Okay. Hello, everyone. I'm Pascal Soriot, CEO of AstraZeneca. Welcome to this Full-Year and Fourth Quarter 2016 Results Presentation for our Investors and Analysts. We are here live in London and we also have a number of people on the phone and following us on the webcast. For those of you who are online, the presentation can be downloaded from our website. We plan to spend about 45 minutes on the presentation and we'd like to leave plenty of time for question and answers as I'm sure you have a number of questions. We have about an hour and a half together. So if you want to ask questions on the phone, you can get in the queue already now by pressing *1. There's also an option to ask questions online as well at the webcast. So please turn to slide 2, this is our usual forward-looking statement that you have here. So if we move to slide 3, I'm very pleased today to be joined by Marc Dunoyer, our CFO; Mark Mallon, our Executive Vice President of Global Product and Portfolio Strategy; and Sean Bohen, our EVP of Global Medicines Development and our Chief Medical Officer. If we move to slide 4, this is the agenda. I'll give you an overview. Mark will cover the growth platforms; the other Marc will cover the financials; Sean, the pipeline; and then we'll close and open for Q&A. Moving to slide 5. So this is the highlights of the year, and the performance in the quarter and the year was in line with our expectations. Total revenue declined in the year reflected, as expected, the loss of exclusivity for CRESTOR, but also the lack of U.S. FluMist sales this season and finally, the tail end of the NEXIUM loss of exclusivity in…

Thanks, Pascal, and good afternoon, all of you. I'm really pleased to have an opportunity to spend a few moments talking to you about our growth platforms. And so, if we turn to slide 11, we can get started. So, our growth platforms have really continued to demonstrate overall growth and success in 2017, and that's despite challenges we faced with Respiratory and the biannual price cuts in Japan. The combined revenue of our six growth platforms represent now two-thirds of total revenue, which is an important milestone. And you can see that from the chart that we've got great momentum with BRILINTA, with Emerging Markets. And actually in 2016, Emerging Markets surpassed Europe in sales for the first time. And then, of course, with New Oncology. Very exciting milestone for us as we're now approaching $1 billion on an annualized basis on sales from our New Oncology products. If we move to slide 12, I'll just start by saying – sorry, what I want to say is I will just touch on briefly Emerging Markets and BRILINTA. But I want to do is spend most of the time together focusing on Respiratory, Diabetes, Japan and New Oncology, where I think maybe most of the questions would be. Slide 13. So starting first quickly on Emerging Markets. The key message here is we remain on track for our long-term growth objective of mid- to high-single-digits. And in fact, you can see from the chart here, we achieved that both in the fourth quarter and for the year in Emerging Markets. And this strong performance in the fourth quarter was driven by our growth platforms: by BRILINTA, by the Forxiga and by Respiratory. We're also excited as we move into 2017 that in China, we're looking at potential approvals for both…

Thanks, Mark, and hello everyone. I'm going to spend the next few minutes to review the performance of 2016 and I will then move onto the guidance for 2017. If you want you turn to slide 21. In this first slide, we have the reported P&L performance for the year-end and for the fourth quarter. While core EPS declined in the year, the 9% increase in reported EPS was mainly driven by credit with respect to the diabetes alliance acquisition, reflecting revaluation of the contingent consideration. Please turn to slide 22. Before we get to the core P&L, we have four core adjustment that I would like you to consider. About half of our restructuring cost in the year related to the streamlining of operations and reduction in cost announced in April last year. We anticipate a full-year annualized net saving from this program of $1.1 billion in 2018, mostly within SG&A. The majority of the $1.3 billion of intangible assets amortization was related to the acquisition of MedImmune 10 years ago as well as the legacy agreement with Merck regarding part of our U.S. business. And as I mentioned a moment ago, you see a credit in the year with respect to diabetes alliance acquisition, reflecting the revaluation of the contingent consideration. The $0.17 that you can see here is in fact the net amount between the intangible amortization and the adjustment of the contingent consideration. To remind you, the payment stream of royalties to BMS will expire in 2025. Finally, the $250 million of other adjustment is partly reflects various legal provision as well as a discount unwind on the Acerta Pharma option liability. We provide this information because we want to be as transparent as we can and by showing you the detail of these core adjustments,…

Thank you, Marc, and thank you all for joining us both here and online. It's a great pleasure to be able to share with you our encouraging progress since our last update. And I would like to emphasize that we're very confident in the molecules and the medicines and the trials underlying our pipeline and very enthusiastic about what this year has to bring us as was evidenced by the fact that Mark Mallon wanted to forgo Marc Dunoyer's time to do the financial results presentation of the year-end results and give that to me, so I appreciate that. If we could go on to slide 28, please. This slide shows you, looking back at Q4 in 2016, our main pipeline highlights from the therapy areas that we focus on. So I'd like to start with Oncology. Obviously, we're very proud of the U.S. regulatory submission acceptance of durvalumab in bladder cancer. This is the first-ever BLA application for AstraZeneca. And FDA, as we had announced, did awarded a priority review designation. So we expect a regulatory decision on durvalumab in bladder cancer in Q2 of this year. As Pascal alluded to, the work that people are doing within the company, the time from data availability to filing was absolutely heroic in this case. So I, too, would like to thank all of the people who dedicate themselves to bringing our medicines to patients. We also have U.S. and EU regulatory submission acceptances for TAGRISSO on AURA3. I'll spend a moment on AURA3. We showed that data in December. I'll spend a moment on that data coming up. That also received a priority review designation from the U.S. FDA. We had regulatory submission acceptances for FASLODEX in the U.S. and EU that – in first-line breast cancer. In CVMD, we…

