葛蘭素史克 (GSK) 2018 Q2 法說會逐字稿

Good morning, and good afternoon, everyone.

Thank you for joining us today for our Q2 2018 results, which were issued earlier today.

You should have received our press release on both results and the 23 announcement and can download our presentations from GSK's website.

The presentations today are also being webcast.

Before we begin, please refer to Slide 2 of our presentation for our cautionary statements.

And with that, I'll now hand you over to our Chief Executive Officer, Emma Walmsley.

So thanks very much, [Seth], and let me reiterate a very warm welcome to everybody.

We're going to be splitting this afternoon into 2 sections.

Firstly, Simon and I are going to be covering our Q2 results.

And then after a break, we'll hear from our new Chief Scientific Officer, Dr. Hal Barron, for the first of his updates on R&D and our pipeline.

For both sessions, we have a full Q&A team in the room, and I know they're also very much looking forward to talking with you all at our reception at the end of the meeting.

So this time last year, I laid out my 3 long-term priorities for GSK: innovation, performance and trust, all to be powered by a necessary change in culture.

When I first laid this out, of course, it was simply a framework, but I do hope you now have a better understanding of the changes we're starting to make.

Innovation was our first priority and is the very heart of GSK's purpose, which is to use our science to develop better medicines, vaccines and consumer healthcare products for patients so they can do more, feel better and live longer.

Now Hal is going to talk to you later about how we're taking our approach to innovation forward.

But on this first priority, we've already made some great progress over the past 12 months.

First, we've seen 3 major approvals in Shingrix, Trelegy and Juluca, and the launches are going well.

We've also taken some significant steps in advancing our early-stage pipeline and have started our first pivotal study for BCMA in multiple myeloma.

On performance, I have said very clearly that GSK has to deliver better results.

We've taken a series of actions to do this, such as reallocating resources to key products and geographic areas that can best deliver profitable growth.

The U.S. and new product launches being at the very top of this priority list.

Now it's obviously people that drive performance, and building the right teams has been key to the changes that we're making, half of our top 125 leaders are new in role, with 75% of these being internal appointments.

And we are all very focused on building a culture that's underpinned by purpose and values but with greater performance edge, which means more focus, agility, accountability, and where appropriate, the courage to take smart risks.

Now culture change, of course, takes time and energy.

And as well as having aligned leadership, we are also supporting the change we need with new metrics and incentives.

Trust is built first and foremost by the quality, differentiated innovation we bring and the impact we have on millions of peoples' health and well-being, be it through our latest innovative launches or the access we provide globally to needed vaccines, medicines and consumer healthcare products.

Now I want us to build trust in the long term by being a responsible company and a modern employer, and our new 6-monthly check-ins and our global workforce engagement are showing good progress.

We've also developed a new approach to global health that's going to be one that's more science-led and more sustainably funded.

And our priority focus for global health is accelerating access to those in extreme need to prevent and treat infectious diseases that affect children and adolescents in the developing world.

That's particularly HIV, TB and malaria.

And we were really thrilled last week with the FDA approval of Krintafel, the brand name for tafenoquine, the very first new medicine for the radical cure of P. vivax malaria in more than 60 years.

So overall, I'm pleased with our progress over the last year on these 3 long-term priorities.

But of course, today, our first agenda item is the immediate topic of our performance over the last 3 months, the Q2 results.

So let me turn to them.

We delivered group sales growth of 4% in constant exchange rate terms in the second quarter.

Our Pharma business grew at 1% CER during the quarter, driven primarily by the performance of our HIV business.

Our new Respiratory portfolio grew at 37%, including a GBP 26 million contribution from Trelegy.

In HIV, we also continued to deliver double-digit growth, driven by sales of our dolutegravir portfolio, including Juluca, the first of our new 2-drug regimens.

Vaccine sales were up 16%, with a continued strong demand for Shingrix and a good performance in our U.S. business.

Sales of our meningitis vaccines declined primarily due to prior vaccination catch-ups in Europe, but we do continue to see meningitis growth demand over the longer term.

In Consumer Healthcare, we delivered 3% growth with good performances in oral health and skin health, partly offset by slower growth in wellness and nutrition.

Group operating margins this quarter were up 80 basis points with lower spend in R&D, reflecting the priority review voucher last year as well as the benefits of prioritization of R&D expenditure.

Earnings were up 10% CER, reflecting a 7% increase in operating profits as well as a reduced noncontrolling interest allocation of Consumer Healthcare as a result of the Novartis buyout and the reduced adjusted tax rate.

The improvement in operating profit was reflected in our free cash flow of GBP 492 million.

So we have today upgraded our guidance to 7% to 10% growth in adjusted earnings at constant exchange rates if no generic Advair is approved this year.

The upgrade reflects the positive momentum we're seeing from our new launches, the vaccine Shingrix and the buyout of the Novartis stake in our Consumer Health business, offset by the pricing pressure we continue to see in our inhaled respiratory portfolio.

Now in the event of a generic Advair being available from October 1, we would expect the adjusted earnings growth to be between 4% and 7%.

So turning to our Respiratory business in a bit more detail.

We continue to see good update for our once-daily closed triple therapy for COPD Trelegy.

This quarter, we had the landmark data from the IMPACT study presented at the ATS conference, and we received approval for a broader label for the product.

And we're now able to promote it to this broader patient group beyond simply conversion from Ellipta, and we've prioritized our sales and marketing resources behind the product.

Nucala is continuing to build momentum, too, with sales growth of over 100%, and we've increased our frontline investment in this growing market.

Uptake in Europe is particularly strong, and we see further opportunity as we continue the launch rollouts.

We presented the long-term safety data from the COLUMBA study at ATS, which demonstrated consistent reductions in exacerbations and improvements in asthma control in patients over an average treatment period of 3.5 years.

The consistency and durability of the response they're seeing with Nucala are gaining good tractions with physicians.

And of course, we have the potential for some further upside in additional indications.

So overall, the performance of our new Respiratory portfolio remains strong despite the continued pricing pressure we are seeing within the ICS/LABA category, and we are very focused on delivering competitive commercial execution.

So moving now to HIV.

Our U.S. market share for our leading products, Tivicay and Triumeq, is holding steady despite the increased level of competition in the market.

The first of our new 2-drug regimens, Juluca, which now has a 3.7% share of new-to-brand prescriptions, generated sales of GBP 24 million in the quarter.

Juluca weekly TRx in the U.S. is now over 1,000 prescriptions, with approximately 1,400 physicians having prescribed so far.

And yesterday, we presented data to a packed venue at the IAS Conference from our pivotal GEMINI studies.

This is for the second of our important 2-drug regimens for both naïve and switch patients.

And the study showed that the 2-drug regimen, dolutegravir plus lamivudine, led to a non-inferior virologic outcome compared to a triple regimen of dolutegravir + 2 NRTIs at 48 weeks.

And the success rates we saw were high regardless of baseline viral load count.

And very importantly, we did not see any treatment emergent resistance in either arm of the study.

So with this data, we are confident in the potency and safety of the dolutegravir plus lamivudine combination, and we expect to make regulatory submissions before the end of the year.

Now with more patients with HIV living normal life expectancies, we believe it is very important that they have the opportunity to reduce the number of medicines they're taking, to stay well while reducing the risk of side effects and drug-drug interactions.

So moving to our new vaccine, Shingrix, which was launched towards the end of 2017 in the U.S. and Canada.

Last year, we received a preferential recommendation from ASIP, giving a target universe of over 100 million patients in the U.S. alone.

In Q2 this year, that was echoed by a strong recommendation in Canada.

And as you know, Shingrix represents a new standard of prevention with more than 90% efficacy in the prevention of shingles, and the demand for this vaccine is high from both distributors and patients.

More than 3 million doses have been administered in the U.S. since launch to date, and we expect to vaccinate significantly more patients this year than were vaccinated in total by both us and our competitor in 2017.

We saw sales of GBP 167 million in Q2 and now expect sales for the full year to be broadly in the range of between GBP 600 million and GBP 650 million, this reflecting our growing levels of supply.

So today, our growth is being driven by the medicines and vaccines I've talked about and the 3 key launches identified as critical for our 2018 to 2020 performance: Shingrix, Trelegy and Juluca.

This growth is supported by the optimization of our base business and the solid performance we're still seeing in consumer health.

In the period to 2020, a number of new growth drivers will come into focus with launches in the near term for, if, of course, successfully approved, the 2-drug regimens dolutegravir plus lamivudine and the long acting cabotegravir plus ripilvirine; and of course, our most advanced oncology asset, BCMA; and potentially further indications for our Respiratory products.

And then, in the longer term, we anticipate seeing our earliest stage pipeline start to have an impact on our growth outlook from 2021, as illustrated here on this chart.

And this, of course, will evolve as the data reads out and we supplement with business development, too.

But we have a portfolio of assets with the potential to launch in the subsequent 5 years, and obviously Hal is going to talk to you a lot about that in more detail later.

It is also worth noting our limited exposure to patent expiries in this time frame once Advair has been genericized.

Now last summer, I laid out our capital allocation priorities, which remain the same.

The first is investing in the growth of our business.

And within that, pharma and its R&D pipeline is the most important commitment.

During Q2, we completed the buyout of Novartis's stake in our Consumer Health JV, and this will allow GSK shareholders to capture the full value of a business we believe is well positioned to deliver future sales growth and continued operating margin improvements and bring certainty, too, to future capital allocation planning.

Now I'm just going to take a moment here to comment on what I know will be a question later on the recent press speculation regarding our Consumer business.

The Board's position on the group structure is unchanged.

We believe that the 3-business structure of the group offers significant opportunities in the current health care environment and provides GSK with more stability in our earnings and helps in free cash flow generation.

But as we have also consistently said, this is subject to each business continuing to perform competitively and having access to capital.

So for Consumer, we continue to see very good potential for growth and performance, and hence our decision to increase our margin target for the business to approach the mid-20s by 2022 at 2017 constant exchange rates.

We're also investing in vaccines capacity, and this is to support the rollout of Shingrix and our other vaccines, including our meningitis portfolio.

And then after investing in the business, our next priority for capital allocation is shareholder returns.

We know the dividend is important to our shareholders, and we continue to expect to pay 80p in 2018 and focus on rebuilding cash flow over time before returning the dividend to growth.

And then finally comes scale M&A beyond pharma pipeline, BD activity, where we will maintain a strict discipline on returns.

So how are we going to fund our future growth?

Well, with the recent new product launches, the development of the new R&D approach and the successful buyout of our Consumer business, we've evaluated the group's cost base and determined what is required to deliver competitive, long-term growth and performance in each of the 3 businesses.

Simon is going to give you a lot more detail in a moment.

But today, we are announcing a new major restructuring program.

Savings from the program will be fully reinvested into the group to help fund targeted increases in R&D and support commercialization of new products.

So this is going to supplement the other ongoing changes we're already making in our business to increase cost and cash discipline and to allocate resources better with more focus on growth and profit generation.

So with that, I am now going to hand you over to Simon, who is going to give you a lot more detail.

Thank you.

Thank you, Emma.

Overall, today's second quarter results demonstrate encouraging progress towards our key strategic objectives.

We continue to grow sales across the business and deliver operating margin improvements while investing behind new product launches.

Based on this momentum, we are confident in our delivery for the rest of the year and have upgraded our guidance for constant currency adjusted earnings per share growth for 2018.

Our earnings release provides an extensive amount of information, so I'm going to focus on major points, our expectations for the rest of 2018 and important comparisons to take note of within your modeling.

I'll also provide greater detail on the new major restructuring program we're launching today.

As usual, my comments will be on a constant exchange rate basis except where I specify otherwise, and I'll cover both total and adjusted results.

Starting with the headline results.

Group sales, up 4% to GBP 7.3 billion.

Total EPS, GBP 9p.

And adjusted EPS, GBP 28.1p, up 10%.

Total operating profit was GBP 0.8 billion, up over 100% compared to the small operating loss in Q2 2017.

And adjusted operating profit grew at 7%, ahead of sales, with profit growth in all 3 businesses contributing despite significant investments behind new products in Respiratory, HIV and Vaccines, especially Shingrix.

On currency, the strengthening of sterling compared with last year, particularly against the dollar, resulted in a headwind of 4% on sales and 7% to adjusted EPS.

If exchange rates remain in line with the rates at the end of the second quarter, we would expect the full year headwind from currency to be approximately 6% to adjusted EPS.

Total results for the quarter show a significant improvement on Q2 2017, with reductions in a number of the more significant adjusting items.

Firstly, intangible impairments were materially lower this year, remembering that last year reflected our decision to withdraw Tanzeum.

Secondly, major restructuring costs showed a further step down compared to last year as the existing program nears completion.

And finally, transaction-related adjustments were lower, primarily due to the buy-in of Novartis's interest in the Consumer Healthcare joint venture during the quarter.

Partly offsetting this was an increase in the charge for the remeasurement of the ViiV contingent consideration liabilities driven by both exchange movements and changes to sales forecasts following the recently completed GEMINI study.

The rest of my comments will be on our adjusted results.

Turning to the top line.

This quarter's growth of 4% was driven by continued momentum in all 3 businesses and by particularly strong contributions from HIV and Vaccines.

Sales within the Pharma business were up 1%, driven primarily by the HIV portfolio, which grew 11% in the quarter.

Respiratory sales declined 2%.

We've grown momentum in Europe and international as the regions switched to the new products, offset by continued competitive and pricing pressures in the U.S. Trelegy and Nucala performed strongly, with Trelegy benefiting from share gains after an expanded U.S. label.

Nucala continued its global rollout and also benefited in the U.S. from market expansion as well as some restocking after a destock in Q1.

Seretide/Advair continued to decline, and we transitioned -- as we transition our Respiratory portfolio globally but also impacted by the step-up in U.S. pricing pressures I highlighted in Q1.

The pricing comparative for Advair should be slightly easier in the balance of the year as the increased pricing pressure really started in the second half of last year before then again picking up in Q1.

I still expect an overall decline in Advair for the year of around 30%, assuming no generic entry in 2018, while that might be a little worse depending on how the anticipation of an imminent generic impacts year-end stocking positions.

I'll come back to our expectations for a generic Advair and its impact on our guidance shortly.

Breo grew 4% in the quarter, with strong growth outside the U.S. being offset by a decline in the U.S. This reflects another quarter of the RAR catch-up that we flagged at Q1.

Overall volume growth in the U.S. was strong at around 30%.

With the catch-up now largely done, we expect to be back to good net sales growth in the second half.

Established Pharmaceuticals declined by 5%, although the quarter benefited from favorable RAR adjustments for Lamictal and some post-divestment inventory sales.

I anticipate that the decline will be steeper in the second half and, overall, continue to expect the performance for Established Pharmaceuticals to be a decline in sales of mid- to high single digits over the full year, including the impact of divestments.

Despite the pricing pressures we're experiencing in our Pharmaceuticals business, the momentum we have from our new products and the recent investments we've made mean we remain confident that we will deliver overall sales growth in the low single digits, assuming we do not see a meaningful Advair generic before the start of Q4.

Turning to Vaccines, sales up 16%, primarily driven by further acceleration of Shingrix as well as growth in hepatitis, which benefited from a competitor being out of stock.

Emma highlighted earlier the successful Shingrix launch, and we continue to work on increasing supply.

But I can confirm that we have plans now in place to deliver sufficient doses to fulfill Shingrix sales in the range of GBP 600 million to GBP 650 million for 2018 as a whole.

Previous vaccination patents for shingles vaccine suggest we will see stronger seasonal demand in Q3 than Q4, and I expect to see this reflected in the phasing of sales in the second half.

The meningitis franchise remains an important growth driver for the business even though this quarter saw a decline of 3%.

Bexsero was particularly impacted by the completion of cohort catch-up vaccination programs in Europe, while Menveo was impacted by minor supply constraints that have now been resolved.

I expect a positive second half, and we're well prepared as we head into the back-to-school season.

As I've said previously, vaccine sales will continue to be lumpy due to tenders and the impact of CDC stockpile movements, and this is demonstrated in the quarterly results for a number of vaccines, including Synflorix and Rotarix as well as our overall international sales.

As a reminder, Q3 last year was very strong, driven by flu sales.

And it's difficult to predict precise ordering patterns between Q3 and Q4, but current bookings suggest it might be a bit more weighted to Q4 than last year.

The momentum in the business continues to give us confidence in the mid- to high single-digit outlook for sales CAGR over the medium term, though 2018 is likely to show higher growth than this as a result of the Shingrix performance.

Turning to Consumer.

Sales up 3% despite a 1 percentage drag from the combined impact of the divestment of nonstrategic brands, GST in India and generic competition for TDS in the U.S. Quarter 3 is expected to be the last quarter impacted by GST.

And although the decline in TDS may spread into next year, it will no longer be a material factor in 2019.

A strong performance from oral health was partly offset by a weaker quarter for wellness due to an abbreviated allergy season in the U.S. as well as tougher competitive pressures in the pain category, particularly in Europe.

Importantly, the business continued to achieve a good balance of growth between price and volume.

Power brand growth dipped in the quarter due to stocking patterns in the phasing of some promotional activities.

I anticipate that, that growth will pick up in the second half, and they are expected to contribute strongly overall for the year.

We remain confident in delivering low single-digit growth for Consumer for 2018.

Turning to our operating profit.

Our adjusted margin of 28.8% was up 30 basis points at actual rates and 80 basis points on a constant currency basis.

