Asset-centric is a term coined by Medicxi founding partner Francesco De Rubertis to describe an investment strategy he pioneered, together with Kevin Johnson and Michele Ollier, in the early years of the 21st Century. The key was focusing investment on a single asset to avoid the problem with pipelines in early-stage biotech companies. While it seemed superficially attractive to hedge against failure of the lead asset by having a second or a third program in the wings, the reality usually turned out differently: when the lead failed and the pivot came, far too frequently the follow-on assets didn’t pass scrutiny. “In short, we ended up developing things we would never have chosen to invest in had they been a lead asset” De Rubertis explains.
Over the next few years, the practice of asset-centric investing was honed through various iterations all designed to counter the inherent progression bias that damages returns in pharmaceutical R&D. Biology is unavoidably complex and always provides a plausible explanation for any failure, allowing managers with a natural aversion to crystalising losses to keep trying. Maybe it was the wrong patient population? The wrong dose? Not long enough treatment? The wrong end-point? Of course, Occam’s razor suggests the simplest explanation is likely to be right: the drug just doesn’t work well enough – but many millions are often burnt before the inevitable conclusion is reached.
One of the principal causes of progression bias is the creation of infrastructure – if you have a team of employees, a lease on a building and so forth it can be painful to call a halt even if the enthusiasm is draining away. As long as there is a plausible route forward, and there nearly always is, then further investment become almost inevitable – a phenomenon often described as “the hand in the mangle” (referencing a long-defunct apparatus comprised of two rollers used to squeeze water from clothes during laundry, which, if it caught your hand, it was almost impossible to escape from without a painful injury).
Asset-centric investment used out-sourced infrastructure to reduce the drag of progression-seeking behaviours, and over a decade delivered the ultimate asset-centric company: the zero-person biotech. Capital was dripped in slowly, so none was “trapped” inside a company that had begun to deviate from a stellar trajectory. Over the years, Medicxi switched the progression heuristic from “nothing has yet killed the opportunity” to “we are still excited and want to further back the opportunity”. It sounds subtle, but the impact is enormous.
Other benefits became apparent, as much by chance as by design. For example, out-sourcing (at least if done well) can cut costs, even while the owner of the infrastructure earns a return on their capital – because it makes the cost of each additional piece of data explicit to the people running the program. With an internal team, it is easy to keep people busy generating information that’s “nice to have”; but when you see the direct cost of that knowledge in black and white, quite often it seems less valuable. Some exquisite Monte Carlo modelling proved that, in a low-validity environment such as early-stage drug discovery and development, it rarely pays to buy more information.
Buoyed by the string of successes in early discovery delivered by this approach, attention inevitably turned to how we might leverage the same insights to improve return on capital in the rest of the long path through drug development. As a drug progress through development, costs rise exponentially so unlocking improved efficiency in later stages of the process would have an even bigger impact.
But how to proceed? The asset-centric “playbook” that worked so well in the early stages of discovery and development are not obviously transferable: virtual teams and drip-feeding capital seem more likely to hurt than to help later stage development.
Look a little deeper, however, and it quickly becomes clear that techniques such as virtualisation and highly-tranched financing are only the “how” of asset-centricity, and not the “why”. To really understand how asset-centric thinking can revolutionise later-stage development in the same way it has for discovery, we needed to return to the principles: what is the problem we are trying to solve? Doing that allowed the creation of new tools that are appropriate to deploy much later in the pharmaceutical R&D process.
Asset-centricity works because it minimises the impact of unintended biases on the key decision-making framework. For example, a company that owns its own discovery infrastructure bears an unintended bias to continue work on an asset when the excitement has drained away. Employing a large management team means there are people whose livelihoods depends on the continued progress of the company, however unpromising that may look. Having a lot of capital in the bank enables further expensive attempts to solve a problem even after the prospect of success has fallen below any objective threshold for positive returns.
Separating infrastructure from assets, virtualisation, and the “zero-person” biotech company are therefore all stage-appropriate tools to minimise these undesirable biases towards progression of assets that would otherwise be unattractive. DrugBaron has written extensively about progression bias before, and the importance of creating decision-making frameworks that do as much as possible to limit its malign influence. In effect, these tools are trying to ensure that the decision what to do next is dominated by the data specific to the particular asset, and not on “irrelevant” factors unrelated to it.
Most bad decisions (and the worst are spending more money on assets that will eventually fail) usually trace their origins to the dominance of global considerations (how much money we have, how to keep expensive infrastructure deployed, avoiding painful downsizing and so forth) over local ones (the data indicate that the asset it less exciting than we previously thought). And such is the complexity of drug development (and indeed biology generally) that it doesn’t take much effort for management to “spin” the available data to match the chosen narrative. Examples, such as post-hoc analyses of failed trials, are today too numerous to even be notorious.
The spirit of asset-centricity is therefore no more complex than creating decision frameworks that are data-driven and insulated as far as possible from unintended biases.
Learning to translate these insights to the capital-intensive later stage pharma R&D process, therefore, requires an understanding of the strongest biases that blinker data-driven decision-making, recognising that they are likely quite different to those that bedevilled early-stage decisions.
DrugBaron has identified two important ones, though there are likely many others. Uncovering and minimising the impact of unintended biases is not a once-and-done intervention, but an evolving journey. The key to success is being on the plane.
