Good day, and welcome to the Bristol-Myers Squibb 2017 Second Quarter Results Conference Call. Today's conference is being recorded. At this time, I'd like to turn the conference over to Mr. John Elicker, Senior Vice President, Public Affairs and Investor Relations. Please go ahead, sir.

Thank you, Hannah, and good morning, everybody, and thanks for joining our call to discuss our second quarter earnings. With me this morning are, Giovanni Caforio, our Chief Executive Officer, Charlie Bancroft, our Chief Financial Officer. Both Giovanni and Charlie will have prepared remarks. Also joining are Murdo Gordon, our Chief Commercial Officer, and Tom Lynch, our Chief Scientific Officer, who, obviously, will be here for Q&A. And before I turn it over to Giovanni, I'll read the Safe Harbor language. During the call, we'll make statements about the company's future plans and prospects that constitute forward-looking statements. Actual results may differ materially from those indicated by these forward-looking statements as a result of various important factors, including those discussed in the company's SEC filings. These forward-looking statements represent our estimates as of today and should not be relied upon as representing our estimates as of any future date. We specifically disclaim any obligation to update forward-looking statements, even if our estimates change. We'll focus our comments on our non-GAAP financial measures which are adjusted to exclude certain specified items. Reconciliations of these non-GAAP measures to the most comparable GAAP measures are available at our website. Giovanni?

Thank you, John. Good morning, everyone. I'm really pleased that we just finished another strong quarter. Our non-GAAP earnings per share was $0.74, which is a 7% increase over last year. And we grew our revenue by 6% with double-digit growth across our key growth drivers. Now, before I go into more detail, we all know that a competitor announced data this morning that is clearly important to the market. So let me make a few comments. First, our CheckMate 227 is a first-line, non-small cell lung cancer program, not just one trial, investigating several important scientific questions. In study 227 we have at least three discrete opportunities for success. We will be able to evaluate the combination of Opdivo plus Yervoy, we will evaluate Opdivo plus chemo in PDL-1 negative patients and we will be available to evaluate Opdivo plus chemo in all comers. Additionally, as you know, we are testing two cycles of chemo with the combination of Opdivo and Yervoy. It's also important to recognize that MYSTIC trial and CheckMate 227 are very different trials. First, the dose and schedules are different. In 227, we believe we have optimized the dose and the schedule. Second, the trial sizes are very different. 227 enrolled over 2,200 patients, with 1,200 patients in the PDL-1 positive portion alone. In contrast, MYSTIC enrolled roughly 1,100 patients in all comers, and its primary endpoint was evaluated in a subset of that population. While the MYSTIC results are important data and we look forward to seeing more, it's very difficult to read across trials. So let me now get back to our results this quarter. Globally, Opdivo sales were very strong. In the U.S. Opdivo delivered strong performance in an increasingly competitive market. Shares in the second-line lung cancer market remained stable, and…

Good morning, everyone. As Giovanni described, we delivered strong results during the quarter driven by solid execution across the company. Let me first provide some color on the performance of our key products, starting with Opdivo. During the quarter, we saw continued growth for Opdivo in the U.S., with a year-over-year increase of 19%. We're very pleased with this performance, particularly given the increased competition in the market across the range of approved indications. Specifically, in second-line lung cancer, we maintained roughly 40% share as we exited Q2. Outside of lung cancer in the U.S., the trends in renal cell remain strong with approximately 50% share, and we are seeing early signs of acceleration in head and neck. Given the current trends and strength of our business in the first half of the year, we believe that Opdivo will grow in the U.S. compared to 2016, despite pressure in the second half of this year due to the competitive dynamics in both first and second-line lung cancer. Internationally, sales of Opdivo were up 17% sequentially as we continue to see adoption across second-line, non-small cell lung cancer and renal. Outside of these tumors, we are in the process of launching Opdivo in head and neck and bladder in several markets where we already have secured reimbursement, including Germany, Belgium and the Netherlands. With respect to the combination of Opdivo and Yervoy in melanoma, in the U.S., share for the combination remains stable, with use in roughly one-third of first-line patients. Internationally, access and reimbursement approvals for the combination facilitated greater adoption, resulting in 24% year-over-year growth in Yervoy sales. As Giovanni mentioned, Eliquis delivered strong growth and is leading new-to-brand share in both the NOAC and OAC class. Given the strength of our data, including the real-world outcomes data presented…

Thanks, Charlie; and thanks, Giovanni, for the comments. Hannah, I think we're ready to go to the Q&A.

