Clovis Oncology's CEO Hosts R&D Day (Transcript)

Feb. 1.14 | About: Clovis Oncology (CLVS)

Clovis Oncology (NASDAQ:CLVS)

R&D Day Conference Call

January 27, 2014 11:00 AM ET


Patrick Mahaffy - President and CEO

Dr. Ruth Plummer - Clinical Professor of Experimental Cancer Medicine at the Northern Institute for Cancer Research, Newcastle University

Dr. Andrew Allen - EVP, Clinical and Pre-Clinical Development and Chief Medical Officer

Dr. Carlos Arteaga - Professor of Medicine and Cancer Biology; Associate Director for Clinical Research; Director, Center for Cancer Targeted Therapies; Director, Breast Cancer Program; Vanderbilt-Ingram Cancer Center; Vanderbilt University

Dr. Lindsey Rolfe - VP, Global Clinical Development

Dr. Lecia V. Sequist - Mary B. Saltonstall Endowed Chair in Oncology, Massachusetts General Hospital Cancer Center; Associate Professor of Medicine, Harvard Medical School.

Dr. Thomas Harding - Senior Director, Translational Medicine.


Robin Rankin - Credit Suisse

Yaron Werber - Citigroup

Cory Kasimov - JP Morgan

Brian Klein - Stifel

Peter Lawson - Mizuho Securities

Marko Kozul - Leerink Swann

Dr. Patrick Mahaffy

All right. Good morning everybody. We have actually a very full agenda, three hours. So I am going to get us started very quickly, and then turn it over to the scientific team. First of all, you know that slide you have seen is 8 trillion times, but just to remind you, things change. So these are your forward-looking statements.

Today's agenda is an overview of each of our programs, with an introduction that that's kind of the state of the field today in terms of clinical approach by three clinicians who are expert in the field, and I'd like right now to thank each of them for coming. We have Ruth Plummer, from University of Newcastle; Carlos Arteaga from Vanderbilt; and Lecia Sequist from MGH. And so it's really a great event, they have taken their time and make themselves available to you as well.

Briefly, I'd like to introduce some of the team as well from Clovis just sitting over there; Gillian Ivers-Read, heads up Regulatory and Technical Operations; Anna and Breanna, all of you know, who have done a spectacular job getting this organized. Erle Mast, Chief Financial Officer; Steve Hoerter heads up Commercial; who've I missed? Andrew Allen -- Andrew Allen, our Chief Financial Officer, you can see a lot of him today. I'd like to introduce two, who will be speaking, who you've not met, I think before; Lindsay Rolfe, our VP Clinical, who actually has the primary lead, particularly in 1686; and Thomas Harding, and Tom runs our preclinical and translational group.

And with that, I am going to disappear and turn it over to Ruth Plummer, who is Professor at the University of Newcastle. Ruth actually, in addition to being involved in our development of Rucaparib, is the physician who dosed not just the first of the first three patients with Rucaparib. So she has real experience. Ruth, thank you.

Ruth Plummer

Well thank you very much and I apologize you upfront if I talk too fast, it's a failing of mine and I always do. Well Andrew asked to outline similar to sort of clinical space, but also the science behind PARP. I am totally a PARP enthusiast, being described in slightly relative terms. I do work at Newcastle University, and actually run the early phase trials unit there, which is called the Bobby Robson Unit.

You can see that I have been working with PARP for a while. Those are my disclosures, and in terms of clinical trials and early phase trials, I have worked with a number of the agents in the class, over a number of years.

What Andrew asked me to do, is to firstly set the theme in terms of the ovarian cancer, because of the focus of where Rucaparib is, the development of that drug is likely to go. And as you can see on this slide from a US perspective, ovarian cancer is in the top 10 female cancer, it creeps up the rankings, if you actually look at cause of death, and that will become clear. That's because, it's one of the cancers, positively pancreatic cancer, but presents late, and therefore it's hard to cure surgically.

Worldwide, it again is in the top 10 female cancers -- and cause of female cancer deaths. Drops down a little bit in that ranking, if you can call it a ranking, simply because worldwide, cervical cancer, because of lack of screening in some third world countries, is higher on a worldwide basis, compared to the, what's called the developed, the US, Japan and Europe, where it would drop down a little bit in terms of cervical cancer.

In terms of the five year prevalent cases, it is predicted that the incidents of ovarian cancer will rise. This graph affects 2011 basis in terms of number of cases, with the US, because of its larger population, having a similar number of cases compared to the sort of four or five major European countries. This predicted increase is totally related to the fact that what we would hope, if you get better treatments and patients surviving longer, but also with a global aging population, this is a disease of the elderly, and therefore it is expected that the number of cases of ovarian cancer worldwide will continue to rise over this period.

This, as I said, is a cancer that usually presents; the majority of cases presenting stage-IV disease, where the disease is spread outside the ovary and actually there are seedlings throughout the -- what's called the peritoneal cavity within the abdomen, and these patients therefore cannot be surgically cured, because you can't cure the disease.

The survival is reflected in terms of 5-year survival, but rather poor presentation pattern with ovarian cancer, because there is only about 15% of patients diagnosed with truly localized disease. It's hugely spread, in a majority of cases to regional disease. The distant spread is a lower percentage, only about 25%, that actually is relatively rare for ovarian cancer to spread outside the abdomen, and even in patients with liver metastases, they tend to be on the purpose of the liver, not buried within it. But as you can see, the percentage of people diagnosed with that later stage disease, represents the majority of this, and those patients are incurable, and would inevitably need systemic therapy throughout the rest of their life.

This, I have been practicing oncology for 15 to 16 years now and this hasn't changed a great deal in that period. Patients hugely present a symptomatic disease. There has been screening programs, but is not well established or successful. The patients usually presents with increasing weight gain and the site is reflecting the late stage disease. The backbone is primary surgery, sometimes with chemotherapy first. It's always been a platinum and the big additions, just as I would start you out with the additions of taxanes; and carboplatin/paclitaxel remains the mainstay of treatment. Second with laparotomy is that acronym SLL, with patients after initial chemotherapy, that tend to be responsive to surgeons who sometimes go back in to try and resect any remaining disease. There are some trials and maintenance treatments, but then on possession, its second line chemotherapy, and this disease, thinking it through -- I am not the hematologist, but even in hematological malignancies, stands alone in this. You get repeated responses to platinum-based chemotherapy, which is pertinent to close this focus on the group of patients that they might give Rucaparib to be used. But this is very unusual in the oncology world for us to keep going back to the same drug. You get quite a high percentage of patients have a second response, its duration is usually shorter, but you can keep going back to the same drug, which is unusual and it's the platinum that is the backbone of doing that.

In terms of recent advances, both Andrew and I looked at that recently. It's a little bit depressing it seems to me. There is a lot of interest in the angiogenesis inhibitors. I think if you are a patient, its nice to see an increase in progression free survival, but those on average, are a couple of months. There has not been big breakthroughs since we got the taxane out, it's the platinum in all of the studies have looked at either adding in a VEGF inhibitor, both the antibodies and the small molecule inhibitors, with some improvement, but increase in toxicities is certainly the maintenance disease, the quality of life improvement was not there, because of the toxicity of the maintenance treatment.

In terms of chemotherapy, (inaudible) is used as the chemical sequence by some physicians, the big hope was adding in the (inaudible) the Doxorubicins, again you can see the improvement in progression-free survival, not really very impressive, just a matter of a few months. So this is a sort of state-of-the-art of ovarian cancer treatment currently.

And now, and I promised Andrew I won't talk for too long, to talk about PARP and DNA repair, and I am (inaudible) to go through this slide. The reason I put up a cartoon is, this is the majority of the major features of the two of the five pathways of DNA repair. We call can repair our DNA, its very important for protecting our genome and those single strand breaks, SSB, is a break across one of the pairs, within the DNA. We get tens of thousands of those spontaneously happening a day in our cells, and they are repaired very efficiently by basic excision repair, a pathway in which PARP participates.

DNA single strand breaks, is not repaired as the cell divides, will actually become a DNA double strand break and I have a further cartoon of this, and there are two pathways, the double strand break repair, because otherwise, this is a catastrophe for the cells, and they will actually die. So we protect ourselves very strongly, with double strand break repair. One of the reasons for putting this up with the various means of the proteins, is here circled in the red color. You will see on the subsequent slide, proteins that are commonly mutated in ovarian cancer. So there is a signal in ovarian cancer, this is a very important pathway, and we are looking to emphasize you, as we move to talk to you about the BRCA. But familiar in breast and ovarian cancer patients group, is that the BRCA1 and BRCA2 proteins that are mutated in this germ-line inherited form of cancer, actually fit in this either signaling that with the double strand break, or involved in the repair of double strand breaks in dividing cells and cancer cells are inherently divided. So that's why, the BRCA pathway is actually important.

This goes to that the -- the cancer genome act, plus here showing in. If you look at the most common form of ovarian cancer, the high grade serous ovarian cancer that the majority of our patients have. 50% of cases have mutations, alterations in those signaling proteins that I highlighted in the DNA repair pathway. So we know that in patients with blood-related cancers, there are a proportion of patients, a significant proportion of patients with ovarian cancer, who are likely to have a defect in that pathway, and therefore, potentially vulnerable to a PARP inhibitor treatment.

As we told you about PARP; PARP is an enzyme -- PARP1 and PARP2 are enzymes that sit in the nucleus and act as molecular (inaudible). They signal that the DNA single strand breaks. We have about one (inaudible) DNA in the nucleus, that's inactive until there is a break in the DNA. They bind, they activate the enzymes that uses NAD to make polymers or poly ADP-ribose, which are negatively charged and help open up the breaks.

All of the current potent PARP inhibitors in the clinic use this nicotinamide molecule here almost as a template. And that -- they have all been designed, as that fits in the binding side. So you will see here, the hydroxy-aiming group here of nicotinamide. Those binds in the active site. The third compound, the potent one, was actually a Newcastle University compound, out of which the theory that Rucaparib is part of [kane]. But you can see also, looking at the Biomarin compounds, Tesaro, the Merck-Tesaro compound, Olaparib from AstraZeneca and Veliparib, the Abbott, AbbVie compound. Those are similar areas. These compounds are all, kind of different degrees, biding in that same area.

In terms of the idea about synthetic lethality that has been much in the published press about. In cells, where homologous recombination, should the double strand break repair is positioned or worked. If you got PARP and the cell tries to divide, it would simply use double strand break repair to mend the lesion and be able to remain alive and be healthy.

If the cell is deficient in homologous recombination, it gets if you block PARP. So these elements that I mentioned, including BRCA1 and BRCA2, will actually process. With the PARP inhibitor prevalent in cells that are proficient in homologous recombination. Whereas cells that are deficient in, particularly in BRCA1 or BRCA2, there is largely preclinical data and now clinical data as well, showing that those cells and in these patients that is their tumor alone, if it's not in the rest of the patients, will not be able to process a double strand break, if single strand break repair is blocked.

So tomorrow, we get to work with Thomas Helleday, who was (inaudible), with us and now back at the Karolinska, was one of the Rucaparib theories, with the first application of this synthetic lethality, alongside debate with Alan Ashworth's group at the ICR in London, and the concept here is that, we do that, tens of thousands of single strand breaks a day. In non-dividing cells, you can wait to have those repaired in dividing cells, as you peel apart, and I was saying to Andrew earlier, its almost like two ply wool, but most of the audience wouldn't follow that analogy. It's like a piece of string with two strands. If one the strands is cut, that string is intact. But if you were to peel a part of the strand, you will end up with a snap, unless you can repair that break. So usually, these single strand breaks are just repaired all the time by PARP. If you block PARP in a dividing cell, the single strand break of the DNA, yield the PARP to copy will become a double strand break. In normal tissues, that will be repaired extremely easily, but if there is a BRCA1 or BRCA2 defect or if the tissue have a homologous recombination defect, that will actually become a double strand break and collapse replication fork, which is a very potent stimulus to apoptosis.

So that was a concept in terms of synthetic lethality of using PARP inhibitors as single agent, with a very strong rationale also, but the rest of the patient is protected, even on a germ-line BRCA background, by having one gene, one function allele of double strand break repair. That was the background on which -- we felt that we folks could go forward, in a relative non-toxic, but potent manner.

Can we exploit that in the clinic world? Yes, we can and there is preliminary data showing that. We also have inhibitors of homologous recombination. So you potentially could just block both pathways, by combining them, and I think there are very exciting opportunities to do that, when you would have tumor vulnerability. The larger genetic alterations, ovarian, breast, pancreatic, probably prostate cancer as well on a BRCA background, where you could exploit this, just on the germ-line background, and also, we know that a lot of tumors -- that's a vulnerable pathway, a lot of tumors, and particularly high-grade serous ovarian cancer, significant proportion of carrying mutations by chance in this pathway, of a tumor generating lesion and the new stability lesion, and therefore is a potentially patient population, where the tumor will be sensitive to a PARP inhibitors, and obviously as we all know nowadays, in terms of molecular profiling, to do this right. Patient selection is key, which is why to me, the model of Clovis with a companion diagnostic is what would be a very compelling one, in terms of somebody, who has spent their life in drug development.

I think Andrew, I think that's it. That takes time.

Dr. Andrew Allen

Great. Thank you, Ruth. I'd just like to remind everybody that this is being live webcast, and if you wish to review afterwards, it will be available on our website. Let me go to Q&A, which we will do at the end of each drug section, there will be a handheld microphone, which I will ask you to hold your question until you have that microphone in your hand, so that your question is then audible on the webcast.

Okay. So very much introduction to PARP. Rucaparib is, a PARP inhibitor as you all know, which we licensed from Pfizer in 2011, and we selected recently, the dose that we will be moving forward with, in our Phase 2 and 3 studies.

The PK Rucaparib is very pleasing from a drug developer's point of view, with most type PK, and low ranges of exposure with dose proportionate of PK, as you can see on this slide here. So all of the patients dose about 360 milligrams BID achieved the target trough concentrations in plasma, and you can see here, just looking at AUC, which is a decent sort of integrative measure of exposure, to have this nice dose proportionality.

In terms of adverse events, which is a very important attribute, when we are talking about drugs being used with maintenance therapy. Rucaparib seems to be extremely well tolerated. Now this is all dose levels, with all adverse events which is being related to Rucaparib and described in at least 5% of patients from the clinical studies to date, which is over 100 patients, and you can see that there is some fairly toxicities of mostly Grade 1, nausea, fatigue and vomiting, that you see in the context of ovarian cancer, can sometimes be hard to tease out with the regulated adverse events from the disease related adverse events. But nonetheless, obviously you can see, that this is a very well-tolerated therapeutic.

Now we don't see it so clearly on this slide, that anemia and neutropenia are seen reasonably frequently, and actually was a major driver of selecting the appropriate dose. I then turn to emerging cycle one, they do tend to occur cycle two, cycle three of the cumulative toxicity. They are not typically captured on adverse events, but they are found in the laboratories abnormalities database. So as we've seen with the rest of the class, we do see myelosuppression with continued dosing.