COGS as a percentage of sales increased, primarily reflecting an adverse year-on-year comparison for vaccines, which benefited from a GBP 45 million settlement in Q2 2017 for a loss of third-party supply volume.

Aside from this, we saw a broadly flat COGS performance as favorable mix and continuing supply chain efficiencies offset U.S. respiratory pricing pressures.

SG&A was up by 6% in the quarter as we invested significantly behind driving new products in Respiratory, HIV and vaccines, particularly Shingrix.

This was partly offset by reductions in back-office and other noncustomer facing resources, and I would expect SG&A growth to slow over the second half, although we need to continue to support key seasonal products in Q3.

R&D cost, down 15%, down 6% excluding the impact of the PRV.

And as I mentioned at Q1, the first half was expected to see reduced spend as the savings from last year's portfolio choices kicked in, but we also still expect to see those savings start going back into R&D over the second half when we should see pharma R&D spend start to grow again, accelerating into Q4.

Royalties down 23% due to payments from the sales of Cialis, which ended in 2017.

And we continue to expect around GBP 200 million for the full year.

In the bottom half of the P&L, we continue to manage our funding costs carefully.

Net financing costs in the quarter included the additional costs of the Novartis buy-in, which came in from June 1. With the additional debt funding for the buy-in and some of our older debt now refinanced at lower rates, I expect funding costs for the year as a whole to be around GBP 725 million.

On tax, the adjusted rate was 20% in the quarter, and we continue to expect a rate of 19% to 20% for the full year.

The charge for minorities in the quarter was GBP 170 million compared to GBP 174 million a year ago.

The ViiV minority interest increased due to the share of the PRV cost that impacted the minority last year, and this was offset by a 2 months' saving on the Consumer Healthcare minority interest as we recognized 100% of the profits from when the agreement to acquire full ownership became unconditional, on approval of shareholders on 3rd of May.

Moving to cash generation and net debt, we remain very focused on driving greater cash discipline across the group and improving our cash conversion.

Free cash flow for the group during the first half of the year was GBP 0.8 billion, up GBP 0.4 billion compared with last year.

This increase driven by improved operating profit reduced restructuring spend and tighter control of capital expenditures as well as the comparison with the cost of the PRV in Q2 2017.

This progress was partly offset by the Vaccines milestone payment to Novartis at the beginning of this year, foreign currency movements and a greater increase in working capital than the first half last year.

The working capital increase reflected the usual seasonal build we see in the first half, but also additional inventory is going in behind new products in Respiratory, HIV and Shingrix.

And the rapid take-up of Shingrix has also led to a step-up in receivables, which we would expect to collect during the balance of the year.

Like last year, I expect cash flows overall to be weighted to the second half.

Net debt now stands at GBP 23.9 billion after the GBP 9.3 billion impact of the Novartis buy-in, dividend payments of GBP 2.1 billion and adverse currency impacts of GBP 0.4 billion, partly offset by the increased free cash flow and some small disposals.

Our credit ratings have recently been confirmed unchanged, and we remain comfortable with our balance sheet capacity to support future investment requirements in the business.

Our results demonstrate the benefits of having clear financial goals as set out in our financial architecture.

Rigorous benchmarking is used to target a competitive cost structure for each of our businesses while ensuring that we prioritize the investments that will generate the most attractive long-term returns.

Ongoing efficiency improvements are particularly focused on streamlining back-office activities where we can take advantage of our recent systems improvements to automate and drive down costs.

We've also made significant changes in procurement with a new global organization driving top quartile savings.

In the commercial space, advanced data and analytic tools are helping us target more precisely our product and market investments.

We've also unlocked deeper larger scale structural changes in our cost base through major restructuring, including the integrations around the Novartis Vaccines business and the formation of the Consumer joint venture as well as the reshaping of our pharmaceutical commercial footprint that we initiated last year.

Through these twin efforts, we've resized our cost base to allow each of our businesses to be competitive in the markets in which they operate and fund the investments needed to deliver our key future growth drivers.

But this is clearly something we need to keep under review.

And with the new product launches underway, the recent buy-in of Novartis's stake in the Consumer joint venture and the new approach to R&D in place, we've evaluated again the group's cost base and the requirements of each of the 3 businesses to deliver competitive long-term growth.

As a result, we are today announcing a new major restructuring program, which aims to deliver GBP 0.4 billion of annual savings by 2021.

The new program is expected to cost a total of GBP 1.7 billion, comprising cash costs of GBP 0.8 billion and noncash costs of GBP 0.9 billion.

This new program aims to significantly improve the competitiveness and efficiency of the group's cost base, with savings delivered primarily through supply chain optimization and reductions in administrative costs.

Savings from the program will be fully reinvested into the group to help fund targeted increases in pharma R&D and support commercialization of new products.

As you think about your models and future spend levels for R&D, adding most of the savings identified to the regular low single-digit increases in R&D gives you a guide as to how we are thinking we will need to step up R&D spend.

The cash cost of the program will be self-funded through improved cash conversion as well as disposals of fixed assets.

The program is not expected to have any impact on our dividend policy.

Moving on to expectations for 2018.

Based on an encouraging first half, I'm very pleased to be able to upgrade our guidance for the year.

Assuming no generic Advair, I now expect adjusted EPS growth of 7% to 10% for the year on a constant currency basis.

And in the event of a October 1 entry of generic competition, I expect adjusted EPS growth of 4% to 7%, again on a constant currency basis, with U.S. Advair sales for the year of around GBP 900 million, again at constant exchange rates.

The major moving parts enabling this upgrade include the positive momentum demonstrated by Shingrix and the benefit to earnings of the full ownership of the Consumer Healthcare business.

These more than offset the continuing pricing pressures we're seeing in U.S. Respiratory, which also leads us to continue to expect a steeper decline in Advair sales before generic competition of around 30% for 2018.

The eventual timing of the launch of a generic Advair will clearly impact the growth rates we can expect to see across 2018 and '19.

So in conclusion, it's been a positive first half.

Our new product launches are going well.

We're working hard to drive cost and cash discipline across the company and seeing the benefits within the results we've reported.

We've upgraded our 2018 guidance, and we're increasingly confident in our financial outlook for the group of mid- to high single-digit EPS growth over the 5-year period to 2020 even with the new investments we're making in R&D and new products.

Our financial architecture and the new major restructuring program we have announced will provide us with the flexibility to make the increased investments in the new approach to R&D, which you'll be hearing about from Hal shortly.

And with that, we'd like to take your questions related to our Q2 results.

And I'm going to invite Luke, David, Brian and Luc to join Emma and me on the stage to take your questions.

I just -- just ask any R&D-related questions, please save that for Hal later.

Thank you.

Thank you, Simon.

We'll be taking questions here in the room and over the phone and from the micro site.

We would request that those asking questions state their name and institution for asking the question and ask no more than 2 questions this time so that everybody has a chance to participate.

For those in the room, please wait until either myself or a member of my team has passed the microphone to you before you start speaking.

And after you've asked your question, please, can you pass it back again so that we can move it on to the next questioner.

Thank you.

And with that, I'll hand you over to the team.

James?

James Gordon from JPMorgan.

It's a pipeline question, but a financial question rather than a pipeline data question.

Two questions.

One was just the R&D was down this quarter, but how much might R&D grow over the next few years?

And asking that both in terms of the P&L and also in terms of M&A.

So for the P&L, how -- I know you've got some cost savings coming through.

But if I benchmark versus peers the level of investment for the Pharma business, could you go up to high teens as many of your large cap peers are?

And in terms of the R&D not going through the P&L but actually doing some business development, how aggressively might you do that?

So I think previously, you said just bolt-ons.

But could it be quite a big bolt-on?

What is the bolt-on count as in, as there's not a lot of very imminent obvious [faith regard] decisions?

Could a big chunk of the rebuild be external?

Okay.

Well, I'll just answer both of those briefly.

The -- so in terms of percentage of R&D, I'm not a big fan of saying there needs to be a percentage spend in R&D, we will -- and I know that Hal would take this position as well.

We will spend according to the data readouts that come, and we are minus 6%, excluding the PRV, in part because we stopped a whole load of programs.

And as Simon has already said, we expect increases through -- to accelerate [through to] the end of the year.

And for the models, you should assume that the savings program they've put in place is -- assuming that the data justifies it, is added to the spend that we have.

But I'm quite relaxed about having a -- even a potentially a quite lumpy R&D spend according to the assets that we're putting in place.

And in terms of BD, I'm not going to sort of set a numeric criteria around what qualifies as a bolt-on.

We have -- I would anchor us in, we're feeling good about some of the organic momentum that we have, whether it's the recent launches that we have or indeed some potential, assuming they're approved, pending ones in that period, '18 to '20, for our outlook, whether it's the new dual therapies in HIV or indeed BCMA that you're going to hear a lot more about.

And then we have the whole new cohort that we'll be looking at and Hal will highlight some of, later today.

But we absolutely do expect to do more business development to strengthen that pipeline.

It's the first priority in capital allocation.

In the room, we've got Kevin Sin, who's been here for all of 3 weeks.

So he won't be outlining his full and complete strategy for that yet.

But you will hear from Hal on how he thinks about BD, and by the way, obviously, focusing on the priorities that he's laying out in terms of assets or technologies, partnerships as well.

But it's not just going to be in-licensing.

We're also looking at out-licensing and creating more flexibility for funding in that way.

We just announced tapinarof.

It may also be in other parts of the broader portfolio.

As you know, we have a review ongoing with Horlicks.

So we are quite comfortable that with the strategy that we are laying out today, organization that we have the balance and the capacity to pursue that.

But a lot is going to depend on the data that reads out in the next couple of years in terms of how much we need to accelerate that or not.

Just to add to that.

I mean, remember also when you look at R&D spend, the 3 businesses have very different characteristics, so looking at the group number is not very useful.

If you look at pharma R&D to pharma spend and the overall shape of that P&L, I think you can kind of run your own models through the numbers we've just talked about, and that's a relevant benchmark, I think we're pretty competitive as to what we expect.

And we obviously have factored in some element of BD because that's a central part of the strategy.

Clearly, there's a degree of unpredictability around that, that we'll have to deal with when we get there.

It's Graham Parry from Bank of America Merrill Lynch.

So you've upgraded 2018 guidance, and I think consensus, looks like it's running a little bit below the range at the moment.

But next year, you're still going to be facing slowing HIV, potential Advair generic coming maybe possibly even at the turn of the year, the increasing in R&D, which -- that you got some savings to help.

But when you put all that together, how comfortable are you with where consensus numbers are sitting for 2019 and, in particular, the current assumption of flat margins into next year?

And then, secondly, a question for David on the dolutegravir.

So NBRx absolutely have now dropped around 28% since Biktarvy launched, and Biktarvy's equaling you roughly on NBRx.

TRx look like they're flattening off in terms of absolute.

So can you perhaps just run us through the dynamics that you're seeing in the market?

How long you think it takes for new-to-brand prescriptions to actually translate into total prescriptions?

And if you're seeing any switching from dolutegravir regimens or is this just all the new patient share gain at the moment?

So David, that one is coming to you in a second.

But just let me answer the first one.

We're saying we're obviously not going to or we don't comment on consensus.

We're very pleased with the 2018 upgrade.

And as Simon said, we're feeling increasingly confident about our 2020 outlooks, whether that's because of the momentum, Shingrix, the consumer buy-outs, U.S. tax reform.

And we are absolutely expecting to digest the Advair genericization, and I'd remind everyone our 2020 outlook is within a range, so we're feeling increasingly optimistic about that.

But David, do you want to pick up for HIV?

Yes.

Thanks, Graham.

So as Emma said, first of all, globally, overall, dolutegravir and dolutegravir-based regimes grew 18% in the quarter.

I think if you go into the U.S., as we showed, TRx for DTG overall up to around about 37,000, clearly varies week to week, but 37k a week.

And our market share of the way we define it, of core and STR, has grown a little bit, is just about 28% or so.

So we're pretty pleased with that.

I would say when you drill down into it, Tivicay particularly looks absolutely rock solid.

So we're seeing almost no switching from Tivicay.

And indeed, some prescribing of Tivicay is still into new patients.

Quite a lot of Tivicay, as we've said before, somewhere in the 40% to 45% range as with Descovy.

But that business looks very, very stable.

Triumeq is a bit more competitive.

We are seeing some switching around the edges.

Some of that is undoubtedly going to the competitors, some of it is actually going to Juluca.

So there is some switching from Triumeq, perhaps not as much as some might have anticipated.

As Emma showed, we're actually quite pleased with the way Juluca has launched, up to over 1,000 scripts a week.

We've got about 1,400 to 1,500 physicians now in the U.S. prescribing.

And I think that paves the way very nicely for GEMINI and dolutegravir, lamivudine.

There is clearly appetite for 2-drug regimes.

So that's where -- I don't think any of those trends are actually that different from Q1.

It's really been very much a continuation, and I think we're pleased with that.

Michael Leuchten from UBS.

I have 2 questions on the financial side.

The upfront costs for the restructuring program seem high relative to the savings.

So what makes this program different from what maybe you have done previously?

And then on the free cash flow definition, I think you've now moved disposal gains from intangibles into free cash flow.

Just your thinking behind that and what that means.

Simon, do you want to take this?

So I think what is different about this program is the particular focus on the supply chain.

And the moment you get anywhere near our fixed asset footprint, you end up with reasonably sizable noncash write-offs.

And you can either take the view you'll never go near those or if you want to confront them and address the overall complexity we have in our supply chain, then that's an issue.

I know we've debated with all of you a number of times.

We felt that was justified.

The cash costs of GBP 400 million against cash costs of GBP 800 million, so very similar paybacks to what we've seen in the other programs we've done.

But it's really driven by that particular focus on the consumer supply chain post the Novartis buy-in, where we can really get after that business in a way that we couldn't with our partner.

And now a year on with commercial focus being much clearer, R&D in place, we can look at the pharma supply chain, see what do we really need to go forward and get up both of those in one program.

That's really why it's a bit different.

And then on the free cash flow, all we've done is a small tweak to the definition to bring in disposals of intangible assets, not businesses, intangible assets because we have the costs in and not the disposal benefits.

Whereas on the fixed side, we have both in.

And I think that, therefore, it's just a question of balancing.

It was about GBP 18 million in the quarter and about GBP 40 million last year.

So it's not a big number.

Emmanuel Papadakis from Barclays.

Maybe I could just take the inevitable Shingrix capacity question.

You said you had enough to make to sell the GBP 600 million, GBP 650 million figure that you've kind of provided us.

Does that represent a ceiling on what you can make?

Or -- and if not, where is the ceiling on what you can make this year?

And what might be the additional flexibility between now and the end of the year on that?

Okay.

Well, thanks for that question.

And the line also is picking up one that's come through from Danny from AXA as well.

So I won't pick that up for a second time.

But obviously, we are delighted with the early start to Shingrix.

Obviously, it's also in line with that preferential recommendation, which was beyond initial expectations.

We did significantly mobilize supply and continue to do so, and I'm not going to comment beyond that confirmed increased outlook of GBP 600 million, GBP 650 million.

But we do expect to continue growing this business through next year and for it to be a very material contributor to growth for the company.

Supply is a bit bumpy, but we're working very tightly with the CDC and with wholesalers and particularly focused on that second injection and very confident considering we're also looking at the geographic rollout pacing that we can maximize this launch in the U.S. I don't know, Luc, whether there's anything you'd add to that?

No, I mean, it's what you said.

It's clear it's a key growth contributor also for the future, and that's why Simon said we keep on investing and we supply more going forward.

And maybe if I could just take a second question.

The contingent consideration liability increased as a consequence to positive GEMINI study.

Is there anything you can tell us about what we should think that might translate to -- into in terms of sales potential?

Okay...

Well, I don't think we're going to give forecasts, but I think you can sense from the increase, our degree of confidence improved.

And what we've really changed is the probability of success because, clearly, that has gone up a lot.

I think you have a question now on the phone, if you're ready.

Is that Steve?

Yes, it should be.

Steve Scala, are you there?

Oh, I'm sorry.

Yes, I am.

So 2 questions.

In the past, a number of guidance elements have been provided for 2016 through '20 beyond earnings, but they have not been reiterated so far today.

So these include for Respiratory, vaccines, pharma, consumer in total turnover.

Should we assume that all of those elements are intact or not intact since they haven't been reiterated?

And secondly, is Horlicks still likely to be divested in 2018?

And what impact would that have on earnings if it is divested?

Thanks.

Well, I'll let you take the earnings.

But the short answer to the question whether all the other guidance elements are intact is yes.

And on Horlicks, I think it depends when it happens.

The plan is still to get it to a conclusion during the course of 2018.

But obviously, we need to balance that.

We're making sure we get the right price.

So I think when we conclude that, we can give you some more specific help on that.

It's Keyur Parekh from Goldman.

Apologies if you addressed this in your opening remarks, Emma, but there's been some recent press speculation about the broader structure of the group.

As you go through kind of the Shingrix launch rest of the world is help us think about, conceptually, would this be a good time for you to think about something different or is as is kind of a good place to be in for the next couple of years?

So thanks very much, Keyur.

But I'm not going to repeat what I did say in my introductory remarks, which is the Board's position on this is unchanged.

We like the structure of the group for its continuity and balance of the cash flow as long as all 3 businesses continue to perform competitively and have sufficient access to capital.

And on that basis, we're unchanged for now.

Thank you.

I was going to ask this question to the Chairman, but in his absence, it's going to go to you.

It's Andrew Baum from Citi, by the way.

So he's clearly indicated an understandable willingness to reassess the group structure in light of the context which you alluded to.