Thank you. And we'll take our first question from Jami Rubin with Goldman Sachs. Jami Rubin - Goldman Sachs & Co. LLC: Thank you. Tom, I have a question for you. And, Giovanni, I appreciate your spelling out the differences between the MYSTIC trial and CheckMate -227. But, Tom, just curious to know your level of confidence in just the general CTLA-4 thesis in light of the MYSTIC results. I note that there are a couple of other really important front-line studies with Yervoy and Opdivo that could report later this year renal and next year head and neck. If you could comment generally about the role of CTLA-4 and your confidence in this asset. It's very hard to walk away from MYSTIC feeling warm and fuzzy about 227. So I'm just wondering if you could talk big picture about that particular strategy. Thanks very much.

Well, Jami, thanks so much for your question. And what I'd like to do is start off by making a couple comments about lung cancer in general and then address specifically the thoughts about CTLA-4 because I do think it's an important data to emphasize how we look at CTLA-4. So as I've said many times, Jami, lung cancer is an unbelievably difficult disease to treat. I've been at it for 30 years and I continually am struck by how challenging it is to make progress in this disease. I think as Giovanni outlined in his comments, one of the key things about Bristol-Myers is we have a comprehensive program in lung cancer. We're not wedded to any one approach. We have optionality with monotherapy in first-line. We have combination of Opdivo and Yervoy in first-line lung cancer. We also have Opdivo-Yervoy combined with chemotherapy, as well Opdivo combined with chemotherapy in first-line lung cancer. So study 227, along with its new cousin, study LA-9 (sic) [9LA], are really a family of studies that we hope are going to demonstrate what the best opportunity in lung cancer is in the first-line. So let me just comment on CTLA-4 as a mechanism. As an oncologist, what really matters to me are drugs that impact overall survival. And CTLA-4 is one of the two drugs in immuno-oncology that has been shown to impact overall survival. And let's just think about it. In Phase III we had data this year that shows in melanoma that Opdivo and Yervoy has a survival advantage in patients with melanoma. The NCCN has endorsed the combination therapy in small cell lung cancer and in mesothelioma. We have Phase II studies that shows strong signals in renal cell cancer and MSI-high colorectal cancer. And, again, in the next 12 months, you're going to be seeing nearly five large Phase III trials where we're looking at CTLA-4 plus Opdivo in renal cell cancer, head and neck cancer and small cell cancer in addition to what we were talking about in non-small cell lung cancer. So we believe that CTLA-4 is an important mechanism. In fact, Jami, not only do we think that Yervoy itself is important, but we've also invested in a non-fucosylated new formation of CTLA-4 and we're partnering with CytomX to look at a Probody for CTLA-4. So I think as an oncologist, as a clinical researcher, when you see an agent that's associated with a survival advantage, I think it's important to find out what is the optimal way to use this drug and how can we improve outcome for our patients. And one of the things that we've become very aware of is dose and schedule are important when you combine I-O/I-O drugs. And I think that we feel, certainly in 227, that we have a good dose and schedule for our patients.

Thanks for the question, Jami. Hannah, can we go to the next question, please?

We'll go next to Chris Schott with JPMorgan.

Great. Thanks very much. And just a couple additional follow-ups given the MYSTIC update. So just a couple here. So first, when you reflect on the various differences between the studies specific to CTLA-4 and PD-1 combos, what are the really critical ones, in your view, that could lead to a different outcome from what we saw with MYSTIC today? Specifically, do you see this as issues of MYSTIC were about study powering and size? Or is there something specific with the agents and their dosing that led to this outcome? And if you're willing to comment on that. My second question, anything from MYSTIC that would impact any of your thoughts around the 227 design, or more specifically, statistical powering (18:08) statistical plans? And then finally, how are you thinking about PFS versus OS as a key endpoint as we think about the first-line lung market and I-O studies? Thanks very much.

Chris, this is Giovanni. Let me just start and then I'll ask Tom to give you any other comments. I think many of the – of course, it's very difficult to speculate on any comparison of the two studies because we don't know really a lot about the data that was communicated today. With respect to the design of the study, first of all, as you mentioned, size is really important. Second, we should remember that the studies include two different medicines, both in terms of the PD-1 and PDL-1 and in terms of the CTLA-4. Dose and schedule are extremely important. And also, as we've said already, you have to remember that 227 really is a program and gives us the ability and optionality to look at multiple combination strategies in different patient populations. But let me ask Tom to give you more comments about that.