Now we've selected a dose of 600 milligrams BID. Obviously, we have a larger group of patients treated at more dose levels, and so we have aggregated some of the efficacy data. And if we look at the event, and the patients specifically who had BRCA mutations, and these are nearly all, germ-line BRCA mutations. But for patients treated at 360 milligrams BID or above, the adjusted response rate, which includes RECIST and CA-125 response, is 45%; and obviously, it's a very small run, so we are not going to make a big deal out of it. But at 600 milligrams BID; so far, we have four patients who have reached the first evaluation, and two of those who have achieved adjusted responses as well. So we can see that clearly, in this approach with a targeted population, we have a very effective drug, as one would expect. This is not a model observation.

We have also seen activity in germ-line BRCA mutant breast cancer, and here you can see, in our CT scan, showing example of that. And we have disease control at 24 weeks of 50% in the BRCA mutants with breast cancer, which again is a very compelling result for a heavily pre-treated population, who have come on to study -- a phase-1 study, of course having exhausted all typical approved and available therapies. And again, if we look at the scope of patients treated at 360 milligrams BID or higher, we have a 57% objective response rate.

We also are very excited to see a pancreatic cancer responder, which is approaching the germ line BRCA2 mutation, which is typical in pancreatic cancer. This patient progressed very quickly, on the biggest (inaudible) that we had, in terms of chemotherapy, which is Folfirinox. This poor gentlemen progressed after four cycles of full dose Folfirinox, and then entered a beautiful response on 360 milligrams BID of Rucaparib. His overall duration of response was just north of six months. He was on study drug for about 11 months, and obviously with the patients who have just progressed with Folfirinox, this is a really wonderful result for this gentleman.

So let's take a look at our registration approach. There has been much discussion about this, because obviously we have some competition, and we have all taken slightly paths. Here with the guiding principles we have in front, our development strategy and our registrational strategy.

So first of all, we have learned from the (inaudible) frankly. We have got combinations of cytotoxic chemotherapy plus PARP inhibitor were very troublesome, because the myelosuppression is negatively additive or even synergistic and we end up reducing the dose of either the chemo or the PARP inhibitor and we end up with a very messy combination, where you compromise (inaudible) dose of one of the molecules or both, and thus you end up not really knowing, whether you are actually helping at all, but in fact combination is really achieving what you could have achieved with full dose chemotherapy for example.

That complexity, was one we wanted to completely avoid; and so we elected to pursue monotherapy registrational strategies. And of course, Ruth wanted to go -- had clinical data suggesting activity, and therefore de-risking.

Ovarian cancer is very attractive when we follow these two very simple principles; firstly, because we knew that PARP inhibitors as monotherapy have a lot of activity in ovarian cancer. Both have objective responses and prolonged stable disease; and also, because the monotherapy maintenance study, after cytotoxic chemotherapy, is an acceptable registrational strategy in ovarian cancer, with a placebo control.

So Ruth showed you the slide, the standard treatment of ovarian, you get tablet chemotherapy typically and you finish your six cycles, and you stop, and patients have nothing and they just wait, and unfortunately for most of them, they are just waiting for the disease to come back. That window, that interval of no therapy, is your maintenance opportunity, to randomize to PARP inhibitor versus placebo, and obviously that is a very straightforward registrational track, and is one that we have chosen to pursue.

Pancreatic cancer is attractive, because in most BRCA mutant patients, objective responses are unheard of, without therapy. So objective response with therapy, potentially, is an accelerated approval endpoint, in a selected group of patients, who have no alternative therapies, who have done badly on the frontline chemo, which has been the approved regimen in pancreatic cancer. So that's obviously an alternative strategy for accelerated approval of the drug.

The challenge in breast cancer is, first of all, yes we know that there is activity in germ line BRCA mutants. But beyond that, there is much less clout as to whether the drugs work, and if so, in whom? Secondly, single arm studies with response rates in breast cancer, are unlikely to be acceptable for accelerated approval. Well obviously an example we would all point to, is the T-DM1, which was a very effective, very safe targeted therapeutic, which was not acceptable for accelerated approval.

Thirdly, if you do a comparative study against cytotoxic chemotherapy in the germ-line BRCA population. We've run the risk that the chemotherapy is going to be very effective in germ-line BRCA mutant patients, and that was well shown in ovarian cancer, in a randomized study in platinum-resistant patients, germ-line BRCA mutants randomized to a lapra versus Doxil. The good result that the lapra works very well. The bad news was, that Doxil worked very well as well, and so that was actually a negative study, because chemo works well in BRCA mutants, and so the control arm is challenging in comparative studies. Overall, we are bored of breast cancers. It's trickier at this point.

Focus then on the ovarian cancer strategy; we were fortunate to have an opportunity to learn from some work with AstraZeneca due to the lapra, and this is a key study done by Jonathan Ledermann, who is now principal investigator on our Phase-III study. This is a study, recently published in the New England Journal, and it was in that randomized maintenance setting of ovarian cancer. So these are 265 women, with high grade serous ovarian cancer, who have had at least one previous response to platinum based chemotherapy and have now had their second all subsequent treatment with platinum based chemotherapy, and would enter this early remission, either complete or partial remission. And those women were randomized to receive placebo, or a lapro at the end of the cytotoxic chemotherapy and they were followed to PFS as the primary endpoint. And in the overall population of 265 women, the hazard ratio of the PFS was 0.35. So obviously very compelling results.

Now, retrospective analysis presented at ASCO last year, went to look back that the germ-line BRCA status of the patients, and they had about 122 wildtype and 96 mutants, and in that population, the hazard ratio for Olaparib was even stronger than the overall population, at 0.17. Furthermore, in ovarian cancer, because of the typical primary surgical procedure generates an abundant amount of formerly fixed paraffin-embedded tissue. Those archives of tissue are available in most women with this disease, and they are able to go back and get the tissue BRCA status, the 252 out of the 265 women, and they found roughly 40 additional patients, who have normal germ-line BRCA genes, but would have required somatic BRCA mutation, just in their tumors; and those somatic BRCA mutations have exactly the same biological consequence as a germ-line mutation. The overall hazard ratio in this population is 0.11, and you can see that the medium in the Olaparib treated group did not budge, telling you that the broad lever expensed in the tissue BRCA mutants, the somatic BRCA mutants was exactly the same, that in the germ-line BRCA mutants.

So this has been informative to us, and itself that, you are missing an opportunity to treatment and the benefit from drug, if you just look at the germ-line BRCA, and the germ-line BRCA testing is not the right path, in this context. It's fine, if you try to screen people for risk, which is of course how the germ-line BRCA test was originally developed by Myriad. But when we have a patient with established cancer, it's not the right test. You should be testing the actual tumor, looking for a tissue BRCA mutation, which integrates both the online and somatic mutations.

Now, an even more interesting analysis from my perspective was, this breakdown that Jonathan presented. Looking at the BRCA mutants on the left here; and as I mentioned here, the hazard ratio was 0.18 in this slightly later analysis, 1.2 versus 4.3 months. And this BRCA mutation, was the tissue BRCA mutant, so this includes both germ-line and somatic, and we are looking at this line here, the Olaparib recipients versus the black line, the placebo recipients. But very interestingly, in the patients with normal BRCA genes, we still saw a very meaningful benefit, in terms of progression-free survival, with a hazard ratio of 0.53. That's the blue lines versus the yellow lines.

So, you may think that perhaps platinum sensitivity is a very usable tool for finding patients that benefit from PARP inhibitors, we didn't need to do sophisticated genetic analysis. But I would suggest, that that's not the case. If you looked at the early immature survival analysis that was presented from this trough, up to 38% of patients had died. Overall, there was absolutely no benefit in terms of survival.

If you then go to subset analysis, influenced by the tissue and somatic analysis I just described, the germ-line BRCA patients actually has, suggestion of benefit, with obviously the 95% confidence intervals cross one, so you can't claim statistical significance here, but obviously a suggestion that emerging benefit in the BRCA mutants. But in the non-BRCA mutants, again, I wouldn't make a big claim to say that this benefit, but absolutely no evidence of benefit of drugs. And I think what this is telling us is, that they have a group of patients, who are platinum sensitive, who really don't derive any benefit from PARP inhibitors. They are dying quickly and they are dying symmetrically on the Olaparib arm versus the placebo arm. So platinum sensitivity is not correctly calling all patients who can benefit from drugs.

So can it be better? We have a way to take this speculation of patients who receive Olaparib in the past, and for those who are truly, if you like, the existing for PARP inhibitors, who derived no benefit from it, versus those who actually derived meaningful benefit from PARP inhibitor, and pushed this blue line further down towards the black line of the BRCA mutants.

So let's step back to some of the bars that Ruth was describing. The insight that has emerged from the Cancer Genome Atlas work, is that there are many homologous recombination defects at a genetic level in ovarian cancer; meaning that some of those proteins involved in homologous recombination have mutations that lead to defective proteins. Therefore, homologous recombination doesn't work, they have homologous recombination, and as we have outlined, they will therefore develop synthetic lethality when treated with a PARP inhibitor. BRCA is a good example, but then many other genes involved, and she showed you Fanconi anemia family gene, APM, APR, RAD51, POLD2, there is a whole list of them. They have been mutated in ovarian cancer. They have been moving to the same genotype, such that when you treat them with a PARP inhibitor, they develop apoptosis.

But we are not smart enough to really understand all of the reasons why people develop homologous recombination defects; and of course, increasingly these days, we recognize

that there are epigenetic drivers of gene silencing, which is effectively the same genotype. Perhaps even my gene can be completely mutated and dissected, or it can just be silenced, than with low protein presence, and I may as well have a defective gene, because there is no protein allowed to do the job.

Methylation is a well-described mechanism of gene silencing, and is also being well-described in ovarian cancer, that BRCA methylation is a common cause of BRCA gene silencing, even though sequences of BRCA gene is normal, it can be functionally silenced, not expressed.

There is also growing interest in micro-RNAs mechanism of gene silencing. All of these approaches leads to defective HR pathway expression, and it sends genotype of HR deficiency. So if we are trying to give credit, we can look for homologous recombination defects for the genetic level. But we are going to miss them, unless we start really complicated, multi-modality assays. And its (inaudible) there is a much easier way for us to approach this problem, which is to look downstream. The common consequence has been HR deficient, is that the tumor cell develops genomic scars. It has abnormalities in the genetic material, that are very characteristic of HR deficiency, so called genetic scarring. That's the key to our approach to finding the right patients to treat with the PARP inhibitor.

[Mainly], we have two clusters of each gene, and there are different types of genomic styles that can occur. You can get mutations, obviously as we have discussed, deletions, amplifications, translocations, and sometimes you get these long stretches of one copy of a gene and that leaves you with only one copy, and that's termed loss of heterozygosity, LoH.

Now interestingly, different mutations, different scars, are associated with different DNA repair and analysis. You've probably heard of mismatched repairs, which is common in colorectal cancer, that leads to one particular type of genomic scar. HR deficiency, interesting one for us, leads to LoH. So this is the scar that is characteristic of HR deficiency, this is the scar we are interested in, loss of heterozygosity. And Foundation Medicine were able to address this question with their platform, not only sequences gene, they will sequence a bunch of SNPs as well. And by looking at SNPs scattered regularly through the genome, you're able to assess, whether there's LoH through that tumor site genome. And so the standard FM approach would lead to quantification of LoH, which is our definition of the genomic scar of interest, and thus, we can define those HR deficient genotypes.

Now in addition, as you well know, Foundation Medicine sequences genes and they are sequencing the BRCA genes as well, at the tissue level. So by working with Foundation Medicine, we were able to take direct results of BRCA which is really important as we have seen, and we sequence it in the tissue, capturing both GEN1 somatic, but also, we can quantify the LoH, which is that downstream common outcome of being HR deficient. So its two ways to find the responsive patients to PARP inhibitors.

Now if we apply this approach to the TCGA Dataset, which is now several hundred women with high grade serous ovarian cancer; and if you just ask the question, let's look at the overall survival of these women, and let's break them into two groups, broken by those who are positive on our tests and negative. So our test with Foundation is to look at either if there is [serious] BRCA mutation, or positive HRD assay, meaning you've got less of LoH. Either way, you can become positive; and you can see a very meaningful separation of the survival curves, when you just do that test. While the hazard ratio is 0.53, and of course, much of the survival difference is driven by differential response to platinum based chemotherapy. But the one thing that all these women are getting, is platinum-based chemotherapy for one or typically, several cycles.

So that's the first interesting analysis. Now, it's not just about BRCA. On the left, you can see that -- if you actually look at people work BRCA mutations, so women with tissue BRCA mutations, and why do you ask the question, how many of them had this LoH signature and how many don't? You can see, roughly that's a two to one split of being HRD signature positive or negative, and as you might expect now, those women, who had both the BRCA mutation and the HRD signature, do better than those who are BRCA mutant, but don't have the HRD signature.

And if we look in the non-BRCA mutant patients, again, you see this nice separation of the curves. Further evidence that this signature that is developed with Foundation, is over the need to identify patients, who do well in platinum, and would seem likely to benefit, from PARP inhibitor therapy. And importantly, we are identifying women, who don't do so well in PARP inhibitors, who probably should be receiving some other form of chemotherapy.

So to sum this up, if we look at the entire population of high grade serous ovarian cancer patients, we think that Rucaparib is likely to have potential activity in different groups of patients, based on their degree of HR deficiency. The BRCA mutants, as far as we can tell, these are function -- and so whether its somatic BRCA mutant, which is about 8% of all patients at diagnosis of germ-line BRCA mutants is at 15%. We think the biology of the same, when we analyze this group together, it's tissue BRCA mutant.

Then we have this group who are non-BRCA mutant, but have the HRD signature, this LoH signature. That, we think will also do very well on Rucaparib. And then there is a group we think, who are HR proficients, who don't do so well on Rucaparib. Now these statistics are given at the outset of disease, at the point of first presentation. Bear in mind, typically offset is, is we will be looking at platinum sensitive women, which actually enriches for these top groups. So these women down here, are typically where the platinum resistant patients tend to lie, and those are patients that quickly get weeded out of their typical trials, and so there is a relative enrichment of these groups up here.

So our overall development strategy in ovarian, we started out thinking, let's just do the pivotal study, let's repeat Jonathan Ledderman's study, maintenance design with Rucaparib in women with platinum-sensitive ovarian cancer, who have responded to their second or third or fourth platinum, that had got chemo and then you can randomize them to a placebo or Rucaparib. But the signature that I talked about was to find HR deficiency, so far has been largely developed in-vitro using PARP inhibitors, or in clinical data using platinum-sensitivity as a proxy, and as I said, its not a perfect proxy. So we decided to actually run a Phase 2 study in parallel with the Phase 3.

So our goal is to prospectively assess efficacy in Phase 3, but to get there safely and securely, we are running a Phase 3 biomarker study that's now open, where we are taking women with platinum sensitive, who have recovered, and are all receiving single-agent treatment with PARP inhibitor. We get their archived biopsies, we get fresh biopsies, and we assess the efficacy of Rucaparib in those patients with the genotype and the phenotype of the patients. So it's a parallel track, two trial strategy, taking us to the finish line.