Thank you, Sean. So let me just say a few words of conclusion before we open for Q&A. So if I can have the last slide, please. The messages I want to leave you with today are: number one, our financials on track and we met our guidance for the year. Number two, we have 12 new potential medicines in Phase III and under registration. So a very, very strong pipeline that is starting to deliver. Importantly, beyond the pipeline, we also have tremendous commercial organization around the world. And in particular, in the Emerging Markets, we have a business that has been steadily growing. We have delivered consistent mid- to high-single-digit growth rate. In fact, sometimes double-digit growth rate overall. And in China, in particular, we have consistently delivered double-digit growth rate and outpaced the market growth rate. We've made faster progress in Oncology. For obvious reasons, we have, of course, great medicines, but also, as you all know, you can develop products in Oncology faster than in some other therapy areas. TAGRISSO has been a great launch so far and it's continuing to do very well, and we have great data to support further growth this year. The immuno-oncology programs, of course, are making progress and this year is going to be an important year for our immuno-oncology portfolio. And over the next 12 months, we have a very busy news flow. So, at the end of the day, net-net is we're really getting to an inflection point here as we end this patent cliff, which often has been a little bit exhausting for our teams around the world. We're getting to the end of a patent cliff, the pipeline is starting to deliver and we see the new AstraZeneca emerge. So we have another year or year-and-a-half…

Sorry, go ahead. Just behind session (55:12). We're just going other on this wall. Is that okay? Matthew Weston - Credit Suisse Securities (Europe) Ltd.: Thank you. It's Matthew Weston from Credit Suisse. Can I ask three, please? The first on the durva-treme and MYSTIC, Sean. It's fitting that we start there. There's been a lot of debate since you changed the endpoints as to whether or not the combination on these to show superiority to durva mono in order for the trial to be fileable. I think we all just love your views on that in terms of how you see the filing strategy and absolutely what's necessary. Secondly, Marc, on the numbers. You gave us a trend for what we should expect for R&D over the course of 2017, but you didn't really give us clarity on SG&A. If you could give us some indication as to where those levels are going to settle down, the incremental savings possible in 2017, I think that would be very helpful. And then, Pascal, a big picture question. I think you set out the pipeline all the extremely exciting opportunities at the company. But, clearly, what we have seen in recent months is that a number of members of the executive team have been attracted away from the company. So, I would very much welcome your views as to why that is, and whether that impacts your ability to deliver commercially and on the pipeline going into 2017 and in the mid-term.

So should we start with the D plus T question? Sean, go ahead.

Sure. Thank you for the question. So for the trial to be positive, durva mono or durva-treme combo has to be standard of care chemotherapy, okay, and that will enable filing. The question of durva mono versus the combo is a little more nuanced in that it will be a benefit risk assessment. So, we are quite confident based on data that there will be more toxicity of the combination, in particular, autoimmune toxicities common with IO. And so, it will have to have a benefit above durva mono in order for the combo to be fileable and that benefit would have to offset, in our assessment, the assessment of regulators any toxicity that's seen. So it will be data-dependent in that way. It does not have to beat durva mono to be fileable. Now, I talked a little bit about it before. Our original hypothesis was that what durva-treme combo would bring us is an opportunity to bring IO therapy to patients who weren't benefiting it, that mean PD-L1 low and non-expressers. I think as data has emerged in the field, the question of whether durva-treme combo might also bring benefit to PD-L1 high expressers is a question that is out there. And so, we've enabled the trial to look at those populations, and we'll have to wait and see what the data says. I hope – did that answer your question? Good.

Yeah.

Just a comment on the question on SG&A for 2017. When we were one year ago in 2016, early 2016, we indicated a material reduction on this SG&A line and we, in fact, delivered 9%. So, I will say we delivered probably a bit more than what we had expected. We are continuing our efforts on SG&A cost reduction and you can anticipate another reduction in 2017. I'm not going to quantify the reduction, but it's going to be another reduction for the year.

Okay, Matt. Thanks for the last question. Let me just cover this question about talent movement. First of all, I don't know how far you go when you say over the last few months. In the recent few months, I mean two of our senior leaders have moved on and I have to say they've moved to fantastic jobs, both of them. And it makes me very proud that they actually could get such great job and it's a reflection that many ways that we have a very deep talent bench at AstraZeneca and we have people that are able to take those jobs. I know that some people have said, well, do they know something that we don't know? Does it mean the pipeline is not working out ? So, I know some of you will be at the breakfast meeting tomorrow that we've organized and Mondher, in particular, will be there. So you can – it's kind of a farewell party for him. Mondher has been a great friend of mine for the last 15 years. Actually, he's recruited by the person who recommended him to me 15 years ago. So it's kind of almost a family affair. So you can ask him questions about what he believes of the pipeline. But net-net is they've really been able to grow. So that's one thing I would say. The second is we have a very deep talent pool at AstraZeneca, and we have other people who will grow and take advantage of these opportunities. Jamie Freedman, who is replacing Mondher, and you will meet him tomorrow, was at MEDI maybe for the last four, five years. Before that, at Merck and GSK. He's incredibly talented oncologist, who has already started making a big impact on our Oncology business. So,…

Sachin Jain, Bank of America. A few questions, please. So, firstly, for Marc. You haven't given product sales guidance, but I guess it's implied lower than consensus given the higher one-off expectations versus consensus. Just two related questions. Any color on how you're thinking about product sales growth for 2017? And I guess, more importantly, how much the delta versus consensus? There's a weaker base business versus lost sales that you're externalizing. Second question still on the one-offs, you've commented on one-offs for this year being on a similar level to 2016, which was $3.5 billion. So $3.5 billion and $3.5 billion is getting into $7 billion. If I remember rightly the discussion at this time last year, it was roughly $2 billion for both years, but you're obviously substantially higher. Is that front loading of what you planned over the mid-term or do we expect this high level to continue? And then, the final question is for Sean on MYSTIC. I'm not sure you're expecting this issue. I think Pascal referenced in his introductory comments confidence in the combination despite external factors. I wonder if you could address what we've seen from Bristol more directly and your perspectives on the delayed filing and their less than effusive commentary on the combination in the first-line. And I guess the background to this is you have a lot more data, Phase I, than we've seen from 006. So is there any chance of us seeing 006 prior to MYSTIC? Thanks.