Does the Board's open-mindedness also relate to reassessments of the physician engagement policy that GSK is currently running with?

You recently hired a new General Counsel.

Several of us think it's uncompetitive versus your peers, and I suspect the same may be true of your internal staff.

So whether there's any intention to renew that and what it would take to renew that.

And then a second question for Simon that relates to business development.

You ceased R&D in your dermatologic creams area, I noticed recently.

Given your in-market sales are substantial and there are now some private equity players coming in for picking up some of these established brands, are there potential opportunities to generate capital to reinvest elsewhere in the business, particularly within derms?

Thanks very much, Andrew.

So I'll take the first question.

It's an important one.

We have been investing heavily, as you know, internally in terms of our medical engagement capability, and we do continue to look at this.

We look at it through the lens of very much what is in the interest of the patient in terms of new data coming through and also what's going to help the [HDPs] that we work with, especially as the innovative pipeline develops.

What matters to us is patient interest and transparency.

And on that principle, we continue to look at it.

But no new news today.

And on the asset side, I mean, I think as we've said a number of times, we continue to review the Established Pharmaceutical business to see whether there are better opportunities in disposing of those assets.

Generally, the portfolio is a very strong cash and margin contributor.

We've got a number of disposals that are washing through the numbers this year, which -- and some genericization, which, as we go into '19 and beyond, reduce the drag significantly.

But I think we're very open-minded where there is real value.

But we are now managing that business much more discreetly as a distinct entity to lever margin and cash into funding some of the other objectives that we've described today.

So typically, when we benchmark those alternative proposals, they don't stack up very well.

But occasionally, they do.

And some of the disposals we made to Aspen would be a good example of that.

But we keep looking.

And if a good value opportunity comes up, we're very open-minded.

I'm just going to add, Andrew.

The bulk of the value we can create is one, through a more efficient and focused supply chain, better forecasting.

And then overlaying that, a greater concentration of commercial effort on a smaller number of products because we're quite dispersed in markets such as China and India.

And I think there's a fair amount of work that's advanced that should yield benefit over the next couple of years.

I think we're going to take a question online now from Danny at AXA on 2020 guidance.

Yes.

So Simon, perhaps you can answer this.

The question is related to what the levers and drivers are to get earnings to the upper end of the range in 2020, in terms of 2020 outlook.

Well, I think that we've highlighted a couple in particular in the upgrade that we gave for 2018, and those we certainly expect to continue to play significantly as we go through to the 2020 period and complete that, that outlook that we gave back at the time we closed the Novartis transaction.

But alongside that, continued growth in our new Respiratory products, growth in the meningitis franchise and, as I just touched on, very different contributions from some of the Established Portfolios as well, coupled with continued cost drivers and operating leverage and good focus on the bottom half of the P&L.

Put all that together, along with improved cash conversion, and I think you can see how we get to the 2020 outlooks we've previously given and fund a very significant step-up in R&D spend that we're planning to put behind the new approach to R&D as well as supporting some of the commercial priorities that we've identified.

So it's really more of the same, is the kind of, probably the one line answer to that.

It's Kerry Holford from Exane.

Two questions, please, firstly, on Shingrix.

Can you confirm whether you've started your DTC campaign on that yet?

And given the speed of the ramp, do you think this vaccine could now be more profitable earlier than you previously anticipated?

And also, can you just remind us which ex-U.

S. regions you've now launched the vaccine in?

Trelegy is my second question and more broadly into Respiratory.

So you spoke a little about -- previously about commercial prioritization.

Can you detail what that actually entails?

Have you increased the number of reps?

And can you talk about the relative positioning in that promotional process of Trelegy and/or of Breo?

And how you're now promoting each of those products to doctors?

So since Luke is also leading the execution of Shingrix around the world, why don't you pick up both those questions, please?

Sure.

So the good news is we've not actually had to commence any DTC.

We did originally have quite a significant investment assigned for DTC.

But as you're no doubt aware, the press coverage, New York Times, Washington Post, et cetera has meant that we haven't need to engage in that at that point, which has meant we've been able to deploy that effort to products like Bexsero.

Outside of the U.S., we've launched in Canada.

It's very interesting.

So the NACI recommendation in Canada is not as strong as the U.S. recommendation.

But what's striking, and it's only just come in, what's striking is we've seen a ninefold increase in the number of GPs who are regularly writing Shingrix.

And we've got a market that was declining at about 30% for the last 3 years with ZOSTAVAX, is now more than 100% growth.

So the demand is there, and I think that's encouraging as we look into Europe and other places.

The other place that we've got a very narrow launch is in Germany, and that's very deliberate.

We want to establish a pattern of use in a subset of patients.

Germany, of course, it's really influenced by the presence of STIKO and the recommendation with STIKO.

So that bill will be longer.

And then we have a small allocation going to Japan for the same type of reason there.

In terms of Trelegy, and I think this is a key shift in the culture.

I mean, Simon has made the point in terms of reduction in resources in areas.

And historically, if we adjust a sales force is that's where the costs tended to come out because those people are very visible, you could do that.

What we've tried to do to fund Trelegy and really be quite bold in terms of our investment on Trelegy is to reduce the back office and reallocate those resources.

So the positioning of Trelegy, we're going after people who have had one or more exacerbations in line with the label.

With Anoro, we position it for COPD patients who are symptomatic.

And if you're exacerbating, there's Trelegy.

And that's essentially the strategy in a very simple way.

We split the teams.

Historically, in some markets, we'd have people selling 2 or 3 products.

The net result, when we looked at market research, was physician confusion, which is not the objective.

So we've split that.

We try and avoid that at all costs.

We've split that.

And then consistently, when markets launch, I want almost the entire sales force on Trelegy at launch, so that we really inject energy around that.

And you can see those uptakes, whether you look at the U.K., you look at Germany, you look at Australia.

Canada, it's not fully reimbursed yet.

But if we look at the patient program, again these trends are very strong, and now we just need to keep that going.

Laura Sutcliffe, Berenberg.

Another Shingrix question, please.

Clearly, Shingrix is going very well.

It's more expensive than some of the other more commoditized vaccines products.

We've not seen any DTCs done, for the reasons you just mentioned.

So what are you thinking about in terms of margin in the Vaccines division going forward?

And when we might see some of the benefits of that?

Simon.

So we're not going to get into a breakdown of margin by product, but I think we've very clearly and consistently had a target of plus 30%.

And also, the caveat that we would move the margin around quarter-to-quarter depending on when the investment requirements were, were most acute, and we've seen that this quarter.

But if you look at the shape of the business with the growth drivers we've got, meningitis, Shingrix and the Established Vaccines business, there's nothing to say that we couldn't deliver a sustained performance in the 30%, 35% range that we've typically had for our own business pre the acquisition of the Novartis assets historically, and that's probably the right sort of guide.

Yes, I would just [like it's in] the results, there is quite a bit of pressure with the DTPa in Europe as well as Synflorix in Emerging Markets through Gavi.

So that's -- and sorry, I didn't answer then your question.

Jo Walton, Crédit Suisse.

Two questions, please.

If we look at the gross to net in the U.S., you can see that Respiratory is one of the biggest areas of rebating.

I'm intrigued as to what you think might happen if rebates were rolled back and you had to move to some sort of net pricing.

I mean, effectively, your speculation, your war gaming, whatever it might be on how you think that might impact you, because rebates are a big part of the Respiratory business.

They are.

And I'm going to ask Luke to comment.

We're not going to be in the business of speculating.

Certainly, we're not speculating publicly on what's going to happen in terms of the follow-up from blueprint.

But you're right on net pricing.

If you look at our net pricing in pharma, already we're at a CAGR of 5 years at minus 1, last year at minus 5. And in fact, this quarter, a little bit more than that.

So it is a therapy area under a degree of pressure, but that's why we're so focused on the new Respiratory because it's quite focused on ICS/LABA.

In terms of policy, we have obviously responded ourselves to the request.

We are very supportive of anything that's going to still encourage innovation but actually help with the out of pocket for patients, and bringing more transparency to that huge differential between gross and net is actually a good thing.

We wouldn't necessarily support rebates to be removed completely.

But we do think that's -- more transparency in the value chain is a positive move.

I don't know, Luke, whether you want to add anything in terms of...

Yes, I would just add, the discussions are comprehensive, and there are some clear markets for the administration out there in terms of rebates and safe harbor, and we've been involved through pharma in giving our views.

You can actually read our submission on the AAH's website.

And I think, to Emma's point, it's too early to say.

But I would expect that there will be some changes in the structure.

And the second question, it's a consumer question but in a different way.

I wonder if you could just reprise for us what synergies you think there are between pharma and consumer.

Are we moving to a more consumer-led prescription market, where some of your consumer insights are valuable?

I mean, I understand that you want to keep businesses that perform well in their own right, but the debate is how much synergy there is between any of the individual businesses that adds more.

So I'll let Brian talk about the synergies that he sees from the consumer end of it.

But I'll just make a couple of points on the reverse side of it.

First of all, the Shingrix launch is a consumer launch.

Most of its distribution is through the retail environments in the U.S., where we are very familiar operators.

I think it's 60%, is that right, of the actual -- 65% of the distribution.

So a lot of the planning initially around how to launch this, and it's been, most would agree, a good start, was actually in partnership with the Consumer teams.

Effectively, you're asking an adult to self-present, to go and print out themselves.

It's not a baby that's literally carried in whether they like it or not.

So it is -- we do see that as being a relevant capability.

And the second thing that's been very interesting to me from -- and this is -- I think you'd agree, a long-term view, is when you think about the deal we announced today with 23andMe, which is potentially extremely exciting in terms of identifying these genetically-validated targets, and therefore, addressing this fundamental challenge in the industry about probability of success.

It's also very interesting from a consumer empowerment point of view, whether you've got 5 million people that are signed up now there'll be 10 million very shortly, who are massively motivated around their personal health.

And 80% of whom are -- at the moment, I think that's correct, are voting for being able to participate in research.

So there's -- I think over the long term, and we shouldn't overstate this, we will start to see patient power emerging a bit more strongly.

It's one of the only industries where the end beneficiary is neither in charge of the paying or the choice that's involved.

And over time, with the transparency of information and more value demonstration, I think it's going to be relevant.

Now that does not mean we don't need to prove the other criteria we've laid out as critical to a group structure right now, that the whole group -- the whole board is supportive of.

But Brian, maybe you can talk about...

Maybe just quick on Shingrix, both Luke and I engage quite a bit on that.

And our teams actually work quite closely on the Shingrix plan.

We actually transferred some folks from the consumer marketing group into Luke's group.

I think the classic advantages we've had so far is we talked about Flonase, which is a $5 million product in the U.S. We switched it to OTC, GBP 200 million in 12 months.

And now we're actually launching that around the world, and we're seeing good growth on that.

Most recently, the pharma business did quite a bit of work on a CRM program.

That -- their engagement with HCPs is obviously their expertise.

We do a lot of that with dental professionals and doctors around the world.

We were able to take that off the shelf and launch it in 80 countries around the world last year, something we could never do if not part of the group structure.

So we continually look for those opportunities to take advantage of being part of GSK, and they're there.

Yes, I'd say the next one is Bexsero.

I mean, the upside of having a delay in the Shingrix availability ex U.S. is that I think we can do a lot more in Bexsero, particularly in Europe, where I think initially, we had supply constraints.

And then, of course, we started to think, well, is this a [UMV] market?

I think the conclusion we've reached right now is really, it's going to be a private market.

And that again lends itself to the skill set that's in Brian's team.

And we have another exchange going on in that dimension.

So thank you very much.

I think I'm going to bring us -- an end to the Q&A around Q2.

I know Q2 is very exciting, but wait to see what's coming next.

Let's take a 12-minute break.

We're going to have a -- back at 3:30 sharp, please.

Thank you very much.

(Break)

As I said, innovation is the first of our 3 priorities for the company.

But most importantly, it really is the driver of the other 2. If we innovate successfully, we will perform, and we do become a more trusted contributor to society for the impact we have on people's lives.

So it's absolutely core to setting GSK on the pathway to success.

Last year, I stood in front of you and said we needed to do a better job of, first of all, prioritizing innovation, our pipeline and R&D.

But we really need to do a better job of focusing on fewer assets with bigger potential.

We need to focus on improving our development capability, speed things up, get sharper at decision-making and really sort out the alignment with the commercial organization.

We'd already made some reasonably quick, no-regrets decisions on some of the portfolio to stop things or indeed divest some things.

We hired Luke Miels, who has a long track record of partnering effectively with R&D and not least with this guy right here.

And that's already had a kind of big impact in terms of bringing the commercial voice in earlier.

But by far the most important appointment that I've made in terms of the impact of innovation is, of course, that of our new Chief Scientific Officer.

His job is to transform our pipeline and to reignite GSK's reputation as an innovator.

That is going to take some time.

But I can assure you, he is already having a very meaningful impact in just 6 months.

So I'm delighted to introduce our new Chief Scientific Officer, Dr. Hal Barron.

Thank you.

Thank you.

Thank you, Emma, and good afternoon.

Thank you, everyone, here for taking time out of your busy schedules to come hear about the new R&D approach.

And hello to everybody back in -- at the ranch, as we say, at GSK House and other places.

And good morning to the folks in UP and other places in the United States.

Today, really, is the first, I think, hopefully, of quite a few R&D updates.

Our goal is to be much more transparent and give you some insights as to how the portfolio's progressing over the next every 6 months or so.

I'll spend about an hour giving you an overview of how we're thinking about focusing our efforts in R&D, and then we'll have an hour for Q&A.

What we're going to do is focus really on 3 areas: science, which I'll go into; technology; and culture.

And we'll spend about 20 minutes on each.

Before I get to the new stuff -- let's see if I can figure out how to work this.

That's good.

Well, skip that.

I think I was supposed to read something there, but you've probably seen that slide before -- is I thought I'd go and give a little background about the company, the history.

Because I think it's really important, as you think about an organization that you're coming into new, like myself, to spend time getting to know the people in the organization.

And I spent about 100 days really spending time with folks.

I had about 40 or 50 of these meetings with about 20 or 25 people and listening to what they thought was going well, what they were proud of the company, what they thought we should do more of, and areas that they thought we could do a better job at.

And I think, actually, after a while, the themes became very consistent.

And that was the backbone, really, of this strategy to a lot of extent.

And it's very clear, when you talk to people and look at the history of GSK, there's 4 areas that GSK has been -- had a real leadership position in and very inspiring nature.

The first is leadership in respiratory.

And you've heard a lot about this.

But it's important to go back 1969, when Ventolin was approved for asthma.

And over the ensuing almost 50 years, a number of significant advances, LAMA/LABA, the first anti-IL-5, Nucala and most recently, Trelegy, the closed triple inhaler with data as mentioned in the IMPACT study really suggesting that there's a big impact of that new medicine launched recently.

Vaccines, it's a group that is different than pharma in many respects.

But what is so inspiring to me about GSK and the Vaccines, which is the largest and most important vaccines group in the world, is that we talk in pharma about -- we used to talk about improving outcomes.

And at some point, we were able to say we're actually increasing survival.

And occasionally, now we're even talking about cures in certain places.

Vaccines prevent disease.

It's really quite inspiring.

It's very hard to do this with medicines, if not impossible.

So in Vaccines, as I said, the #1 vaccines player.

And what we have is what, 2 million doses of vaccines given every day; 13 different vaccines in the pipeline; and Shingrix, as you heard a lot about.

We're leaders in global health.

This is a very important component to the company's culture.

We've got one of the most outstanding pipelines in global health.

We care deeply about patients, not just those that have insurance, but those around the globe.

And we've done, I think, an amazing job at ensuring those patients have access to outstanding medicines.

And it's very inspiring to me, and it was very inspiring to the group that that's part of the culture.

I think HIV/AIDS, the therapies that have been developed is a very personal thing for me.

And about this time 29 years ago, 1989, I finished medical school and started on the wards in San Francisco General Hospital at a time when the AIDS epidemic was just becoming an epidemic, really.

And I was, my first day on the wards, my first day as a doctor, I was in the county hospital.

My attending was the Head of Infectious Diseases, Merle Sande.

And I would say at that time, probably 1 in every 3 -- every 4 admissions I would see was a patient with AIDS, usually a cachectic male who had Kaposi's sarcoma all over their body and usually admitted with shortness of breath, having most likely pneumocystis pneumonia.

And it was sad.

But we both knew, the patient and I knew that this was most likely their last admission.

They would usually die on the services.

Maybe they'd make it out after being on a ventilator for a while, but they would probably die the next admission.

And to think that what you heard today, last night even, about the data from GSK is that now, we're talking about 1 pill, 2 drugs, complete viral suppression, these patients living a normal lifespan.

And we're talking about compliance and safety, which is just -- it just, to me, it's so inspiring that what the industry can do when we get our act together.

I think I got quoted yesterday saying I loved big pharma.

This is what I love about it, is when we get together and do something great, you change medicine.

So very inspiring to me, and I want to see more of that.

So you've seen this slide.

I think I'm going to skip most of it.

Those areas are our growth today.

They're growing nicely.

They'll be growing, these 4 areas, nicely for a few more years.

The question, though, is what's coming next?

What do we have?

And these are 19 different assets that could launch during the '21 to '26 time frame.

Of course, not all of them will.

I'm going to talk about a few of them today.

But this portfolio is changing.

As Emma said, we're going to be seeing the introduction of a new oncology medicine, BCMA, at the end of '20.