So a little bit more detail with the Phase 2 study, the biomarker study that's called ARIEL2. Its open for enrolment, has been there for a couple of months. High grade serous ovarian cancer, at least one prior platinum; they have to be platinum sensitive; they have to have adequate tissue for us to be able to do the fresh biopsy; we can't have had a PARP inhibitor previously. Everybody gets 600 milligrams BID Rucaparib until disease progression, and we look at efficacy in the HRD subgroups. This is a total of 180 patients and they have capped the number of germ-line BRCA patients pretty aggressively, because we know that germ-line BRCA, we don't need a lot of those patients to inform getting it right, differentiating between HR deficient and HR proficient patients, who have normal BRCA genes.

The Phase 3 study, very similar to that Jonathan Ledermann study, and again, Jonathan is running this study first in Europe, Rob Coleman with MD Anderson, our PI over here in the States; and the eligibility, very similar. High grade ovarian cancer, at least two prior platinums; platinum sensitive; you responded to your last platinum and we have archival tissue. You take patients that are stratified upon randomization, 540 patients total; its a two to one randomization to a Rucaparib plus the placebo, and the primary endpoint is PFS. That sounds just like the study I showed you, the one key difference is this one, but we are not doing an all (inaudible) analysis off the bat, we are doing a step-down analysis, through those different genetic groups that I have described to you.

So of the blind break, the first efficacy analysis is PFS in the tissue BRCA mutants, so that's both the somatic and germ-line BRCA patients. If we pass that test, we can then step-down, and preserve out with this range, if you step-down, your second test is then in the HRD population which is both BRCA mutant, but also the BRCA wildtype to have this genetic signature. And obviously, we are very confident that this will be a robust positive result as well.

Through those wall, you can then do the -- in terms of truth analysis, but of course, the agency are not fools, and if in fact, the drug doesn't work very well indeed in HR proficient patients, then we would not get the label in those patients, and they should be treated with something else, if the drug doesn't work there, and that survival analysis I showed you earlier, does address that there is indeed a group of patients who are HR proficient, who really don't get much benefit from PARPs. So let's try and identify those prospectively, thus improve the benefit risk profile of the product.

Okay, the trials have been running parallel, but because we have all three studies, it's a straight pool of single arm study that will finish faster. We can then look out the results of that. We can determine that the -- if we want to tweak our signature at all and the product respectively for the ARIEL3 study. So you can (inaudible) ARIEL3 as we are doing now, without having to find that signature definitively, you can leave that action open, as long as we of course, never look at the data from the ARIEL3 Phase 3 study. So we are completely blinded to this study, its enrolment that we are blinded, we can use this study to inform the analysis plan.

Okay, our development strategy is distinct from our competitors. You can see that the competitors have basically focused on germ-line BRCA mutations, as you all know, and probably have taken a different path, for reasons outlined.

Finally now, switching route to pancreatic cancer. It's become clear, that pancreatic cancer commonly has a genetic signature that's very similar, provided the ovarian cancer patients have just described. Recent data at the Nature described a particular pattern of mutation and in breast cancer, ovarian cancer, and pancreatic cancer; and may be BRCA mutations are reasonably common in pancreatic cancer, and their study suggests that somewhere between 10% and 20% of patients with pancreatic cancer, will have a germ-line BRCA mutation. We also know, that a group of patients with the exact figures probably between 5% and 10% will have a somatic BRCA mutation. This is new data that is just coming out now from sequencing work in pancreatic cancer.

So we have designed this study that Pat announced at JP Morgan, a very simple study, up to 100 patients, single arm in BRCA mutants with pancreatic cancer, who progressed on one or two lines of monotherapy, measurable disease, they get Rucaparib until progression, and the primary endpoint is reach a subjective response rate.

So the overall registrational strategy is to go for initial approval in the US group, obviously a simultaneous drug on diagnostic approvals, and (inaudible) think the design cancer strategy, with ARIEL3 as the leading study in ARIEL2 is supported, and obviously will be significant to companion diagnostic test of an NGS-based tissue test, and clearly the recent approval of the [MISIC] tests are very encouraging, suggests that the FDA is trying to encourage this type of NGS pathway.

Pancreatic cancer is pretty straightforward. If we are successful and we have a response rate of 25% or greater, we have (inaudible) authorities, and we have a very decent chance of getting accelerated approval for this enormous unmet need in that terrible disease of pancreatic cancer. We are obviously going to be working on the companion test, which again will be NGS-based.

Specifically, this drug is somewhat complex, because the original composition of master patent is quite old, but we have a lot of new information, particularly with some of the newer salts that we have been using, and so variously, the patent protection extends to 2020 to 2031, depending upon obviously which patent is issued, and of course there is also regulatory data exclusivity in the US and in Europe, and we have orphan drug in ovarian cancer, and we will be submitting it in pancreatic as well.

Okay, so in sum, Rucaparib is a potent inhibitor of PARP. Good over interactivity in germ-line BRCA mutants in Phase 1, a very differentiated development program, particularly in ovarian, but now recently in pancreatic cancer as well, and a really robust collaboration with our good friends at foundation medicine.

Okay, thank you for your attention. Now we will move to questions. Yes please. Can you have the microphone, we have it. Over here please. Thanks. I will pull in you, Ruth, as required.

Question-and-Answer Session

Robin Rankin - Credit Suisse

Thank you, Andrew and Ruth. Robin Rankin, Credit Suisse. Perhaps first question for Ruth, can you just describe the mechanism, whereby tumors of the patients rather get resensitivities to the platinum agents and linked in with the second part of that question, probably for you Andrew. How should we think about the sequencing of using PARP, its first or second or third round of platinum agents?

Dr. Ruth Plummer

I think, ovarian cancer is very unusual, as I said. I don't think it’s a recent high dosage of platinum, I think they remain sensitive. It’s a cytotoxic chemotherapy, with chemos of toxicity. So once you've given a patient six to eight cycles, they need a break, so that's why you stop. That's where, in my opinion, as some of you treat the patients, the PARP inhibitors are extremely attractive, because they are far less toxic than cytotoxic chemotherapy, and you actually can do the very protected treatment. It's not the same as giving repeated cycles of cytotoxic.

The other problem with platinum, as all those patients remain sensitive, a significant proportion of them actually become allergic to platinum and this crossover between carbo and cisplatin. So it's not something you can keep doing, even if the patient remains sensitive.

Dr. Andrew Allen

Second question obviously. So patients hopefully will do well on Rucaparib, for a long period of time, probably eventually (inaudible). At that point obviously, typically you then give them platinum again, because they are platinum sensitive by definition. As I mentioned, combination is really tricky, and so probably the prudent thing to do, is we stop the PARP, give them the platinum again, and essentially restart the PARP as maintenance once more. We should put it to trial to kind of affirm that that made sense.

The early concern, was that actually you might have modified the response to subsequent platinum, because it would consume the use of PARP inhibitors. Those were converged with, say the last year or so they have pretty much put that concern to rest, by looking at that duration of response to the subsequent round of chemo, seems to be just the same. So I think, that notion of sighting is probably what we will start to do.

There was a study that actually could combine PARP with keynote, followed by maintenance. This was done out of Canada, and when Kaplan-Meier posed a study, it's during in the combination phase, the PFS, Kaplan-Meier is absolutely on top of each other, and they are only separated, once you enter the maintenance phase. And of course, at the end, there was a compromise on the base of the chemo during that combination phase. I think your evidence is pretty clear, that combining is not a successful strategy, it’s a maintenance monotherapy where you can get best value out of these drugs.

Robin Rankin - Credit Suisse

Thank you.

Marko Kozul - Leerink Swann

Thanks. Marko Kozul, Leerink Partners. Can you talk a little bit about pancreatic cancer, the kind of durability you'd like to see? Then also maybe, what registrational strategy for this pancreatic side?

Dr. Andrew Allen

Sure. I think (inaudible) months, and again, given the context, given how awful pancreatic cancer is, as we all know, in the second line setting, which is the group that we are studying here. No group therapy, and typical PFS is noted in a few weeks in pancreatic cancer. So in a (inaudible), you will get negative six week PFS in the second line setting. So the bar is pretty low, depressingly. We are obviously optimistic, given the experience we have so far, that we will exceed that bar, and for that, we need to accelerate the approval.

Then the conventional study, your follow-up question, we had mapped that out. It presumably will be a comparative study. Whether we do that in a frontline setting or in the second line setting, is a conversation we'd have with the agency. But you know, the confirmatory studies, you don't have to work in exactly the same context as you accelerate the approval. There is a little bit of flexibility there.

Combination doesn't seem very fruitful. But one could imagine, for example, I think an attractive approach might be to go into BRCA patients, who are in the adjuvant context, give them cytotoxic chemotherapy with platinum, for the BRCA mutants is adjuvant and then maintenance with PARP. That for example, would be an interesting strategy, but these are conversations we have not had yet.

Yes, question over here please.

Marko Kozul - Leerink Swann

Just pancreatic. I think you mentioned this Andrew, just on the percentage of patients you expect to have kind of HR deficiency or BRCAness and I mean, just as a follow-on. When is the next data that you are likely to see for the pancreatic patients?

Dr. Andrew Allen

So it's a single arm study, open label. So if you will be getting data through the course of the study, starting the first half of this year. So we all had access to (inaudible), obviously, we probably won't be sharing that, and then you move (inaudible) for the inroad.

You asked the question about the frequency of HR deficiency or BRCA in this. I don't use that term, deliberately trying not to confuse, but sometimes people refer to the HR deficiency that's not related to a BRCA mutation, its called BRCAness, it’s the same thing. I didn't mention that, because there is lack of clarity right now, as to how common that is. As you may know, we ran a pancreatic program with mandatory metastasis biopsies. We have very good clinical annotation of those cases, and so that has been a very helpful resource for some of our sequencing collaborators, who are doing work right now. If we look at the frequency, both of our somatic BRCA mutations, which is not being well described, is it’s a 5% to 10% of the estimate. But also, it's a question of, is the HRD signature I described to you in ovarian, is that positive and present in pancreatic cancer.

That is (inaudible), but obviously is a pretty robust (inaudible) there. How common it is, I don't know. We will find out. So I think, to be optimistic, if we pass the test in the BRCA mutants, we can then think about broadening the population to include, both HIV positive, pancreatic cancers. But here, we will sufficiently and certainly have decided to stick it, first of all to the low hanging fruit of the BRCA mutants. Although again, we were blending both germ-line and somatic. As you are probably aware, normal patients with severe pancreatic cancer nearly diagnosed, are seeking sequencing of their gene, because the prognosis is very bleak, if you just kind of take standard chemotherapy, it's a pretty rough road, but that's moving survival, still short of a year.

Yes sir.

Yaron Werber - Citigroup

Hi Andrew. It's Yaron Werber from Citi. So just clearly, two questions also. One, well you guys were talking about the overall response rate of 45%, can you break out the actual RECIST versus standard kind of -- standard non-progressive disease? I am trying to get the kind of apples-to-apples of the competition? I mean, secondly, when you guys look at your HRD tests, and you gained a hazard ratio of 0.53, which is not similar to what Ledermann saw on the unselected BRCA population. Help us understand kind of apples-to-apples how to think about that? Because I was hoping you wouldn't get to even a better hazard ratio.

Dr. Andrew Allen

So first question, we don't have a big enough end to reduce robust comparisons, because you didn't require measurable disease to enter our Phase 1 stage. Which is why, to give you a decent and new kind of blended stage groups. The majority of them are, at least as responsive not 125. In terms of the hazard ratios, as you say (inaudible) 0.53, and then we broke out the PTGA curves, using our HRD signature, we got, as you said, fairly similar hazard ratio. Now I would say, that’s in response of platinum, that’s a survival care with no PARP inhibitor involved, and then you got therapy, this is mostly platinum. I think we’d like to be much better than that, in patients who have received PARP inhibitor. So I don’t think that our final hazard ratio, in patients receiving Rucaparib is going to be 0.53, I think its going to be meaningfully better than that.

Cory Kasimov - JP Morgan

Thank you. Cory Kasimov, JP Morgan. How does AV filing for approval, potentially, also the regulatory outlook for Rucaparib, if at all?

Patrick Mahaffy

Hi. Just pushing through this, we can’t know for sure. We would be surprised, particularly in United States, if it achieved an approval, although it could. As you know, it’s a retrospective analysis in a non-prespecified subset populations. So this is an aggressive approach. If they do get it, its going to be limited to, I think, the germ-line BRCA population. We of course are going for the larger population of germ-line BRCA plus BRCAness, if you will. We don’t believe it will change our plans and finally I will say, based on the comparative studies, sort of an accelerated approval path, it is a full approval.

I don't think it changes things, and I think that we will remain the only drug, if approved, that has both the germ-line BRCA and the BRCAness label, which we think is a meaningful differentiator.

Dr. Andrew Allen

Great. I think we need to wrap up. Its time for an expedited serving of lunch. There is foot outside. But obviously, we are not going to be sitting outside. So please grab your plates, please bring it back. It is now 11:55 am. We are going to restart at 12:05 pm. So you have 10 minutes, to get your food. You can eat during the presentation. So all you have to do is, get it in 10 minutes. You don't have to eat it in 10 minutes. Okay. See you in 10. Thank you.

Okay. There are still a couple of people outside, but they were just too slow. Okay, so we are going to move on now to Lucitanib, which is a new acquisition. It's a potent oral inhibitor of FGF receptors 1 and 2; VEGF receptors 1-3 and PDGF receptor alpha and beta. We just had some very nice evidence of activity status in selected breast cancer patients, which we will be talking about right now.

But to setup the breast cancer conversation, I am really delighted to be able to have the privilege of introducing Professor Carlos Arteaga. Carlos is the President of AACR and Professor of Medicine and Cancer Biology, with a bunch of other titles that you can read, at the Vanderbilt-Ingram Cancer Center, Vanderbilt, in Tennessee. And Carlos is going to give you an introduction to breast cancer, HGF aberrations, particularly in breast cancer, and some historical context around therapy for selected breast cancer patients, that we will setup, since we are all talking about clinical data with Lucitanib. Thank you, Carlos.

Dr. Carlos Arteaga

Thank you very much Andrew and thank you to the colleagues at Clovis for the invitation and the opportunity to speak here. Again, I will -- these are my disclosures, and considering I will start with this slide, but this is just to remind me that breast cancer is the most common female cancer in the US, and the second most frequent cause of death. And this is important, because if for some reason, 10% of breast cancers had a phenotype or a genotype that happens to be the target of a specific drug we are discussing today, and that's going to be more than pancreatic and ovarian cancers combined, and one quarter of all the other tumor sites combined. These are very common disease, where again, because of these numbers, we can generate clinical trials quite expeditiously.

For the last several decades actually, we have used clinical markers to treat breast cancer. We used ER and PR to (inaudible) hormone, tumor hormone dependent and treating with anti-estrogen, endocrine therapy. HER2 positive tumors as called by IHC and gene amplification, and these are by FISH, and the recommendation is for them to be treated with HER2 targeted therapy. There are four targeted therapies against HER2 as we speak, and then the triple negative group, which is what is the [block] detection of any of these receptors, and those are the tumors treated with cytotoxic chemotherapy.

Molecular classification is increasingly used to inform treatment decisions. Now we use gene expression profile, like PAM50 for example, to categorize tumors like luminal A, B, HER2, and region, basal like and normal. And there is also OncotypeDx and Mammaprint with gene expression to guide adjuvant treatment decisions, mostly chemotherapy yes or chemotherapy no. But I would like to emphasize one point, none of these assays tell you about actionable targetable alterations, like the type of methods I want to cover in the next few slides.