Thanks, Sachin. Before Marc and Sean address your questions, let me just say a bit earlier when I said we're confident, we run the studies because we don't know the answer. Of course, we run MYSTIC to have the answer to whether combination – what's better than monotherapy. So, I'm not confident to the extent I know what the results of MYSTIC will be. What I meant to say is that there's nothing new that we have seen that would change our level of confidence and in fact, our data are not really changing any of our previous views. And Sean will cover this, but the BMS development is, to a great extent, something we were expecting for reasons Sean will be explaining in a minute. But maybe we should start with the financial questions, Marc.

Yeah. So, first of all, on the guidance for product sales. we do not provide guidance for product sales, but we do provide a guidance on the total revenues and I have explained this to you earlier on. Regarding your second question, which is part of the total revenue because it contains externalization revenue, the guidance I gave one year ago was that it would be an increase of externalization revenue and other income versus the level of 2015. And the total of external revenue and other income in 2015 was $2.5 billion. So, I do not remember giving you a guidance for two years in a row $2 billion. If I did, I will have misspoken, but if you could look at it. So, what I can confirm today that we have given you a trend for an increase of external revenue and other income for 2016 and you have now quoted the numbers, $3.4 billion, $3.5 billion. We are going to have a greater amount in the year 2017. So greater than $3.4 billion, $3.5 billion.

Thanks, Marc. Sean, do you want to cover the other question?

Yeah, yeah, yeah. Sachin, thanks. Thanks for the question. Let me start with the BMS combo question and if we see a read-through to MYSTIC and durva-treme combo. So, we were asked and I talked to some of you around Q3 about BMS's announcement around this approach. And at that time, prior to hearing anything else, we said we thought it was very low probability of regulatory success. And the reasons that we felt that was when you file on a single-arm trial, you really have to demonstrate an extraordinary treatment effect over the standard of care. Now, if the standard of care is really ineffective, then not much data can actually indicate you have more activity than it. But when you're moving to the first-line of non-small cell lung cancer, the standard of care is doublet chemotherapy, which has a demonstrated PFS and OS benefit. It doesn't cure the disease; it's still an unmet need. But to get a single-arm data, that gives regulators confidence that they should put your new medicine out there as a big bar. And so that's why we thought that. The other reason we thought that as we've discussed many, many times, is we believe that overall response rate, which is primarily the data you will have, underestimates the treatment benefit of IO therapy and, of course, IOIO. And so, again, that's primarily what you'd be relying on. So that combination of things made us, at that time, believe it was unlikely. Now, we've heard it's not happening. It doesn't change anything for us in terms of that. Again, we're – I want to reiterate what Pascal said. If we knew the outcome of MYSTIC, we wouldn't have to do it. But what we're confident in is the power, the design, the analysis plan, and that as the field has moved, we've incorporated what is known as well as we possibly can into that plan. 006. Prior to MYSTIC, we did expand 006. We expanded 006 with second-line patients. We added some first-line patients. 006 enrollment of that expansion was completed late last year. So, we do what all companies do, which is as a deadline for a submission for an abstract comes up, since it's an unblinded study, we take a look at it and we ask is it mature enough. And to be a meaningful presentation and our conclusion for 006 for ASCO was, no, it's not mature enough to be in a meaningful presentation. So we will look again and aim for meetings later in the year to see if we're ready to present.

Yeah. Sachin, I'm sure you will remember very well that if you're looking for accelerated approval on the basis of a single-arm study, you have to be the standard of care not only with the median, but also the lower end of your confidence interval has to be higher than the upper band of the confidence interval of the standard of care. So, that brings the bar very, very high. That's what we always doubted that a single-arm combination study would able to deliver an accelerated approval. It seems to have borne out as we expected, but that's what we can say at this stage here.

Hi. Richard Parkes from Deutsche Bank. I've just got three questions as well. And one just to be more sort of specific on MYSTIC. I think people saw nervousness around BMS's decisions not just on their decision not to file early, but also just the overall confidence in their approach seems to be – have reduced from where it was a few months ago. So, I just wondered if you could reassure investors that the response rate you're seeing from the 006 study coming through is consistent with the 35% response rate you saw from that initial cohort. That's the first question. Then a couple of financial ones. The gross margin, I remember in the third quarter, you were highlighting that you had broadly stable gross margin despite impact from CRESTOR and I think you pretty much absorbed the full impact there, and I think the reason given was the shifting towards more sort of specialty care. But now in the fourth quarter, you've had a 200-basis-point negative impact. I'm just wondering what changed there. And then, the final question. Looking out to 2018, given the base in 2017, the sort of externalization/other income is now a pretty big proportion, which I'm assuming won't be sustainable at that level. Should we be thinking about a return to top line growth in 2018, but the earnings growth not materializing until 2019?

So should we start with the gross margin financial questions, Marc? And then, Sean can cover MYSTIC and 006.

Okay. So I think the – I guess the impact of CRESTOR is real. Over time, it's going to disappear, and the transition to a more specialty care type of portfolio is going to compensate for it. I mean, on the last quarter of the year 2016 and for the year 2017, the reduction of gross margin will be felt. So the gross margin in 2017 will be lower than what we see for the year 2016. That's clear. Maybe in the third quarter, the impact of CRESTOR was not – I mean, for the – if you total the three quarters, we're not that high, but as we see the first quarter, there was a difference in the percentage of gross margin. And we have reported on the report there was, on the first quarter, 260 basis points on the first quarter alone at CER and 130 basis points at actual rates. So we are, in a way, helped a little bit by the depreciation of the sterling and other similar currencies.