And if you look at the potential launches in 2021 to '26, we've got a lot of different molecules, much more heavily oriented towards biologics, a lot more oncology, a lot more immunology.

And we need to think about how to frame up those years, and how do we decide where to resource, what is our strategy going to be.

I like this article.

This was from Paul Nunes and Tim Breen from the Harvard Business Review, and it's entitled Reinvent Your Business Before It's Too Late.

And I thought I'd just read this little paragraph because it, I think, summarized to me exactly how to think about this.

"Sooner or later, all businesses, even the most successful, run out of room to grow.

Faced with this unpleasant reality, they're compelled to reinvent themselves periodically.

The ability to pull off this difficult feat, to jump from the maturity stage of one business to the growth stage of the next, is what separates high performers from those whose time at the top is all too brief."

And this is an ideal time to be thinking about reinventing R&D, because we're doing well.

And it's not a time when we're falling off our S curve; we're growing.

And we need to be thinking about where does the future hold possibilities.

So the key thing in thinking about how you reinvent yourself is what questions are you trying to tackle?

What are the key questions you think you want your strategy to solve?

And what are the levers you have?

So we spent some time thinking about this, and we evolved to 2 questions that we think the industry and GSK, in particular, needs to focus on.

The first is the probability of success of a medicine entering the clinic is about 10%.

Despite all these brilliant scientists working all of the time thinking they've got the ideal target, when it gets to the clinic, 9 out of 10 never make it through.

And this is really, really a very challenging concept in this industry.

It's very high risk.

And if we can go from 10% to 20%, that's a pretty big deal.

You can do twice as many programs.

The programs cost half as much to develop, however you want to think about it.

And the second piece, and it relates back to what I said about HIV.

We need really innovative medicines, ones that are going to be very transformative, ones that don't stop after their first indications, sometimes have a broad life cycle, ones that help the most number of patients.

And when they help them, they help them in a fundamental way, not just symptom relief, but get at the fundamental biology and disease-modifying components.

So those are the 2 questions.

And the 3 levers we thought we'd have access to is: Science, how do you find and focus on the best science?

The science that you think is going to take us forward.

What's the technologies that you think you can leverage?

Because technology is a great driver of innovation, not just in the health sciences but in all fields.

Betting on the right technology can unmask lots of signals in the science data that you couldn't see otherwise.

And culture.

Culture is really important.

And we put this science times technology times culture to imply that it's not just the average of the 3. It's not like it would be nice to do 2. If you have the wrong culture, it's 0 times 2 things.

If you have missed the technologies, you're really not going to win.

It's really science times technology times culture.

And so that's what we'll talk about for an hour.

So the industry needs more innovative medicines that might be intuitive.

How do you find those?

One way we decided was to look back wherever the innovative medicine's been.

Look forward from the literature.

Where does the science seem like it's going?

And then look inward, what does GSK do well?

So when we look backwards, we came up with -- we try to be focused because there's lots of different conclusions you could draw.

Ours was that looking backwards, drugs that have modulated the immune system starting with steroids back 70 years ago, have had pretty profound effects.

They had lots of toxicities in the steroid era, and that led to a more targeted era where we see now antibodies developed to inhibit B-cells, like rituximab and Benlysta and omalizumab; or antibodies that block cytokines, like TNF with Humira and Remicade and Enbrel; or antibodies that block IL-6, IL-5 like Nucala, IL-4, IL-13, da da da -- there's a lot of really disease-modifying therapies that have had really pretty significant benefit, and they've had life cycles beyond the first indication.

I mean, rituximab's such a good example.

That starts out as a lymphoma drug and evolves to RA and then studies in ankylosing spondylitis and ANCA-associated vasculitis and ultimately, in multiple sclerosis.

And this is why.

Because fundamentally, the immune system is involved in lots of different diseases.

And now we're seeing the immune system clearly having a role in oncology.

So immuno-oncology is a relatively recent field.

But we're seeing antibodies to PD-1 and PD-L1 that are having really pronounced effects.

And we're seeing T-cells being engineered and reintroduced to patients, modulating the immune system in a very different way, having pretty profound effects like the CAR T, the Juno-type CAR T therapies that we're seeing.

And we think there's a lot of opportunity for immunology to advance our understanding of cancer and treat patients more effectively and get to places where we're talking about cures.

But moving forward, it's very clear that if you look at the literature, we're seeing more and more immune modulators being explored and diseases that hadn't been thought of as inflammatory.

We're seeing IL-1 beta looking like it works in cardiovascular disease.

We're seeing NLRP3 as a target for the inflammasome for -- as a metabolic sensor.

We're seeing complement as being potentially involved in Alzheimer's, microglia playing an important role in neurodegeneration, neuroinflammation and pain, something we'll talk about briefly in a few minutes.

And even getting back to the, can we leverage our understanding of immunology to make the vaccines group even more effective?

We're looking at immunomodulators that could be adjuvants for vaccines and now enabling the immune system to be more effective in mounting its response.

Even in aging, there's evidence that suggest that as we age, the inflammation is causing some of the problems.

So that's forward.

We heard backwards.

That's forward.

What do we do?

This is, to some, a reasonably striking figure.

We have 43 molecules in the clinic, 27 of which actually are immunomodulators.

It's sometimes hard to tell if something exactly the only mechanism is immune modulation.

But the vast majority of the 27 are pure immunomodulators and some of them have dual actions.

In addition, as I said, the 13 projects in the Vaccines portfolio, which are also, to some extent, immune modulators.

And I'll talk about 7 or 8 of these over the next hour.

The first one that I'd like to talk about, and I think it's a pretty fascinating molecule.

It's GSK'165.

It's an antibody to GM-CSF.

And it's for -- being developed for rheumatoid arthritis.

This is an important -- to remember.

We talk about all these amazing drugs for rheumatoid arthritis, which have had a big impact on these patients.

But the disease remains very significant with about 50% of patients, who will remain symptomatic after a year of just TNF inhibitor therapy.

And of course, IL-6 and JAK inhibitor molecules have been developed, and they're adding value.

But almost half of the patients will still experience significant pain, 25% of whom actually transfer their therapy to something new because of the pain.

So this cytokine that we're blocking was -- it has some effects in inducing proliferation of granulocytes and macrophages.

It's one of the first cytokines that was identified in the synovial in patients with RA.

And the preclinical data is pretty compelling.

I won't go into it because we have clinical data.

This is a fully humanized antibody.

There's nothing really terribly fancy about that part.

But we've gone through the development and recently, looked at the IIb data.

And it'll be presented at the ACR, hopefully, or some congress meeting soon.

And I think what you'll find is the data's very compelling.

There's a significant treatment effect.

There's a reasonably rapid onset of symptoms.

And what's particularly interesting is these scores that people use to assess the severity of rheumatoid arthritis often have a pain component and an inflammatory marker component, like CRP.

And what this molecule seems to have is a much greater effect on reducing pain than we think other cytokine inhibitors do, other therapies.

And we think this could be a very unique feature.

And in fact, I won't get into it unless there's questions in the Q&A and Gijs is here.

But we think we understand some of the biology behind that, that has to do with CCL17 and the upregulation of certain receptors that might be driving some of this pain phenomenon.

I'm going to skip that.

This is -- moving on to immuno-oncology.

And in immuno-oncology, we have a very interesting late-stage project, BCMA, but we have 9 other projects that are in the clinic or about to get in the clinic.

And we've divided them into therapies that are cellular, like NY-ESO, which I'll talk about, a T-cell therapy against the NY ESO protein that's expressed in certain cancers; as well as some programs in the epigenetic modulation area, where they modulate the chromatin structure to alter transcriptional profiles; and immune modulators that can be both antibodies that agonize to stimulate T-cells, to be more aggressive at doing their job that they're supposed to do, which is eradicating tumors, as well as antibodies that block the inhibitor receptors to allow the T-cells to be more effective at getting in the tumor and destroying it.

As well as some other molecules that are involved in immunomodulation through pathways such as PI3K beta, STING, et cetera, and the Toll receptor 4.

So let me tell you a little bit about BCMA.

It's the most advanced compound.

We have pretty striking data, and it's a very unique antibody.

Multiple myeloma is a plasma cell dyscrasia.

It's a cancer, basically, of the plasma cell.

And this is a very important disease affecting almost 500,000 people, and over 100,000 people are dying from it.

The treatments have gotten better, but it's still reasonably clear that if you get this, you're going to have a very, very rocky course and almost assuredly die from it.

BCMA is a protein expressed on plasma cells and signals the plasma cells to continue to live.

And so this antibody we have is actually quite unique.

It has 4 different modes of action: First of all, it's an antibody.

It blocks BCMA, the protein, from doing that signaling.

That's terrific.

It's also interestingly -- and we have quite a few molecules in our portfolio that we've engineered to do different things.

This one is so-called afucosylated.

It's not made with any fucose, so you don't have any fucosylation of the antibody.

And this enhances something called ADCC.

And what that is, is antibody-dependent cytotoxicity.

So the Fc portion of the antibody, the bottom part, if you will, binds the NK cells.

And that T-cell antibody can actually destroy tumors.

And by having this afucosylation, you get much more activity.

And the third piece that makes this a very effective therapy is that there's a -- conjugated to the antibody, linked to the antibody is a toxin called MMAF, a very potent chemotherapeutic toxin.

It's so potent that it can't be given systemically because it would be too toxic.

But if you could link it to an antibody and have the antibody deliver it just to the cell that you're trying to kill, in this case, using BCMA to get it right into the plasma cell, you can have some very potent effects.

And in addition, and I won't get into this, there is an additional component, this immunogenic cell death that if there's questions, we can get into later.

But this really unique molecule has seen -- been in the clinic.

And you've probably seen this data, but I want to put it in context.

In very heavily pretreated patients, you can see on the left that there's an overall response rate of 60%, very impressive for this really severe group of patients.

And when you compare that to Amgen's Kyprolis, which was studied in third line, not as severe a population, a refractory population, you see that it's over twice as effective in terms of response rate.

And when you look at progression-free survival, again, over double the progression-free survival.

And that's true even of Darzalex, which is, as monotherapy, a drug being developed by Janssen.

And they're fourth line, not quite as refractory a population again, only 29% response rate with a progression-free survival of 3.7 months.

So we think when we're trying to do apples-to-apples -- and again, these are even more severe, 40% actually hadn't received Darzalex.

That we're seeing overall, as I said, 60% response rate, 43% in the prior dara-exposed patients with a 7.9-month overall median progression-free survival and 6.8 in the dara group.

And we're seeing that we haven't even reached the overall -- median overall survival.

So we'll have that data at some point to be able to compare as well.

But we think this is going to be a very active molecule.

And the key thing with this molecule, once you see activity, once you see -- going back to the 1 in 10 works, this is the one.

And this is the kind of -- when you see this kind of data, you want to jump on this.

You want to move as quickly as possible.

And this is -- when I got here, we really took a deep dive, and Axel and his team saying how can we really jump-start?

How can we move it as aggressively as possible?

And I won't go into all these different combination trials and moving in from fourth line to third line, to second line to frontline, and the enabling studies and the combinations with novel reagents that will -- like PD-1 to see if we can bring out additional synergy that was in that fourth bucket.

But all of these different studies are going to allow us to move from a population that is reasonably large, 36,000 patients, to a much larger population, and more importantly, be able to help more patients.

Because the earlier in the disease you treat, the more likely you are going to have a meaningful effect.

And we can go through these later, if you want.

But a lot of exciting trials, and we're moving them aggressively forward.

We don't have time to go through all the oncology assets, but these potential medicines are very unique and are already demonstrating single-agent monotherapy activity.

We have an agonist antibody to ICOS.

Again, agonist antibodies are very unique.

These are the ones that bind and stimulate the receptor.

ICOS, we think, is something that once stimulated, makes those T-cells particularly aggressive, more active, and they are better suited to shrink a tumor.

This is an example of a 64-year-old male who had head and neck cancer, where you can see there's -- the yellow is not part of the MR. It's to highlight the differences in the tumor size.

And you can see a pretty -- a dramatic reduction in tumor size.

The first in-human trial's ongoing across several tumor types.

And we're seeing, as I said, clinical activity observed, both as monotherapy and with PD-1 combinations with pembro.

So trying to, again, identify the ideal combinations.

Similar, with OX40, we're seeing -- we're having a dose escalation with both monotherapy and PD-1 combinations.

And we've seen, again, clinical activity.

This is a 66-year-old female with a liposarcoma, and you can see the tumor shrinking.

And we have a very interesting -- based on some really interesting preclinical data, TLR4/OX40 combination, which might be particularly exciting.

PRMT5 inhibitor, first in class, potential for broad application.

Again, we're seeing responses.

And a BET inhibitor, an oral epigenetic-targeted drug, which is being developed for a broad range of tumors and clear activity in that midline carcinoma, where the BET mutation is known to be driving, but also other activity has been seen, particularly in HER positive breast cancer.

So a long list of other immune modulators that are in the clinic that we have -- that we're excited about.

These are -- those are some of these that might potentially have data readouts soon and that ultimately might end up becoming medicines.

So I'll move on to that.

We'll have plenty of time in the Q&A to go into more detail if you should have questions on that.

But I wanted to talk about the problem that I think I probably have the most passion around is, how does we -- how do we get this 1 in 10 -- up from 1 in 10 that succeed to some higher number?

Maybe ideally 2 in 10?

And one of the drivers for this is that GSK has been a leader in understanding how genetically validated targets contribute to the probability of success, paper that was written by, on the right here, Matt Nelson, in our group at GSK in R&D.

I went back and looked at the literature and found that genetically validated targets, on average, have about twice the likelihood of becoming medicines.

And this is a big deal.

Going from 10 to 20 would go from being average to best-in-class.

It would go from whatever it costs to develop a drug, $2.5 billion or whatever the most recent number is, to $1.25 billion.

Or said a different way, again, you could do twice as many.

And to show you how this works, I just thought it would be fun to take a few minutes, maybe 10 minutes and explain how this genetic stuff works.

So there's something called a GWAS, which is a genetic -- genome-wide -- so the entire genome-wide association studies, okay.

And what we do is we take patients who have a disease, let's say in this example, diabetes, and a control group, people who don't have diabetes, and we do a chip test on their genome until we go through every gene essentially, oversimplifying a little bit.

And we see all the different variations that might occur.

And they're all going to be probably reasonably randomly distributed for disease and non-disease.

But occasionally you'll see one that's really enriched in either the patients or the controls.

And you have a good suspicion that this is involved in the driving of the disease, the underlying biology.

And so these are GWAS hits, if you will, genetically validated targets.

And when we advance those, we get medicines against those targets that have twice the likelihood of succeeding.

An example of this, just to put it in perspective -- and you probably know this one.

But back in 2006, Helen Hobbs and her group presented in the New England Journal about a -- essentially a GWAS finding.

That there are people, not very common, but some people who have a genetic mutation that -- where they don't have almost any PCSK9, a protein that we make.

And she asked, "So what's different about these people?" And what turns out, they have a very, very low risk of having cardiovascular disease.

So if you imagine not having this, that we could pharmacologically try to mimic that, we can make an antibody for that protein, take the antibody away.

We can mimic the genetic condition.

If you did that, one might imagine that you'd reduce the risk of cardiovascular disease.

And in fact, that's what people did.

Made an antibody.

11 years later, through lots of stuff that we do in drug development, you see it works.

And it doesn't always work this nicely, but this is a common theme that when you have a genetically validated target, you spend much less time and you have a higher probability of success.

We think there's a lot more of these out there, and we want to pursue those because those are much better than the targets we think that are generated from understanding preclinical models, where there's lot of similarity with humans, but there's a lot of differences.

And I think those differences could be misleading us and causing, in part, this attrition.

Now complementing that is something that's only recently been able to be done, because we haven't had the data sets that have been large enough.

But it's something called a PheWAS.

And like a GWAS, a PheWAS is a phenome-wide, a phenotype-wide association study.

Now what that means is we take all the different phenotypes, like diabetes or cardiovascular disease or osteoporosis or osteoarthritis or neurodegeneration, any kind of disease, and we ask, if you have a gene of interest, let's say PCSK9, besides reducing the risk of cardiovascular disease, does it do anything else?

And why is this useful?

So folks -- Erik Ingelsson and folks looked at this.

And what they found was that they look to see does it reduce Alzheimer's?

Because that would be nice to know.

You could put your molecule on a development for Alzheimer's [have] reduced it.

Or does it reduce something else?

Turned out the one thing they could find was that it reduced the risk of ischemic stroke.

So from this, you can predict that if you did a study to prevent the risk of ischemic stroke, you're probably going to have a successful trial.

So this PheWAS allows you to think about how to do second and third indications.

And in addition, you can imagine that it doesn't just show what it prevents, but you can imagine it might tell you what safety problem you might experience.

It might have worsened the ischemic stroke, in which case you'd predict that the side effect of the drug would be that.

And sometimes, the side effects might be so significant that you decide not to even develop the drug.

So that's part of why it probably increases probability of success because you have so much more information about what the likely effects of these drugs might be.

So faster, more likely to work, you understand the life cycle better and you can understand toxicity through this GWAS and PheWAS.

Hopefully, that makes sense.

So we've been leaders in this.

We have first started with the collaboration called Open Targets back in 2015.

We subsequently moved with -- in the U.K, with the U.K. Biobank, which is 500,000 people where they're deeply phenotyped.

They're imaged frequently.

They have blood work done.

They're followed serially.

It's one of the -- probably the most important genomic databases that exists.

And the question is, where could we have taken that next?