And this is to show, again, that the OncotypeDx21 gene RT-PCR assay again, is recommended by NCCN guidelines for tumors that are more than 0.5 centimeters and they are lymph node negative. We use this in the clinic, and based on the recurrence scores of low, intermediate or high, we make decisions and recommendations to patients in terms of having or not having adjuvant chemotherapy, in addition to adjuvant endocrine therapy.

Now this is the new information; this is data from the TCGA, The Cancer Genome Atlas, which again examine whole gene, whole exome analysis, RNA-SiC RPPA, and a number of methods. It's several hundred breast cancer. This is the test. This is the clinical data that we still use actually. This shows that tumor divided by ER, PR, and HER2 status. Please note that the ER, PR tumors; again some are Luminal As and Luminal B gene expressions. These are poorly behaved, compared to the better behaved luminal As. Then you have the HER2 positives, clinically, that either can (inaudible) in the luminal group, the other half in the HER2 enriched group, and its becoming increasingly apparent that the HER2 enriched and the HER2 luminals are very different diseases. There are some somatic alterations that categorizes group compared to the others, and I think in the future will be dividing clinically, for the sake of clinical investigation, this HER2 positive groups, along these lines of Luminal and HER2-enriched.

Then here you have the basal-like group, the majority of this are triple-negative, but not all, which have a very high mutational rate for KD as shown here, but not all of them do, but the majority do. The new information and the exciting is here, a number of somatic alterations, mutations, amplifications, insertions, deletions that again the low sides, and many of these confer dependencies, identify vulnerabilities of these cancers that can be exploited therapeutically, one of them is amplification of the FGFR one, and amplification of – FGFR ligands, which I will be discussing today. So this is a new information, that is going to have a great impact in our ability, not only to do clinical investigation, but also in terms of the – these types of discussions we are going to have with patients in our offices from now on.

Again, Next-Gen Sequencing, swiftly coming to impact breast cancer therapy. Again, these are some of the platforms that we are currently using at my institution. We use a lot of FoundationOne, again, which uses a capture assay that examines with some depth, close to 250 cancer associated genes. I think, in the future, this number is going to increase; and if we fast forward five years or 10, most institutions will have hundreds of FoundationOne type assays available at their institutions, and currently at my institution, anybody that has stage-IV disease with breast cancer, that is eligible for clinical request, we request a FoundationOne assay, and this is going to basically expand in the years to come.

One example is of the application of this technology is SAFIR01 trial, is a trial led by Fabrice Andre a colleague of mine. This is basically a national project, where patients that have metastatic disease anywhere in France, are consented for a provision of a biopsy, that is subjected to hold genome or hold exome analysis; and based on the information that patients are randomized either to action, I mean, to a therapeutic approach based on the genomic information generated by the assay, or standard of care. Again, this has been very successful. In the majority of these tumors, molecular information have been generated. Again, in most cases, the problem is that in some cases, they have low similarity, which again, does not permit generation of actionable data. But, in many cases, one gets very important information, which is one example of the type of the alteration they generate, i.e., mutation, amplification of PIK3CA, amplification of FGFR, amplification of FGF4, one of the ligands, mutations of FGF part II, etcetera.

And I think as in five years, slides like this would be ridiculous, because we are going to have much more information and many more optionability, many more, again, information from this type of analysis.

This shows the prevalence of the alterations in the FGF pathway, which is again the target of Lucitanib. This shows again, that there are many chances that have amplifications of FGFR1, this is breast cancer. Also, there are other tumors that have amplifications, 11q, which is the location of the low side of FGFR3, 4 and 19, these are ligands that activate the FGF receptor. But the overlap is not complete. You can see here, there are some tumors that have amplification of 11q and the ligands, the ligands locii, but again, do not have amplification of the receptor. So again, this suggests -- again, this alteration is relatively frequent, or not uncommon in many breast cancers, amplification of the FGF receptor platform.

And this is a distribution, of course, of different subtypes, ER positive, the HER2 positive and the triple negative. The frequency of amplifications of the FGFR gene, and amplification of 11q, where the ligands exist, the ligands locii exist. Again, its about the same. But because of the frequency of ER positives, again, numerically, this is going to be an increased number of FGF altered tumors among the estrogen-receptor positive tumors.

How do we treat ER-positive breast cancers? Well, as I said before, we have endocrine therapy, estrogen-receptor targeted therapy, Tamoxifen, Fulvestrant, aromatase inhibitors, again, are the standard of care for this type of disease. The first combination approved in this setting happened last year. That was the combination with exemestane, with Everolimus, and mTOR1 inhibitor. And again, and in general, these patients now, after they progress on these treatments, they go on to chemotherapy. Chemotherapy is not very effective in this type of disease. We use it for palliation mainly, and if we can delay it, as long as we can, we are all happy and the patients are happy. Chemotherapy is not the treatment of choice of hormone-dependent disease, its positive.

And this space between endocrine therapy and chemotherapy is going to expand. Its going to expand, because we now have a number of drugs, like PI3 kinase inhibitors, HDAC inhibitor, CDK4 inhibitors, combinations of these inhibitors, that again are being tested in clinical trials, with current hormonal therapies, i.e. aromatase inhibitors of Fulvestrant. There is also development of oral ER-down regulators in this space. So I think, patients with ER positive breast cancer have a lot of options in this journey before chemo. And this is where again, the FGF receptor targeted therapies will have some opportunity for investigation.

Now the FGF receptor, the FGF signaling can endow, can facilitate tumor progression, not only by a tumor cell autonomous mechanism, this will be an amplified receptor in the service of a tumor cell, or a mutation of the ITF receptor that endows the receptor, with again, a function. That's going to endow the tumor with a tumor cell autonomous mechanisms of progression. But the over-expression of ligands is going to allow for the tumor to signal to the stroma, recruit the stroma, again, which in turn may have some other; again, a function effects on the cancer. In addition, in the stroma, FGFs are going to cooperate with PDGFs, with VEGFs and these are all ligands that are important for this angiogenic phenotype that tumor needs to generate in order to progress and metastatize.

In five year examples, from congenic mouse models that this is the retag model, the receptor for insulin promoter, P antigen model. These mice develop angiogenic, pancreatic tumors, that if they are treated with a VEGF-R inhibitor they respond. But if one wants continuous therapy, eventually they progress, because they amplify the FGF-R; and again one can treat with an EGFR inhibitor, in addition to a VEGF inhibitor, one can delay the progression of these cancers. And there are examples in the literature, again, that also the PDGF pathway, also amplifies upon the acquisition of the resistance to VEGF receptor targeted therapies, and there are examples that targeting these three stromo-ligands, FGF, PDGF and VEGF has improved the anti-angiogenic response in some of these tumors that are highly vascular.

Now, I will come back to that point in a moment. I am going to now show you some data publicly available, with two pure FGF receptor inhibitors, AZD4547 and BGJ398. These are again, high activity in-vitro against the FGF receptor 1; and again, unfortunately, the early data would suggest very limited activity or no activity of these inhibitors in patients with breast cancer, even particularly in these cohorts, that included a fair amount of tumors with FGF receptor amplification. So again, this purity of our inhibitors don't seem to have much early activity, even in FGFR amplified breast cancers.

This is longer table of VEGF receptor inhibitor, showing that again, these ones have little to no activity in breast cancer monotherapy. You can look at the timetable of over three months. This is again, very-very short. One worth highlighting is Dovitinib, which also is PDGFR and FGFR. Again, all these three receptors that I was telling you, can drive this angiogenic phenotype in some cancers. But still, with a very-very unimpressive time to progression, and there are reasons to speculate that Dovitinib may not inhibit VEGF receptor 2 as well in that. It doesn't cause a lot of hypertension, which is a pharmacodynamic biomarker of inhibition of the VEGF receptor. Again, this is a very low rate of hypertension. Again, which in the clinic, if we don't see hypertension with somebody that is getting a VEGF-receptor targeted therapy. We are a little concerned, that may be, that drug is not hitting its target as well.

Further -- and again in this space to study, in several breast cancer genotypes, zero of 22, FGF receptor amplified breast cancer, did not benefit, whatsoever from dovitinib. The majority of these were hormonal receptor positive breast cancer, consisting with the frequency across the subtypes of FGF receptor amplification that I showed you earlier.

So this is the biochemical profile of dovitinib. It inhibits at nanomolar concentrations. Again, these FGFR1, 2 and 3, FGF receptor 2 and PDGF receptor alpha and beta. Again, these are reasonably low IC50s, that suggests that cellularly, it is going to be very effective.

So this is my last slide. I think the Lucitanib tests would test -- if tested in breast cancer, would explore two key hypothesis for the first time. The first one being that, FGF amplification may not create a dominant oncogenic effect causing pathway addiction. It is, again an important oncogene, but one that they need to be inhibited, in addition to -- with another target at the same time. The other one is that, VEGF receptor inhibition alone has low utility in breast cancer. Again, I show you data in support of that. But, combined inhibition of other molecular targets may be synergistic. Therefore, two questions I'd like to pose is, is there clinical value in combining potent inhibition of both VEGFR2 and FGFR1 and 2 with Lucitanib? Is there clinical value in selected FGF-aberrant breast cancer for treatment? I think my tentative answer to both of those questions is yes, but we will hear more from the next speaker, and then you will agree with me or challenge, again, whether yes is the answer to these questions.

Thank you very much for your attention.

Dr. Lindsey Rolfe

Thanks Carlos and good afternoon everyone. I am really excited to be here and to discuss the more clinical data of our newest acquisition, Lucitanib. So, as Carlos showed you, Lucitanib is an orally available potent and selective inhibitor of VEGF receptors 1-3; PGF receptors alpha and beta, and FGF receptors 1 and 2.

I have just said it was selective, and these kinoing profiles on this slide, illustrate the point really well. Just to orient you a little bit, the red dots represent the kinases that each molecule inhibits, and the size of the dot represents the extent of the inhibition. So these experiments are done, using a panel of some 450 kinases and the drugs are administered at quite high concentrations, 100 nanomolar. So you can see, the profile of Lucitanib is really clean. The red dot is limited and concentrated in one area of the [kinome]. The other molecules that we looked at, selected pan FGF inhibitors, Dovitinib and AZD compound less selected than Lucitanib; and then on the bottom, the other molecules, Foretinib and Ponatinib are really much more promiscuous.

So this is some animal data that just explores the effect of Lucitanib -- exclusive effects of Lucitanib on the FGF pathway in a little more detail. So on the left, we have cell line data, a panel of 78 non-small cell lung cancer cell lines. On the far left of that graph, there are several lines that are highlighted with arrows. Those are cell lines that have (inaudible) in the FGF pathway, specifically amplifications of the FGF receptor 1 gene. As you would expect, Lucitanib is most active here, that's no surprise.

On the right genograph data, again in the non-small cell lung cancer line with an FGF receptor 1 amplifications, three doses of Lucitanib were tested, with activity at each dose level. At the highest dose level tested, you can just about say, it’s the green line on the bottom, and we feel good tumor regressions with Lucitanib.

This is some more non-clinical data that examines the effect of Lucitanib on the VEGF access. On the left, again, we have some cell line data, and this is a system whereby, ordinary HUVEC cells have their -- and VEGF pathway switch done by administering exogenous ligands of VEGF-A. The cells were then treated with Lucitanib, and you can see that Lucitanib switches off the signaling through VEGF receptor 2 very potently, the IC50 there, and this is in so -- its 5 nanomolar. So really very potent indeed.

One the right, small genograph data. This is an angiogenesis relevant xenograft renal cell line. You can see, Lucitanib is the green line on the bottom, again, effectively inhibits tumor growth. The comparison here is sunitinib, and that does not perform so well in this cell line.

So moving on to the clinical data; we are just completing a Phase 1 First in Man study of Lucitanib in Europe. In this study, we treated 104 patients. This study started off with some dose escalation cohorts, as you would expect, and in that part of the study, we treated dosage ranging from 5 to 30 milligrams. And after that, we did some expansion cohorts, firstly in tumor types of interest, so abnormal FGF tumors, angiogenesis sensitive tumors, and then we looked at some novel dosing schedules, with intermittent dosing regimens. I should add here, that of these 104 patients, more than half were treated at or above the recommended Phase 3 dose, which is 60 milligrams per day.

So this shows the side effect profile. Now, as you'd expect from the non-clinical data I showed you, these are dominated by the effects of VEGF inhibition. So we see hypertension, proteinuria and hypothyroidism. The effects of the adverse effect profile of VEGF inhibitors is well understood. So these effects are expected, and are very predictable, and they are manageable. In addition of the commonly adverse effect, typical of -- or the administered tyrosine-kinase inhibitors, so we have the mild to moderate GI effect and tiredness.

So moving on to the activity data; and this waterfall plot shows the experience in patients with advanced breast cancer, receiving Lucitanib hormonal therapy, and these are the patients with FGF abnormalities in their tumors. The color-coding shows which FGF abnormalities, which the blue dots represent patients with anticipation of receptor 1 and the red bars represent the patients with the 11q anticipation. So as Carlos showed you in his slide, those are the patients who are overexpressing the FGF ligand.

The activity data are really robust here. We have had 12 evaluated patients, and of those, six have achieved RECIST partial responses. And you can see here that the group that's heavily pretreated, I don't think anybody has had less than three previous lines of therapy for advanced disease, and you can also see that several had previously been exposed to other FGF inhibitors.

So, this slide tells a story of a 63 year old woman, who is still on study. She had an ER positive tumor, that also had FGF receptor 1 anticipation. When she went on study, she had had 15 previous lines of therapy for advanced disease in the previous 10 years. So you can do the simple math, and work out on average, each of her previous therapy lines have been affected for less than one year. She joined our study at the 20 milligram dose level. Subsequently, dose reduced twice. At baseline, she had really bulky plural disease. But by the end of cycle two, she had responded well, and as I mentioned, she has retained that response for more than two years now. I did the calculation this morning to be more precise, and she has actually been on study for over 800 days.

So our Phase 1 study is wrapping up. We have several key questions to answer in the near term, with our later stage development program. Firstly, we really want to understand, which patients benefit most from Lucitanib. So is it the patients who have amplification of the gene, that codes to the FGF receptor 1? Is it the patients who over-express the lie downs, FGF3, 4, 19? Or is the activity of Lucitanib in breast cancer broader than this, and not limited to those FGF subsets of patients? And our second question that needs addressing, is what is the optimal dosing regimen for Lucitanib? We know that 60 milligrams per day is a good dose, but its an active dose. But we also know that many of the patients do end up reducing that dose, somewhere along their treatment course. So, what we are going to find out is whether if we start with a lower dose, say 10 milligrams, whether we get good robust activity at that dose too, or whether the 15 milligram starting dose, really is what's necessary to drive those responses in the initial cycle?

So in our response to breast programs, you know of course, it is a partnership, a joint development program with our colleagues at Servier in Europe. Servier are sponsoring a European study to answer the question about which patient population benefits. That study, which is enrolling now, is called FINESSE, and it looks at three different cohorts of patients. As I have mentioned, these are the cohorts with the receptor 1 amplifications, with the ligand over expression or without those abnormalities.