Thank you, Giovanni. And, Chris, I think, again, it's important to reflect and think about these studies. There's much about MYSTIC that we don't know. All we've seen is the same press release that you saw this morning. But I think just to re-emphasize, first, the CTLA-4 drugs are different. They're different drugs. And dose and schedule may be important. We've spent a lot of time with Yervoy optimizing its dosing schedule across different tumor types. And we found that there are differences across that. The second is the PD-1 target is different. We're a PD-1 drug, that's a PDL-1 drug. That could be when you combine with CTLA-4 potentially a difference between these trials. And you pointed out and Giovanni emphasized, the power and size are different between the two trials. So I think it's very difficult to read across the two studies. You did ask a specific question about 227. And as you know, we have not disclosed our statistical plan for this study. And, again, I think that the data today certainly is one piece of the information, but I think that the ability to understand what the ultimate best way to treat patients will require more data from MYSTIC and certainly the results of both 227 and 9LA.

Thanks, Chris. Hannah, can we go to the next question, please?

Yes. We'll go next to Vamil Divan with Credit Suisse. Vamil K. Divan - Credit Suisse Securities (USA) LLC: Great. Thanks so much for taking the question. So I guess just a couple here. You mentioned the dose and schedule a couple times there in terms of differences here between 227 and MYSTIC. But I was also intrigued by the filing you had with the fixed dose, every four-week dose and you're relatively quiet in terms of talking about that with the Street before you submitted it. So just can you share a little bit more about the data you have around that new formulation and what gets you comfortable that this fixed dose will be appropriate across all tumor types? Then maybe one, just shifting gear more to the results for the quarter. You raised the bottom end of your guidance range on EPS. Your sales guidance looks like it's essentially unchanged. Can you just share some thoughts on your views on the top line? And obviously, the moving parts in the back end of the year with the Keynote-189 study. What is it that's keeping you not changing sales guidance, even though you had a good quarter here and you are raising the bottom end of the EPS guidance? Thanks.

Thank you. So this is Giovanni. Thank you. I'll try to cover some of your points at a high level, and then I'll ask Charlie to make some comments on guidance and the rest of the year, and Tom and Murdo to give you some perspective on dosing. So first of all, let me say, our performance in the second quarter and actually in the first half of the year is very strong. Commercial execution continues to be very, very strong and trends are good across products and across geographies. So we see good momentum. Obviously there is a degree of uncertainty in lung cancer, but here the U.S., similarly we are at the beginning of our launch of hepatitis C in China. So those are trends that are a little more difficult for us to comment. But, overall, I would say we feel we are in a really good position from a commercial execution perspective. With respect to dosing, I think that, as Tom was mentioning, we are working really as a priority to continue to optimize the schedule of our assets across indications, across lines of therapy, in combination. And what you've seen with 227 is really a lot of work that went into the finding of that dose. Similarly, we've been working on the four weekly dosing for Opdivo monotherapy. And I'll have just Charlie first, and then Tom and Murdo, give you more comments about that.

Yeah, not much to add to what Giovanni said. We are pleased with the first half performance. There is uncertainty, as I mentioned in my comments, regarding Opdivo as we look at the second half of the year. And we're roughly two weeks into our hepatitis C opportunity in China.

So I think your question about our fixed-dose sBLA I think is a good one. And we think about how to use these drugs. We want to think about giving doctors and patients the most flexibility and optionality in the way they use these drugs for convenience and for potential efficacy when we combine them in different settings. So as you know, we have the 3mg per kg dose approved and the fixed dose at 240. So the ability to have a q4 week dose might be appealing for many patients and many doctors. And also to note, we also have in 227 the three-week fixed dose, so that can combine better with a Q3 week chemotherapy. I think what patients and doctors want is the ability to use our drugs in a way that's most convenient for our patients and for physicians as we put together multiple combinations. I think as you're going to find, as combination therapy continues to play a greater role as we move forward, that this kind of flexibility in dosing will be very important as we move forward.

Yeah. Vamil, the only thing I would add is, this could be particularly helpful in parts of the market where infusion infrastructure is a little more constrained. For example, academic medical centers, large regional hospitals. We have seen definitely some preference there for longer duration or longer interval between dosing. Whereas in the community setting, the q2 week dosing schedule we've had until now has been fairly well-accepted. So that q4 will have an advantage, definitely in the academic setting, definitely in the regional hospital setting. And then we are interested, as Tom mentioned, in pursuing the most convenient dosing across indications. And that may vary depending on combinations, as Tom mentioned, as we evolve our portfolio.

Thanks, Vamil, for the question. Hannah, can we go to the next one, please?