2018, Marc?

For 2018?

2018, yeah.

So I think the gross margin is going to be lower in 2017 than in 2016. For 2018, I think we should have a very similar gross margin...

No, the question – sorry to interrupt. The question was external revenue and OI.

I see. So we're not going to provide guidance to FY 2018.

Just a little bit.

We gave you as much as we could. I mean a lot of details about 2017, different lines of our P&L, an indication on the total of externalization revenue and other income. I mean, I think I'm going to stop there. I think I can't provide you anymore detail for 2018.

Yeah.

But I can certainly say that the return to growth is imminent and it could happen in the course of – at the end of 2017 or early 2018.

You would also – I mean, Richard, you also, I'm sure, appreciate the fact that this will be very much influenced by the mix in the portfolio, right? And as a result, be influenced by the clinical news flow and what comes out. We've got so many new clinical data coming out over the next 12 months. That will influence the product mix of 2018, and therefore, the speed at which we grow, but also the gross margin and our profitability. So I really think that to be able to be more specific on 2018, we need to have a better understanding of what the portfolio will look like actually. Sean?

Yeah. So with regard to update on 006 data, it's really same as the answer to Sachin's question, which is when the data is at a place where it's meaningful – a meaningful publication, we will submit it and we'll update, but we don't do it at an analyst meeting. And as I said, our confidence in MYSTIC is unchanged. It's a good point. The durva-treme therapeutic hypothesis is not yet proven in first-line or second-line non-small cell lung cancer. And so, it is an experiment, but we believe we're doing the right experiment and the trial is well designed. You also asked about – and I know it's a – I think Pascal said it, our enthusiasm for the trial just really hasn't changed. The BMS announcement – part of what you're asking is what did BMS see. I don't know any more than you do about what BMS saw. So, I can't say whether their change in tone is reflecting something in the data that if we knew it, would we (01:15:37). I'm sorry, I just can't answer that question. I will say that, as I said, the – it was our assumption that that accelerated approval path wouldn't work. So to now have confirmation that it's not working again doesn't change our confidence because that's what we thought was going to happen anyway.

One other thing that we will surrender not us as a company, but as an industry I believe is over the last maybe 12, 18 months is that overall survival is really the end game in immuno-oncology. And the other thing that we have more recently reconfirmed that what happened to BMS is really – it's not necessarily the best thing to do to kind of look at your data every week or months or something and then publish things because then you have a lot of viability on what you publish. So, our approach is we want to get to a stage where we have a sufficient number of patients, maturity in the data, overall survival data, and then publish that. And so that will be probably second half of this year at the earliest conference we can actually publish that. But we really want to get a mature dataset and then share that at that point. Other reports are kind of sharing data and doing that too often. And so, the basis of the adjustment of the plans around MYSTIC was our own internal data, but also the external data. As we always said, we would look at (01:17:04), present the ideas more. We would look at our own data. And then based on that, inform the final statistical analysis of MYSTIC and essentially looking at OS as a critical endpoint, but also looking at the cut-off points, and that's basically what we've done. Another question in the room and maybe we'll go online for one or two questions. Nicolas Guyon-Gellin - Morgan Stanley & Co. International Plc: Nicolas Guyon-Gellin, Morgan Stanley. Three MYSTIC-related questions, I'm afraid, all for Sean. The first one is around the recent changes. I noted a slight change in the wording of the press release, where you now refer to PD-L1-expressing tumors rather than positive or high expressions. Does that influence your thinking about PD-L1 expression and all your cut-off in any way? Second, with regards to the readout and the timelines. I mean, the first patient dose in MYSTIC was in Q3 2015 and the last patient commenced dosing in early Q3 2016, which means that you have a minimum of follow-up of at least 12 months up to 24. So basic average assuming straight line recruitment, 18 months. This is much more than what Merck had in KEYNOTE-024, in which they demonstrated overall survival. So are we missing something or – but basically what precludes you from showing OS data as early as mid 2017? And the final one is with regards to the dosing regimen of the CTLA-4. Just conceptually, I would like to hear your thoughts on the two different approaches, the one of BMS and your approach. As far as I understand, you will dose treme only four cycles, whereas BMS may dose until progression. I mean what are the pros and cons? Thank you.

Three questions, Sean.

Okay. Yeah, yeah, yeah, yeah, yeah. There's somebody, I'm sure, has a non-MYSTIC question, but maybe we'll get to that.

I'd love to get to it. Let's start first with PD-L1 expressers.

So, let me try and take them backwards and then I'll then make you remind me of the first one. So the question about the overall survival endpoint and maturity, these are event-driven analyses. And what we share with you is when the final analysis will be done, we think, right , because if the event rate is slower or faster than we forecast, then the trial matures at a different rate. And overall survival will be mature in 2018. As we get closer, we'll be able to narrow it down to a quarter. As always with these trials, there are IDMCs, there are interim analyses, and we just don't share timing for that. So, PFS is midyear. We're getting close enough that we can say that. And for overall survival we're still giving a big range.

Just adding to that...

Yeah.

...actually, if I may, is that in term of the level of maturity that you're referring to is keep in mind that – or maybe we didn't communicate that clear enough before is that we increased the size of MYSTIC as Sean explained.

Yeah.

And we had a team that did a fantastic job accelerating the recruitment. But the end result of this was that the recruitment accelerated toward the end of the trial and the last proportion of the patients were recruited in the last two, three months of the trial. So I can't exactly remember the percentages, you may have them, or maybe we don't want to communicate them. I don't know, Sean. It's up to you, really. But a lot of those patients were recruited in sort of June, July, August. And therefore, we don't have as much follow-up as you may think we have, but certainly, there would be quite a number of patients that would have a reasonable follow-up by May, June this year for sure.

