Session Notes: The R&D Productivity Challenge

[0:16] **Vera Örså**: So good morning everyone and thanks to uh my summits for your kind invitation to speak. I do intend to be very diligent with my time. I now understand I have 20.0 minutes. Let's check. Can you always check it? It's time. five. Yes. The adrenaline is good, this one thing is not this one. This one's a little bit more than a little bit. So we knew how to tap that as I would see. Okay, okay. No worries, no worries. So 20 minutes start from that. I'll be myself honest.

[2:06] **Vera Örså**: So I'd like to talk to you about the RE challenges that we have in Black Party today. Which quite frankly, if I were giving this presentation three decades ago, would not quickly. Ribble topics I'd like to take you through the R and U final. Looking at decades of technological advances and the burgeoning cost of getting to the registrable profit. Chemical proof of concept and area that I work uh within a very long period of time, I simply uh called the clinical value of data. And looking at who was managed to uh proper that chasm.

[2:53] **Vera Örså**: There are going to be case studies, very abbreviated ones of three companies that appear to have done so at least via publication. This is not, in fact, an opinion. It simply uh takes information of the literature. And in fact, lessons should be learned from emerging biofarm companies. Gate in decision making and how poorly we apply.

[3:18] **Vera Örså**: the stage gate process that was discovered by Kump and Hadrick decades ago, and everyone has a gated system, and yet we continue to make a preponderance of type 1 and type 2 errors that is false positives and false negatives, and we hate terminating anything. And then I have some modest recommendations. So let's look at decades of technological refiners. We have not stood still over. the past several decades. If you look at drug discovery, and then in the 80s and 90s, we had the industrialization of screening and molecular biology.

[3:56] **Vera Örså**: I started off my career as a molecular biologist. So glad to see them there. And then in the world of clinical development, we had standardization writings of frameworks. Look at how much we have advanced technologically over the past few decades. Now everything is AI. or every other word is AI. There's automation and new modalities in drug discovery and AI-driven data-centric trials infinitely developed. So we have really moved the very following. One might expect, prima facile, that given this, that we would now be incredibly more productive than we were, let's say, in the HSM 9 types.

[4:37] **Vera Örså**: In fact, on average, and I hate to stress the term unaverage. unaverage or not. I'll get back to the unaverage. There's a slide like this that I created about three decades ago. Not much has changed. I haven't had to alter much. Yes, I do update much on one slide. This one was easy. The only thing that really changes the number of inputs at this level. So on average, it takes about 23 to 36 times. targeted molecules to enter this part of the funnel. And so you folks over here don't feel that we're being ignored.

[5:19] **Vera Örså**: 23 targeted molecules that get you between 8 and 14 pretendable molecules or entrants that result in one or two registrable products. That results in about 170 million dollars worth. of out-of-pocket expenses or over P. Now, what do I mean by out-of-pocket? It's simply cash outlay, adjusted for present value, but has nothing to do with your project failures along the way, nor does it account for the opportunity cost of capital. That, when it is accounted for both project failures or attrition, as well as the opportunity cost of capital, turns into roughly 1.2. 1.9 billion.

[6:07] **Vera Örså**: That's the expected out-of-pound cost. Now, those are my numbers that in fact I generated from literature that are not the same as others like the NASI, slightly higher when they just did for the uh um $2025, not the same as uh Stephen Paul at Lily, and some of you may remember. remember the the absolute brilliant seminal article that he and his colleagues wrote in 2010 that's followed Roger. Uh in fact where his estimate now converged to 2025 model inflated at 2.62 percent annually is roughly 2.6 bills.

[6:53] **Vera Örså**: We've all got different numbers which aren't really easy here. the picture does look rather dry. It says if I spend on an expected or risk-adjusted basis and account for the opportunity to cost the capital, on average, I'll get one product coming out. It's not as dismal as it sounds. In reality, this is what happens. If you start with 23 MMEs, There's a distribution of a number of molecules that you'll get that are entering phase one. And that distribution could be as low as zero, not eight, not twelve, and it could be as high as 19.