Each cohort will be analyzed independently, so each could succeed or fail in its own right, and each has a Simon 2-stage design. So if we see two or more responses, RECIST responses in the initial 21 patients in each cohort. The study will then expand to complete the second stage of the assignment.

All patients in this study will start with the 15 milligram once-daily dose. Everyone gets [parts] during screening, will be measuring FGF data centrally using FISH, and as I mentioned, this is a Servier sponsored study, not in the US, taking place in Europe and Canada, and enrolling patients now.

So Clovis will be addressing that second question, what's the best dose of Lucitanib. We are planning a study with a slightly different design, so we are going to randomize Phase 3 study, in patients with FGF aberrant advanced breast cancer. As it's a randomized study, we are able to have a PFS endpoint, which is highly relevant in breast cancer, and will enable us to enroll a broad population of patients.

Patients will be randomized one-to-one to receive by the 15 milligrams or 10 milligrams of Lucitanib. We are going to recruit 160 women, and as I said, this is sponsored by Clovis, US-only; and back at the (inaudible) in Boulder, San Francisco and in Italy, we have a team of people working really hard to get this opened next quarter.

So looking to the future; we are already thinking about our later stage development program in breast cancer, and we will be focusing on two disease areas, both in advanced disease. Firstly, in earlier stage advanced disease. As Carlos mentioned, as the mainstay of therapy here and for ER positive patients is estrogen directed therapy, and we'll be looking to combine Lucitanib with these therapies in this disease stage. Ultimately in Phase 3, of course, we will be performing a randomized study of Lucitanib, combined with estrogen-directed therapy versus the estrogen therapy alone. In that earlier stage disease, we will have the PFS endpoint. To set us up for this, Servier are going to be sponsoring a Phase 1b study to start looking at the tolerability of the combination later this year.

Then, the second area of focus will be in more advanced disease. Lucitanib will be given here as hormonal therapy. So these will be patients who are more like those that you saw on the previous waterfall plot. Again, this will be a randomized study. We will compare Lucitanib with standard of care chemotherapy. The endpoint here could be PFS or it could be overall survival.

So, moving on to our second sponsored indication, which is the development program in non-small cell lung cancer. As you know, squamous cell non-small cell lung cancer is very often associated with abnormalities in the FGF pathways. I mean, if you see data here from an analysis of the TGCA data set showing that a large number of patients with squamous cell non-small cell lung cancer have these abnormalities. That's really exciting, because these patients don't often express the other target that has proved so actionable previously in non-small cell lung cancer. So squamous patients don't tend to have EGFR mutations or out-mutations, and so its exciting to be on the cusp of perhaps having a targeted therapy that will be effective here. And then of course, the abnormalities are also found in adenocarcinoma patients with less frequency. There isn't a robust clinical data set yet, but there are encouraging anecdotes of efficacy from clinical trials of other FGF inhibitors.

So, here is our Phase 2 study. And again, this will be a Clovis-sponsored study. This time its not limited to the US, it’s a global study. It's a single arm Phase 2 study; again, there is a Simon 2-stage design, with a maximum of 40 patients to be enrolled. We will be recruiting people with advanced squamous non-small cell lung cancer, who also have the amplification of the receptor 1 gene, and the primary endpoint is overall response rate. This study is gearing up to open imminently.

So, like other programs, you will have a companion diagnostic to go with Lucitanib. In our Phase 2 program, we are using the fluorescent FISH approach. In the background, we are working on other potential strategies, including FISH, shown on the right there, the chromogenic approach, and also, next gen sequencing. We will nail down our companion diagnostic strategy by the end of Phase 2, and validate that during our Phase 3 program.

So just brief words on intellectual property; is that it is well protected in the US. Note that we have composition of matter through 2030, as well as a broad range of patents. Servier of course controls the portfolio in that territory.

So just to finish, Lucitanib is a selected potent inhibitor of PDGF, VEGF and FGF receptor kinases. We have got robust data from our Phase 1 study. We are embarking on a sponsored program in non-small cell lung cancer and in breast cancer. We are partnered with Servier, and they cover the cost of the first $110 million of our development activities. We will have a companion diagnostic. We have strong IP and commercialization rights in US and Japan. So thanks very much.

Dr. Andrew Allen

Now some questions on the Lucitanib program.

Unidentified Analyst

On the 160 patient breast cancer study, can you just run us through the criteria for step-down of those? And you didn't talk about the doses in the lung cancer study, could you just give us those?

Dr. Lindsey Rolfe

Sure. So in all our protocols, we have very simple plans to management of those common adverse events. We have criteria set out to guide physician management, and at the moment, the dose reductions that would be 15 milligrams and for 10 milligrams which is the -- this test should be taken in Phase 1.

Unidentified Analyst

And the same doses for the lung cancer?

Dr. Lindsey Rolfe

Yes, the lung cancer, they will have the 15 milligram dose.

Unidentified Analyst

Thanks. Can you talk about investigator interest here in the US, given stage up until now, then conducted in Europe? And then second, given a lot of these patients were later aligned, can you talk about potential for the safety profiles improve in the ongoing or future studies? And then last, should we talk a little bit about the neoadjuvant setting and the interest there as well? Thanks.

Dr. Lindsey Rolfe

I will answer the safety question first. Obviously, as Andrew mentioned, Rucaparib, sometimes in very advanced disease, as you know, its difficult to tease out the events that are caused by the drugs and those that are caused by the disease. I wouldn't expect the VEGF access affects the change too much, as we move into early stage disease. But some of the more soft event frequencies may change, but we don't have any evidence around that yet.

In terms of investigator interest, Carlos will be leading our US breastwork and I think he's probably in a better place than I am to comment on investigation --

Dr. Carlos Arteaga

Sure. There is huge interest to do a trial like this. Genotype driven trials are the future. Here we have a real surviving set of alterations that are going to allow you to select patients, so I think that drives the interest of investigators like me, to a vast degree, by way of selecting patients. Talking about in the neoadjuvant space, I think this is what considering -- as you know, the FDA published an important paper last year, where basically it states that its now conceivable to do trials in the neoadjuvant setting leading to registration.

Now ER positive tumors are probably not that late to start with that approach; because pathologic complete response will be the surrogate use to call those trials positive. Again, have less significance in ER positive disease. Now that's not to say that pathologic complete response may still mean something in some select cohorts of patients, and I think it wouldn't be unreasonable to start thinking of a new adjuvant trial in patients with newly diagnosed ER positive breast cancer, selective for FGFR alteration in the receptor of the ligand of a hormonal therapy plus/minus FGFR Lucitanib, powered to see some limited number of pathologic complete responses. I think, is that a registration study? Probably not. But its going to allow you to identify the mechanisms of resistance to your combination approach, because eventually you may need to combine Lucitanib with something else. There's going to be plenty of tissue left at the completion of the treatment, that is going to allow you to do an unbiased, open-ended discovery approach of what those markers and effectors of that resistance are.

And also, you never know. You may see some pathologic complete responses with hormonal therapy alone are unheard of, and that will be a good signal that your drug has potential legs for the adjuvant setting.

Dr. Andrew Allen

Just to put a bit of a corporate overlay on that response and as Carlos said, its just lack of clarity from the FDA about the utility of neoadjuvant as a registrational track in hormone positive disease, and so that's not where we sort of go off the bat. But scientifically, as Carlos outlined, very interesting as a strategy for understanding your drug better.

Dr. Carlos Arteaga

There's a very positive reception from academic centers, to do a study like that.

Brian Klein - Stifel

Thanks. Brian Klein, Stifel. Two questions for Dr. Arteaga; first, was your contention that inhibiting VEGF is a key component of looking at the utility of FGFR inhibition? Is there any data to suggest that combining dovitinib or some of the other specific FGFR inhibitors with an Avastin, would yield the same time above responses that we see here. And then secondly, can you talk a little bit about the escape mechanisms in inhibiting FGFR and how that might impact your ability of response? Thanks.

Dr. Carlos Arteaga

Well, I have no data of dovitinib with bevacizumab. I guess in that case, I would imagine you would be blocking ligand and also the receptor for VEGF at the same time. But I don't know that that data exists. In terms of -- I think that inhibiting VEGF receptor in some settings maybe important. Again, the fact that, I can see a situation with tumors that make a lot of FGF, that's going to include stromal elements, its going to reduce angiogenesis. FGF alone cannot do it. So if FGF are reducing that phenotype, FGF is going to need another partners, like VEGF, PDGF. So in that context, I think I can intellectualize a situation where Lucitanib, hitting both EGFR and FGFR, VEGF-R, it is working better. That in a pure FGFR inhibitor.

So in terms of acquired mechanisms of resistance to FGFR receptor inhibitors, we just have to discover those. We have all the tools, we just have to do the right clinical experiment and rebiopsies to study that. But I cannot speculate -- I cannot make something up. We just don't have the data. We have the tools to study it.

Dr. Andrew Allen

Time for one more question. Yes sir.

Peter Lawson - Mizuho Securities

Peter Lawson from Mizuho. Just regarding the patients that come off or reduce the level of drug, what drives that? How many patients reduced their drug? And then what side effects are you expected to see on the VEGF side?

Dr. Lindsey Rolfe

Well at a 15 milligram dose, most patients end up dose reducing. Although, few patients discontinue because of toxicity. With your second question, what kind of --?

Peter Lawson - Mizuho Securities

[Question Inaudible].

Dr. Lindsey Rolfe

This is a late stage population. It's a whole bunch of things. Some are related to the VEGF effect, but a whole bunch are not.

Peter Lawson - Mizuho Securities

[Question Inaudible]. FGF side effects you're likely to see over time?

Dr. Andrew Allen

Well let me take that. So the well-described VEGF related toxicity is hypophosphatemia, and it notably is absent with Lucitanib and largely absent with dovitinib. So I am not sure we know anything about what Dovitinib actually inhibits in the clinic, if anything. And the mechanism is probably related to combining inhibition of FGFR1 and 4. Lucitanib does not inhibit FGFR 4. I think that's why we do not see hypophosphatemia with Lucitanib, but you do see it with pure FGFR inhibitors like AZD4547 and BGJ398. The mechanism relates to blocking renal phosphate excretion and it seems as though, you need to hit both 1 and 4 to achieve that. That was recently shown in the knockout mouse. You just knock out 1 or 2 or 3 or 4, we don't see that phenotype emerging. If you hit 1 and 4, both of which are expressed in the renal tubule, then you see hypophosphatemia in that double conditional kidney limited kidney limited knockout.

Okay, I think we should move on. Make sure we stay ahead of time. We have scheduled a brief break of four minutes, and counting. We will see you back at 12:55. Thank you.

Okay. Four minutes are up. Sorry, I am the time Nazi today. Okay. We have saved 1686 to the last presentation. You all know 1686. I am hoping today, we are going to give you a little bit of scientific depth that we don't usually have the time to share with you. So as you know, we are now going to be showing you new clinical data today, that's going to be Geneva, the European Lung Cancer Conference at the end of March. But we are going to show you some new data today, which I think you will find interesting.

To set the stage, again, it gives me enormous pleasure to introduce the principal investigator for our Phase 1/2 study with CO-1686, Lecia Sequist, who many of you've met I think. Its not the first time. You certainly met Lecia at ASCO in 2013, when she was presenting the poster of our exciting novel data at that point, with 1686, which I know many of you attended in person. Lecia is the Mary Saltonstall Endowed Chair in Oncology at MGH Cancer Center, and an Associate Professor of Medicine at Harvard and she will give you an outstanding background, I think, on all things with EGFR lung cancer, including some very new therapeutic approaches. Thank you, Lecia.

Dr. Lecia V. Sequist

Thank you, Andrew. Thank you for having me here. It’s a pleasure to let me speak with you. I am going to give you a clinical overview of where we are with EGFR mutation-positive lung cancer. And I think its important to keep in mind, that this field has changed incredibly quickly, and many of the things which are really standard practice in every oncologist's office are really quite new. So it has been an exciting time to be in lung cancer, to say the least.

Here is my disclosure slide. You are all familiar, I think every speaker has started actually today with talking about with their relative cancer and the public health burden, but of course lung cancer is far and away, the highest cause of cancer deaths in the US, and within that large bar, are several different types of lung cancer, non-small cell lung cancer is majority, and of course, the biggest subtype of non-small cell lung cancer is adenocarcinoma. So this is a subtype of lung cancer where we'd say the most advances have been made, as far as targeted therapy. This is what majority, the vast majority of EGFR mutation lung cancer, the ALK, translocated lung cancer, the ROS1 translocated lung cancer are found almost exclusively in adenocarcinoma.

And as I just alluded to, adenocarcinoma really is not an adequate description of what type of lung cancer you're treating anymore. You really need to know the driver mutation, and I hope that Carlos' prediction is correct with these types of pie charts will become passé very soon, as we start to understand that there are many layers of depths of what will determine response to different therapies, and this is just the most simplistic surface view, but it is still a huge advance, than five, definitely 10 years ago. The EGFR mutant diseases is a fairly large chunk, 13% in this particular pie chart. But it ranges depending on the population you look at. But certainly, somewhere between 10% and 20% of lung cancers in the clinic.

And the reason why EGFR mutation is important to detect early in someone's course, ideally, at the time that they are diagnosed with lung cancer, is because treating an oncogene addicted cancer, with an oncogene specific treatment gives you wonderful results, as we will go over some clinical trials that you have all seen scans like this; this is an EGFR mutation patient treated with gefitinib, and the reason why the cancer melts away in a simplistic view is because all of the major downstream pathways, that regulate growth and metastasis, are all uniquely sensitive to signaling from the oncogene addicted to receptor tyrosine kinase, in this case, EGFR. If you block that receptor tyrosine kinase, you block all the downstream signals simultaneously.

This has led to really a change in the way that we practice, and these are the most recent NCCN guidelines, and you can see that when you start off with metastatic lung cancer, the first breakpoint is looking at histology and squamous cell. Currently, we are not recommending or the NCCN is not recommending genotype testing. Although many places are doing it to find FGFR-1 amplification for example, and other markers. But if you have non-squamous lung cancer, it is definitely a recommendation by NCCN, by ASCO, by every major society that puts out recommendations for treatments. You should be testing all of these patients for EGFR mutations and ALKs, and even in NCCN guidelines, they mention during (inaudible) testing for other oncogenes as well, and that is important to do right away, because then the treatment recommendations are divergent. If you don't have either of these mutations, you move on to a chemotherapy spreadsheet. But if you do, for example, have EGFR mutation, you go on to this next spreadsheet at NCCN, where an EGFR treatment like Erlotinib or afatinib is recommended in the first line setting.

If you know about mutation prior to certain chemotherapy, and there are some caveats about, if you find out, after chemotherapy has started. But all of these recommendations have this (inaudible). You just introduce to EGFR treatments early.

Now, why do we have these recommendations? You are probably familiar with many of these studies. But I really want to highlight to you what a big practice change this was. EGFR mutations were discovered in 2004, and for the first couple of years, they were thought of as just neat little thing that you could do if you are in an academic medical center, and it was kind of -- and with (inaudible). But it was never really thought to be something that would be important for everyday community practicing oncologists until the IPASS study. Now part of the reason why, is that there some clinical characteristics that distinguish these patients, such as they were never smokers, may be they were of Asian background; and so oncologists thought they could sort of intuit when their patients want in the room, if they would have one of these mutations, and may be a good candidate for these types of treatments.