And I think you've read the press release.

I'm very excited to say that the next place we go is to a collaboration with 23andMe.

So we are going to have access to a large number of genotyped patients.

And this is really probably one of the most significant milestones, I think, for us as a company.

And we're -- we've decided to commemorate this partnership with 23andMe with something really very special to treat -- we haven't shared this with you yet.

But today, right after my talk, actually, we have -- we're actually bringing in a very -- a rock star, a very famous person, who's going to entertain us.

He's a household name.

He's going to sign autographs.

Any guesses?

The person -- Richard Scheller.

You're right, Richard Scheller.

The guy next to Bruno Mars is coming here today, and Richard's in the front row.

And he'll be here to answer questions.

I'm going to take you through a little bit of what the 23andMe opportunity represents.

And Rich will be here for questions, and he's going to do a little routine later, too.

He's going to sing a song.

So 23andMe has a data set of over 5 million people.

And I don't know if we're going to give the most updated number, but it's growing fast.

Richard shared with me the number of new customers that just came on Amazon Prime Day.

It's pretty shocking.

What's interesting about the database is not just its size, because it's over 5 million.

And all of these people, of course, are genotyped like we saw on the GWAS slide.

But 80% have committed to provide their data for research and importantly, to be recontacted.

There's over 1.5 billion survey questions answered.

So you could think about the wealth of information.

And they provide data on what diseases they have, whether they have diabetes, what medications they take.

Richard was telling me one earlier.

Do they -- if they get a mosquito bite, is it actually an intense experience?

Or is it mild?

And I can tell you something even about their immune system.

There's an enormous amount of information collected that we can link to their genotypes.

And we think this could be -- combined, actually, with Open Targets and with the UK Biobank, a very, very effective resource.

And we've nominated the first target.

And I want to walk you through this, because it's not only an example of a way we might be able to help Parkinson's patients, but it's a [form first] of what we could do over the next years.

And as you know, Parkinson's is the second-most prevalent neurodegenerative disease, and it's driven by certain genetic drivers.

And one of them is called LRRK2 or leucine-rich repeat kinase 2. And these patients have a mutation.

One of their base pairs is abnormal, leading to a unique amino acid sequence that has an active LRRK2 kinase.

And we can make small molecules to turn that down, and that should, in theory, work.

If the overactive kinase is causing the disease and we can turn it down, one might imagine it would work.

And if it does work, it's possible that, that pathway's involved even in the wild types.

Just like PCSK9 doesn't just work for the patients who have more PCSK9 than normal, it can work in a spectrum of patients.

And that's what the trials borne out, and we'd have to see.

If it worked in the LRRK2 kinase patients, maybe it would work more broadly.

And what's also interesting about genetics, not to get too far ahead of it, is that we know who these patients are before they're symptomatic.

So there's an opportunity to think about, should a drug be safe and effective, can you start moving it more proximally?

Could you start thinking about prevention?

A very exciting component of the genetic validated targets.

Now let me walk you through not just the target discovery component of this, but what's really, I think, very exciting and sometimes missed as what value 23andMe brings to a collaboration like this.

There are 1 million patients in the United States with Parkinson's, 135,000 of which have -- or sorry, 135,000 people with a LRRK2 carrier and about 10,000 people, 10,000 to 15,000, people, who have LRRK2 and Parkinson's.

So you can think about it as being 1%, 1.5% of people with Parkinson's actually have the LRRK2 mutation.

So it's not a common cause of Parkinson's.

It's just when you have that mutation, you're much more likely to get it.

In 23andMe, there's 10,000 recontactable individuals with Parkinson's; 3,000 recontactable LRRK2 carriers, who don't have Parkinson's; and more than 250 patients with Parkinson's and a LRRK2 mutation.

Now if we were to think about trying to do a study in LRRK2-positive patients, we'd have to enroll 100 patients -- not enroll, we'd have to screen 100 patients to find 1, maybe 2 patients with a LRRK2 mutation.

And if you think about operationally how long it would take to do that trial, you might conclude it's not even feasible.

That if you did do it, it certainly would take a long time and be very, very expensive.

We can, once we have the molecule, actually just go right to these patients and identify immediately a large number of people who could, in theory, be interested in participating in this trial and do a trial twice as fast, 3 times, 4 times, not even clear exactly how much faster, but a lot faster, a lot cheaper and give us the confidence should we see a signal in safety and efficacy, to move aggressively, as I just described.

So really, a very different way of doing drug development, a validated target, much faster approach, help the patients who are in 23andMe and others, and really potentially have a new era in how drugs are both discovered, but also developed.

And I'm frankly very excited about this example.

But I'm also excited about what this example is going to lead to in the future in terms of how we do all of this work.

So I've sort of gone through this.

The collaboration offers scale, size, diversity, sustainability for advancing therapeutic programs.

The questionnaires, I should mention, although they're very rich, we have an additional opportunity to go back to patients and ask them questions we'd want to target for those.

So that's a very unique feature of this data set and this collaboration.

And we, as I said, custom surveys and rapid recruitment of trials.

And it improves target selection; higher probability of success; safer; more effective, as I showed you with the PheWAS; allows more efficient and effective identification for recruitment.

One other thing that you might think about is we know where these patients live.

Instead of setting up clinical trial sites where we might want to, we actually can look at where they live and say, "Hey, we're going to put up a clinical trial site so you don't have to travel very far." So we can make it even more convenient for the patients.

And that's what I mean by empowering the patients.

So I want to move to technology.

My timer is not working, so I'm going to see how much behind I am.

4:07.

What time did I start?

3:30, 40 minutes?

All right, I'm going to be on time.

Technology has been a big driver in every industry.

That's for sure.

The challenge has been, how do you find the right technology?

Which ones do you bet on?

But when you find the right one and bet on it and take it to state-of-the-art, sometimes beyond, you can see things that other people just can't see.

You can identify signals in data that just others, looking at similar data sets without that technology, just don't see.

Or you might have access to a way of making a therapy that allows you to do things other people couldn't do.

So pushing technology to the bleeding edge has been a hallmark of success and innovation in all industries, particularly ours.

Now I want to take 5 minutes and try to explain to you this because this is -- the human genetics is phenomenally fascinating.

This is going to take that to the next level, and that's called functional genomics.

Now GWAS, PheWAS, the 23andMe collaboration is really exciting.

It's about structural genetics.

We're looking at base pairs and how the As and Cs and Ts and Gs differ, but it doesn't really tell you about function.

You might guess that if the base pair abnormality results in a protein and a gene, that that's probably the gene.

But most of the differences aren't in genes.

You have to figure out what they're doing.

They might be next to a gene.

They might be in between 3 or 4 genes, and you're not exactly sure what they do.

So there's an area of science and technology called functional genomics that allows you, essentially with either TALENs or CRISPR, to functionally toggle through the genome and do gene-by-gene assessments as to what happens to a cell when you take down or up a protein.

Now a cell can't tell you, I'm fine.

I'm fine.

I'm fine.

Now I have Parkinson's.

I'm fine, I'm fine, I'm fine.

But it can tell you how it looks or what transcripts are made during each of the genetic manipulations.

And that surrogate, which we call an endophenotype, that intermediate that we think is reflective, say, for instance, with Parkinson's, you might get alpha-synuclein, which is a hallmark of the neuron in patients with Parkinson's.

When you toggle through the genome, if knocking out one gene causes that alpha-synuclein, you might think that gene might be involved in causing Parkinson's in patients.

And not only can we just do that gene by gene, but we can imagine taking a cell from the patient who has a disease, and then taking each gene and knocking it out, and see if you can reverse the phenotype of that patient.

So you might have Parkinson's with the mutation that I just described, the gene 29 TS; and gene by gene by gene, seeing if I knock it out, can I normalize that cell?

That's called a modifier screen.

You can even do this with large data sets, and 23andMe is starting to get to the size where you can do these gene modifiers.

But you can do it in cell culture as well.

And one sort of corollary to that is you can take gene A, and maybe you don't see much of an effect when you knock out gene A. And you take gene B, maybe you don't see much of an effect.

But sometimes, when you take out gene A and gene B, the cell's no longer viable, or it has some phenotype -- some other phenotype of interest.

And this is called synthetic lethal if the cell dies or synthetic interaction, a synergy.

And this allows some very interesting things.

Because a lot of times, particularly in immuno-oncology, we're searching for what's the ideal combination.

We might see drug A doesn't have much activity as single agent, and drug B doesn't have much single-agent activity, maybe none.

But the combination might be very powerful.

And if you think about the different numbers of combinations, we have 10.

But what if everybody else has another 10, and there's maybe hundreds of these and hundreds of different cancers.

The combinations, in theory, if you started adding them up, there's more combinations almost than there are patients with cancer.

So we need to think about how to more thoughtfully identify the combinations that are going to work in patients and the synthetic lethals, like screen, can be used to really help us figure out which are the ideal combinations.

And we have a couple of combination studies ongoing, particularly TLR4 and OX40, where we see this synthetic interaction that might be very powerful.

But when you think about it, gene A times gene B, right, there's 20,000 genes times 20,000 genes, some are essential.

But essentially, you get 200 million combinations, and that's in one cell type.

So if you think about hundreds of cell types and a whole bunch of other assays you might want to run, the data gets enormous.

So we've got the patient data, 5 million patients.

We've got this functional genomic data.

It's definitely overwhelming for any human to think about how to deal with this.

But we now have the second technology, which I want to introduce, is the use of machine learning with deep, deep learning, neural networks, lots of different analytic tools that allow all of this data to get understood: the relationships between the genes; the sort of underlying semantic representation, the sort of like language of the cell.

And once you can figure out this new language, you can start deconstructing the patterns that cause disease, and the number of targets could be substantial.

And again, they'd be hidden in this data if you didn't really do human genetics, functional genomics and machine learning and marry them and push them to state-of-the-art.

And we think we're going to see tons and tons of very interesting signals that could ultimately result in genetically validated targets, not just in the cell, but we can then can go back and take those findings and test and see, is that true in humans by these data sets that we're collaborating with the organizations on.

So I love this quote: "Artificial intelligence is the new electricity and is changing industry after industry." This is from Andrew Ng at Stanford.

And I think while it can potentially change R&D and pharma in lots of different ways, this is one of the most exciting.

We can probably turn medicine and science from really a biology and a clinically oriented field and really push it towards a -- being a data-driven, almost a data science, if you will, for higher-quality targets, faster development, better success rates.

The third technology is -- that I wanted to highlight today, and one that I think reflects both the potential for being very disruptive to our industry and [now looks], in some respects that the way antibodies were to small molecules, is a new technology that allows you to target things that you couldn't have targeted before and have effects that you hadn't been able to see with other sort of modalities is cell therapy.

We talked a little bit about CAR Ts.

Our approach has been to use T-cells and use genetic editing to introduce a new T-cell receptor that would have higher affinity and be more specific to the cancer.

And I'll give you an example of that in a second.

And I'm going to not go into all of the specific manufacturing components of this, which are quite complex and actually very expensive.

But we have actually made a lot of advances and think we can do this a lot better than lots, maybe everyone.

And Tony, if you have a question, Tony Wood will be addressing that.

This is our first program, it's called NY ESO because the T-cell receptor that's genetically introduced has a recognition site for a protein fragment that's presented on a tumor cell.

And it's, in fact, presented classically on sarcoma cells, but it's also present in a lot of different types of cells, myeloma, lung cancer, non-small-cell lung cancer and a few other ones.

And we've affinity-matured these.

It's not -- really affinity-enhanced, I should say, so that they recognize the NY ESO significantly more potently than our own T-cells.

So we're making them super T-cells.

They're recognized at almost picomolar versus 10 micromolar.

Very weak versus very -- 10 micromolars are very weak binding, and picomolars are very tight binding.

And we think that as well is revving them up in different ways.

When we reintroduce the cells, we take them from the patients, reengineer them, throw them back up, submit them back to the patients, and they can have a very profound antitumor effect.

And what we've seen with this collaboration with Adaptimmune, who did the program, that when we give these cells back to patients, after they've been engineered, we're seeing clearly dramatic effects.

Now you might think, okay, this is sarcoma.

Wait until we get the lung cancer data, something else.

And we'll be very excited when we see responses -- if we see responses in lung cancer.

But I think the important thing to see here is that CAR T therapy, the other cell therapy that's been so exciting to the world, because there's been cures in lots of liquid tumors, this is the first example of a cell therapy for solid tumors that CAR Ts don't seem to be working in solid tumors.

And this is probably an example of why we think this could be a very disruptive technology.

And if we start seeing responses in other tumors, there's no reason to think that this might not happen with other targets.

And we have 5 targets with Adaptimmune, 4 others.

And maybe a wealth of other targets emerge should this therapy become as impactful as we think it could be, with this data being sort of an example.

And when you look at the response rates I've shown over here, when you look at overall survival, and you compare it to what -- these cross-trial comparisons.

But when you compare it to what's existing, this looks very, very active.

So the last thing I want to talk about, before we get to the culture part, is that all of the portfolio can be always augmented through external innovation.

Only a very small fraction of the biology and the innovation that occurs in the life sciences are going to come from within GSK.

There's a wealth of opportunities, both in academia and other industry partners, to do business development.

And Kevin is -- who was introduced earlier, comes with a wealth of experience, and is going to really be forming a business development strategy that will be an important part of our growth of the pipeline.

And it will be focused on immunology, focused on genetically validated targets, focused on finding platforms and technologies that augment what I've just described and also, importantly, to free up resources.

Sometimes, we'll have maybe targets, medicines, potential medicines that are better off in other companies, as was mentioned earlier.

And we can use those to free up resources, to generate opportunities that we could apply other places, but also importantly to make sure these medicines get to patients in an effective way.

So I want to end with culture.

Someone sent me a little quote last night, what did it say, something like culture eats strategy for breakfast, something like that.

And the point is, is that culture is really an important part of making this work.

And in some respects, as exciting and as challenging as it is to deliver, to execute on the science and technology, having a culture of innovation is incredibly important, and one that I'm really excited about focusing on.

We've divided it into 5 different areas, really following the science.

And by that, I really mean -- and I want to highlight this piece -- is that sometimes, particularly with genetically validated targets, but also in immunology, we need to make sure when the science sort of speaks and says, "Look, I'm a target, and I'm supposed to go into neurodegeneration," that we don't say, "That's too bad.

We're a company that focuses only on these 3 areas, and we'll force you to look like you work in one of those."

So what I mean by follow the science is to do so in research, as you're discovering targets and trying to figure out what a disease is, in a therapeutically agnostic way.

Follow the science.

If it tells you to go into an area, don't force it to try to work where it's not supposed to.

And we have to make sure that when we do that, we're not taking molecules into diseases where there's not very much unmet medical need, that's not commercially viable.

But we need to make sure we're not pushing it into places where the science tells us it won't work.

Probably the most exciting area for me in culture is ensuring we have a culture where we're incentivizing smart risk.

And by that, I mean making people feel appreciated for making courageous decisions, and taking risk and not fearing failure.

And I'll talk about that in a second.

The third is something that I've basically grown in -- up in, the biotech culture of having single people identified to make decisions, people who have the context needed and the skill set needed and the courage really to make the right decision, and not simply take a vote, not simply to see what everybody can live with because consensus will get everybody happy.

There'll be nobody who disagrees or at least disagrees violently.

But it's almost rarely -- it's rarely, I should say, the best answer.

And it's -- I think it's part of a culture that drives innovation, is feeling that you can be bold and be courageous and be rewarded for taking smart decisions even when they're wrong.

Focus, focus, focus.

We're not going to be successful if we don't identify those projects that are most likely to work and fund them aggressively at the expense of the ones that probably won't work or clearly won't work.

It sounds simple.

It doesn't always happen.

And lastly, a bit cliché, but really takes outstanding people in this environment with the right tools and resources to really drive innovation.

And we're going to demand that we have these outstanding people.

We're going to make sure we develop them, and do everything we can to retain them because outstanding talent attracts outstanding talent.

Let me quickly go through this grid.

I've shown this grid, I don't know, 30 times in my -- when giving talks.

It's easy to show, a little harder to operationalize, but you'll get the concept.

I divide decisions that people make into either good or bad decisions and then either right or wrong.

In other words, is there a good outcome or a bad outcome?

And then, of course, you get a 2-by-2 grid with 4 different options.

Good decision that was right, we don't have to talk about that.

Everybody gets happy about that.

But as we celebrate a bad decision that was wrong, that's not good.

At best, that's a learning opportunity.

I want to make sure you've got the people in the right roles.

The 2 boxes that you can't get wrong for an innovative culture are what you do with a bad decision that had a good outcome, that's called lucky.

Do not reward lucky, because what you're doing is you're telling people that we only care about the outcome.

And if you just reward luck, I can guarantee you, over the long term, that's not going to work.

Luck is not a good strategy.

But you laugh.

But people reward luck all the time, okay?

Now even more challenging sometimes, what if you make a really good decision that's wrong, that has a bad outcome?

We need to celebrate that as much.

Otherwise, we're going to teach people, only make those decisions that work.

So what do you do?

You incentivize a very conservative sort of nature.

And over time, that's not going to be terribly innovative.

In fact, I would argue that over the long term, that's going to destroy a company.

So you need to make sure that you put incentives and reward people when they make good decisions even when they're wrong.

If you think about it, if we came up with this great strategy that was going to double the probability of success, and I told you my first 3 failed, I would say, "Well, 80% failure rate.

The first 3 just by -- that's not inconsistent with a 20% success rate, 0 out of 3." We have to celebrate that if that's a good decision and not say, "Ooh, I wonder if we got it wrong.