[7:50] **Vera Örså**: If you reproduce the simulation, you'll get more or less the same results. Now, if you look at the number of launches. launch products that emerge from the 23 targeted hints, they translated into on average 8 getting into phase 1. There is again a distribution. But let's examine this distribution a little closely. There's about a 38% likelihood of getting exactly one. We shouldn't be interested in that number. should be interested in getting at least one, and that is one minus the likelihood of getting zero. The likelihood of getting zero is 35%.

[8:37] **Vera Örså**: So the averages are what they are, but they can be quite mismired. In fact, I encourage you when you look at these industry estimates to question the average expectancy, ask the following question: what does the actual distribution look like? And in this case, you could get as many as seven, obviously, with very low probability. But the highest likelihood, yes, that's in fact, it's about 37 or 38 percent. But there's a zero likelihood, there's a 35% likelihood that you could get nothing from output, and that is reality.

[9:22] **Vera Örså**: Now, along with my colleague in a book that we'll be publishing in June or July this year, we developed a modified version of the original Stephen Paul uh pipeline proof of the model, where he started the chart of the hit. And it's important that I reference where I got this data from. Ball's data from Junior Henne, the wonderful propagation by Sir Kayak. her colleagues in 2024. So the first set of data comes from Stephen Paul. The clinical data comes from Sir Kaya, and I've adjusted everything to $20, $25.

[10:02] **Vera Örså**: There's a lot of data here, I just need you to focus here. The average out-of-pocket spend or cash out data at one molecule out is about 171 million on an expected basis. it's roughly 1.9 billion. But again, that is on average. What I was most interested in is looking at a sensitivity analysis to each of these individual parameters. So cost and time, you see, cycle time here as well, were adjusted at uh plus. plus or minus 25% of the basal value or base value.

[10:51] **Vera Örså**: Whereas risk or probability of phase-specific success was adjusted by plus or minus 15 absolute percentage points, meaning that this sensitivity analysis, and look at how a good old friend that rises all the way to the top, probability of success in phase two A. which reminds me that I've always complained about these industry analyses that treat phase two as phase two. And I'm of the belief that phase two is not phase two. It's a combination of higher risk phase two A, POC, and lower risk phase two B, those final.

[11:34] **Vera Örså**: And I could have taken what I call an idiot's guide to probability assessment and simply taken. square root of the industry average, but that would not have been a defensible position. Simply because, in my experience, you fail far more often in 2A than you do in 2B. And so I made my own adjustments, made my own assumptions, and so you see what the sensitivity analysis looks like. Note, not surprisingly, that the greatest sensitivity factors all factor associated. to the phase specific probabilities of success. Okay, clinical new scene.

[12:22] **Vera Örså**: This again comes from uh in fact a very nice article from Cert Py novice. And though there's a lot of data, I'll simply ask you to look at the averages. I've also looked at four the therapeutic areas because they represent different extremes of the act. So you interpret this as follows. Starting with non-clinical at a cost of roughly 12 million, the cumulative probability of success, that is to a launch, is 8.5%, and it takes 11.8 million to complete that phase. phase.

[13:07] **Vera Örså**: If you spend another 7.1 in phase 1, you will resolve another 4 percentage points, making it cost equal to 19 million. That is equal to 8.5 to 12.5. That's how you agree this chart. Now let's look at the graphic. This is not clinical, you start at 8.5, you spend 11. And then as you resolve risk. you spend a bit more, or rather, as you spend a bit more, you hope you'll resolve more risk. That is the average of all of the therapeutic areas.

[13:44] **Vera Örså**: That is of less interest to me than looking at individual therapeutic areas among themselves. And only in the interests of time will I move through this as quickly as possible. Look at oncology specifically. This very steep time. indicates that there is a lot of unresolved risk in early clinical development that gets carried forward into later clinical development that is unlike every other therapeutic area that includes mythology, excuse me, anti-infectivists, and CNS. In fact, based on industry averages only, roughly 10% of all poverty projects at the end of phase two.