And I think what IPASS really did, especially in United States, to show people that you can't tell just by clinical characteristics of the patient, and it really does make a difference, if you guessed wrong. So IPASS was a large 1,200 person study that was done in East Asia, for reasons that we still don't understand, the EGFR mutation's activity time is more frequent than they are in western population. Everyone in the study was kind of that type of patient, that oncologists thought they could guess about. They were Asian, they were never smokers, they had adenocarcinoma; and they were randomized, the first one gefitinib, or standard chemotherapy of that day, which was carbotaxel, and the primary endpoint was progression-free survival.

So here is the intent to treat population. It doesn't look like a positive study. The curve is actually cross. But if you look on the right hand side there, the hazard ratio, it is highly positive. It actually crossed the threshold for not only non-interiority, but achieved superiority for the gefitinib. And so the reason why you have such a positive study with curves that crossed, is that there are actually two very different sub-populations within ITC population. Those that have the EGFR mutation and those that didn't.

Not everybody in the IPASS study had tissue available for genotyping. But of those that did, you can see the EGFR mutation positive patients had an even stronger benefit to the first lines of treatment, with a hazard ratio of 0.48. So what really made a difference in practice, is the EGFR mutation negative side of the slide. So here are patients who are Asian, who are never smokers, who have adenocarcinoma. But if they don't have the underlying genotype of interest, the gefitinib line does quite poorly. So after this, testing really became accepted, and now is standard, as I mentioned.

So in just the -- I guess it has been five years now since IPASS was first presented, and there are now eight randomized trials. So as I said, this field has moved very quickly.

These are all frontline trials comparing EGFR TKI to various standard chemotherapy regimens. They have been done in various countries around the world, with various TKIs, Gefitinib, Erlotinib, and the LUX-Lung 3 and 6 studies at the bottom are with Afatinib. They are compared against Taxane-based chemotherapy. They are all platinum based chemotherapy, but the other drug in the double, it has been taxanes, it has been gemcitabine and (inaudible). So really kind of all combinations of what you would look for, and it essentially all had very similar results. The response rate are two to three times to the EGFR directed therapy, compared to chemotherapy in medium progression free survival is generally longer by anywhere from three to six months, and the overall survival, by and large, has not been statistically significant. This is thought to be, because of the good effects that patients randomize the chemo, can later get from EGFR inhibitors. But not all the studies have matured yet.

So, this has been a wonderful event for patients, and as I said, a standard process, but the responses don't last forever, I will get to that in a minute, I thought that was the next slide. First, talk about the adverse events with FDA approved TKIs. So we have two now, we have Erlotinib, which has been approved for a while, but just this past year got an additional indication for frontline EGFR patients; and then Afatinib, that is initial approval for frontline EGFR patients of past year. And I would say that the side effects between these two drugs are relatively similar. Having given both drugs to a lot of patients in practice, they are relatively similar. A lot of patients will have rash and diarrhea and other EGFR wildtype side effects. Not many patients will have serious side effects, but some, especially grade 3-4 side effects you see gets up to about 15% for Afatinib, and fatigue is probably the most common grade 3 adverse event for Erlotinib. So these drugs are definitely tolerable, but definitely have their share of side effects.

And as I was about to say before, responses are relatively short-lived. It’s a huge advance, but as we step back and look at it from a perspective of a patient, perhaps a 50-year old patient who wants to walk their daughter down the aisle, giving them nine months is not, at the end of the day, long enough. So this is a set of CAT scans on the bottom of a patient of mine actually, who had a beautiful response to frontline Erlotinib, then by 16 months, had grown right back to where it had started, and this is relatively common. You could see the IPASS, EURTAC and LUX Lung-3 median PFS times for an example of what you get with Gefitinib, Erlotinib and Afatinib.

So what happens, why this resistance developed? Well, we have learned a lot about this. Initially from cell line models, but I would say more illustrative has been biopsies on patients. So what we know about resistance could be broken into two basic categories, and there are some other boutique things, such as a small cell lung cancer transformation that have been seen, that don't quite fit this model. The boutique basic things that we see are target alterations. So for example in the case of EGFR, you can have some kind of change in EGFR, RTK itself that disrupts from the sensitive situation in the center of the diagram where you have the -- all the down chain pathways being inhibited by the drug. The drug is triangle in this diagram. We can have, for example, a mutation RTK, such as T790M, where the drug no longer binds as well. You can have amplification of the RTK. In any case, target alteration is where the tumor on the downstream signals are still mostly reliant on EGFR. You can also have bypass tracks, where EGFR is being effectively inhibited by your drug still, but another detour, another RTK comes into play and starts stealing the downstream signaling, so that blocking EGFR is no longer effective enough to block cell growth.

And we and others have seen this in patients, when we have biopsies and pie chart on the left hand side of the slide is the MGH experience. We had a set of initial, about 35 patients that we very carefully compared their pre-resistance and post-resistance cancer, and more recently this year, Memorial Sloan-Kettering did a similar thing with a larger group institutions at their institutions, and the results are really quite similar across the two institutions. T790M, the resistance mutation in EGFR is by far the lion's share of what causes acquired resistance in patients anywhere from 50 to 60, some are partitioned up to 65% of patients have this target alteration as a mechanism of resistance. So this is where EGFR is still the main driver oncogene, but the RTK is altered in a way that drugs like Gefitinib or Erlotinib or probably Afatinib just no longer inhibits it. We also see bypass tracks that shows (inaudible) mutations, many amplifications, and then we have seen this transformation to small cell, which we won't talk about today.

As far as how patients do, when they acquire resistance, this is also from the Memorial experience. You can see that in general, looking at breaking it out simply as the T790 patients and the non-T790 patients. The T790M patients tend to do a little bit better. It is, in some cases, have been described as more indolent resistance. But when you do look over a larger population of patients, you see that the overall survival is not very different. Median overall survival is only 0.3 years difference, call it months.

So T790M was described actually very quickly, only about a year after the initial activating EGFR mutation, and this is a paper from Eunice Kwak of crizotinib fame, when she was a post-doc in the in the (inaudible) at MGH, showing that some of the second generation, irreversibly binding EGFR drugs, in this case was HKI-272, may be very effective in treating these patients; and (inaudible) at the time is a huge breakthrough, and this is going to change the future for patients.

What we have seen clinically has been disappointing to say the least, is that all three of these second generation drugs, Neratinib, which is now the name for HKI-272, Afatinib and Dacomitinib, all are very effective at treating T790M in the Petri dish, and all were very disappointing in patients in the clinic, with response rates of 2% and 7% reported in TKI resistant patients. So we don't have actually the perfect explanation for the discrepancy between the Petri dish and the clinic, but we think that it may have to do with tolerance of these drugs, and then ability to dose them high enough to inhibit the resistant cells. All of these drugs do have EGFR wildtype side effects, such as rash and diarrhea. So you are limited in the dose that you can give clinically, and the clinical dose may not be enough to really get around in T790.

So this was a problem without really any great solution, until about two years ago, when the Afatinib/Cetuximab data was reported. So this study was really the first one to actually show some clinical promise, in patients with resistance to Gefitinib and Erlotinib. This study was based on some elegant mouse model work done by William Pao, where he showed that the combination of the irreversible inhibitor, Afatinib, and the monoclonal antibody against EGFR, Cetuximab, should work. In the mouse models, it was actually predicted to work solely in those patients with T790M as the mechanism of resistance; and so, this Phase 1 study was started and grew to include quite a large number of patients. And a couple of things were striking from a clinical point of view; one, they did see responses. So in this case, the preclinical model did predict for the patient responses.

But two, it was not just the T790M positive patients that responded, their response was actually about equal, regardless of T790 status. The third thing that has, I would say, become evident about this treatment strategy is that, while it does work, the toxicities are difficult. This is data that has been presented most recently at ESMO 2012. We saw the waterfall plot here, the PFS curve with the median of 4.7 months. So again, this is not as durable as the first line treatment for EGFR patients. But the toxicity was definitely higher than you'd see with, say Afatinib alone. And this makes sense, because both agents alone cause quite a bit of skin toxicity, and so when you combine this together, you get even more skin toxicity.

But nevertheless, this was the first breakthrough for patients with acquired resistance to a first generation EGFR inhibitor. TO the other hint of a breakthrough that has been presented is, response to the HSP90 inhibitor, AUY922. There have been several HSP90 inhibitors that have been looked at in EGFR patients, and I would say, the results have been rather mediocre, with the exception of AUY922. Where there has been -- this was the data presented in Australia this year. There have been good number of responses to this HSP90 inhibitor.

So in the last minute, I will talk about where we are with our patients in the clinic, patients that have EGFR acquired resistance. I think one important thing that goes into your interpretation of trial results as well, is knowing that many patients are treated beyond progression. We have seen that sometimes EGFR progression can be indolent and slow and many patients will be continued, in spite of slight increases in the tumor measurements on their CAT scan. We have learned from experience that is often quite safe to continue patients on single agent Erlotinib, or to add chemotherapy into their Erlotinib, and this is just one example of a study presented by Geoff Oxnard at ASCO a couple of years ago, showing the timeline of patients treated with first line Erlotinib. The solid red bar represents how long they were on Erlotinib without radiographic progression, and then when it changes to candy stripe, that's when the patients actually got to the resist criteria progression. But you can see that the length of the candy stripe line is how long they have stayed on post progression; and then some of the circles that say S and R, talk about localized treatment options to a single area of trouble, either surgery or radiation, to an area that was growing.

So this has become quite a common strategy. One of the other things driving this strategy is this disease where concepts that was initially described by Memorial publication by Jamie Chaft, where she found that 25% of patients with acquired resistance will flare, when you take away their EGFR-TKI, and her definition of flare, was actually being hospitalized or being dead. So very significant, not just in existing symptoms, but significant change in their health status, and then may be in time from -- that they observed this was eight days. So this is something that has become very important in clinical trials, such as the CO-1686 has really minimizing that washout time period to reduce the potential flare. And the reason for this flare, is that probably we think that each cell doesn't, all of a sudden, develop resistance at the exact moment. Its an evolving process within the heterogeneous tumor. So when you start to see radiographic progression, which are probably seen at the tip of the iceberg, those are the TKI resistant cells, there are many other cells below the surface that are still sensitive to the Erlotinib over the TKI they are on. So when you remove that drug, then everything else starts to grow too, and you can get a lot more tumor growth.

So in summary, genotype-directed therapy for EGFR mutations has really revolutionized the way we take care of non-small cell lung cancer things are very different over the last 10 years in our approach to patients. Major clinical obstacles for EGFR directed therapy is setting of acquired resistance. You have two very good treatments for first line, but nothing that is really, I would say, mainstream available and ready for use for acquired resistance. Biopsies have been very important in our understanding of the biology of these situations and will continue to greatly supplement lab-based research and drug development, and of all the biopsies, especially biopsy in the lung and lung cancer patients can be hazardous, and so we do need less invasive options for accurate genotyping.

That is it. Next is Tom Harding.

Dr. Thomas Harding

Okay. I am going to give a preclinical introduction to 1686. Here on the left, you can see the 1686 molecule, its permitting and one of the important things to note is up here, is this acrylamide residue.

On the right you can see the ATP binding pocket of a mutant EGFR molecule, and in this is (inaudible) is 1686 molecule here, and notably, you see the formation of a covalent bond between this acrylamide and also this Cys797 residue. So it forms a irresistible covalent bond in the ATP pocket of EGFR.

If we talk about selectivity of this molecule, we can look at the whole kinome, and we can say, how many -- 434 kinases within the kinome that 1686 actually hits; and to look at this, we have taken the color coding of red being potent, yellow being less potent and green being less so. And you can see immediately, the most predominant potent here that 1686 inhibits is EGFR/T790M. That also includes the frontline mutation, such as deletion 19 or l858R.

In addition, we do see inhibition of some other kinases, such as JAK3 or HER4, in comparison to T790M is far reduced. We can look at this question in another way, and say which kinases have a cysteine in the same position to covalently bond with the 1686 molecule; and in addition to EGFR, there are nine additional kinases, which have a cysteine in the same position. What we find is that, again, mutant T790M is bound with the highest affinity, and in about (inaudible) affinity, we see EGFR. Then down from EGFR, we see other members, such as JAK3 or HER4, which have a cysteine position, which are inhibited, but to a lesser extent. So the molecule is selective for T790M, or mutant EGFR, be it the del19 or (inaudible) mutation.

If we move this along the preclinical conveyor belt, you then say what is the activity in cells? So in order to look at this, we can look at the amount of drugs required to inhibit cell proliferation 50%. This is the GI50. The lower this number, the less drug is required to actually inhibit that cell proliferation. You can see, for a panel of four different lung cancer cell lines with EGFR mutations, a drug required to inhibit that proliferation is very low. For example, here is two cell lines with a deletion 19 mutation, which require somewhere in the region, about 10 nanomolar of drug to inhibit them. The same is true if you take a cell line with l858R and a resistant T790M mutation, somewhere in the region of about 10 to 20 nanomolar of drug is required to inhibit cell proliferation.

Conversely, if you have wildtype EGFR, consistent with the kinase profile, the amount of drug required to inhibit proliferation is far more. For example, somewhere in the region of 1.5 to 3 logs more drug is required to inhibit that cell proliferation.

Moving along again, you can then take cell line derived xenograft. So this is where we are taking cell line. We are planting it into the back of the mouse. We are allowing the tumor to grow, and then one such tumor is preestablished and growing, we can then treat those mice with different doses of 1686 and observe the impact on tumor growth.

What we find in here is a frontline model, with Deletion 19, 1686 of 10, 30 or 100 mgs per kg. These are dose responsive reduction to chemo growth. In comparison, Afatinib and Erlotinib have a comparable level of tumor aggression compared to 1686. The important thing being though, Afatinib and Erlotinib in these mice, due to the wild type EGFR inhibition, [grows at] a 15% reduction in body weight. That is not seen with 1686, because its bearing the wild type EGFR.

If we then look at second line model, where the T790M resistance mutation is developed. We see the same results with 1686, a dose responsive reduction tumor growth. However if you looked at Erlotinib in this case, as you'd expect, the (inaudible) present in the T790M mutation, means that they are no longer inhibited by this drug and it basically has the same vehicle in terms of tumor growth rate. Afatinib does have some effect in this model. However, it's specifically inferior compared to 1686.

We can then take a patient derived xenograft model, this is where the tumor is being taken directly out of patients and placed into the back of a mouse, which is thought in some cases to be more translational irrelevant in the cell line implantation. This is a PDX carrying l858R and the T790M mutation, and again, 1686 causes regression. Erlotinib has no impact compared to the vehicle.

So here is some unpublished data. Of course -- as well as working in the T790M population, we are interested, whether or not we could move 1686 into a frontline setting. We know that the majority of somewhere in between 50% to 60% of patients will develop T790M resistance mutations. If we treated a patient upfront with 1686 and we delayed the emergence of T790M, all these T790M resistance disease using that drug, would that allow us to extend the PFS of that patient?