You didn't have enough data." This is probably still a good decision.

But a lot of people are rewarded 0 out of 3, people in my job, 0 out of some number.

You're not going to be in your job very long.

What you really have to ask yourself, is this a good decision?

Focus.

And I love focus.

I think this is critical.

I love these 2 quotes.

David Packard, who founded Hewlett-Packard, spent a lot of time [in the barrier] advising companies.

And he said this to all of them: More organizations die of indigestion than starvation.

And you might even think that's intuitive.

But again, these companies believe not only that, that wasn't true, but believe the opposite, that they would likely die of starvation.

And they eat too much.

They have too many projects.

And that's why they die because none of them are adequately resourced.

And one of the things I heard over and over, when I met with the folks in R&D, was that despite spending a lot of money, we did not have many teams saying I'm adequately resourced.

And I think that what we did was we didn't get rid of the least likely to work, to fund aggressively the ones that are most likely to work.

I like how Steve Jobs said it.

I'm as proud of many of the things we haven't done as the things we have done.

Innovation is saying no to a thousand things.

And supposedly, in Apple, there's a little sign that says, simplify, simplify, simplify, with the last 2 simplifies crossed out.

And I think this is the kind of culture that I would love to have at GSK, and R&D in particular, so that we can really incentivize people to focus.

Now we have been focusing.

As Emma said, the -- when you saw the numbers earlier, R&D spend is down.

And that's not surprising.

We have made 65 decisions to terminate partner or divest programs since April 2017.

42 programs were in the clinical phase, and the remainder were preclinical.

There was more than 400 FTEs freed up to be able to work on programs, like I said, with BCMA and GM-CSF, and other programs that we want to push aggressively to do this.

And we're going to be continuously looking at the portfolio to find opportunities to see things that are working and aggressively move them forward, and things that are less likely to work, and removing them.

This is the pipeline.

There's a lot of upcoming milestones that will inform our progress.

Again, my commitment to you is that every 6 months or so, we'll be very transparent about what decisions we've made, what progress we've made, what things haven't worked for efficacy or safety or whatever that are being removed.

Sometimes we can tell you right away.

Sometimes, we have to wait for the data to mature and be presented at a meeting.

But we're going to be much more transparent about this, so that you can actually see, is this strategy bringing value as measured by the kind of assets that you see in the pipeline?

So the new approach, really, will go from an organization that we believe was spread quite thin across many different programs.

A consensus-driven making organization and an organization where R&D and commercial are a bit siloed, and where we have limited business development activity, to an organization where we aggressively back the best potential medicines, removing those that don't look promising.

Create a culture of accountability, where smart risk-taking and courageous decisions are made by individuals.

And that's not to say we won't have teams, but teams will have leaders, who are accountable for making a decision.

And those decisions are rewarded when they're smart risks.

We'll have robust governance models with scientific peer review, commercial input.

Emma mentioned that [Luke] and I -- one of the fun things about my job is I get to work very closely with Luke . We've reorganized the portfolio investment board.

We've got really terrific analytics that help us make good decisions from Kate Priestman's group.

And I think we're together seeing how to optimize the portfolio because that's how you help the most number of patients and provide shareholder value.

And of course, you heard that we're going to be investing significantly in business development to further optimize the portfolio.

Where and what depending on data readouts.

So again, the strategy, science, the 2 areas that we talked about, seeking to understand how the immune system is causing dysfunction, has been, is currently and likely will be in the future, a very important area of science that we think we can identify some important targets that will have broad implication for a large number of diseases as well as using human genetics to really redefine how we identify targets and pursue them.

I talked about all the advantages there.

And really leverage technology, the 3 bets we're making, and we might have some more.

These might not be the right ones.

But right now, we think this is really clear that functional genomics, machine learning, and this focus on cells as medicines are 3 technologies that we want to push to the bleeding edge or beyond.

And of course, create this culture that I just described, so that we get higher-quality targets with higher success rates, so we benefit more patients and have more medicines out there.

Faster development.

I think the LRRK2 example will be not unique.

And more life cycle options through lots of different strategies I just talked about to ultimately have very, very transformative therapies like -- so that maybe some of the diseases that we start tackling will be like what HIV is today.

So thank you, and we'll, I guess, go to question-and-answer.

So do we want to bring up the people who are going to answer questions?

Yes.

(inaudible) The R&D people.

Thank you.

Okay.

A very distinguished crowd.

Rather than introduce everybody, I think you can see their names.

I will take the questions, and then dish them off most likely to somebody who's a lot smarter than me.

Andrew?

It's Andrew Baum from Citi.

When I look at 2 recent R&D turnarounds, Merck U.S. and AstraZeneca, one thing they had in common was speed, but also the depth of the upgrades in the leaders in that organization.

And on the commercial side, Emma has highlighted the replacements of the top 200 leaders, I think, and what percentage she's done.

So my question to you is, how many of your key appointments or reports inside GSK since you joined have been replaced, and if you'd give us a sense of how many of the replacements have come from internal versus external?

So that's the first question.

The second question is on the extent to which, rather than waiting for it discover new targets and drugs and bring them into clinic, which is great, and I understand where you're going with that.

But where do you have drugs right now which are promising mechanistically, but are actually sitting in the wrong indication, which are commercially unattractive?

And I guess, what I'm thinking about is potential for repurposing your CXCR2, your kinase inhibitor and so on and so forth.

So are those potential targets repurposing in other indications?

Should we expect to see that in the near future?

Yes.

Great.

Let me write these down as we get them.

The first question is about the team.

So I think 50% of the people on my leadership team are new as of the last 9 months.

There might -- that number might be too high.

It might be too low.

But that's the number.

I think the vast majority, I'm trying to look around, I think some of them were internal.

I think many -- the head of regulatory hasn't joined yet, but she's from BI, Sabine Luik.

Tony is from Pfizer.

We've got -- Kevin, obviously, is brand new.

Who else am I forgetting?

Anybody?

Well, Kate is from within, but in her job for what, 9 months as the Chief Medical Officer.

So this is, I wouldn't say, a brand-new team, but about as new a team as I've had in a long time.

So I'm -- that's just a fact.

I think there's about at least 15% of the officers, the vice presidents and above, in R&D that are new.

I don't actually have the breakdown of how many from outside are within.

And I think that number might evolve as we see areas that we need to invest more heavily in and get more talent from the outside.

I know Axel can comment on this.

We're really aggressively looking for additional talent in Oncology to rebuild an organization that's going to be incredibly important.

Some of the talent left when -- in 2014 when we had the agreement with Novartis.

So I can turn it over to Axel to comment in a second.

But I think we have a pretty new team and, I think, a very talented team.

Axel, do you want to comment at all on the -- how many of your folks are in oncology?

Yes.

So we're in the middle of hiring new talent to the organization.

That was a natural need after the Novartis transaction.

So we lost a lot of talent, particularly in development, and actually also on the commercial side, just to be clear about that.

So we're committed to rebuild this.

Some talent has come onboard already from other prestigious pharma organizations.

So we're attracting again from larger players, but we're not where we need to be yet.

So there's a lot more to be done, but I see the trend going in the right direction.

So your second question was about repurposing.

And I was trying to get across 2 things about the value of human genetics.

And one of them is that I think you're right that, sometimes, we may have taken molecules and sort of pushed them towards a disease because of where they were therapeutically discovered.

And had we had maybe a disease-agnostic approach, even further complemented by human genetics, that might not have been the first indication.

I hope in the future, we don't call that repurposing, but the first purpose.

And I think there was examples of that.

You mentioned CXCR2, which is being developed in pulmonary disease.

But there's data suggesting it might be active in preclinical models in oncology.

And I know Axel is -- and his team are looking at this carefully in terms of some preclinical models.

Let's see, another good example might be RIP1, where it's being developed in psoriasis, RA and inflammatory bowel disease.

But we're realizing that there might be opportunities outside in neuro, an area that we weren't interested in as much.

But that has been changed.

And we have a small molecule that we're now aggressively trying to move forward that's penetrant into the brain, which is -- requires a new molecule, but we have that.

And John or others can talk about that.

But that, hopefully, will be able to be developed, I think, as an example of repurposings, of what you're talking about.

And I think there'll be more of that.

Benlysta is a little bit like that.

We're doing a combination with Rituxan for lupus.

I wouldn't call it repurposing it, but I might have imagined that, that should have been a life cycle opportunity we did before now, earlier on.

So being more aggressive about where the molecules should go, and be quick to decide that.

And also aggressively moving them forward when we see activity is going to be a hallmark, I think, of the new development organization now.

Yes?

I don't have a microphone but I'll just talk loud.

The 23andMe database can also be very useful, particularly for PheWAS studies because we have 1.5 billion data points on hundreds of different diseases.

So we look at snips in a gene, and we see what diseases associate with that gene.

So hopefully, we'll take a look at the GSK molecules in development and in the pipeline, and see if we can generate some ideas for either repurposing or just broadening indications.

Yes, exactly.

Good.

How about the next question?

Yes.

Maybe some other, I think, really good examples on a marketed asset, which is Nucala IL-5, which that team did a very nice job to expand the potential value by looking at other indications, including EGPA and nasal polyposis and, most importantly, COPD as well.

So the thinking is there.

We just need to embed it more.

And now we've got a great tool with 23andMe.

I agree.

And we just saw an example.

I won't tell you which one it is, because we're -- a PheWAS identified a disease we're not in, where it looks very promising.

It's just 2 days ago or maybe was that yesterday.

And that looked like a real opportunity to maybe even move up the development time line, so that, that could be even the first approval.

So there's -- we're definitely thinking along those lines.

Let me get to the next question.

And I'm sorry, I don't know everybody's name.

It's Richard Parkes from Deutsche Bank.

I've got one on R&D productivity and one product-specific question.

So I think you noted that you were quoted as saying that you love large cap pharma in the article last night.

But it feels like a lot of the innovation at the moment is coming from smaller companies, and it feels also like large cap pharma has lost some of its leverage, given that a lot of the opportunities are very concentrated in terms of patient population.

So you don't need the deep pockets of pharma either from kind of development spend or sales and marketing infrastructure.

So that seems to be leading to companies having to pay very high prices to access some of the next-generation technologies.

So I just wondered, do you agree with that statement?

Or do you think large cap pharma brings something else into collaborations that's still going to give them that leverage?

And do you feel that you've been given enough in terms of a business development budget to continue to access enough external innovation to be competitive?

So that's the first question.

I've got a follow-up on a product but maybe take them one by one.

Well, let me answer the first one, and others can chime if you have thoughts.

Having spent time at a small company and time at -- what was a small company who got bigger, and then got bought by big pharma, the reason I said I like the idea of big pharma is that I do think that there's an enormous amount of value a small, innovative biotech company can get out of a collaboration.

And I think the 23andMe-GSK collaboration is a perfect example of where both groups are going to win enormously and, most importantly, the patients that are going to benefit from that because, yes, Richard can build fermenters and make antibodies and do all of the stuff or we can really -- he could really triple down on doing human genetics well.

And if you have to be a biotech and do all of the stuff that distracts you from what you're really good at and why people wanted to work with you, you will be distracted if you can't -- now you can make a phone call and have a collaboration and have someone allow you to now start seeing opportunities in small molecules, which maybe 23andMe would have passed on because you can't do everything, or other targets that were just beyond the scope of something they could think about.

So big pharma can bring a lot to that opportunity of turning targets into medicines.

I don't think they all have enough money.

They have enough money to do some of the things they want to do, but they can -- but when they're successful, they're going to need to grow.

And I think that the opportunities that we can have by working with these really innovative people can allow us to get more innovative, too.

So one comment on talent.

I mean, it's very important to have outstanding internal talent like we do.

But I'm really excited to get to work with people like Richard and his team.

I know a few of the team members, and they're outstanding.

And so the idea of having a collaboration with a biotech, I think when you have the collaboration ideally formed -- I'm not saying it's like they're part of GSK, but it's analogous to that, in which we're all going to be a big team doing human genetics.

And I'd like to see us do more of the innovative deals with small biotechs, who could really benefit from what we bring, and we could benefit from what they bring.

And I think you'll be seeing more of those.

Okay.

And a follow-up is on the BCMA program, which you've obviously highlighted for accelerated investment and moving into earlier lines of therapy.

So we're seeing very impressive response rates in the refractory setting, but the other notable thing about that clinical data was the ocular toxicity.

So I'm just wondering how -- as that data's evolving, how confident you are that you're not seeing any cumulative toxicity in that setting that might become more important as you move into earlier lines of therapy with patients on [trip] for much longer periods of time.

Axel, you just want to just take that directly?

Yes.

So it's an important observation.

This is a unique type of toxicity.

What we have seen so far is it's mostly low-grade.

It's very well manageable, with simple measures like steroid eyedrops, and potentially those modifications for individual patients.

So we're managing the toxicity well.

And it becomes a matter of education of the treating physician and the patient, as it's given -- or as the agent is given to more and more patients.

So as you go into earlier lines of therapy, you're right.

Toxicity is more important in earlier lines.

Nevertheless, those patients are used to receiving 3 or 4 drug regimens with more severe toxicities than the mild ocular toxicities that is a characteristic for the BCMA agent.

So we believe, ultimately, if we manage this well, it should not be a problem.

And it has been -- this is a view shared by many of the treating doctors in the first trial.

And that's what market research tells us.

So we're pretty confident we can manage this.

And one other tidbit here that's important is that when you look at what -- how could toxicities with other drugs could enhance the toxicity with our drug when you start combining them, which is what we're just beginning to do, we don't see synergistic toxicities so far.

There's mechanistically no reason to believe that the MMAF, which is the conjugate to the antibody, when that goes into the eye or into the cornea, it's a diffusion mechanism.

It's not driven by BCMA.

There's no BCMA in the eye.

And it disappears.

It gets washed out again relatively quickly.

So we don't see any other drug that's being used in myeloma at the moment that actually would enhance that kind of mechanism.

So, so far, we're okay.

But the proof, of course, is always in the data.

So we'll do the trials that may get several of those begin this year or then more advanced studies next year, where we will be able to answer this definitively.

So I guess, who has the microphone?

Yes, question?

It's Ben Yeoh at RBC Asset Management.

I have a question on the culture piece.

Large organizations particularly, we've seen it's very difficult to do cultural change.

So I was kind of wondering what are the kind of metrics that you're looking about, which might be showing you that the cultural change is on track, and if there's any early signs if what you're doing has some traction?

And so the subpart to that because, how do you incentivize people to say no?

It's a kind of almost mindset thing.

It's quite a difficult thing to kind of get right in the process.

And we hear about it somewhat, and we see organizations who do it well, but turning an organization to think about that is quite a challenge.

Yes.

Terrific question.

So first of all, just backing up, we - I -- all believe strongly that we need metrics to figure out how we're doing.

So that if we're not doing as well as we'd like to, we can figure out what's missing.

The metric specifically on culture is hard to imagine how we could tackle it at all.

But one simple way is we do -- we have HR surveys to see how engaged people are, how well they think we're doing at decision-making, how fast we are at moving things forward.

A number of different questions.

And we've been benchmarking for different reasons, but we've been benchmarking this over the last couple of years.

Others who know more could comment.

And so I would expect that we can look at those questions and pull out prospectively those that we think are going to reflect the cultural change and use them, to some extent, at a very high level, but as a metric.

The second one is just asking people to do these things I don't personally think is going to work.

I think you actually have to ensure that you have things in place that make it easy to do that, in fact, that incentivize you to do that.

For example, if we have a metric called put 8 targets into Phase I next year, we will have at least probably 8 targets in Phase I. Now that is a progression-seeking culture rather than one that's going to necessarily be incentivized to kill things that deserve -- that don't deserve to move forward.

So when we set goals, I think we need to be mindful of goals that incentivize what we want.

For example, and I don't mean that we should do this for every goal, but certain goals, I think, should say at the end of Phase II -- I'll make this part up, at the end of Phase II, we should take no more than 3 weeks to assess the quality of the data, its impact, and whether we want to move forward.

Now that might be really easy because it's negative or it might be really hard because it's gray or it may be really easy because it's super positive.

But incentivizing to do things quickly rather than incentivizing to make it go to Phase III -- because if we incentivize to make it go to Phase III, we're not telling people to look carefully at the data and do whatever you think is right.

There's a subtle bias towards wanting to progress things.

So I think it comes in measurable from goals -- I mean, measurable from the surveys, but also creating goals that will incentivize that, and doing lots of talking with employees and seeing how it's going.

And I'm sure we'll come up with other metrics that we think are useful in assessing this.

I don't know if others on the panel have thoughts they want to add or if any people who know more about this than me want to chime in?

Yes?

Oh, sorry.

I keep calling (inaudible) [without a mic.] There you go.

So it's Michael from -- Mike Leuchten from UBS.

I've just got one.

So a question, Hal.

Thank you for outlining the gestalt of where you want to take the business in terms of decision-making and culture, but can you outline an example so far where you have applied that framework decision versus outcome, and it actually has resulted in, are we actually going to kick this asset out?

So what I didn't see in your presentation, in your slides is this was the wrong way to think about it, and we're actually going to reprioritize.

Or is it just too early to talk about that now?

And then the follow-up question, I was just wondering if Richard was going to add on to partnering large cap versus biotech, whether you had any additional thoughts on that.

Good.

I've got to be careful.

So I think that we have had a couple of programs that we've reviewed that I think we'll be stopping as an example of courageous difficult decisions.