Yeah. It's probably more than half are in the shorter follow-up period just because of the shape of the recruitment curve. Cut-offs. Yes, you definitely caught a change in the wording and it does have to do with how we've informed our definition of PD-L1 positive in the trial and enable the analysis plan to look at that population, again both for durva monotherapy and for durva plus treme combination. We aren't communicating more details around that. Obviously, very competitive landscape. So, for us, it's important that we keep some of these things to ourselves until we're ready to share information with you from the trial. What was the first one?

Dosing for CTLA-4...

Oh, yeah, yeah. Great. Excellent. Yes. So, if you look at the biology of CTLA-4 versus PD-1/PD-L1 as a checkpoint inhibitor, it's a little bit of the PD-1/PD-L1 pressing on the gas and CTLA-4 being a brake. And it is our belief from preclinical data and the mechanism of action of CTLA-4 that once you have removed that suppressive signal, that you should – over the months that you would have CTLA-4 inhibition, you should allow the stimulation of the immune system as necessary to fight the cancer, if it's going to be able to. And once they're activated, they aren't subject to CTLA-4 suppression anymore. And so that's the basis for that. It's a small dataset, but if you look at what's probably the best pharmacodynamic marker of immunotherapy activity is actually autoimmune toxicity. And if you look at the two regimens, they actually look quite similar. The one that BMS finished with, not the one they started with because they started with a much higher dose of CTLA-4 with much more toxicity. If you go to the interval and dose they finished with, the autoimmune AEs look quite similar. Okay.

Good. So maybe before we come back to the room, we'll take one question online, Seamus Fernandez at Leerink. Seamus, go ahead. Yeah. Go ahead.

(01:23:38-01:25:08)

Thanks, Seamus. So the first two questions, Sean, they're for you. And I'm sure you'll enjoy the first one because you love TAGRISSO and there is so much focus on MYSTIC these days that this little child called TAGRISSO, which has a very, very promising future, tends to be overlooked a little bit. Go ahead.

Yeah. Therefore, thank you for the question. I love the introduction to it because I am very enthusiastic about FLAURA. I'm going to get Mark Mallon's help with this too from – because there was a question about what's the threshold with regard to payers and access to the market. First of all, I'll just talk about the trial a little bit. As I mentioned, frontline EGFR-mutated IRESSA versus TAGRISSO. The things that I'll mention is our – few things we're going to look at there that can really distinguish. Obviously, progression-free survival, we need to show that TAGRISSO is better, that it has greater activity. We have a couple of things that lead us to think that that's a possibility. One, as I mentioned, is the CNS penetration, the ability to cross the blood-brain barrier. We know that in the range of 40% of patients when they progress on a first-generation TKI for EGFR progress in the central nervous system, and those first-generation medicines just don't penetrate, so they're not being treated there. So that's one opportunity. Second is the obvious thing, it was designed to suppress T790M mutation, which is the most common mutation that leads to resistance of the first-generation EGFR inhibitors and so this suppresses that. So those are opportunities. In terms of difference and how it relates to the market and payer.

Yeah. I will just say a couple of comments. I think – and you probably won't be satisfied with this, but it's going to be – it's variable, right, because the actual first-line treatment of lung cancer is quite variable across the world, right? So, you're going to have – if you're in a health technology assessment market, there will be high expectations about the progression-free survival. They're going to be really challenging in getting out into the details of what the economic implications of that are and that will be probably a higher bar than say something in the U.S. where the really primarily focus on this is, is this really extending progression a reasonable amount for patients and therefore that will be a driver. And then, of course, outside of the U.S. and Europe, where you have a lot of this, this is in out-of-pocket markets. Some cases, you're talking it's still largely – there are markets where chemo is still the primary first-line treatment; it is not even the TKIs. And I wouldn't rule those out as good opportunities for FLAURA because people may say, why we'd go to this interim step? Let me jump to the better therapy if I'm going to be paying out of pocket. So, we're going to have to work on this in a case sort of like, let's say, healthcare system by healthcare system model assuming and once we see the data. So that's the mindset or the framework. We're going to have to think about this.

The other thing I would add is that, as Mark said, it will be valuable and as Sean said, it will depend on the strengths of the data. But there's one place, one type of patients where I really do believe it would be hard to not prescribe TAGRISSO in an EGFR mutation situation, even who has had the T790M mutation, and that's the first-line patient that has brain metastasis. Now, think about that patient. You have an EGFR-mutated lung cancer and you have brain metastasis. What are you going to do? A first-generation EGFR do not penetrate the blood-brain barrier. And I just had recently the example, a friend of my wife had this and then she progressed – and her lung disease was controlled and she progressed in the brain very rapidly. This is a type of patient who actually will need TAGRISSO. And then beyond this group of patients, the question, of course, will be what is the strengths of the clinical data coming out of FLAURA. ARCTIC, Sean, do you want to cover that?

Sure. So the question was ARCTIC results being material, I believe. So, unfortunately, the materiality assessment is made in the context of looking at the result of the trial and what would its implications be for the company, and do we think that that would move the share price and variety of things that we talk about. We can't talk about it except in the context of the data. And so, that's a determination. It's like really a question, I think, that we can't answer at this point.

And SG&A, Marc, can you talk about that?

Yes, Sean. I think in 2016, I gave you the percentage of the SG&A ratio of 35.5% and also mentioned that we were going to continue our effort to reduce the SG&A cost. So, I think one can summarize that the ratio will continue to go down, maybe not at the same speed as it has gone down in 2016, where in actual terms, the percentage decrease was about 2%. So I think there'll be a further reduction, but maybe not as fast as what we saw in 2016. But I think on a medium and long term, I think this percentage is going to reduce over time.