So here we have taken a PC9 model, just like the deletion 19, it represents one of these first line models. What we have done is, we have then taken that tumor, this is an ongoing experiment, with either vehicle, 1686 or Erlotinib. And what we find for the first 25 days, is that the tumor growth inhibitions between Erlotinib and 1686, are completely comparable. However, once you get to about day 30, you start to see the emergence of resistant tumors in the Erlotinib group, and you can actually quantify the difference between the Erlotinib group here and 1686, which maintains inhibition. But at the day 40 level, here you have Erlotinib and here you have 1686, and you can see these are maintained at this much smaller tumor volume, which is statistically significant compared to the Erlotinib treated groups.

Here, we are treating mice 150 mgs per kg. This is comparable to the PK data that we have for the hydrobromide [cell] of 500 mg BID. So we are using a relevant dose here to treat these animals.

This is ongoing. What we hope to see in the future is, whether or not this resistance here to Erlotinib is T790M mediated; and B, if we switch these mice over to a 1686, whether or not we see an ongoing response.

In relation to wild type EGFR inhibition, we looked at the gold standard molecule for looking at EGFR signaling in an unmutated form, which is the A431 model. This has wild type EGFR. The growth of this tumor is dependent on wild type EGFR signaling, and the growth of this is used as the standard to assess the real wild type inhibition of a molecule.

For our vehicle, tumors derived from the cell line in mice grow relatively quickly, over 20 days. 1686 does reduce tumor volume to about a region of about 36%, comparing this value here to here. However, if we treat those with Erlotinib or afatinib, you can see that that greatly increases to about 95% to 100%. If you also look at other third generation molecules, such as the AZD9291 molecule, this molecule, as reported by AstraZeneca from the [triple] meeting does inhibit wild type EGFR at a low dose of 5 mgs per kg, a region about 80%, which will be [somewhere] in the region around here on this graph.

So it does seem from a preclinical data, consistent with the preclinical development that this 1686 is wild type EGFR sparing. So one of my jobs in translational medicine, is to try and look out to see what kind of things would mediate resistance to 1686? One way in which you can do this, is you can look across the literature and say from what's published, what could we predict mechanisms that's resistant to our molecule may be. And the reason we are interested in this is we are thinking about how can we extend the period of treatment that patients would be on our drugs, and how could we increase the proportion of patients that may respond.

One way, as I think Lecia indicated, was the activation of alternative receptor tyrosine kinases. One of these is the activation of the MET receptor, which is the receptor for hepatocyte growth factor. Another way of doing this, may be impairing apoptotic response by the radiation of proteins such as BIM. You can even think of a direct modification of a cysteine residue at 797, so the 1686 molecule can no longer bind to (inaudible) that site.

You can even think about downstream signal activation as being shown, this is AKT with NFkB. You may even have complete transformation of the phenotype of the cell, by transformations with small cell phenotype or even in epithelial mesenchymal transition. I must indicate, we have a preclinical mechanism. None of these have actually been observed with 1686 in-patients, and so this is something that we will have to look, as our clinical data matures.

As we published, with in-vitro, we can generate resistance to 1686. If we take cell lines grown on plastics, and we expose them to drugs for a month's time; and we did this for one of our cell types, which is this 1975 model at the T790M mutation. We basically made these resistant to 1686, we call these CO-1686 resistant or core cells; and what we find is that, these cells have hundredfold increased in the amount required to inhibit proliferation to 1686.

These cell lines, which are resistant to 1686, are also resistant to Afatinib and Erlotinib as you'd expect, and what was particularly interesting, when we saw this resistance, was the further resistance of the EGFR access was not observed in the cell lines. What we are seeing in these cell lines, is it appears to be an epithelial mesenchymal transition of the cells from a completely epithelial to mesenchymal phenotype.

So this means that, in the case of 1686, it appears to be fully turned off the EGFR access, and we are not seeing further mutations, as with further T790M or even as predicted cysteine 797F mutations. Instead, we are seeing epithelial mesenchymal transition, and also, as we published in Cancer Discovery, we also saw activation of the AKT pathway. What this means when we look at resistance is, we can use an AKT inhibitor, such as MK-2206 or GDC-0068 to restore sensitivity to our molecule. Meaning that, if you take an AKT inhibitor and 1686 in combination, you can reverse, basically the resistance of the 1686 by the combination of two drugs together.

Another way in which resistance may be mediated, is that may be the activation of the MET receptor through hepatocyte growth factor. Preclinical studies from (inaudible) lab indicated that when you take hepatocyte growth factor and add it to a cell in culture, this mediates resistance to a broad range of TKI inhibited in the clinics. We repeated this data from Dr. [Kettleman] using our system, where we took again a 1975 cell line with a T790M mutation, or the HCC827 del 19 cell line; and what we find is that, when you add hepatocyte growth factor, comparing here is the NIE cell, here is the cell line with hepatocyte growth factor added; we see, that the amount of drug required is dramatically increase, i.e., they become resistant to our drug.

The good think about the net for hepatocyte growth factor signaling pathway, is there is a drug already approved, that does inhibit MET. This is Crizotinib that is approved for using ALK or in MET positive disease; and we can see, that as a single agent, it does nothing. However, if you use both 1686 and Crizotinib in combination, you restore the sensitivity, even in the presence of hepatocyte growth factor.

The same is true in another cell type, this is the HCC827, which is again lung, but has a deletion 19. So even though we see resistance in the case of the MET, there is at least a clinical path forward, where we could think of a combination treatment, where we could use both these drugs together.

While this is particularly applicable to 1686, given the fact that the drug is wild type sparing, it means your ability to do combination trial is greatly increased. Here is the paper that came out at the end of last year, and in this paper, we looked at the combination of Crizotinib with two different EGFR inhibitors. One was Afatinib and one WZ4002, which is a molecule closely related to 1686, made by gatekeeper. The interesting thing that came out of this study, was that when you use Afatinib and Crizotinib in combination, you see a rapid decrease in the bodyweight of the animal; and this was due to intestinal damage in the small and large intestines. However, if you use WZ4002 to inhibit EGFR, this with wild type sparing and Crizotinib, there was no impact in bodyweight, and there was not the same degree of intestinal damage. But it seems to indicate, that if use our molecule, which is wild type sparing, and our ability to use combinations with molecules like Crizotinib may be greatly improved.

We can actually test the idea of using a combination of Crizotinib and 1686, using a patient-derived xenograft model. Here is a (inaudible) patient-derived xenograft, and we can see for two of these xenografts, two of them have a large amplification of the MET gene. So this the MET gene copy number. Here are the individual PDXs and two of them stand out, having about 15 copies of the MET gene. One of these, which is the 858 model here, also has an EGFR mutation. So not only is it mutated for EGFR, also have MET amplification at the same time.

If we place it into a mouse model, we can see that 1686 by itself has no impact when MET is amplified with l858R over Crizotinib, does have an impact. However using both of the agents in combination, causes tumor regression. There is no significant bodyweight difference between the vehicle and the combination group, consistent with the publication that came out previously, at the end of the last year, and indicates that a small number of patients, which are probably in the low single digit, would still have an amplification and l858R, which you do see, but are relatively rare. You can at least think about how to expand the use of 1686 by using a combination treatment in this population.

Another way that you can do these kind of resistance studies, is to use an unbiased approach. We do drug-drug combination screens; and so here we have taken 1686, and we have screened it against 300 compounds, which are currently in Phase 1, Phase 2 or approved drugs.

What we then do, is take a selection of 1686 resistant cell lines, which here are called our core 1686 resistant cells here. Here you have the wild type 1686 sensitive cell line; and what you can see is, for the Pc-9 wild type, its sensitive to our drug. However, when we (inaudible) in-vitro with that drug, you can pull out these resistant clones, which are resistant to 1686, above the micromolar level.

What we then found as a single agent, no particular -- in this case, we are looking at aurora kinase. No particular aurora kinase inhibitor was potent as a single agent. However, if we took our resistant clone and we used 1686 plus any one of five aurora kinase inhibitors, which are now paneled, from the Millennium which is far along in the clinical trials, or other ones, which is more research tool based. We see the combination for both 1686 and aurora kinase inhibitor, result in the resensitization of the cell, 1686.

This provides you a completely novel combination. If you think about in the patients, should they become resistant to 1686? You could think about a combination trial between the two.

Over to Dr. Allen to give (inaudible).

Dr. Andrew Allen

Thank you, Tom. Very clear. For those of you watching on the webcast, you have a slightly older version of the presentation that you are going for. That will be corrected and the webcast version on our website will be using this presentation, in about two or three hours from now. So apologies for those watching on webcast. These slides will be available as a PDF on our website at the end of this meeting. Obviously, the funky animation will be sadly absent, but you will probably get over it.

Okay. So let's move to the clinical. I am going to move quickly. I want to leave some time for Q&A. So I am just going swiftly over data that many of you have seen before, because its all in public domain. This was our Phase 1 dose escalation study, that we most recently presented simply at the World Lung Meeting. This is the dose escalation, using the freebase formulation of our drug. Tumor biopsy at screening to get central EGFR genotyping on all patients. Tumor assessed by RECIST 1.1 and we went up to 900 milligrams BID.

The toxicity we presented previously, there was a little bit of nausea and diarrhea and fatigue, no rash. This is a waterfall chart, an earlier version was shown at ASCO, and this is the version we showed in Sydney, and you can see several things. First of all, of the nine evaluable patients treated at the 900 milligram BID dose level of the freebase formulation of drug -- nine evaluated patients, six of those had partial responses.

Importantly, all of these patients that had previous EGFR-TKI and eight of the nine with the astrocytes, have come off of their TKI, immediately before beginning 1686; and this is important, because of the phenomena of re-treating effect, which is observed with TKIs. If you treat somebody coming up with chemo, you can get a second response to re-treating with TKI. There was this nice slide in our World Lung Presentation that explains that. You can see the duration of the responses of the weeks on treatment here. This is obviously from Sydney, and we will be updating this of course, in Geneva.

Now as you also know, we have moved over to a hydrobromide formulation of the drug, and here is why we did that primarily. TKIs classically are rather insoluble molecules, and often are really very soluble in highly acidic environments. And so 1686 fell pretty classically into that group, and here you could see the solubility of the freebase in blue, which is pretty good at very low pH, but obviously falls off fairly rapidly; and of course, in cancer patients, many of those who take TKIs or H2 antagonists for a variety of reasons. Drastic pH, is going to be up at five or six, and so you are not getting terribly to solubility of the freebase, and obviously, this will limit absorption.

The hydrobromide salt was identified in a screen. We did various salts as being very different in terms of its pH dependent solubility, and you can see this dramatic shift here of one to 1.5 logs, that led to a desired transition over to the hydrobromide, in the hope of seeing better absorption and higher PK exposures for a particular given dose.

We have dosed four patient cohorts. This slide is out of our JP Morgan presentation, and so I won't go through this in great detail. We have seen nice dose proportional PK with the higher stages that we wanted. Its better tolerated, with very little of that diarrhea that we saw a bit of with the freebase, and hyperglycemia was the dose limiting toxicity, dose-related, usually asymptomatic and obviously typically managed by dose reduction in some cases, or hypoglycemic therapy. We have seen no rash.

Here is a new slide showing just some PK data from the hydrobromide at the four different dose levels that we had treated, and comparing it with the freebase. So we are looking here at the AUC, and we have got the 500 milligram, 625, 750, and 1000; and you can see essentially dose linear, dose proportional pharmacokinetics. So as we go from a dose of 500 to a dose of 1000, we are roughly doubling the exposure from 20,000 to 40,000, hence dose linear. Otherwise, it just is exactly the kind of behavior that you want to see with an oral anti-cancer therapeutic, with predictable absorption that the patients are getting good exposures; and one of the challenges with freebase is, we just had these patients who are clustered down at the bottom here.

We transitioned patents completely from the freebase over to the hydrobromide, and here are a set of those patients, these are all of the patients that we've transitioned; and you can see the -- on the left, we got the Cmax from the 900 freebase on the BID, going over to the 500 milligram hydrobromide, and you can see that patients who are at the lower end, went up, and those patients with decent exposures seemed to stay pretty much flat.

So, we have begun planning for the TIGER program, which will begin in the first half of this year. We are about to set the dose and then we will be ready to submit the final protocols, which are all written, and all that's missing from them is the does. So that will drop in very soon, and then we will submit the protocols and begin enrolment into TIGER1, TIGER2, and TIGER3.

Pat described all of these at JP Morgan, and I am going to skip to the next slide to kind of outline them, in a perhaps more useful context. Again, a typical treatment paradigm.

So we sure outline these days, if you have looked at EGFR lung cancer guide notes, you will receive a TKI typically as frontline. Sadly, progression will happen, and either you are going to switch over to a chemo doublet, so of course typically these patients have been thinking of therapy before, and we may increasingly, continue on TKI. When progression occurs on the doublet, you may re-treat the TKI, which does elicit responses, as I mentioned earlier; and this is why its important to know whether the 1686 are we treating here or are we treating here? Because, retreatment effects happen, and have nothing to do with T790M.

Once you progress the second line chemo, if you're not going to re-treat, you go to third line single agent, and then you continue the experimental therapy, or perhaps a re-treatment of this later stage.

Now the TIGER1 study is a frontline study, this is Erlotinib; and I don't want to describe each of these studies in a little more detail. But we tried to number them in a very, easy to remember fashion, TIGER1 first line. TIGER1 is a clean second-line study in T790M positive patients, after their initial TKI, that's TIGER2. TIGER3 is a slightly more mixed population after one or more lines of chemo and TKI, and so obviously we are providing a home for patients, who are further down this treatment cascade.

Now those are the three priority studies for us in the first half of this year. In the second half of the year, we will begin TIGER4, which is a study in patients analogous to the TIGER3 population, but rather than requiring biopsies, we will actually be assessing the T790M using plasma alone, and I will come on to the potential attractiveness of the plasma testing.

Then finally, TIGER5 will be our randomized confirmatory later line study, against probably single agent chemotherapy as our post-approval commitment study.

So TIGER1, a Phase 2/3 global study in frontline EGFR mutant on small cell. Beginning with the Phase 2 portion, about 150 patients, and then continuously and seamlessly enrolling and then sizing the Phase 3 population, based upon the Phase 2 results. So there is kind of a rolling continuous enrolment, but we reserve the right to determine Phase 3 size, based on the observed data; because today, we just don't know what that hazard ratio will look like. One-to-one randomization against Erlotinib, primary endpoint of PFS; and we will try to use efficacy biomarkers to accelerate a similar start with Phase 3 if we can. So let me explain a little, what we mean by that. How can we do this?

There are a couple of approaches that may help us in the frontline setting. The first approach is that, it is conceivable that you can identify treatment no wwith patients, who are more likely to progress on Erlotinib, let's say, through the T790M mechanism. And if we can identify those patients a priori, of course, they are likely to increase our target hazard ratio, thus reducing trial size and increasing speed. So the key question obviously is, can you identify what's oftentimes de novo T790M positive patients? i.e., patients who have never been treated, newly diagnosed, as you can find the active rate of mutation and you can find T790M, and I will talk about this in some detail in a second. Can we find them, and how common are they.