Yes. We'll go next to Gregg Gilbert with Deutsche Bank.

Thanks. First I was hoping, Tom, you could comment on your perspective of the PD-1 or PD-L1 agent hitting on OS after missing on PFS. Secondly, can you share your thoughts around the company's appetite to play in the PARP space? Pretty interesting collaboration announced earlier that could be material in the industry. And lastly, you've begun to talk more about NASH this year. At this point, have you met with the FDA yet to plan a pivotal study and when could that start? Thanks.

Thanks, Gregg. And let's break these down into three parts. And I think I might start with the NASH first and then we'll talk about PD-1 and PARP. As you know, non-alcoholic steatohepatitis is an important clinical disease and it leads to cirrhosis. It leads to potential hepatocellular carcinoma down the road. We're very interested in our FGF21 agent. Because our FGF21 agent is able to approach all three elements that seem to be important in NASH and the pathophysiology of NASH. First, it approaches the metabolic derangement that we often see in NASH in terms of fat accumulation. Second, it has the ability to reduce the inflammation that's important in the pathophysiology of NASH. And finally, it has anti-fibrotic impacts as well. And so we presented Phase II data on FGF21 at EASL this year that showed that we were able to reduce hepatic fat in a patient population of people with NASH, giving us a clear signal that this is something we wanted to take forward into Phase III. And we have been discussing, to answer your question directly, we have been discussing our NASH data with regulatory authorities. And we hope to begin Phase III trials and a program by the end of this year, depending upon timing and how we're able to accomplish our timing. So this is an area that we definitely want to pursue. As we mentioned many times on these calls, we're not just a cancer company. Cardiology, fibrosis, immuno-science, are very important to Bristol-Myers Squibb. The second question about PD-1 versus PDL-1. I think that's a terrific question. And if you go back two to three years, there were people who thought that PD-L1 drugs might turn out to be better than PD-1 drugs. And I think now, I think again, we…

No, I would just say we think it's an interesting class and particularly it's synergistic with PD-1s and it looks like it may have applicability in tumors like ovarian and prostate and triple-negative breast. And it's something we continue to look at.

Thanks, Gregg. Can we go to the next question, please, Hannah?

Yes. We'll go next to Tim Anderson with Bernstein. Timothy Minton Anderson - Sanford C. Bernstein & Co. LLC: Thank you. A couple of questions. On 227, it's not terribly clear to me why you can't give some elements of statistical design. It's not like there's a wealth of CTLA-4 competitors that are going to suddenly go out and change their trials. So with that as my comment, I'm hoping you can at least talk about whether the analysis is going be hierarchical in nature, and maybe you can reveal what the starting cut point would be for that analysis. And then on this first readout potentially by year-end, are you expecting that you'll have both PFS data and mature OS data at the same time? Or could the OSPs come a lot later, like it will in MYSTIC? The reason I ask that is that MYSTIC today shows us that maybe PFS is not a great measurement. So if your OS data isn't mature, I'm wondering if there's risk to this initial readout. Thank you.

Tim, this is Giovanni. First of all, I would say we've said all along that we have optionality with our statistical plan. We don't really see a need to make any change there. And with respect to the time lines for the study, the primary completion, as you said, is the first quarter of next year. We do have an opportunity for an interim on OS by the end of this year.

And just to add, Tim, to your question. I think that we have not disclosed when the interim is. We've been saying all along that we expect to have some data in the first half of 2018. Whether that data will be PFS or OS, as Giovanni just mentioned, we have the optionality to look at both co-primary endpoints in the study, and we look forward to seeing the data.

Thanks, Tim. Can we go to the next question, please, Hannah?

Yes. We'll go next to Umer Raffat with Evercore ISI.

Hi. Thanks so much for taking my question. I wanted to focus on (32:17) CTLA-4 combo, but outside of lung for a minute. So perhaps starting off in CheckMate -214, the renal trial. I was curious about the (32:25) dose selection every three weeks – every six weeks deployed in 227, and if there have been interims in that trial? And then also on CheckMate -651 in head and neck, are (32:35). Thank you.