Some we've partnered -- you've heard about the many partnered assets.

We're in the midst of doing this.

So it's a tiny bit early, but not that early.

I think you'll be hearing about it very soon, programs that we think from a data-driven perspective don't fit, from a value perspective don't fit, so that we can do more.

I think that there's probably been 2 examples in the other direction where there was some debate as to how aggressively we should pursue it.

And BCMA was the one that we just, Luc and I sat down with the Portfolio Investment Committee and said, "Let's go." And the teams were excited.

GM-CSF, I think, is another example of let's be aggressive.

The other few that I can't really comment on right now, but you'll be hearing about where we'll be stopping stuff.

John, you've been involved in this for -- you probably have some examples that you think are good examples of what he's asking for.

Well, I think they're embedded in the 65 you referred to in the slide, where, clearly, we had to make stop decisions taking into account all the information.

Some of them were very far along, some of them we've extracted the value.

As a [pin or off] example, the Orchard divestitures, the -- I'd say it's a cultural element, too, for really getting all of the teams to be thinking how they articulate the value and understand that it's in a portfolio context when we have these discussions.

And I think we're making significant progress on that.

And I wanted to tie it back to the last question about how to incentivize people to be comfortable with no.

And one of the incentives is that when your project is yes, it's going to be adequately resourced.

And it's something Hal talked about before.

So I think over time, people will get comfortable, okay, I don't have to push my thing forward all the time, and I'll have confidence when my thing is going forward that it'll be adequately resourced to win.

So it's a cultural change, and I think it's progressing.

Richard, do you want to comment on the...

Sure.

At 23andMe, we weren't really necessarily looking for a partner at this point in time.

But as Hal and I often talk to each other, after his move to GSK we continued to talk.

And as we talked, it just became clear that GSK could bring this, 23andMe could bring that, back and forth, back and forth, such that the collaboration made so much sense for both parties that it was just clear that this was the way for us at the time to proceed.

So I think Hal gave the example.

And we're very good at 23andMe in making antibodies.

A number of people from my previous life have moved to 23andMe, and we're terrific at that.

We find probably more targets that are small molecule targets rather more innovative modalities that 23andMe -- sorry, that GSK works on that we just didn't have the wherewithal to pursue at the time.

So by working closely with GSK, we can potentially start some of those projects, manufacturing antibodies.

For goodness sakes, of course, 23andMe has no ability to manufacture antibodies.

Of course, you can outsource that to China, to Lanza, to other places, but it made sense to work with a partner that has those manufacturing capabilities.

So I don't want to line up all the pluses and minuses for you here.

But just to sort of reiterate, we weren't looking for a partner.

But as we discussed it, it just made so much sense that, that was clearly the way to go.

Thank you.

Okay.

At the back.

It's Keyur Parekh, Goldman.

And 2 big picture questions for you.

The first one is on kind of genetics and 23andMe.

We've seen some of the other people in the industry, including your ex employers, kind of like big investments in foundation medicine, [flatiron health], blah, blah, blah.

You guys have had this collaboration with Biobank for the last kind of 3 years.

And I think there's -- it's unclear what benefits those have actually driven.

So can you help us think about what is it that we should be looking at externally as to, have you made the right decision by going down this road?

What are going to be your benchmarks?

And what are you going to do different to make this collaboration successful?

That's question number one.

And I'll wait for question number two later on.

So I think it's a great point.

So as I said, for culture, metric's important.

For the science and technology, these metrics are going to be equally important.

I think there's lots of ways to look at this.

I would -- I think that the reason -- the timing is just ideal for a collaboration like this.

And we're seeing this, and Richard, you should comment on this, that there's a nonlinear scale phenomenon to the size of a database and its ability to identify targets.

There's been many studies where they do the GWAS , and there's no hits.

And then that's 25,000 people, and you say, "Hm, that's weird.

It seems very genetic because you have heritability studies, and yet no GWAS hits." And then you do 50,000, no hits.

But by the time you get to 100,000 patients, you go from 0, 0, 0, and then all of a sudden, you've got 30 hits.

And it comes down to the sample size and the power.

And so I think that now we're actually having a scale -- taking advantage of the scale.

And I think what we're going to see, but again, Richard should comment, that when we've gone to 500,000 with UK Biobank, a lot of things came up.

But 5 million, I think we're going to be able to see a lot of things that we couldn't, including gene-gene interactions, which I personally believe are going to be -- there's going to be a lot of nuggets there.

And you just have no power for gene-gene interactions, gene modifier the way Richard said it.

And so I think that size times, again, the technology, functional genomics, the ability to tile through the genome, and maybe Tony can talk about this, with TALENs or gene-by-gene in the CRISPRa or CRISPRi up down the protein, is just going to give us an enormous wealth of information that, frankly, even 3 years ago, we wouldn't have known what to do with had we had it.

But again, with machine learning, I think we're just going to be able to put this whole thing together in a way that allows us to see things we just couldn't see 3 or 4 years ago.

And -- but the question is, okay, you said that already.

How are we going to know?

I think it's about the targets we enter into the clinic, the pipeline and it's the value and the quality of the targets and the stories that end up getting told about the value.

So LRRK2 is a great example.

What's the next one?

How do we do this?

And I think you should be looking for more of those targets entering into the clinic eventually as a metric.

And if it's not coming in any faster than it was or all of this isn't translating into the kind of value that I've described, then I would say it's failing.

On the other hand, if it is, I think that's a good sign that we're onto something really important.

Tony, do you want to talk about functional genomics?

Is that -- would that -- just maybe a sense or 2 on how you see this field building up.

Let me just add a point about our early experiences both with TALEN, which you can think about as the precision editor, if you like.

And we've been working with the Altius Institute to really understand what is required to get that precision right and, more importantly, as well to have an analytical cycle so that you know you've achieved what you want to achieve.

We've got early experience through the Open Targets collaboration, doing sort of large-scale synthetic lethal screen that Hal mentioned with CRISPR.

So I think one of the things that I'm really excited about here is we can bring insights from both the precision world and the large-scale world, and I think we're now at a point where we can begin to properly industrialize those approaches and fundamentally change the early shape of our portfolio so that we can have larger numbers of targets interrogated in this way and make better decisions as we narrow down and then focus our resources.

And one other metric that we've talked about internally that I forgot to mention is kind of the quality of the collaborations that we form.

If the world sees us as compelling, some of the best and brightest in these areas are going to want to work with us.

They're going to see the value, too.

And if everyone says, "nah, I don't know if it's ready.

I don't think that's going to work," that could, again, be a metric.

So I think you should judge us on the collaborations and how business development is able to augment the strategy.

Are you impressed or not?

I think that's -- we're going to put a high bar, but I think that'll be another way of evaluating us.

Just a final comment.

I think that Hal hit on it.

It's really scale.

And with the over 5 million customers, and I'm not supposed to say this, but Emma did, soon to be 10 million, so I guess we have to deliver on that.

We've been able to see associations that other people haven't been able to find.

There have been many, many attempts to find the genetic basis or components of the genetic basis for depression in Europeans that have all failed.

But with 200,000 people in our database with depression, we were able to publish the first associations, and it's all because of scale.

And this was with another pharmaceutical company, but all of those studies going forward will be done with GSK.

My second question was some of us had the experience of listening to some of your predecessors talk 3 years back about a lot of targets.

And as you spoke about a bunch of things that seemed interesting to you about half of those were highlighted in 2015 as well and a bunch of those were meant to be filed or have proof of concept between 2018 and 2020 and now seem to have pushed out by the last -- by at least 2 to 3 years, including kind of ICOS, OX-40, really impressing targets.

So with a not -- with the history that you mentioned, just help us think about what's gone wrong the last couple of years.

And again, as you change the organization culturally, what will you do to prevent that from happening again?

Let me throw out 2 ideas and maybe see if Axel, having lived it, wants to throw in a third.

I think you -- first of all, I think you're right that the time lines have slipped.

And I think if you look back at why things are taking longer than they should, the so-called white space, the diagnosis -- at least 1 diagnosis, I think, that is reasonably straightforward to make, it's not the teams and it's not even the execution.

It's the governance.

From the time data was available to the time a governance body actually made a decision sometimes was, frankly, shockingly long.

And I think that was in part driven by a little -- a lack of clarity and who's actually accountable for things.

Teams get confused, and that's why I think -- I feel so strongly about ensuring, particularly at the portfolio investment board, that when we see data, we can move because we understand who's making decisions and move.

So I think that's one component.

I've seen a couple of examples where we fixed that by having these [accountabilities].

We would see that and the next day we'd say go.

And I don't think that was common before, including things that required -- and this is the second piece, doing things where you're really not being democratic, meaning there's 10 molecules and we're going to give all of the money to 2 and not any to the other 8. In the past, I think there was more of a sense of being more equitable, and I think that's not the best way of doing it.

And so I think part of that was driven by the lack of the best analytics, to be able to figure out which of the 2 we really want to back, having an interface between R&D and commercial to be resolute about which ones to back.

Frankly, if I was to say what's the biggest change in the last 6 months, I think we've got that going well.

I think we have analytics.

I think we have that R&D-commercial interface.

I think we see data, we move.

I think, frankly, we've moved, in my experience, shockingly fast.

In fact, the single accountable decision-making, we made a commitment to, what was it, like 3 months from the idea when we said, "Let's maybe do a deal with 23andMe." I think we said 3 months, right?

And everyone said, "Well, that seems impossible." I mean something -- you even said you were dealing with a smaller company and it was -- well, maybe I shouldn't have said that.

But it's -- sometimes, it takes a long time, and we said, "We're going to do it." And in fact, I think even our own teams were like shocked, wow, we actually could do this.

And I think it was driven by clear accountability.

So I really do think that's a serious part of what we're changing, and I think it's taking hold.

The specific assets -- sometimes, things are delayed because of data.

Sometimes, studies are less feasible, but I do think that's the significant driver.

Axel, do you want to -- since 2 or 3 of the molecules that were named were in Onc, do you have other thoughts as to what might be -- have slowed us down in the past?

Yes.

There is, of course, the root causes for things, and culture is at the heart of it, as we have already identified.

The 2 things that I have personally experienced, one is governance and the way governance runs.

And we had several delays in some of our oncology molecules because of governance.

And I'm not going into depth of what happened, but that's just at the heart of some of the issues.

And then another is setting realistic expectations.

And we're equally responsible in terms of expectations from the company side, but also from the analyst side because we're being pushed sometimes to deliver something faster than is reasonable because, otherwise, it's not competitive.

And the best example is probably in the agonist antibody space.

After PD-1 success, everybody expected that the next checkpoint antibody will be equally successful like PD-1.

If it isn't, it's not interesting.

And we are now dealing with different biology.

ICOS, OX-40 are completely different biology.

You cannot expect a PD-1-like effect.

And we have seen OX-40 discontinuations because they did not deliver PD-1-like effects, and it doesn't account for the biology or the ability to deliver clinical benefit at all.

So if you see some changes in time lines, in part, they come from reactivity to what has happened in the space.

And in part, they come from us internally adjusting our own expectations.

But in principle, we've been moving quite quickly on a variety of things.

If you think about it, the oncology portfolio is entirely homegrown, with few exceptions like the Adaptimmune deal, was an early adoption of a technology that the rest of the industry did not pay attention to.

CAR T -- CD19 CAR Ts were attractive in cell therapy.

TCRTs were not yet of interest even though those are the entry point for solid tumor therapy.

And we have access to that by making an investment of double-digit millions as compared to an acquisition of double-digit billions to get access to a CD19 CAR T, which is much smaller in terms of commercial value.

So it all depends on how you look at things.

But I'm pretty convinced that with new leadership, with new governance, we can actually accelerate things even faster.

And the BCMA story is probably a good example for that.

And let me just make a commitment because I think it's a great opportunity to sort of highlight what I was trying to get across in terms of transparency.

I don't think you should have to wait a few years to ask that question.

I think that we'll be doing these frequently enough that when things change, I'll say, "Well, nothing went wrong.

We just had the wrong date" or "Here's why it's delayed.

It turns out it's more complicated because of the dosing or because of the biomarkers" or whatever it is and just be transparent about it.

And you may think we're doing a bad job or you may think we're doing a great job, but it will be transparent.

And I think that's what we're hoping to do by having more frequent interactions with you like this, where we can show you, "Look, here's what we showed last time.

Here's what's moved, here's why" and get input.

Okay.

I think we're taking the next question from over the phone.

Steve Scala from Cowen is on the line.

Okay.

So I have a couple of questions.

First is a follow-up to the last question.

Can you compare and contrast at a practical level the use of genomics within GSK to that of Roche?

And what are the advantages and disadvantages of each approach?

For instance, Roche argues that they have scale as well, which was one of the advantages you just cited.

And secondly, this is a longer-looking question.

But when he was head of SmithKline Beecham research in the early 1990s, George Poste made a then pioneering investment in HGS to specifically harness genomics to -- and now I'm reading a 1995 note -- develop new therapeutics, diagnostics and gene therapy vaccines.

30 years later, the world has made huge advances, and GSK really hasn't in this regard.

What conclusion should we draw from this, if any?

Is the answer all culture?

Or is there something else at work?

Let me take the second one first.

And you would think that between me and Richard, we should probably be able to answer the first one, but I'm not going to.

I'm not allowed to.

Yes.

So I think it's really interesting.

And I didn't know about George's comments, and I think they're phenomenally interesting.

And I thought a lot about it.

And I really do think the answer -- I think he was ahead of his time.

I think the answer is that -- and this was not intuitive at the time because we weren't sure how frequent these polymorphic variants were going to be in the population, and Richard, you could probably comment on this more, that if it turns out that common variants, things that occur in 20%, 30% of us, really explain the disease, you're going to see it with a relatively small sample size.

If it turns out that the rare variants or very rare variants are your clues, you're just not going to see it until you get to scale.

And I think what was -- some people in retrospect were saying, "I told you it was going to be rare." But the conventional wisdom at the time, if you go back to the literature, was it was going to be the common variants that were going to explain the answer, and that's not true.

It's not the common variants.

I mean, there are some examples.

But the vast majority of targets, genetically identified, genomically identified targets, I think, are coming from the more rare variants and sometimes, very rare.

And I think that's what explains why it's taken so long.

I don't actually think that it was culture, there's a bit of technology that was needed to get more variants.

I mean, Rich, we -- you measure 700,000 SNPs or whatever, but you can impute 25 million now.

There's a whole bunch of much more sophisticated things you can do.

There's way more datasets.

There's deep phenotyping with UK Biobank that's very complementary.

You have machine learning, functional genomics, I just think, that, that wasn't available when George was very prophetically sort of predicting this would matter.

But I think the most likely explanation was, the common variants didn't explain the disease.

Well, I'm old enough to have been around when that statement was made.

I'm a little older than Hal.

And what was happening then was that the set of human genes were being defined, and that's what people were doing when they were sequencing the human genome and sequencing cDNAs.

We were saying, "Oh, here's a new gene." What we've been able to do over the intervening 30 years and what's so exciting about human genetics today is we've known all the genes now for a decade, 15 years.

There's 21,137 plus or minus, but we're able now to know from human genetics what they do, what diseases they're involved in.

And that's new.

That's today.

That's the last few years.

So what George was -- I would say, what George was commenting about was something totally different.

What are the genes?

Today, we're excited about understanding what they do and what diseases they're involved in.

So it's really a statement sort of comparing apples to oranges in my opinion.

It's Sam Fazeli from Bloomberg Intelligence.

Two questions, I think, is the limit.

One is if you look at the idea of looking at immunity or immunology or whichever way you want to go, that can take you into basically every possible disease area, which is, I think, what you've highlighted.

Do we -- should we expect that there'd be a therapeutic focus emerging out of this based on the success of the molecules that come through over 1 year, 2 year, 3 years?

Or will it continue to be, "Whatever is successful, we'll pursue it," be it neurologic, be it bone fracture, be it whatever it is that immunology touches?

So that's question 1. Question 2 is with regards to AI.

It's a -- I come from a company that has been working with AI forever, as long as it's existed in whatever form.

The -- where do you expect -- where do you actually see a real impact in the next few months, year, or 2 years?

Is it in chemistry?

Is it in automation of reactions, in choosing the best reactions and much more (inaudible) -- where is it?

And where do you expect to see a meaningful impact in the R&D process?

Okay.

Let me try and take that.

I'll -- Tony, maybe you can think about -- beyond what I'm going to say about target discovery, where else we see.

GSK has a lot of effort in some of the medicinal chemistry stuff that Tony can talk about.

I think the therapeutically agnostic component of the strategy is in research.

And as the target matures and we get more confident that it's both likely to work, likely to be safe, likely to hit a significant unmet medical need, likely to be commercially viable or whatever, we're advocating for moving it forward.

It might fit nicely into a therapeutic area that we already have expertise in that's represented here, or it might not.

I just want to tell one quick story about a drug that I was privileged to work on actually with -- it's great that Richard's here, which was a discovery in the research labs by Napoleone Ferrara, who discovered VEGF.

And it became clear that protein was involved in the angiogenesis in tumors, and we made an antibody to that to try to be a cancer drug.

But data from humans showed that VEGF was overexpressed in the eyes of patients with diabetic macular edema and diabetic retinopathy.

And there was changes in the eyes that were angiogenic, so it's very likely to be involved.

We didn't have any knowledge of ophthalmology.

We had no history of ophthalmology.

We had nothing.

And people said to me many times, "How did you guys decide strategically to become the #1 ophthalmology company?" That's not how it worked.

We were focused on the science.