[14:30] **Vera Örså**: even though right now you know it I don't like the acronym in phase two, only 10% of them actually make it to launch. And if you happen to get to the beginning of phase three, only about 35 or 36 percent of them get through to launch. So though we have made tremendous advances in the uh therapy to gear on quality, it still remains a very high risk theory because of the fact that we do not resolve risk early on. early on. So taking control of your destiny.

[15:02] **Vera Örså**: What have three of the bigger folks done to try to cross that chasm that we call the clinical value of death? Many of you familiar with the work Paul and Owens in 2010 and 2015 when they launched their course program. There's a lot of detail here. So basically trying to front load clinical data. data acquisition to accelerate go-no-go decisions and limit downstream capital exposure with the risk that it could make type 2 beta errors by not first acquiring sufficient information with which to make a no-noble decision. The retrospective evaluation is quite significant.

[15:54] **Vera Örså**: that is over a decade 2002 to 2012. When their PGRS remain constant, that's the probability of success all the way to launch, 6%, over both development paradigms, meaning traditional versus coarse. Reduction in transitions to phase two. So far fewer molecules getting over the phase. phase one earth, getting into phase two. However, they did bring with them much higher quality, leading to a probability of transitional success in phase two of 34%. In fact, a fifth of the percent in corus as opposed to 34% in traditional development.

[16:46] **Vera Örså**: This lean to proof of concept was achieved at a a cost of about 6 million in 2010, as opposed to 22 million in 2010 for conventional approaches. And L2POC was achieved in only six months longer than traditional development. sacrificing long-term success rates. However, based on work by Morgan, who was published in 2012, they published three pillars of survival for improving phase two success. One, exposure to the target site of action over a desired period of time, two, binding to the pharmacological target as expectation.

[17:46] **Vera Örså**: expected for some of action three expression of pharmacological activity commensurate with the demonstrative target exposure and target binding. What did this do for FINSIN? I'm unable to advance so I'll go back and try the line through again and it works good. So a review of 44 NMEs that reach phase two between 05 and 09. Of 14 programs have meant all three pillars with both high confidence in both pharmacology and expoke. 12 of them achieved a positive clinical POC, and eight advanced to phase three. Let's just quickly move to the other extreme.

[18:38] **Vera Örså**: 12 programs that actually failed to make a full definition. of all three pillars, none of them advance to phase three. So there is at least internal evidence that if you were to meet each of these phase three pillars with high confidence, then it would result in a high-likeli and both possible QC, as well as advancing to phase three. Then looking at the 2010 to 2020 period, one Fernandez worked.

[19:10] **Vera Örså**: worked, they separately reported that from a low PTRS of 4% in 2016, Pfizer's PTRS rose to 21% in 2020, and compared to industry averages of 8 and 11% over the same time specifically. This was due to phase two success, which rose from 15% in 2016 to 52% in 2020. That to me is tremendously successful. simple to see. And lastly asked reset. Not three pillars, but five Rs, which are now six Rs. Okay, because a digital R is not enough. Target, right TCU, I safety, and right commercial potential.

[19:59] **Vera Örså**: They did a review of 142 projects that are active between creating a billion PC2 between O5 and 10. And then they ran a comparison between 05 and 010 and 23 and specifically phase two success rates rose from 15 to 43 percent when compared to industry mediums of 22 and 18 percent over the same time.

[20:26] **Vera Örså**: So the primary value of the 5R framework, according to folks at AstraZeneca, is that they were able to to systematically surface the most consequential sources of technical risk so that they could be explicitly tested, mitigated, and used as a basis for early determination. Emerging biopharmaceutical companies, at least some of them appear to have learned from some of their bigger brother. What's the definition? or early commercial enterprise of leverages cutting into biological, chemical, or computational pathogens to generate differentiated therapeutic candidates.

[21:12] **Vera Örså**: And generally an annual RD expenditure of less than 200 million, and some with no revenues at all. Portfolio structure typically concentrated and very asymmetric. Lean and outsource organization. organizational structure. And I have some examples here that were never intended to be collectively exhausted. If you work for your emerging biopharmaceutical company and you're not listed here, that was very intention to exclude you or anyone else. It was simply an intent to give you an education. In 2024, they meaning emerging biopharma companies, found with 85% of the core anemones that were approved in the Urestal level.