The only thing I would add to this is that I will not use the industry as a benchmark because in the industry, you have all sorts of companies with different portfolios. So the benchmark we tend to use is made of companies that have a more similar portfolio to ours, and those are companies like Lilly and some others. Still is that we have potential for decrease of our SG&A ratio, there's no question about it. You got to keep in mind that we have quite a unique position in the emerging markets and we have been driving this pretty hard, in particular, in China. And in China, we are now getting into a stage where we have a totally unique position. Our presentation across the entire value chain, we are number two; we have 12,000 people; we had 6,000 people three, four years ago; and we've established ourselves in the place that is quite unique for the future. But, of course, it's recoiled investment. And so, keep all of these points in mind. We'll continue reducing SG&A, both in value and as a percentage of revenue. And again, maybe that's not a satisfactory response, but the speed at which we can evolve that will very much depend on the portfolio mix we have in the second – by next year essentially. So, again, it takes us back to the clinical news flow this year, what are the projects that work. I would love to believe everything is going to work, but some things will work better than others as always. And depending on what we have in our hands, we'll be able to allocate our resources in shift and move the SG&A ratio at a different speed. Should we go back to the room? Yeah, there was a question here we'll go to.

Okay. Jack Scannell, UBS. Two questions. One is around the diabetes market in the U.S. So, you implied that the GLP-1 market is having modest negative pricing. In some ways, it looks like a pretty competitive market, right? There's aggressive contracting already forward management a number of players. But other insulin segments would love to have mildly negative pricing, which implies in some ways the pricing power that you've been able to maintain in the GLP-1 segment is better than, for example, the insulin player C (01:32:37). Is there something structural about the GLP-1 segment that makes it different and do you expect that to continue? That's the first question. The second question is around Alzheimer's. Our view is that U.S. payers have got in the last few years very, very good at stamping on drugs with high budget impact even if they might save money in the long run and that is because in the long run, someone else insures the patient, right? So it's hard for them to take a long-term view. Do you think that matters for assessing the commercial prospects of Alzheimer's projects? And if so, how do you handicap that in your thinking when you're thinking about whether to engage in Alzheimer's or not?

Mark, do you want to cover those two questions, starting with diabetes versus GLP-1?

Yes. In terms of the class, I mean I think there's a number of differences between the injectable business and GLP-1 and oral business, right? You can think about obviously the severity of the patients. So, as physicians are thinking about how readily they are willing to switch the therapies, certainly the – even just the mix of the business, they can be a bit different between say the Medicare Part D versus commercial and there are different pricing pressures there. And also, really, there's some overlap with the competitors, but there are some different ones and they have different approaches. So, I think overall in diabetes, we're going to deal with pricing pressure. It's going to continue to increase. Right now, this is part of the reason we're staying interested and focused in both because we want to see how this continues to evolve. Right now, we feel very confident with what we've got in the Forxiga and we're able to drive very significant growth and continue to lead that class. And we're continuing to – we've got some exciting things coming for BYDUREON, which we think will be a differentiator and allow us to compete in the future in that as well. So another silver bullet or one particular thing that causes the difference at least in my view.

Keep in mind that the insulin prices have been going up very substantially over the last 10 years and payers knew that, and there was – they were quite high and then they had a long way potentially to go down. So the potential to go down was large to start with. And secondly, you have the additional dimension of the introduction of analogues and Lilly have been aggressive in launching the analogue of basal insulin. So the dynamics are a little bit different from what you can see in GLP-1, yeah. With your question about Alzheimer, I think you're right that to some extent, not completely right, but to some extent, it is true that sometimes payer worry less about things that will prevent events in the future. In the case of Alzheimer, again, it will depend on the strength of the clinical data. But I think if you can actually influence the course of the disease and keep patients stable and at home as opposed to being hospitalized, then essentially payers would be able to see the savings quickly. And in fact, the payer in that instance is going to be the government because most of these patients are going to be on Medicare Part D. So I suspect, if we – well, if we, as an industry – those companies developing products for Alzheimer, if we get good clinical data, we should be able to gain access for those treatments. Clear? By the way, Thomas proposed that we extent to 2:15, so – because we saw there were so many questions. Yeah, go ahead.

So I've got two questions. I'll ask the first one in two parts. One for you, Sean, and one for you, Pascal. So, let's assume – you're all kind of cognizant of how confident and comfortable you are in detail to what may be the hypotheses. But in the unfortunate event that we end up sometime middle of this year, you opened up the data box and at least at PFS, neither the two arms are successful. Sean, the question for you is, what does that mean in terms of the trial mechanics? What happens the day after two patients on the study? And Pascal, the question for you is, assuming that hypothesis fails at OS, what's plan B for returning Astra to growth?

You mean if...

So if...

...if OS doesn't read out, a mistake.

Yeah. If there were no more it's not positive at OS as well, then what's plan B for returning Astra to growth?

Okay, okay. So first question for Sean.

Sure. So, couple of things. When we do the PFS analysis, we're doing it when PFS is mature, right? So, at that point, there will have been many progressions. That's how the maturity of that occurs. So, the patients who have progressed, if they're eligible, will have already gone on to another therapy. It will be a relatively, I think, modest problem. What happens to the remainder of the patients, whether now you've set it up as negative a scenario, I personally would not agree.

(01:37:56)

Not very. Oh, yeah. And the positive scenario, I think it depends upon what you're looking at. But then you may – depending upon how positive to end up in the situation of having to cross patients over, right, and give that chemo patients the IO. In the negative scenario, it will be a physician and patient assessment at that point in time what they feel that they should do should patients go to chemotherapy, who were randomized to the IO. And I think it really will depend upon the details of the data. But again, it's not a large subset of patients because PFS will already be mature at that point in time. So does that answer your question?

