy in melanoma, and the utility of this marker as well as PD-1 is the subject of considerable discussion. When the ORR is 40%, it is helpful to select patients prospectively. We can take a close look at one of the smaller cohort studies to get a good look at this. In a study of responsiveness to pembro, Richard Kefford et al (abstract #3005) used an analysis of PD-L1 expression to demonstrate a remarkable difference in clinical response between patients who had > 1% tumor PD-L1 expression versus those who were PD-L1 negative. Biopsy was required in the 2 months preceding the start of pembro therapy; tumor PD-L1 expression was assessed by immunohistochemical staining. Patients received pembro at either 10 mg/kg Q2W, 10 mg/kg Q3W or 2 mg/kg Q3W. With a median treatment time of 23 weeks and ≥13 months follow-up, ORR was 41%, median PFS was 31 weeks and median OS was not reached. The 1-year survival rate was 81%, so this was a terrific cohort within the larger pembro study, likely due to the higher doses used. PD-L1 expression was associated with improved ORR by (51% vs 6%), PFS (median 12 vs 3 months) and 1-year survival rate (84% vs 69%). Note that while there were no treatment-related deaths; 14% of patients experienced drug-related SAEs (grade 3/4) again reflecting the aggressive dosing schedule.
In the large trial of ipi-naive patients treated with nivo, PD-L1 positive tumor staining was associated with ORR, but only weakly with PFS and OS. Why the data are less robust than the Kefford study is unclear. What is abundantly clear however is that there were profound responses in patients scored as PD-L1 negative, as shown in this screen grab from Dr Weber’s Discussant review of the melanoma oral poster session:
These data suggest that caution should be exercised in the use of PD-L1 staining as a prognostic tool, and the search for better biomarkers of response continues.
We will revisit some of these issues as we move on to NSCLC, RCC, bladder, ovarian and solid tumors more generally.
Introduction
Anyone attending the immunotherapy sessions at ASCO earlier this month would have heard several distinct messages about PD-1 pathway inhibition in oncology. PD-1 appears to be a central control point for curtailing T cell responses in the peripheral tissues, similar to the role that CTLA4 plays in regulating initial T cell activation in secondary lymphoid organs such as the lymph nodes and spleen. Remarkable progress has been made in the 13 years since Gordon Freemen and colleagues first proposed in Nature Immunology that the PD-1 pathway was used by tumor cells as a shield against immune system attack (http://www.ncbi.nlm.nih.gov/pubmed/11224527).
It is clear that PD-1 pathway antagonists show tremendous promise in treating diverse cancers. Less clear is an understanding of why certain patients respond or don’t, what biomarkers might predict response, how to increase response rates, how to accurately measure response, and how to safely combine PD-1 pathway inhibition with other therapies.
Table 1 lists the PD-1 therapeutics in development (some of these therapeutics did not have updates at ASCO).
As the table demonstrates, the PD-1 pathway inhibitors are being developed in diverse tumor types. As late Phase 2 data and Phase 3 data are coming out we can begin to see the real promise of these drugs in clinical responses measured in large numbers of patients. The amount of data presented at ASCO was a bit overwhelming so to simplify the landscape we can address each tumor type individually, when possible. Some terms we will use are given in the table below.
Table 2 defines the RECIST1.1 clinical response parameters and their abbreviations.
To put these terms in perspective we can just consider that a meaningful clinical response is a measureable response to therapy (SD < PR < CR) that is durable and leads to an increase in PFS, which in turn allows a significant increase in OS. There are other terms used to describe clinical responses but these are the most common. We will start with some of the most recent data, and see where that takes us.
Part 1: Immune Checkpoint Combination Treatment of Melanoma
The very first trials of PD-1 pathway inhibitors began with the investigation of nivolumab in metastatic melanoma. As such, there was an impressive amount of progress reported and we now have mature data on different therapeutics. To set the stage, we can consider the benefit shown by nivolumab monotherapy compared to standard of care treatment protocols, and also to ipilimumab (brand name Vervoy) an anti-CTLA4 antibody, also from Bristol-Myers Squibb (BMY). Ipilimumab is approved for the treatment of metastatic melanoma based on Phase 3 clinical trial data in metastatic melanoma patients that had failed prior therapy (a chemotherapy regimen). The trial compared ipilimumab to a tumor vaccine targeting the melanoma antigen gp100. Ipilimumab treatment improved median OS to 10 months versus 6 months with the vaccine treatment (which was no better than standard of care). The 1 year survival rate was 45%. ORR however was low, just about 10%. Also, adverse events (AEs) were a problem, and included autoimmune manifestations (colitis, pituitary inflammation) and some treatment-related deaths (2% of patients). In a separate study of treatment-naive metastatic melanoma patients, ipilimumab therapy was associated with an OS = 11.2 months and a 1 year survival rate of 47%, falling to 21% by year 3. Patients were given ipilimumab or placebo plus chemotherapy (dacarbazine), and then moved to ipilimumab or placebo alone if there was a response measured or if the initial therapy caused toxicity. One consequence of this scheme was that AEs went up dramatically, with 38% of patients experiencing an immune related, grade 3 or 4 severe AE (SAE). We dwell on the anti-CTLA4 antibody ipilimumab because it is the benchmark for other immunotherapies such as nivolumab.