So thanks for your question. Let me address the renal cell question as we think about it. So think about kidney cancer. Kidney cancer is an interesting disease. I can remember when I was training, the only drug that worked in kidney cancer was vinblastine and it really barely worked at all. And yet since then, we've seen the advent of tyrosine kinase inhibitors in kidney cancer and they've shown very good benefit, though not for all patients who get treated with the tyrosine kinase inhibitors. And, clearly, there are patients with more severe adverse disease who don't necessarily seem to do as well with tyrosine kinase inhibitors, but they've made a huge difference in how we approach the disease. And then we've seen the advent of I-O and immuno-oncology agents and the difference they make. As you likely know and are alluding to, we were delighted to be able to publish our study in the Journal of Clinical Oncology our Phase II trial looking at NIVO and IPI in patients with renal cell cancer. And as you know, the combination of NIVO plus IPI had a response rate at 40% with a two-year survival that was between 60% and 65% overall, which we think is very a encouraging data and is the basis for why we've proceeded with the randomized trial 214. Our hope is that we're going have some data from 214 by the end of this year as we move forward. The second part of your question was about head and neck cancer. And, again, I want to emphasize, this has been a busy week for I-O in terms of readouts at various different head and neck trials. All we know from the pembro study is it did not meet its primary endpoint and we don't know…

Thanks, Umer. Hannah, can we go to the next question, please?

We'll go next to Seamus Fernandez with Leerink.

Great. Thanks for the question. So just two questions. First, Tom, can you remind us again your biomarker strategy and how you see this impacting lung? I'm particularly interested in the application of the tumor mutation burden test as a stratifying metric in the new lung study that you guys just announced for the combination. And then the second question is just, can you give us just a general sense of how excited you are and perhaps when we might see data from either the LAG-3 or the Flexus IDO in tumors outside of melanoma? And if you are willing to expand a little bit, do you have plans for registrational trial starts in tumors outside of melanoma for either of those two assets? Thanks.

Okay. So let's get to the meat of the question. So first question, Seamus, that you asked us about, our biomarker strategy. And this is an area of great interest to me and it's an area of great importance to patients across the spectrum. And so if you think about lung cancer, and we've said many times it's a difficult disease to treat. But I think the thing that I've been very impressed with in lung cancer as a lung cancer clinician is how this disease has evolved and how it's fragmented over the past 10 years. If you go back 10 years, 10 years ago we just thought of small cell and non-small cell and that's what you used to make your treatment decisions. Now when a doctor is looking at a patient with lung cancer, they're thinking about EGFR. They're thinking about ALT. They're thinking about ROS1. They're thinking about PD-L1 expression. And I think that they may soon be thinking about TMB as an important marker as well in cancer. And so let's think, what is TMB? TMB is tumor mutation burden. It's a measure of the amount of mutations that are present in protein coding genes in patients' samples. And when you think about it, some people think it's important in and of itself. Others think that TMB is really a reflection of important tumor neoantigens and that one day we might be able to look at the TMB data and define which tumor neoantigens are actually driving the immune response. So we've been happy that we've been able to look back and use TMB in a number of settings. As you know, the data we presented at AACR, we're able to look at TMB in patients in our 026 study and we were able to…

Seamus, maybe if I can add a couple of comments there. I think these developments that Tom articulated for you are really important for us because we've discussed this before. And when you think about our strategic priority, advancing that early pipeline, in oncology and outside of oncology is clearly a very important priority for us. So I'm personally very pleased that we are working between now and the end of the year to start a number of registrational programs. We mentioned FGF21 outside of oncology. Tom mentioned LAG-3, the two IDO programs and nivolumab is another important program. I'm very pleased we are seeing the beginning of a number of registrational programs that are going be really important for us as a company. And even more broadly, I think what we've discussed today is how we are advancing a very broad program in lung cancer which continues to expand and advance. The second pillar is how we are moving forward with a broad number of short-term opportunities for Opdivo and Yervoy within oncology beyond lung cancer. And the third pillar is really the new pipeline. And I think you are seeing that we're making very, very good progress across every one of those three areas.

Thank you, Seamus. Hannah, can we go to the next question, please?

We'll go next to Andrew Baum of Citi.

Hi. Three questions, please. The first one relates to MYSTIC. So your competitors seemed to highlight that a number of trials with I-O have missed PFS and gone on to hit overall survival, and head and neck is obviously one of them for you. You obviously have more experience with I-O than anyone else in this space. Our concerns is the examples of missed PFS that hit OS with I-O all seem to be in trials with short PFS, where patients have been through multiple lines of therapy, which looks to us very different from the situation with first-line non-small cell lung. With that in mind, if you were facing a similar situation with 227, which I hope you won't, where you've missed PFS with the Opdivo Yervoy combination, how confident would you be feeling about making the OS? That's the first question. Second, just going back to the differences between the individual molecules. To what extent is your development plan for non-fucosylated CTLA-4 reflect the fact that you believe effector function is a critical element of the efficacy of the molecule different from your competitor? And then finally, just on patient selection, aside from baseline tumor load, to what extent do you think their learnings in optimizing the patient population using baseline tumor load, performance data, liver mets, which may significantly skew any data you may get from a large clinical trial, and to what extent can that be leveraged in your recruitment criteria? Thank you.

