What do we mean by ‘innovation’? The question of how to define and measure productivity in pharma is one that remains open. There have been a number of incredibly exciting drug approvals over the last year, but the expected sales of these drugs are at their lowest, on average, in recent history. So, if we simply take approvals, or new indications for old drugs, we’re missing the definition of innovation, as distinct from invention – innovation is the realisation of value from invention.
Data from IDEA Pharma’s Pharmaceutical Innovation Index show that if you take the industry’s average R&D spend over the last five years and divide by the number of approvals by the top 30 pharma companies, the average cost per approval is around $6 billion, with some companies exceeding this by a significant margin.
The market has changed substantially in the last 20 years. Once, the philosophy was that if you achieved launch, the world would pay for it, but the market is now much, much more intensive in terms of competition. The industry was initially slow to react to this, but is making up lost ground fast.
Given the number of similar assets in development, the key lies in thinking less about the molecule itself, and more about overall path design. Importantly, you have to do all of that design at a point where you can change your direction, change your trajectory, change the indication you’re looking for, change the measurement tools around it, change the service setting – if you don’t do that in early phase development, it’s never going to happen.
At IDEA Pharma, we still see plenty of companies effectively going down a conveyor belt – the narrow design of phase 2 often dictates a linear progression to a similar, scaled-up phase 3. In other words, what goes on one end comes off the other end – but with opportunities to examine different hypotheses often missed. But if you look at it early enough, there are different conveyor belts you could have sent that same molecule down, and maybe you want to send it down several in parallel. It’s typically hard to change things beyond phase 2: it’s a big risk at that juncture to go off on a tangent.
This is not an easy problem to fix. The various late pre-clinical models that companies are using often aren’t sufficiently predictive of what you see in man, and the challenges of improving this translational strategy are often substantial: some will never be fixed. You need to look for more detail early on in clinical development: more animal model work often adds little to determining whether to take an asset forward, and less the best approach to clinical development.
A great example of this is in immuno-oncology, where rodent models are only a modest surrogate for the human immune system: a significant amount of data generated on PD-1/PD-L1 inhibitors in rodents was poorly predictive for man: both false positives and negatives were observed. Another good example is in CNS development, where models are poor or absent for a number of conditions, and of course where higher cerebral function is often absent.
Much has happened in recent years to build the value of early efficacy studies in man, to not only gain more detailed observations, but also consider a wider range of hypotheses. Although this change has enveloped the industry, it has been slower to start in some companies than others. The key is putting those assets into early-phase development in a way that gets you the right kinds of signals to make decisions about A) whether you can go forward at all (early ‘kill’ decisions), and B) if you are going to go forward, what is the selectivity of the patients you’re going to go after? In other words, what can the drug deliver selectively within the right patient population? Although this approach can increase the costs in early development, there is evidence that it can not only reduce costly late-stage failure rates, but also deliver assets to market with more competitive profiles.
For example in CNS drug development, where assets may have a number of different potential benefit profiles, more enlightened companies have realised that it is hard, based on pre-clinical investigation, to establish the breadth of potential opportunity: you need to have a more open-minded view of examining a range of possibilities in early stage clinical development. If you ask more of the questions of what the drug could deliver in earlier phase through parallel clinical studies, and really consider the options, you can then start to generate a range of potential signals, as well as eliminating some of the possibilities. This saves significant time in testing the hypotheses in parallel rather than series.
We’ve encountered this when working with one organisation, recently. In the past they tended to pursue a single hypothesis ‘to the death’ rather than looking at the molecule more broadly at the beginning. We worked with them on one asset that they were taking into a new phase 2 trial after observing an interesting signal in an earlier, failed phase 2 study. In effect, it was pure serendipity that this signal was observed: it wasn’t part of the planned hypothesis testing.
Interestingly, the pre-clinical team had knowledge that supported this serendipitous finding: knowledge that wasn’t acted on as part of the early-phase clinical programme. This is an example of where a single pre-clinical hypothesis was pursued ‘to the death’, rather than considering the range of possibilities the mechanism could deliver. This early ‘funnelling’ – which occurred in later pre-clinical development and was followed through into early-phase clinical development – led to a narrowing of thinking that ultimately proved costly both in time and trials expenditure.
An antithesis of that is Merck and its PD-L1 monoclonal antibody Keytruda. Roger Perlmutter, the company’s head of R&D, was presented with a range of options for early-phase development. Bear in mind, this is a mechanism that could have been brought into quite a few different tumour types. When he was asked which should be taken forward, he said: All of them. The insight was that translational science would not answer the question of likelihood of success in clinical studies, given the limited value of animal models. He realised that some ‘spread betting’ was needed to establish clinical value: a more traditional, sequential approach to development was going to burn time and competitive advantage: what was needed was parallel hypothesis testing, and to utilise the power of testing in multiple tumour types to try and examine where the clinical benefit of PD-L1 inhibition lay.