Did we answer your question?

Yeah.

Okay.

So the second one is that your question about – if OS doesn't read in MYSTIC, right, and what you mean by that is the combination.

Either one.

Yeah. I mean, the first point is really – I mean, we believe AstraZeneca will grow and the question is, at what speed will we grow, as I said earlier because we are all – I mean, many people are mesmerized by this MYSTIC study and to a great extent at least, of course. But you got go to look at everything we have. We have TAGRISSO, we have benralizumab to a launch. We are launching many products. But also importantly, the massive drive we've had to deal with is sort of headwinds called patent expiries slowly but surely getting behind us now. Price pressures on SYMBICORT, they're stabilizing Europe. We still have some of that in the U.S. CRESTOR is gone soon enough. So we'll not going to have to deal with that to the some extent, right? So, I think we will go grow. The question is, again, it depends on how many of those projects read that. I mean FLAURA, we don't talk about FLAURA but FLAURA is a really fundamental growth driver for us. The new indications for LYNPARZA. There are also very important launch of benralizumab as I said. So we will grow; the question is, at what speed. If you come MYSTIC, my comment here is that, again, we tend to look at MYSTIC in sort of a binary fashion. But it's not binary, it's not black or white. It is all shades of gray there. The total negative scenario, meaning nothing works, including monotherapy, hopefully you agree with me is possible, of course, but it's sort of low probability, right? So if we have a monotherapy indication, what it means is we have durvalumab and it becomes a marketing battle. And, of course, as we've known all along the way, we are not leading; we are coming later. But remember, in lung cancer, which is the biggest indication, we have a very strong presence, historic presence with IRESSA, more recently with TAGRISSO. We know those physicians. And by the way, just a little piece of information, there's 80% overlap between the treatment of bladder cancer and lung cancer. The physicians who treat bladder cancer are more or less the same as those who treat lung cancer. So from a sales force effectiveness viewpoint, there's a very strong synergy. And in lung cancer, we're very well positioned. We know those physicians. So it's a marketing battle, but we are in – we have a history there. And then, from there, the question is what do we get out of the combination. Do we get benefit versus monotherapy in one group positive, negative or do we get a benefit in all-comers, of course? My answer to you would be we will grow. The question is, at what speed? Yeah.

And then the second question, Marc, I'm going to try my luck at slightly different way. Assuming the pipeline pans out the way you expected to pan out, what contribution do you need from externalization and OI for 2018 EPS to grow?

So I think we – I'm not – I mean, it's already hard enough to answer your questions on externalization in 2017. I'm not going to venture on 2018. I mean, we believe that externalization is part of our business model. So the part of externalization should continue and the recurring part of externalization is going to grow over time. So I don't think externalization is going to go away, whether MYSTIC is positive or negative. Now, the part of other income after a while may dwindle a little bit. So this part of product that we divest may be impacted by MYSTIC and also because we have completely finished the management of our portfolio. But I think the externalization, which attracts more attention, is going to stay. This is part of our business model. We will always have products that we'll discover that we are not best placed to market or market on our own. So this will continue.

And remember, as we've already said, there will be a recurring dimension to those and then the one-offs will over time decline. We've also said there will be a time point when the sum of the one-off and the recurring actually reduces a bit before, hopefully starts growing again based on the recurring revenue. Some of this externalization relates to pipeline projects as you know. Some of it is pure divestments, but it's relatively limited. Those have small products, $40 million, $50 million in sales that would be doing better in the hands of someone else, and we'll divest them and crystallize the value of our invested. But a big chunk is also about commercial partnerships. Plendil is a good example, the (01:43:57). And here with Plendil, what we've done, for instance, is we felt this is a product that has potential in China. We can't focus on it, first of all, because we have too many things to do. Secondly, because you have to cover too many physicians. We have 7,000 medical representatives in China. A Chinese company has 25,000 to cover the entire country. So we've partnered this product with a Chinese company, we've given them a margin and they will cover the whole country. And we're doing this with a number of products. What that means is we have a one-off reset. So we drop, of course, as we give them – or not give them; we sell them a share of this margin. But then the core of the products, all the time, it will grow again. That's sort of the model, so – and we'll keep doing this. Essentially, what we try to do is find value everywhere we can in a portfolio, focusing our investment in our own R&D and our own pipeline and then unlocking value outside everywhere we can and sort of reinvest it. So we continue doing it. Maybe, we'll take one question from Luca Issi at Cowen online. Luca, go ahead. Luca Issi - Cowen & Co. LLC: Oh, thank you so much for taking my question. Luca Issi, Cowen & Company, on behalf of Steve Scala. Three non-MYSTIC questions, I guess. So the first is, I think at ASH, we've seen the first two cases of atrial fibrillation for acalabrutinib in patients that were actually previously on IMBRUVICA. So how confident are you that these cases are driven by prior IMBRUVICA's treatment and not by actually acalabrutinib? And maybe more broadly, how does this impact your ability to differentiate the acid? And then, second question is on FARXIGA. Can you just provide your high-level thoughts on how competitive dynamics will change going forward in the U.S. given that now JARDIANCE has a proper CV indication on their label? And then, third question is on BYDUREON. Can you just remind us if there's any interim look at the cardiovascular outcomes study for BYDUREON? Thank you.

Thanks very much, Luca. Sean, do you want to cover the acalabrutinib ASH question atrial fibrillation and maybe also the BYDUREON cardiovascular data? And Mark, if you could cover the FARXIGA potential.