[22:08] **Vera Örså**: That to me is significant enough to be occupy no more. So they must be doing something that is atypical of traditional pharmaceutical development, especially by the big flocks first, in order to be able to deliver this type of connectivity. Gate in decision making. Am I keeping myself on time? Absolutely. So minimizing time fundamental type QRs, we love the Cooper and excuse me, Cooper and Edge at stage gate process. If you don't have good governance with disciplined decision makers, this is is no more than a pretty picture that does you nothing.

[22:58] **Vera Örså**: We blame competition, regulatory authorities, and everyone else but ourselves for our inability to make input. In fact, to make effective and efficient decisions. And I met many hugely talented heads of RD who would rather have these projects kill themselves that is harm pay. harm patients before they would proactively try to kill some of these products themselves. So that to me is rather unfortunate. Not having a credible business case.

[23:37] **Vera Örså**: Whenever I go into a client and I see projects in their portfolios, that I give in the question and ask for a business case, and yes, I'll get one. The amount of dust most of these business cases is amazing. They were generated about a decade ago. No one's ever chosen to look at them again. And for some of them, the best reason I can get for why they were in the portfolio is because a while ago someone wrote the first check. And that became the impetus for writing subsequent checks.

[24:11] **Vera Örså**: And nothing to do with the technical competitiveness. Competitiveness or or in fact any commercial attractiveness. A simple business case does not have to represent, I don't buy this uh uh traditional cop colour that says we don't know enough. If you were the CEO of your own company, you would never write a check to someone who simply says, Hey, I'm a nice fellow, trust me. Me being the exception, of course, I am a nice fella and I'd like to trust them. Target product profiles.

[24:53] **Vera Örså**: We're still in an age where we generate a base target product profile against which we do our risk assessments, against which we do our risk evaluations of why. You can't force the drug to demonstrate a certain level of efficacy, safety, or tolerability. So why not very self-affected and ask the question? question? If I got a little more successful, what could my TPP look like? And on the other hand, if it were not as successful as I'd like to be, just how bad could it look? Is it still registerable?

[25:27] **Vera Örså**: And even if it is, is it still commercially viable? And of course, I'd like to see that there's a good relationship between a business case, a business problem, and the portfolio fit that begins with asking the question. question: what are we trying to achieve, and what problem are we trying to solve as opposed to our smart scientists have generated and new entity? So, recommendations. And this is my penultimate slide. The RD funnel, the averages are what they are. The real question is: how are your assets performing in? pursuit of your idiosyncratic disease-specific profiles.

[26:14] **Vera Örså**: These industry averages are no more than a guide. Use them very sparingly and try not to anchor yourself just because there's a large N that tells you there's a big amount of data. Clinical POC does not have to be a fate of pomplay. Both the little guys and the big guys have proven other ones. otherwise. So, risk resolution and clinical development, in my view, is dependent more on your risk appetite than it is on the actual state of the nation. Take control of your own destiny, other people can do it, so can you.

[26:52] **Vera Örså**: You'd have to be a big farmer to do it. And emerging biopharmaceutical companies have little choice than to embrace the downside of risk because it is the upside of risk that keeps. them afloat and throttling. A decision makers requires organizational awareness, readiness, and compliance, and not mere rubber stamping. Of course, you'll all say that doesn't happen in my company. Similar modest recommendations: beware of cell as a snake on that. I truly believe it that in fact algorithms have a purpose. in that they inform decision making but make no decisions for us.

[27:39] **Vera Örså**: And avoiding discipline decision making is simply an act of following. While it may keep you afloat for the short term, you just need to sink sooner or later. If you're patient enough to wait until June or July, you can read more about this in chapter one of the books. I was very uh pleased to recruit my collaborators.

[28:02] **Vera Örså**: in Quata Jose Aureliana who works on the Fair in Australia and he's been a wonderful individual who worked with it and that's me and I wish you well thanks for your time and for the vast majority of you just staying away big question to Richard I think you've answered all the questions, Richard. Or we have a break now till half. Just come to 

... [transcript truncated]