The first killer bias is “the story”. This is the unifying narrative that companies, particularly biotechs, tend to weave around their assets to make them palatable to investors. These come in many flavours: it could be a technology thread (editing DNA, protein degraders or engineered stem cells for example), deep expertise in a particular indication (“we are the leaders in endometriosis”), or focus on a particular biological mechanism (antigen-presenting cells, mitophagy, caspases or a million other possibilities). It is hard to think of a 21stCentury biotech that doesn’t have one – and doesn’t leverage the story as hard as possible to build a positive momentum behind its assets.
But the price of doing so is much less obvious than the benefit. The “story” links the assets of the company, and delivers “read through” from other events either within or most likely outside the company. When Amgen shows positive data with its Ras inhibitor, Mirati benefits too. All boats rise together. Even negative data elsewhere can help, as a competitor stumbles a laggard can suddenly find themselves in the lead. But “stories” used in this way create an insidious bias on the key decision-making: if the data on my lead asset suggests it should not progress, killing it may be interpreted as a stain on the whole approach. What the wonderful story has done is create a hidden bias towards progression.
Having a strong story has other negative impacts: you may ignore better opportunities that do not fit within its boundaries. Worse, you may undervalue data that is inconsistent with the narrative. Protecting and polishing the story becomes the dominant incentive to management. Before long, the story rather than the product candidates themselves have become the principal “asset” of the company.
The solution is easy: isolate the assets. If you want to access benefits of scale at the operational level, it’s fine to have multiple assets but they need to be selected purely on the basis of data about each asset separately. While the operations may benefit from being united, the assets themselves will not. In such a setting, decisions about what to do with one asset are isolated as far as possible from decisions about what to do with any others.
In practice, this looks much more like a traditional pharma company than a biotech. Global pharma companies, at least until recently, didn’t have “stories” linking assets but an uncorrelated pipeline of product candidates chosen for their individual value.
Yet pharma companies have their own source of bias often less relevant in biotechs: the presence, and indeed importance, of their commercial infrastructure. Pharma companies, unlike most biotechs, actually sell something – and they are usually phenomenally good at it. Selling drugs, rather than discovering or developing them, is arguably the core competence of the biggest and best in our industry.
But this creates the second important bias DrugBaron identified: much like the negative impact of owning the infrastructure to discover drugs has on early-stage decision-making, so too does the owning commercial infrastructure have on later stage decision-making. In the past decade or so this has manifested itself in even the largest companies declaring their allegiance to a limited number of indications. There are so many synergies in the commercial engine from selling different products to the same people, that the “gravitational pull” from the commercial hub starts to dominate R&D decisions.
If you need to feed a world-class oncology sales force, what do you need? New oncology products. Your R&D team risk choosing to buy or progress inferior assets, contrary to the data about those agents, in order to meet that imperative. Once again those key decisions that are the foundation of value-creation are polluted by, and in the end dominated by, factors other than the data on the prospects of the specific asset.
Again, there is a simple solution: separate the commercial organisation entirely from R&D. If that commercial company merely buys approved products to feed its sales pipeline, it is once again making key decisions for the right reasons. But if, as many do, they try and discover and develop them they risk creating the dreaded progression bias.
So much for the theory. In 2020, Medicxi put this thinking into practice through the creation of Centessa – a biotech company designed from the ground up to enshrine these fundamental principles that underpin asset-centric thinking to deliver data-driven decision-making that is, as far as practical, shielded from external biases. Six months after raising a $250m Series A and merging single-asset companies to create a pipeline with at least 13 product candidates, it completed an IPO, filling a war chest to fund data-driven drug development. By October 2022, DrugBaron had joined as Chief Innovation Officer, with the challenging remit of delivering data-driven drug development in practice.
Will it work? Its still much too early to write the analysis, but the early signs are encouraging. Most obviously, Centessa has already announced the termination of several programs – many of which would likely still be consuming precious resources (both capital and attention of talented scientists and managers) in other companies. These assets had not “failed” and it would be all too easy for a progression bias to have kept them alive had they been linked to a narrative or designed to feed a commercial powerhouse. But the data suggested they were no longer as exciting as we had hoped, and crucially the team was able to listen to the data and focus the capital where the returns are likely to be larger.
Asset-centricity has therefore evolved. Today it is defined not by the tools for its implementation (which are very specific to the particular application), but by its fundamental principle: freeing decision-making so that it depends on the factors you want to dominate that decision – a principle DrugBaron has called #Unshackled
#Unshackled thinking doesn’t just apply to pharma R&D – it applies to decision-making in any industry, but particularly those with complex processes and decadal time-lines. The more important a decision is for the return-on-capital of your business, the more certain you need to be that data – rather than culture, structure, human nature or perverse incentives – remains the dominant factor.
Our adventure at Centessa is the first attempt to supercharge return-on-capital in the trillion-dollar pharmaceutical industry by deploying the #unshackled approach. DrugBaron has never been more optimistic.
The Cambridge Partnership is the only professional services company in the UK exclusively dedicated to supporting companies in the biotechnology industry. We specialize in providing a “one-stop shop” for accountancy, company secretarial, IP management and admin services. The Cambridge Partnership was founded in 2012 to fill a gap. Running a biotechnology company has little …