Andrew, thank you for your questions. And the first question is really a hypothetical. And it really is impossible for me to be able to opine on a potential hypothetical of missing PFS or hitting PFS or missing OS or hitting OS. And I really think we have to wait until we get more data from the MYSTIC study to really understand that. As we've said earlier, we have PFS and OS as co-primary endpoints in 227 and we have not disclosed our final statistical plan in that setting. Your second question regards the new formulation of the non-fucosylated CTLA-4. I think one of the reasons we've looked at these two new CTLA-4 targets, first, we believe in the target. We think the target's important. The target's been shown to have a survival advantage, and as I illustrated in my answer to Jami's question. So we think the target's important. And if we think the target's important. And if we think the target's important, the question then becomes, how do we optimize the therapeutic index? How do we make CTLA-4 a better molecule, an even better molecule than it is? And there's two main approaches. With the CytomX approach, the thought is, the Probody technology will allow a differential exposure of tumors to CTLA-4, allowing us to enhance the immune response within the tumor itself and possibly not see the same degree of side effect profile that you could see with immune side effects in other areas. So that's the thought behind the new formula, behind the CytomX approach. With the new formulation, there's a possibility that there could increased potency of the drug that could actually allow us to have more effectiveness in that setting. And, again, very early. We just started, as we announced at the last earnings call, we just started trials with the non- fucosylated CTLA-4. And so we're going have to see how that evolves. And your third question, Andrew, is around the concept of patient selection. And I think you've hit a very important point is that, as we look at these cancer types, and as I mentioned this about lung cancer earlier, we are seeing increased fragmentation in the patient populations. And we're seeing the biomarker selection is becoming increasingly important as we look at patient populations. Lung cancer, for example, is not just one disease. It's many diseases defined by EGFR and ALK and ROS and possibly Tumor Mutational Burden and PD-L1 positivity. So I think as we go forward, attention to patient selection and attention to thinking are we treating the right patients is something that we take very seriously in our tumor design and our trial design. It's one of the reasons, as Charlie mentioned earlier, we're continuing to up-invest in R&D and up-invest in translational medicine because we think it's important for us to prospectively be selecting the correct patients.

Thank you, Andrew. Hannah, can we go to the next question, please?

We'll go next to Steve Scala with Cowen. Steve Scala - Cowen & Co. LLC: Thank you. I have a couple questions. On CheckMate 227, is the interim in 2017 on PFS or OS? I think Giovanni said OS. I had thought it was also PFS or only PFS, so please clarify. And related to that, why would it make sense to do the interim look in Q4 as opposed to Q3 since the conclusion is in January? Would you agree Q4 doesn't provide much benefit in terms of the timing? So that's the first question. The second question is regarding CheckMate 568. I think Bristol is saying there is a possible update in 2017. What does whether there is an update or not depend on? Thank you.

Hey, Steve. It's John. Real quick on their interim analysis. We have said all along that, like in all of our trials, we've built in optionality, which we have in 227. In terms of timing, everything is event-driven. So we've never committed to January, even though that may be what's on ClinicalTrials.gov. And we've said sometime in the first part of 2018 is when the primary endpoint should read out. So both the interim and the primary analysis are event-driven, and so, obviously, time flexible.

So, Steve, regarding your question on 568, I'll just give you a little bit of color behind 568 and how it got to the place where it is. 568 was designed as practice-enabling study of Opdivo and Yervoy. We initially designed 568 to be about 160, 170-patient study initially. And with more experience, we decided to increase that size to 300 patients. And then last year we decided to actually add a randomized portion to 568 looking at this question of two cycles of chemotherapy followed by Opdivo and Yervoy, and that will be an all-comers trial. Now, after discussions with the Agency, we've actually decided to make that a separate Phase III trial. So that's the trial that's now called 9LA. And so that trial will get going. We hope the first patients will enter that trial. We have a safety lead-in right now, which, to the best of our ability, seems to be going well with the two cycles followed by Opdivo Yervoy. And so we hope to begin accrual for that trial in September of this year if all goes according to plan. As of now, we have not yet decided upon our presentation strategy for 568, but we look forward to sharing that with you when we have.

Thank you, Steve. Hannah, can we go to the next question, please?