And we let it drive us, and we saw a huge unmet medical need and very compelling biology.

So when you think about it that way, it requires a lot of being nimble.

We have to get clinicians, scientists who know how to translate research findings and focus them when things start moving over here and here.

We definitely need to be partnering with commercial so that when we get these assets that they can be commercialized well, and Luke and I have talked about this.

And they rely on big unmet medical needs where the treatment effects are big and the safety profile is robust.

And so the focus part -- because it's really broader research, the way to focus is use these really high-bar criteria and say, 'Look, if it isn't really going to be an innovative therapy and it's outside of what we know, why would we do that?" Now we might have an out-licensing program because somebody else might find that attractive, but -- we're going to have a high bar but follow the science.

Just quickly on AI and then, Tony, jump in.

I personally think that AI can do lots of things.

To me, the question is not what AI could do, it's what is the most important problem for AI to solve because if it solves the problem, it's really not a big problem facing us.

It's interesting, it's kind of cool science, but it's not the fundamental problem that we need solved.

I believe the fundamental problem is getting better targets.

And that's why, to me, leveraging human genetics, functional genomics, this massive dataset to find the best targets is something machine learning is really ideally suited to do, massively, highly dimensional datasets.

But machine learning is effective in a lot of different areas where that's the case.

And we're doing a lot of stuff in medicinal chemistry, and there's lots of other places where there might be shorter wins.

But I'm not so sure that, that alone will be necessary to drive the kind of change that we're talking about.

But Tony, do you want to..

Yes.

So without getting into medicinal chemistry, which is my background, so there's a danger I'll go into a lot of detail there.

I think it is -- in that area, we're about predictive science.

It fundamentally allows us to be more efficient.

It allows us to design with other criteria in mind.

But as Hal indicated, the nature of the data that we're going to be able to generate in functional genomics and the sorts of questions that we can ask there, they're going to be much more impactful.

The rest is about efficiency and about predictive design, as I've said, and it's about ensuring that we have the right data types.

And that science isn't new.

We've -- I was playing around with neural networks in the early 1990s in chemistry.

So what you have to bear in mind is that it will help us make better decisions.

But if the process that we then follow with that is another [rate 71 like] synthesis, where, again, we can apply the same techniques, what we're doing is simply becoming more efficient.

Fundamentally, to me, it's about transforming through making better choices on targets in the first instance.

And then, of course, the opportunity from functional genomic ideas, to go back to Richard and ask him to help us find qualification within human datasets through 23andMe, creates a virtuous circle of high-scale testing, and then validation hopefully in the real world in the disease state.

Good question.

Graham Parry from Bank of America Merrill Lynch.

So first question on BCMA and your second-line studies.

Just could you run through the rationale for not including Darzalex combination in there?

You've referred to it as being in combination with standard of care, but arguably Darzalex is now standard of care there.

Do you think not having that in the arm could actually hamper recruitment?

And then secondly, on Slide 43, you've run through a lot of potential data readouts between now and 2020.

I was interested in particular in the number of proof-of-concept readouts that you still have coming.

And of those, could you highlight anywhere you feel you might have a faster market strategy or where you can go straight to pivotal trials without having -- necessarily having to do further clarification Phase II trials?

Axel, do you want to take the BCMA Darzalex question?

Yes.

So it's actually a mixed bag.

So we do have Darza combinations in the second-line strategy, there's one.

And there's one that does not contain Darza.

And the reason we have done that is simply based on the evolution of the space.

The multiple myeloma space is moving very quickly.

Darzalex combinations are being -- moving up from later lines to earlier lines, and it's different combinations.

So you have pomalidomide as a combination partner in the later line.

You have lenalidomide in second line right now.

In all likelihood, lenalidomide would end up in first line.

And then every -- if a patient receives it in the first line, you're unlikely to give it again in second line.

You will seek out a combination that uses something else.

So we are accounting largely for the dynamics in the space with the strategy that we're currently running, but based in a way around Darzalex.

So we will certainly have Darzalex combinations, meaning working BCMA versus Darzalex combinations.

We'll have that in every line of the strategy, so -- Darzalex, pomalidomide, dexamethasone versus BCMA, pomalidomide, dexamethasone is a second line component of the strategy.

So Darzalex is in there.

And then there's one other important thing to be said.

This space is fast-moving, yes, but it also is -- hasn't delivered yet these massively transformational effects for patients that we have been hoping for and have seen, in non-small cell lung cancer, for example, with, in some ways, curative therapies with immuno modulators.

So what we are looking for is to actually drive this a little bit more aggressively and take a little bit more risk, but also then potentially transform the treatment landscape.

And we're doing that in a three-step approach.

Fast-to-market with monotherapy, that's something you're familiar with.

Darzalex did the same.

A second line strategy to bring the standard of care combinations in the mix, that's basically what I just said.

And then a third step, which goes into first line, will include some standard of care but will also include some combinations with novel agents, immune modulators, eventually our own or that of partners, to completely transform the treatment landscape.

So the 3 steps together could maximize the value of what BCMA can deliver to patients and what it can actually then also do for GSK.

Thank you.

So let me go to your second question.

Is that forward?

No, what number is the slide you...

Slide 43.

43.

How come that says 43?

44.

44, no.

They're numbered funny.

What do I add?

There, so I'm going to give you a few minutes to prepare.

I'm going to give 2. We didn't practice this question, but it's a great question.

I'm going to pick 2. Guys, if you don't like my 2, you can push back.

I'm going to -- because I think that part of your question is essentially, where do you see in Phase I the potential for data that you go 'wow,' right?

So I think there's not many diseases where you can do that.

The 2 places I think you can do that is cancer.

And the question -- and frankly, any of the cancer things can go from being not so 'wow' to 'wow' pretty quickly.

But the one I want to pick out is NY ESO, and I'll explain why.

So maybe you can get ready.

And the other one -- and it's a little bit out there, so -- is I want to say something about HPV ASO because what's really interesting sometimes about doing drug development in virology and you know well, when your endpoint is measured in long reduction of something, you can see things pretty quickly, right?

The reason oncology is kind of cool is because you're getting a volume reduction, which is really pretty interesting.

Sometimes you can get things like MDR and surrogates of tumor clearance.

But the antisense program with our collaborators has the potential.

I mean, it's very -- you could argue how likely it is, but one could imagine that things could come out of that, that make you say 'wow.' And the NY ESO, you'd say, well, don't you have data?

I think if we see data in lung cancer, for instance, where the drug works in lung cancer, again, it's -- these are very high bars, but that's where you say wow.

Okay, it's a solid tumor sarcoma.

Okay, I think I know what to do there.

But if you have responses in the second tumor type, particularly lung, and there's also studies in myeloma, and there's combination stuff, I think that can be -- now you didn't ask how likely is that, but those are 2 where that could happen.

So maybe -- do you want to just briefly tell us what about -- what that study is about and how that's working, the HPV?

The HPV?

Yes.

So it's -- these are antisense oligos.

It's a partnership with IONIS.

We've got 2 different ones.

One is a direct targeting antisense.

The other one is a prodrug that actually allows a lower dose, so it's administered to the liver in a lower dose and gets converted to the active agent.

And we're looking for evidence of suppression of viral antigens over time.

And that is, as you said, a surrogate marker for the potential for a cure.

And the idea is that this would be a treatment not just for, like, the nucleoside analogues that are used in those patients that suppress viral replication but also to reduce the production of the antigen in the liver, which is thought to be the most oncogenic activity.

It's a pretty big opportunity worldwide for sure, but it's also a reasonable opportunity in the U.S. And [Luc] could comment on that if you'd like.

And the study is designed to get a high confidence in the ability to have a good rate of functional cure from that, and we have an option decision next year based on the data.

Do you agree with NY ESO as having that potential and if so could you just expand on that?

Yes.

So NY ESO is actually somewhat of a unique case because it has multiple potential effects on our strategy.

The first one is it's the anchor asset, in the clinic at least, for our cell and gene therapy effort in oncology.

And our focus here is on solid tumors.

So number one, we want to expand what has already been done with CD19 CAR Ts in liquid tumors into the much larger unmet need in solid tumors.

PCRs allow you to access targets that are inside the cell, not just on the surface of the cell, as such they get you much more opportunity to go into solid tumors.

And they're already in the first generation of engineering, clinically active, as you have seen.

The CAR Ts took 3 generations of engineering before they were really clinically useful.

So having said that, the way we run the program is we focus on the small indication for entry, that's sarcoma, expand into larger indications, which is multiple myeloma and non-small cell lung cancer.

In myeloma, we already have data.

In lung cancer, we're just really gearing up towards that, and then we go into this space of combinations.

So there's a PD-1 combo study running in myeloma.

There's a PD-1 combo study planned for non-small cell lung cancer, which is obvious as it's very dominant there.

And then there is another aspect of this, which is very different than any other modality you can use to treat disease, and that is you can reengineer cells.

So that means if putting the T cell receptor into the cell is not enough, and we're already seeing clinical activity that's quite profound, with that single step you can engineer the cell further.

If you would have a small molecule and you don't get the effect you want, you need to make a new one.

Here, we take the same cell that they already made genetic change.

You add another change.

Let's say we put a cytokine gene into the cell, we knock out PD-1, we do whatever makes biologic sense, and you can enhance the potential of the medicine.

So cell and gene therapy is really something where we have a lot more opportunities and NY ESO is our anchor.

We get in with that.

We will expand to other targets.

We will bring in other molecules as we go along.

And more importantly, we mature the platform.

And it goes back to Tony's job, the platform we approach differently than other pharmas have done.

So we spend a little bit more time getting manufacturing ready for getting these products to patients.

So you don't have the scalability problems that you get when you put a product into patients quick, but you're not yet caught up with CMC.

So we're not having that problem.

We're not as fast as we would like to be, but we will actually be commercially ready with a commercial manufacturing process at the time when our pivotal trials start, and that's critical.

That's completely the opposite from anything else that you have seen in the industry so far.

Is that -- Graham, is that good?

Emmanuel Papadakis from Barclays.

Hal, you spent a lot of time on the discovery piece of the R&D equation.

To what extent should we think that will translate into an uptick on the discovery part of the R&D budget?

And I guess I think the next question is to what extent do you avoid unnecessary duplication of the huge amount genomic phenomic work going on in the academic realm of the scientific world?

And then it was just a follow-up on NY ESO.

You seem, Hal, to express some ambiguous confidence about the ability to target solid tumors earlier.

I mean, maybe Axel just answered it, but I was going to ask you to elaborate on that a bit, your degree of confidence that it will translate?

And if not -- if so, why have you not been more aggressive to accelerate the clinical development program more broadly because it's felt a little bit slow?

I think I'll ask Richard to comment on the academic overlap.

I'm actually not sure how much this will impact the research component, the discovery component of the budget.

I think probably the way I'm thinking about this is that all -- a lot of this effort will be at the -- we'll be able to fund this through things we won't be doing.

I think where this is going to have a bigger impact is on the development side when things start working.

But it's early days, and it'll be data driven.

When I said -- if I led you to believe that I didn't think it was likely to work in solid tumors, that's wrong.

It does work in solid tumors.

We have the sarcoma.

What I said was we don't have data outside of that solid tumor.

And I'm extremely pessimistic, if you will, that CAR Ts will work in solid tumors, so maybe that got confused.

I don't think CAR Ts, at least in their current format are likely to work in solid tumors.

That's why we are so excited about the opportunity for T cells in solid tumors, and it works in sarcoma, which is a solid tumor.

There'll be, I think, a significant inflection if we can find a second...

It's more a question of translation from sarcomas into other solid tumors, it seems to me.

Yes.

No, I mean, I didn't mean to be pessimistic at all.

It's just we don't have data, and you -- it'd be wonderful.

Should it work?

Yes, it should.

If it doesn't, one could imagine we need to develop better diagnostics to identify where NY ESO is overexpressed, and we're working on that.

That actually isn't as straightforward a problem as you might expect, and then finding out which tumor types has the most people with overexpressed antigen, to identify them so you can enroll them in trials.

But I actually think I'm optimistic that we'll be able to find other tumors, solid tumors where NY ESO works, particularly when you think about how the [re-engine] in the cells in the second and third generations really got this down.

It's a bit of a longer view, I think, to be excited about a disruptive technology.

This is not something that's, we're done, let's move on.

This is going to be a complicated but potentially very disruptive concept.

I'll add one more thing just to make it easier to follow.

Synovial sarcoma is a subtype of sarcoma.

That's the data you just saw on the slide.

That's where we have the first, what we believe, transformational response rate of about 50%.

This year at ASCO, when the program was still with Adaptimmune, they reported on a different histology still within sarcoma called [mixed-rate lung cell] liposarcoma.

Same response rate basically, like we saw in synovial sarcoma.

So that's a different type of tumor.

It's in the same realm of sarcoma but different histology.

So if you can go from sarcoma 1 to sarcoma 2, maybe you can also go to other solid tumors.

It makes sense.

So we're pushing towards that.

But your point is well taken.

We still have to deliver the data.

So with respect to overlap between 23andMe and the academic community, and I would say other large genetic programs, there is some overlap.

But as Hal mentioned, largely in the more frequent, the more abundant SNPs.

Since our database is so much larger than any other database, we're able to see less common SNPs that other folks just aren't powered to see.

And I think one of the -- I think perhaps the greatest telling thing about our data compared to the academic world is we have hundreds of requests to collaborate with academia.

We've published over 100 papers, most of which are in collaboration with academia.

So the academics are actually vying to work with us because of, mostly, the size of our database but also the great variety of phenotypic information that we have.

So yes, we will continue those academic collaborations where it makes competitive sense to do so.

Let me just add one more point and then we'll go to the last question.

I'm very excited actually about the complementarity of some of the other efforts we have ongoing, in particular the UK Biobank.

Because, while it doesn't have the same size, as I mentioned earlier, it's so deeply phenotyped.

It's so rich that I think the 2 together actually are way bigger than the sum of the parts.

And I think we'll hopefully be able to show that.

So one last question.

[Jane Wilson] from Crédit Suisse.

I'm going to go back to this slide that you have here.

You've told us a lot about how you're reengineering the early stage of research and some of the failings and what you're going to do to deal with it.

If I look at the comments that the AdCom documents from [Ukala where] yesterday, they showed some failings or at least they perceived some failings in your clinical trial design.

Maybe you could comment on that?

But I'm more interested in your proof-of-concept studies here.

You've got 6 due in the first half of next year, 6 in the second half, 5 or 6 anyway.

Do you think we should look for an industry-standard success rate there?

You must have looked through these programs.

And with your fresh eyes, do you think that they're correctly powered to give you the decision points that you need to move quickly?

Or were they half-assed studies that are going to give you something where you have to redo them?

That I think is what Graham was getting at, so that we can see -- I think we all understand you're going to have fantastic science in 2021, 2022.

But some of us investors want to see real progress in the shorter term and your analysis of how good those programs are and what sort of success rate and progression rate we should look for from that would be very helpful.

That's a great question.

It was hard, and you made it even harder.

But I'll give you my very transparent answer.

I think some of the studies here, if they have a huge treatment effect, will emerge.

I think the current way of designing some of these trials was that some of them didn't have as much robust power to -- as needed to see a signal.

Now the danger in that is twofold.

One is that you can be fooled by a trend which isn't real, and you advance them.

I think that's a worse problem to have than deciding that the trend is probably negative and discarding an agent that might have been active.

And I say that because I think very Bayesian, the prior for any molecule entering the clinic working is only 10%, so I don't have a very high probability.

If it doesn't look like it worked in Phase II, I'm not -- it's not like it had a 90% chance of working.

It had 10%.

So when it looks worse than I expected, I think it's important to kill it.

But I think moving forward, we need to make sure that we power the Phase II studies to be able to see whether it works so we don't advance things into Phase IIb and certainly in Phase III where we don't have the confidence that we know that we should have.

So I think some of these are powered, but there are definitely ones that I'm worried that the approach had been to be a little bit too democratic and give enough money to all of them to make it go forward without powering the ones we are most excited about.

And so I think it's a little of both, but there's no doubt there are some that are probably underpowered.

And again, I don't think that was a lack of insight on how to design trials.

I think it was going back to this, have we really thought carefully about how to really aggressively pick the right targets and fund those really well, at the expense of a couple others that might be 0, rather than everybody getting 80% of what they need, because that can be lost.

And so I wouldn't be surprised if some of these might be underpowered.

John, I don't know if you want to add anything to that.

Well, maybe it's just a corollary to your last comment, which is I think we can reassure you very strongly that we're not going to progress them just to progress them.

It's an obvious statement, but we're going to look at them with a very critical eye and the order of magnitude of change indicated by the study is going to have to be consistent with what we think is going to be a commercially important and clinically important effect.

So if they're not good enough, we're not going to go forward with them.

And the ones that we are confident, we're going to put more and more resource behind, so that, I think, you'll see as a significant change in way of working.

And I do think a metric going forward is that our Phase II proof-of-concept is designed to actually ensure we have clarity in the right decision moving forward, and that that's going to require that they're adequately powered and have more money associated with the studies to do that at the expense of other programs that we just don't think are as valuable.

And that's the kind of smart risk-taking, courageous decision that I think is fundamental to making that happen.

So I really appreciate that.

It wraps up a lot of what we're trying to do and say with the culture piece as a way of optimizing development.

I know we've run out of time but we have an opportunity to get together socially after this.

So thank you very much for your time.

It was fun for us and enjoyed hearing your questions.

There will be drinks upstairs with the R&D team.