Yeah. So, I'll start with the BYDUREON. When I think the – so these large cardiovascular outcomes, trials have again independent data monitoring committees and interim analyses, and then we don't give detail about those things when they're happening. Obviously, it's hard because they're also event-driven and then IDMCs don't communicate anything to us if they don't feel a need to, but it's got all the normal structures that a large CV outcomes trial will have. Yeah. With regard to acalabrutinib and, I would say, its tolerability profile in general. We remain committed that acalabrutinib has a superior tolerability profile to ibrutinib, if only because there are patients who are not progressing on ibrutinib who discontinue it because they have toxicity. Some of that is atrial fibrillation; some of that is a variety of other things. We outlined them in the slide deck actually when we did the deal. When you go into an older patient population such as those that have CLL, you are going to have some atrial fibrillation occur, and the dataset we presented at ASH is very small. And what is really going to be important is the longer term chronic therapy dataset that we get. And recalling as well that we are doing a head-to-head trial of ibrutinib versus acalabrutinib, which is really going to be the gold standard for looking at tolerability, also potentially efficacy in that trial, and that's what's going to tell us what happens. But the idea that in CLL patients, you were never going to see an atrial fibrillation is probably not very realistic given their comorbidities and the age of onset of the disease. That's it.

Mark? Thanks, Sean.

Yeah. So, it's a great question, one we're certainly spending a lot of time thinking about. And so the first thing I would say as we think about sort of the data that our competitor has and eventually there could be more data from all of the products, and actually two, is that it's important data. Obviously, in the limited patient group that that was focused on, which is a minority of the diabetes patients, it had important benefit. But I guess the question we've been asking ourselves is why hasn't that had a bigger impact on the overall SGLT2 class growth rate. And also, actually, even it's been uneven with these share impacts that we've seen on the brain. And I think you have remember that, first and foremost, we're talking about diabetes patients, right? And so, for many of these physicians, their first question is, what do I need to do to get the glucose under control? How well-tolerated, how safe this is going to be versus their products that I've been using, which, in this case, is the DPP-4s? And the DPP-4s are really good products, right? The patients are incredibly comfortable with them because they get good glucose reduction and they're very well tolerated. They don't have to worry about that. And so, I think the thing that we feel like is going to be the key dynamic that we need to look at and see how that is going to happen is, what is going to get physicians to really understand that actually the SGLT2 is a better choice after metformin because of the really good glucose control, because of the weight loss, and then because of also the potential of this class has around cardiovascular benefit. And so I think what we'll see evolving…

Thanks, Mark. So, should we take one last question? Yeah. So, Thomas allowed me to take one last question. So I'll take this one here and then we'll have to close.

I'm Marietta Miemietz, Primavenue. Just a few very quick financial questions. On the U.S. pricing pressure, based on your comments and some of your competitors' comments, is it fair assume that for SYMBICORT and maybe some other large drugs, Q1 pricing will be materially below Q4 2016? And what gives you confidence that pricing pressure is going to moderate in the second half, given that we've always heard that some contracts can be reopened at any point in time? A quick question on PULMICORT growth in the emerging markets going forward, if you could just give us sort of a rough skew because I believe that the China growth has very much been driven by the rollout of nebulization centers and just be good to understand where you are at in terms of that. And my final question is to Pascal on your guidance and your comp. I mean, if you literally deliver on your guidance, i.e., you don't over-deliver and there's no massive devaluation of the U.S. dollar, is it fair to assume that that would have some negative implications for your deferred comp? If you just could give us a rough feel, that'd be great. Thank you so much.

So, thank you. I thought it was going to be one question; it's three. Let me deal with the last one quickly so we can on to more interesting questions. We've always said we're committed to the dividend. We've always said that if something has to go because of price pressure from currencies, and that's really what we're experiencing, the dollar is gaining strength day after day and it's putting pressure on our ability to deliver the cover at actual rate, it is clear that our comp will be negatively impacted. It was pretty clear. Beyond that, I cannot comment more. We continue doing the right thing by the company and we'll do what we believe is right to do to make sure we maximize the value of this company because in the long run, we all win. The patients, to start with, us management, and the shareholders win if we do the right thing. So we'll continue trying to achieve our short-term goals, but in the end, if we have to prioritize, we'll prioritize, of course, the dividend in the long term. PULMICORT, Mark, maybe you want to cover this one. And let me just quickly cover the SYMBICORT. We've said that in 2017, we'll see similar price pressure as we've seen in 2016 in the U.S. with some moderation and we've said that this will be worse in the first half than the second half. And I guess maybe we will leave it at that for today. And then PULMICORT, Mark, since you are the expert in the emerging markets.

Yeah. So the short answer is that we still have tremendous amount of work to do to get to all the patients that could benefit from – particularly kids with asthma – a nebulized solution. And I'll say two points on that, so – three points. So, first of all, in China, I think we actually have some sort of sales of PULMICORT RESPULES in over 10,000 health institutions. I'm not sure we're reaching half of them yet, right? It's so massive, the number of hospitals, the number – and we haven't even – we're just beginning to start to communicate about PULMICORT RESPULES with community health centers. There are 24,000 community health centers. These are the primary care clinics. They're just starting to actually initiate therapy. So there is still a huge amount of work to do. Many places in China still do not – kids do not have access to a good option on asthma treatment. The second thing is they shouldn't even be going to the hospital for this treatment. The real place they should be doing it is in-home nebulization, and we really started working on that in the last two years, but that's going to be a multi-year process to educate on that. And then the last point in terms of the rest of emerging markets, we've been spending the last two years basically educating them on what we've done in China and that sort of uplift is just still ahead of us. And we're now seeing nebulization centers come up in Indonesia, in Russia, in other places in the Middle East and Africa. So, asthma is a terrible – there's huge number of kids and people, both asthma and COPD, around the world and this is one solution. I mean, ultimately, we'd like to think the fixed-dose combination inhalers is the way to go. But for the emerging markets where acute care is still the standard, this is a great solution. So, a lot more work to do.

Thanks, Mark. With that, we'll end it here and thank you so much for your attention.