Nivolumab therapy for advanced melanoma has produced impressive data, with median OS = nearly 17 months, and 1 and 2-year survival rates of 62% and 43%. ORR was 33%. AEs were significant if less severe than those seen with ipilimumab. Grade 3-4 treatment-related AEs were seen in 22% of nivolumab-treated patients. Immune-related adverse events (all grades) were seen in 54% of treated patients, and included skin, GI and endocrine disorders. However only 5% of patients experienced immune-related SAEs of grade 3 or 4 and there were no drug-related deaths. These data from Topalian, Sznol et al. from John Hopkins University School of Medicine were presented at ASCO last year and published earlier this year (http://jco.ascopubs.org/content/early/2014/03/03/JCO.2013.53.0105.full.pdf).
So with that as our backdrop lets update the state of PD-1 pathway antagonism in melanoma. One of the obvious next steps in the development of immunotherapy is to combine treatments and we saw dramatic long-term data from the combination trial of ipilimumab plus nivolumab in advanced melanoma. Early trial results presented at ASCO last year introduced 4 cohorts of patients given different doses of nivolumab and ipilimumab in combination, with an ORR across all four cohorts of 40% and a 1 year survival rate of 82%. Median OS had not been reached. SAE rate across the 4 cohorts was 53%. This quickly gets complicated so let’s define the cohorts. Numbers are doses of nivolumab and ipilimumab, respectively, in mg/kg: Cohort 1 (0.3 + 3), Cohort 2 (1 + 3), Cohort 3 (3 + 1), Cohort 4 (3 + 3). No data were presented for Cohorts 6 and 7 so we’ll skip those. Cohort 8 is designed to mimic the dose schedule chosen for later clinical trials.
Note that after the induction phase, patients are moved onto maintenance therapy, as show below.
The slide is taken from the trial update presented at ASCO by Dr Sznol (Abstract #LBA9003). The data updates drove home several critical points. First, at the optimal dose rates of 1 + 3 and 3 + 1 the ORR ranged from 43-53%. The author’s introduce a new classification of clinical response to capture the observation that many patients are experiencing benefit while not strictly meeting RECIST1.1 criteria, this is termed “Aggregate Clinical Activity Rate” and reaches 81-83% in Cohorts 3 and 4 (note that Cohort 4 (3 + 3) was the maximum tolerated dose due to SAEs and will no longer be used). Perhaps more meaningfully, the percent of patients whose tumor burden was reduced by > 80% at 36 weeks was 42% across the cohorts. This is a remarkable number suggesting sustained clinical benefit. Indeed, in those patients who responded, the median DOR in Cohorts 1-3 plus Cohort 8 has not been reached. In Cohorts 1-3, 18/22 patients are still responding and 7 of those had discontinued therapy due to AEs (more on this below).
Dose cohorts were analyzed for impact on 1 and 2 year survival. In Cohorts 2-3 the 1 year OS = 94% and the 2 year OS = 88%. Most stunning of all was this data showing a median OS in Cohorts 1-3 of 40 months. Median OS in Cohort 3 (1 + 3) has not yet been reached.
These data are best-in-class for treating advanced melanoma, and place ipilimumab plus nivolumab at the forefront of therapeutic options for these patients. The one outstanding issue remains that of toxicity. 23% of patients had to discontinue therapy due to toxicity, and one patient died of complications resulting from treatment. While Dr Sznol repeatedly pointed out that the toxicities observed are controlled by standard interventions, the problem is that these standard interventions include cessation of therapy. We have already learned from the ipilizumab experience that responses to immune checkpoint inhibition can take time, and for those patients who have to stop treatment after 1 – 2 doses due to toxicity, time may not be kind. It will certainly be beneficial to reduce SAEs so that more patients can remain on therapy.
Tomorrow we’ll look at other PD-1 pathway therapeutics and combinations in melanoma before moving on to other tumor types.