The second way to be thinking about this, is to identify evidence of efficacy superiority in 1686 over Erlotinib, prior to getting maturity of such data; and that may be achievable, using surrogate markets for progression. And the two -- I will talk about these, there will be tumor growth rate, and circulate the tumor DNA configuration.

So (inaudible) de novo T790M. the biggest opponent of this, is Rafael Rosell out of Barcelona; and Rafael has a diagnostic entity within this organization that develops a proprietary TaqMan PNA assay, peptide nucleic acid; and using that assay, they have analyzed a set of tissues, not the complete set, but as many tissues as they had I think, from the EURTAC Erlotinib frontline study. And interestingly, in the group of patients who are randomized to chemotherapy. So this is a mutant EGFR, lung cancer study, randomized with chemo or Erlotinib in the frontline.

And in patients receiving chemotherapy, it didn't matter whether the TaqMan PNA assay was positive or negative for T790M, the overall PFS was the same. But if you received Erlotinib, there was an apparent difference between PFS in those patients who received -- they had T790M, present the baseline or absent the baseline; and basically, if you had T790M present at the beginning, you did worse on Erlotinib. Obviously the assumption will be, that you'd progress with T790M acquired resistance faster, because of that preexisting clone. So that's the tantalizing data that Rafael has presented in one public forum, in his significant publication I believe. But there is much debate about the veracity of this data, and it all comes down to assay sensitivity and specificity, and so here we collated a lot of publications in the literature, using a variety of techniques, and I won't -- you can read what they are here. The highest fraction of T790M at baseline is found using a very unusual colony hybridization assay.

This is the DXS approach, which is the diagnostic company in Manchester in UK, that was acquired by Qiagen, and they have an amplification system called ARMS, which did have some meaningful positive issues. They generate to a figure of 38%. More recent data using snapshots of Lecia's institution or next-gen sequencing, is to get to much lower rates of de novo T790M.

NGS is not the most sensitive tool, digital PCR is probably the most sensitive tool we have, and so this is still a live issue, and I don't know what the answer is, and we are not -- we don't know enough to commit to this, but its clearly something we are watching very closely, because it could be very relevant to us, and our drug development.

Now, how can we predict whether our drug is better than Erlotinib, before we have mature PFS data? Some very interesting data have come out of Dana-Farber looking at tumor growth rate, after the Nadir, so the patients tumor size on the scans. They started on Erlotinib, they do very well tumor shrinks that come down to the Nadir, and then at some point, the tumor starts growing again. And if you look at the tumor volume over time, post-Nadir, you get this very linear plot, this is a semi-log plot here, and you can see the straight lines of patients, some have been progressing very slowly, some have been progressing very quickly. And it was highly reproducible for individual patients, and the observation was, that if the growth rate was greater than 0.15 per month, you would progress clinically and on CT scan, within two months.

So this may be a useful target endpoint. This one I am showing specific to EGFR mutant lung cancer, which is what I am showing. There is quite a lot of data now, looking at a variety of solid tumors, and this concept of tumor growth rate, as a predictor of progression. So this may be very helpful to us, in terms of being able to get competent early, our drug is better than erlotinib.

The other approach is to look for the acquired drug resistant mutation emerging in the plasma. This was first done, a couple of years ago from a group at Hopkins. This is a patient who had colorectal cancer, analyze KRAS wild type, as you'd expect them to see Cetuximab. But over time, on the study, after about 20 weeks, mutant KRAS alleles, (inaudible) actually, were detectable in the patient's plasma and he became clinically resistant to Cetuximab.

We have started similar observations from some of the patients treated in early cohorts of our study. But obviously, the efficacy was muted, because the doses were inappropriately low -- well, appropriate for Phase 1, but not therapeutic, as we say. And you can see, here are the patients who had a very small shrinkage of tumor lesions, and a reduction in there, circulating plasma allele frequency for the T790M and del 19 mutations, who then clearly progressed and clinically progressed about a month to a month and a half later. Now, we are collecting a lot of these data from all of the patients in our studies. I don't yet know, whether the kind of delta between serologic relapse and radiological clinical relapse is going to be big enough to be meaningful, but obviously this is an area of get interest to us as well.

Okay, so that's TIGER1. TIGER2, very straightforward study, I will move very quickly through this, as we have discussed this I think at length, I will also skip over this slide.

TIGER3, we have made a small, actually a meaningful modification to the thinking that we presented to you two weeks ago. This protocol is just being finalized, and after much discussion with our physicians, we have elected to do this study, and the later stage patients, as I mentioned, requiring TKI failure prior to chemotherapy.

They are going to do mandatory biopsy, but actually, we are not going to only allow T790M positive patients into the study, we will also allow T790M negative patients, and those whose biopsy is inconclusive.

There is enough evidence that, particularly with re-treatment effect and potentially, tumor heterogeneity leading to false negatives on the biopsy, such that we may be missing patients, if we rely upon the T790M positive biopsy result on tissue, to get into the study. So we have opened up the enrolment, and to compensate, we have increased the end to 200. We will have a floor of about 100 T790M positive patients in this study. The TIGER2 and TIGER3, both have response rate as the primary endpoint, with duration of response, obviously important, and both are intended to accelerating approval in the United States, and conversations in Europe and in Japan.

Now as you know, we are working with Qiagen, to develop the allele-specific PTR test for T790M, but we need in the United States, to get approval in a T790M positive population. This assay -- Qiagen has actually been TNA approved, it was the companion for the Afatinib approval in frontline; and importantly, analytic validation of T790M was completed in that application. So that work has been done. We do not have to do analytic validation of T790M in this assay, all we have to do, is clinical validation, i.e., show that patients who are T790M positive using this assay, respond to drug.

Now, we are getting interested in platinum-based testing, because it obviously is attractive to avoid the need for needle-based lung biopsy. So, we have been looking pretty extensively at the performance of plasma assays versus tissue assays in samples that we have match pairs from our various clinical trials. And if we look at the activated mutations, so it’s a del 19 or l858R, you can see that the sensitivity of the assay is 78%, with essentially 100% specificity. There are no false positives on plasma. So if the plasma reads out, its real.

T790M is a rarer beast, there are fewer of the mutations in plasma than the activated mutations. The sensitivity is a little bit lower, as you can see here, and you have tehse two patients here.

So first of all, plasma negatives, tissue positives. Some of those I think are false plasma negatives. So I do think that the plasma test doesn't pick up everybody who has T790M disease. These two however, are not false positives, which is important. We sequenced the plasma of these patients, and we found T790M. So these are real; and the issue, probably one of tumor heterogeneity. You list the T790M positive lesion on the biopsy, so the tissue test comes up as negative, then you take it up in plasma.

Now the team did a nice deeper dive, into the patients and observed that, if you looked at patients who had metastases outside the thoracic activity, (inaudible), then the percent positive agreement increased to 85% from the basic 65%. So what I am saying is, if you just look here in all patient tissues to get an idea, and the percent positive agreement, 65%. But if you select the patients with diffused metastatic disease, it goes up to nearly 90%, and for patients with purely interpulmonary, inter-thoracic disease, that's where the sensitivity drops off, which obviously may make some intuitive sense. So we may be able to select patients, who are good candidates for plasma testing, avoiding the need for biopsy. And we have been doing some extensive work now with the BEAMing folks you have probably heard about, who have a very nice quantitative mutant allele assay that works on plasma. This is a Hopkins spinout group Inostics, recently acquired by the Japanese diagnostic company, Sysmex.

So our registrational strategy is to seek tolerated approval in the US, potentially followed by conditional approvals in the EU and Japan, using the TIGER2 and TIGER3 studies, with the TIGER5 confirmatory study ongoing during the review process, and getting that trial looking for frontline approval, which is TIGER1.

We are obviously interested in combinations, Tom outlined very nicely, some of the non-clinical work that we have been doing. We are working, in particular, with the team at Mass. General to get biopsies from all patients upon progression at the start seeing whether we can find obvious combinations that might be therapeutically useful, and I won't go through them, but you can imagine, the obvious list here and we present a little of data on C-MET and aurora kinase.

The patent protection of this molecule is very good, with composition to release 2031.

So I will stop there, you know the summary, and I will leave five to 10 minutes for questions. Okay. Thank you. So let's see who gets the mike first. Let's give it to Marko, shall we, just for a change. Leerink wins.

Marko Kozul - Leerink Swann

I love the sound of that, thanks. Based on your understanding of 1686 escape mechanisms and in terms of subsequent therapy, would it make sense to immediately recharge patients with 1686 combination or potentially give a TKI holiday for good and then recharge?

Dr. Andrew Allen

That's an interesting question, I think you can make arguments both ways. As Lecia said, I think the argument for keeping the patient on a TKI is getting ever stronger. If they got T790M disease, probably setting back to Erlotinib, and allowing T790M to drive forward, doesn't make a ton of sense. Some patients, based on our plasma work, seem to have a lot of T790M upon their progression. Some patients don't. So there may be heterogeneity there. So I guess though, the time will start try and do the right thing. If you go T790M, you should probably stay on 1686 and bring in the appropriate combination, if you can find something. If you cannot find something, then maybe chemo is a better option at least at this stage. The flipside is, if T790M is gone away, maybe it makes sense to cycle it back to Erlotinib perhaps, and then you've still got T790M therapy available, when that T790M comes back. Lecia, anything you want to add -- commentary to that?

Dr. Lecia V. Sequist

Yeah I think, a lot of these things are also practically, what's available to your patient at that moment in time. So we don't really have the luxury of making all those intricate decisions.

Marko Kozul - Leerink Swann

And so, just a quick follow-up, is it possible to maybe explore this as an amendment to the TIGER1, with patients there?

Dr. Andrew Allen

I think that would make it a very complicated study. You have to kind of break the stuff apart and do piecemeal sensible study. I guess, somewhere down the road, we'd end up doing, effectively a master protocol in patients progressing on 1686. So hopefully, we will have established collaborations with people who have investigational events, who have the right combinations, based on what we have seen as the drivers of resistance, and may be some of those will be available, with a drug like Crizotinib, that you can just buy. So eventually, that would probably be the kind of study that will be convincing, I think the physician's mission. Thanks Marko. Thank you for your patience.

Unidentified Analyst

Thank you, Marko, for leaving me with the easy questions. When will you give us the hydrobromide dosage, and secondly, when you have gone to the FDA, we do a very novel, for a loss of a better word, adapt a strategy with TIGER, or both regulatory authorities. Anything on those conversations would be appreciated.

Dr. Andrew Allen

So we will present clinical data at Geneva, so that's when we will talk about the dose, and we will show obviously updated efficacy data, as well as the PK and the tolerability. So that's the next clinical data set. The TIGER1 design is not that novel. I mean, Phase 3 is Phase 3. All that's changing is, we are retaining a little bit of flexibility to size Phase 3, based on the observed Phase 2. So as long as we're blinded, as soon as that veil of secrecy comes down at the end of the Phase 2 component, we've set it at 150 patients. That veil comes down, we don't have to be blinded, and then we have got the right to resize the study based on the observed Phase 2 data. But, it's the end with that development. There's nothing adaptive within the Phase 3 itself, so I don't think this presents any meaningful regulatory challenge. Jill, do you feel differently? Go on.

Unidentified Analyst

[Question Inaudible].

Dr. Andrew Allen

TIGER2 and TIGER3? Yes, seeking accelerated approval based on response rates, and duration of response.

Unidentified Analyst

[Question Inaudible].

Dr. Andrew Allen

Pat, do you want to speak to the timing of data on TIGER2 and TIGER3?

Unidentified Analyst

[Question Inaudible]. TIGER2 and TIGER3 data, two potential filing dates. Thank you.

Patrick Mahaffy

Always relates to speed of enrolment, all the factors that you know. I would imagine, that we will be fully enrolled in both TIGER2 and TIGER3. If not at the end of this year, then early next year. We had data in the middle of third quarter of 2015, and Jillian has assured me, we are going to file in 2015. Good news for me there. We think what we know right now, and again, we don't know everything. That will be on file by the end of 2015.

Unidentified Analyst

Andrew, thanks. On the amended protocol for TIGER3, what led to such a quick change in thinking on doing this, and can you remind us, how many patients you were originally going to enroll, and out of the 200 you plan to enroll now, is there still a minimum number of T790M positive that you want to bring in?

Dr. Andrew Allen

It’s a complicated answer. We were going to enroll 125 patients in TIGER3, when that was going to be biopsy proven T790M positive only. We are now enrolling 200 patients, with a minimum of 100, who are T790M positive. So that's a numeric answer. The change in thinking was, obviously, we are moving very quickly here, and we put together our synopsis based on the observed data to-date. Then we get continuous data update, subsequent to our trials, and then we share our protocols and thoughts with our positions, to give us their, sort of feedback. And there is a strong sense that, we may be missing patients, who would benefit from the drug, if we only allow T790M positive patients into TIGER3; and the arguments were twofold. First of all, that there is a re-treatment effect. There is value in just giving drug to patients who come up with chemo, and who may or may not be T790M positive at that point. We don't really know about -- when you start T790M positive and then get chemo, how much dilution with that original T790M clone happens, whether you lose positivity in the assay, which depends on the sensitivity of the assay. So that was part of the complexity.

The other part was this evidence that we can generate in showing that you can get false negative tissue results in patients, particularly here, sort of further down the track, because there is some degree of heterogeneity of the genetics of the tumors. Physicians try the biopsy of the growing lesions. Not always possible, they are not always accessible, so sometimes they just biopsy what they can, and so you can miss patients who are T790M positive in plasma and in truth; because you biopsy the negative lesion. So those were the two key drivers of the decision.

Unidentified Analyst

Can you talk about, what work remains to be done for --?

Patrick Mahaffy

They were supposed to pay me. I wasn't got to click through unless I got bribed but nobody stepped up. Sorry.

Unidentified Analyst

Can you talk about what work remains to be done for selecting the dose? You showed a PK slide, with exposure levels, but we are getting the dose at the end of March?

Dr. Andrew Allen

Yeah, its not worked. We will be selecting the dose before the end of March. We will be telling you about it at the end of March. What we need is, just to gather a little more data from the patients, who have been enrolled at those doses; because obviously you are looking for toxicity, you are looking for efficacy, you are looking for PK, the trade-offs inherent in that, and we will be able to explain all that to you in Geneva.

Okay, should we make it last question, we are just past top of the hour.

Unidentified Analyst

[Question Inaudible].

Dr. Andrew Allen

Typically its two or three weeks ahead of the data presentation. Obviously, we will put that on the slide that we presented, to be variable depending upon the data sweeps. Okay. Pat, do you want to close? Great, well thank you for your attention, great questions. Hope that this day was useful and enjoyable for you. It’s a pleasure to have the opportunity to present some of the great science that our team has been doing. And again, thank you to our external physicians for joining us and giving greater clarity. Thanks to all. Safe trips home.

Copyright policy: All transcripts on this site are the copyright of Seeking Alpha. However, we view them as an important resource for bloggers and journalists, and are excited to contribute to the democratization of financial information on the Internet. (Until now investors have had to pay thousands of dollars in subscription fees for transcripts.) So our reproduction policy is as follows: You may quote up to 400 words of any transcript on the condition that you attribute the transcript to Seeking Alpha and either link to the original transcript or to All other use is prohibited.


If you have any additional questions about our online transcripts, please contact us at: Thank you!