We'll go next to David Risinger with Morgan Stanley. David R. Risinger - Morgan Stanley & Co. LLC: Thanks so much. I have three questions. The first is, with respect to U.S. Opdivo in the second quarter, could you just provide the percentage of sales in different cancer types in the second quarter? Second, with respect to your expectation for Yervoy combo Phase III kidney cancer results by the end of the year, could you just help reconcile that with ClinicalTrials.gov? ClinicalTrials.gov indicates a primary completion for that trial on September 30 of 2019. So it makes me wonder why look at any Bristol-Myers dates on ClinicalTrials.gov. And then finally, with respect to your Flexus IDO, when do you expect to present Phase I data on that later this year? Thank you.

Okay. David, let me ask Murdo to start and give you some data on Q2 U.S.

David, for Q2 U.S. about 50% to 55% of our sales are in non-small cell lung cancer for Opdivo. And renal cell carcinoma is between 15% and 20%. And then the other big one for us is melanoma, at 10% to 15%. There are obviously other tumors, with bladder, head and neck and others, but those are the three big ones and that's the split.

For the 214, by the end of this year that is an interim analysis. You may be referring to the final analysis. And let me ask Tom to give you an answer to your question, to your third question.

And your third question regarding the Flexus IDO, we hope to have data, in fact, we're pretty sure we will have data, for the CITIM Meeting this fall.

Thanks, Dave. Can we go to the next one, please, Hannah?

Yes. We'll go next to Geoff Meacham with Barclays.

Geoff, are you there?

Yes, I am. Can you hear me?

Now we can, yes.

Okay. So thanks for the questions. So, Tom, another one on CTLA-4. Obviously data-dependent. But if 227 looks underwhelming, would a strategy be to explore triple combos in lung? I'm just thinking with all the assets and novel mechanisms you guys have in-house, whether you view CTLA-4 as more foundational. And then for Murdo on the quarter, can you give us some perspective on second-line lung share new start trends? Just trying to think whether KEYTRUDA is slowing starts. And then looking forward, would you expect any formal reimbursement limitations on sequential I-O therapies in lung? Thank you.

Geoff, why don't we start with Murdo to give you some perspective on your more commercial questions. And then Tom will address your CTLA-4 question.

Okay. Thanks, Geoff. In second-line lung cancer our shares have actually been very stable. So we run around 40% share of second-line lung cancer. And that market volume has held up quite well so far because the impact of pembro usage in first-line has not yet been felt in a reduction of eligible second-line patients. That will change as we get closer to the end of the year where the shares in second-line lung should hold, but the addressable eligible population of second-line I-O naïve patients will start to shrink. And obviously next year, there will be a smaller pool as well. I will say, our teams have done extremely well in establishing this holding of share despite two competitors in the second-line setting. And ex-U.S., we've done very well given the lead that we've got in second-line in many markets due to fast reimbursement. So the emergence of our lung cancer business ex-U.S. is also driving some nice growth. And then outside of lung cancer, our shares are very good in regimen in melanoma, in head and neck we've seen some recent acceleration. In renal cell, we've held up well. Bladder, we're emerging. Actually the broad usage in other areas where NCCN is updated, like Tom mentioned in mesothelioma, is also helping to drive growth outside of lung. So we feel good about that. In terms of reimbursement limitations, really in the U.S., we have seen very good prior authorization approvals in the mid to high-90% range even in combination settings. And, of course, in melanoma it's really the market where we see the most in combination used today. Now, we are obviously tracking the U.S. environment very closely. And we are, I think, in a very good competitive position should there be further restriction or barriers to combination reimbursement as you see sequential lines of therapy evolve, given that we have multiple targets and multiple combinations to use in those settings.

Tom?

So Geoff, thank you for your question. And this is an area of great interest to me. So the first thing is, I can't really think about hypothetical of 227 and MYSTIC. So I think it's really not helpful to think about hypotheticals. But I do think it's important to think about triples. And something I've been interested in for quite some time is this idea that I really think that to begin to think about how we're going to be driving prolonged survivals in cancers, we've really only begun to scratch the surface in terms of thinking how to use these drugs. And I do think that triple therapy, quadruple therapy, down the road makes sense. Just think about lymphoma. And you think about the fact that we can cure lymphoma now with five drugs. We've made a huge difference for patients with lymphoma in that setting. And it's entirely possible that ultimately the way we approach lung cancer, other cold tumors and even melanoma will be with multiple drugs. So for example, we have announced that we have begun some dose-finding studies in early trials with IDO inhibitor plus Opdivo and Yervoy, looking at that together to try to learn, could there be a signal with that triplet in melanoma. And we also have plans to look at a LAG-3 with IDO and Opdivo as well. And so we are committed to multiple drug combinations. Of course, the difficult thing is as you begin to put together multiple drug combinations, trying to determine the relative impact of the different components becomes very challenging and requires a tremendous amount of creativity and innovation in clinical trial design.