His argument was that the only way we were going to know whether it could do anything across this range of indications was to just get on with it. And he was exactly right. Almost all of those signals turned out to be positive. Some of them weren’t, but they learned a huge amount along the way about what was going to work and what wasn’t. As a result, that propelled them forward and allowed them to start gaining on BMS’ competition in a way that they never would have done if they hadn’t made that decision.
As the old adage suggests, in order of priority, a right answer comes first, a wrong answer second, and no answer is the worst of all. However, companies that wait until a ‘right answer’ presents itself are instead engaged in the ‘no answer’ process.
One of the problems in the industry is that a lot of people believe you can just keep drugs alive for long enough and good things will happen. Sometimes that works out, but there are many more cases where drugs are kept alive as zombies in pipelines for a long time, heading towards a cliff edge because there’s not going to be anything successful to come out of them.
It may be a joke that large companies like their disruption to be gentle, and top down. For larger companies, however, this kind of disruption often needs to come from within their own organisation. Johnson and Johnson’s JLABS and the J&J Innovation Group are a good example of the ‘deliberate disruption’, and AstraZeneca’s ‘5R’s’ framework would be an example of process improvement, instead of disruption.
Some years ago, AZ was in need of change, in terms of its late phase pipeline. They were progressing a lot of medicines into late stage development that were never going to succeed in the end game of getting registration and then, of course, being commercially interesting.
Under Mene Pangalos, AZ’s executive vice-president, R&D, AstraZeneca put in place a framework called the ‘Five R’s’ – Right target, Right tissue, Right patient, Right safety and Right commercial potential. This design thinking approach, trying to address all these goals early on, means that every drug they investigate has a much higher chance of future on-market success, whereas before they were doing relatively small, early-phase studies, getting pretty weak signals and ad hoc guessing where to go next. This has been a major driver of the turnaround in their business.
AstraZeneca have realised that value comes from getting better insight into whether a molecule they are taking forward is worth something to the people they might be giving it to. Now they have a greater focus on creating meaningful medicines rather than just something that works well enough to stay alive in a pipeline.
A lot of pharma companies keep changing the people at the top, but not really changing the approach. That doesn’t help. The companies that will be successful in the future are the ones that are prepared to think about a different way of doing what they do – to embrace agility, adaptability and serendipity within Development. In addition, the history of the Pharmaceutical Innovation Index has been that companies that are closer to their market tend to do better than companies with a single approach.
Because of this, it’s unlikely that true innovation in R&D will come from larger companies.
Ten years ago, no one really knew who Gilead were, yet now they are regarded as a large player in the industry, and the most innovative company on the 2019 Pharmaceutical Innovation Index because of the success they have been able to generate from a game-changing approach to antivirals. The problem that they have is turning from that specialty into a broad player, because their assumption that you can take an antiviral company and turn it into an oncology company, for example, hasn’t worked.
What we’re seeing is a rise of purposeful, individual companies that are smaller and who can out-compete large companies. Start-ups getting to market isn’t actually as hard as it used to be because a lot more is outsourced – such as clinical development, manufacturing, and sales and marketing.
The role of the large pharma company has to be redefined in order for them to stay relevant. Buying talent and assets is one way of doing it to keep the pipeline stocked. The question of whether large pharma will add value to those assets was the salient question that we dug into with the Innovation Index, because you get a lot of companies in the situation of having a product at the beginning of phase 2 and wondering how to make the best of it.
Often small companies actually know their market better than larger companies do. That might sound paradoxical, but what you find in large companies is that they start to understand markets via market research (second-hand knowledge vs. the first-hand knowledge that typically informs biotech) and then they appoint market research people to talk to market research agencies, to talk to people that respond to market research surveys about what markets are.
The questions are usually meaningless, so you end up with very little market insight as you get bigger, while some of the smaller players know what matters at a gut feeling level. Many start-ups began with a bio-scientist who deeply understood where that product would work and where it wouldn’t, and was able to keep it focused and small. They’re agile, and some of that agility is about their deeper understanding of their unmet need that their drugs could be addressing, and being able to pivot if necessary. That will always be potent.
We don’t want an industry full of me-too drugs that offer a 5% improvement on what was there for the last 10 years – we want drugs coming through that are cures for conditions that need them. Look at the mess around Alzheimer’s disease, which is just people hitting their heads against the same brick wall time-after-time, hoping somehow their head’s going to be better at hitting the brick wall than someone else’s was.
You’ve got to change the approach – go around it, use a different tool, do whatever you can to make a difference. At the end of the day, the reason that we all do this is to make a difference to people’s lives, and the lessons of innovation show that new approaches to Development will yield more success than hoping for better molecules, or novel insights into biology.
The positive news is that good, deliberative human decision-making remains the differentiator between innovative companies and less successful companies – and that is changeable for those who want to learn those lessons.