Thanks, Geoff. Hannah, I think we have time for two more questions.

Thanks. We'll go next to Marc Goodman with UBS.

Good morning. So first, Murdo, for Opdivo, can you give us a sense of where the OUS sales are coming from, just like you did with the U.S.? Second, on Eliquis, where is the market share now for the cardiologist specialist? Just give us a sense there? And then third, for Daklinza, can you confirm there were no sales in the second quarter for China, and can you give us a sense of what kind of ramp should we be expecting in China? Obviously, hep C we've seen pretty amazing ramps when these products launched. So give us a sense of how we should expect sales for China. Thanks.

Sure. Thanks, Marc. Ex-U.S. we've seen very nice launch progression in Japan, France, Germany, Italy, Spain. We are very early in our launch trajectories in Canada. We're hopeful that we'll be launching soon in Australia. In the U.K., we've launched in our regimen setting in melanoma, and it's going extremely well. So our business is really broad, and we've got over 20 markets ex-U.S. now reimbursed either fully in lung, renal and melanoma, or at least lung and melanoma. And we're working now on our head and neck and bladder indications to be able to secure reimbursement there. So I think our position in Europe, in particular, is very, very strong because of the timing advantage we've secured with indications there and with the rapid uptake that we've been able to achieve. So OUS I feel very good about, and the teams have done extremely well across medical and commercial to launch. With respect to Eliquis, we are sitting with very strong shares now. We are at roughly 53% Eliquis TRx share. And we're leading – and we've got large shares in NBRx leading that just a few more points. So we're growing very well in Eliquis. And, of course, there's still a lot of warfarin used in total prescriptions in the U.S., with about 47% of the market still in warfarin. And in NBRx, that's down now to roughly 30%. 29%, if I remember. So we still see a lot of headroom for growth on Eliquis, but very, very good share trends there. And leading in our categories in BTE and atrial fibrillation. And with respect to Daklinza in China, we did not book any sales in the second quarter. We have started to book sales this quarter. The uptake curve in China with the dual of Daklinza and Sunvepra is a very difficult thing to predict. We are in the process of securing some reimbursement now in three main provinces in China. We are currently selling primarily in the private pay sector. So more to report on that next quarter. But obviously we're very excited about the opportunity being first in the market and being able to help a very large unmet need and a very large patient population in China.

Thanks, Marc. And, Hannah, can we go to our last question, please?

We'll go to Tony Butler with Guggenheim.

Yes, thanks very much. Two questions, Tom, for you. One, if I may go back to FGF21 and NASH and Phase III. And the question is, could you share the designed, if there is a clinical endpoint, for the Phase III program? Or is it simply looking at tissue histology following sort of end of trial biopsy? That's one. And then second, back to MYSTIC and 227. Is there – I understand dose and schedule and a lot of the comments that you've made. But are there perhaps other characteristics of these antibodies that may be substantially different which could actually allow for some clinical differentiation? For example, one being IgG1 versus IgG2. Don't know if that really matters, but it could. Clearly, you've made some comments that glycosylation could affect binding characteristics for ipilimumab. I'm just looking for something more than simply dose and scheduling, if you think that might be useful. Appreciate the time.

Tony, thank you for your question. The first on FGF21 is the data on the design of those trials will be available on ClinicalTrials.gov. It's something we're discussing with the regulatory agencies now. The second question is a good one. And many people have suggested that IgG1 versus IgG2 could explain potential differences. They are different drugs, so there does exist a possibility that there could be differences. But, again, our job is trying to define the best way to use CTLA-4 based Yervoy and our agents. And really I don't have a real thought as to how important these other factors could be. I mean, there's certainly some science that suggests that there could be some differences. But, again, it's again very early in the process of understanding how they work.

Thank you, Tom. Thanks, Tony. Again, we just completed another strong quarter. I am very pleased with execution across the company. As you've seen, we've delivered strong sales performance and we are continuing to make good progress from a clinical perspective and from a regulatory perspective. Looking ahead, as I said at beginning, I see tremendous opportunity for us to continue to grow and I am very confident that we're well-positioned for the long term. So thanks, everyone, and have a good day.

And this concludes today's conference. Thank you for your participation. You may now disconnect.