It has often been said that one of the key features that distinguishes “big pharma” from biotech is access to the latest technological platforms to aid efficient drug discovery and development.
These platforms range from vast chemical libraries, ultra high throughput screening and huge genetic databases in discovery, to predictive toxicology platforms, cutting-edge ‘omics’ and even deep-seated knowledge of particular therapeutic areas in development.
All these ‘platforms’ have two things in common: they can be used on any (or many) development candidate assets, and they cost huge sums to establish in the first place, and in a few cases each time they are used as well.
Hence their restriction to the largest pharmaceutical companies (and a few of the so-called ‘big biotechs’ that are, in many ways, indistinguishable from the old-guard pharma). Only when you have hundreds of active projects can you justify the cost of creating and operating these platforms.
Or so the mantra goes. It is access to these platforms that keeps the big companies ahead in the race to discovery and develop the best medicines (or at least counterbalance the disadvantages of being large and slow moving, depending on your point of view).
But is that just an assertion? How much evidence is there to support the proposition that the efficiency gains due to these platforms outstrips the cost of creating and maintaining them?
Not for the first time, DrugBaron takes the opposite view: that without proper metrics of the cost versus benefit of such platforms, they rapidly become the “black hole” into which billions of R&D spending disappear. Like an alcoholic opening a bottle of vodka before breakfast, the justification sounds good to themselves but holds little sway with rational analysis. Spending on platforms prevents big pharma properly addressing the cost problem in drug discovery and development and unless it is addressed such platforms may ultimately prove to be their nemesis rather than their saviour.
The solution may lie in an unusual kind of de-merger.
Platforms (or ‘asset-independent technologies’ as DrugBaron prefers to call them, to capture all kinds of capabilities that can be leveraged across many different drug candidate assets rather than just discovery tools that the term ‘platform’ immediately brings to mind) are ubiquitous in modern pharma. They are the product of an arms race, to secure access to best capabilities in key areas.
In the 80s and 90s, that race was focused on screening capability – larger, more diverse libraries together with faster, higher throughput screens. Eventually, all the big players (and even CROs) had sufficient capability to ensure that finding the right molecule was no longer limiting. So for the last decade or so, the focus shifted to technologies to reduce attrition in the clinic: better target validation and improved predictive toxicology. The genetics bubble was perhaps the biggest and best known example of this competition playing out, but there are many other examples: one pharma company recently told DrugBaron they had more than 6,000 preclinical efficacy models of disease in-house; another has developed the most detailed metabolomics platform for predictive toxicology; while every major pharma has a proprietary platform for antibody generation driving their biologics pipeline.
Keeping these technologies “cutting edge” has become so expensive that increasingly we hear pharma companies talking of “pre-competitive” approaches to develop the next generation. A group of companies might develop a platform capability they then share. The principle goal of such initiatives is to access even grander and more expensive tools than individual companies could afford, rather than to dramatically cut costs (although sharing platforms rather than developing the same thing in parallel in each silo should at least keep a lid on rising costs).
But throughout the last three decades, as spending on these platform capabilities has mushroomed, there have been few voices questioning their value. After all, it seems plain obvious that more and better tools will improve discovery and development, leading to more and better drugs and bigger profits.
DrugBaron does not question that assertion, but it is only half the equation.
New information derived from these platforms has a value – but it also has a cost. The question of whether the investment in platform technologies is a boon or a bane for pharma lies in whether the extent of the benefit outweighs the cost of achieving it. Not just the marginal cost of each application of the technology, but an appropriate slice of the fixed cost of creating and maintaining the platform. After all, the cost of developing the platform, even spread over a number of projects, is usually by far the largest component of the cost.
DrugBaron cannot point to an analysis that accurately estimates the incremental benefit of each new platform. Such an analysis is impossible in a world that approves only a few tens of new drugs each year. The dataset is no where near rich enough for a refined assessment of the increased output of pharma as a result of each new technological platform – there are, after all, more platforms than annual approvals!
Nor is it sufficient to say that despite massive increases in spending on platforms generally, productivity (drugs approved each year) has actually declined (at least throughout the period between 2000 and 2010). Productivity might have been even worse without the new platforms.
In the end it has to remain a matter of judgment – how much benefit you believe you get, and therefore how much to spend on platform innovation overall, and in particular which specific areas and technologies you will address.
That the benefits cannot be quantified accurately is not the nub of the problem though: that honour belongs to the thorny issue of accounting for the costs. DrugBaron argues that the different way large and small companies account for these costs represents the single biggest difference in their business models. And it really matters precisely because the incremental benefit derived from using such platform tools is inherently impossible to demonstrate accurately.
In a small company – a very small company, such as the asset-centric drug development companies pioneered by Index Ventures – there are no platforms (there may even be no people), just a single drug candidate. The company operates on its singular asset by accessing suppliers offering each required process (whether they are out-sourced drug developers or CROs). This allows access to many different platform technologies, though in many cases not the “platinum grade” tools proprietary to pharma.
But even the gold-standard tools available to the fee-for-service clients can be very expensive. Ask for a GWAS study for a particular disease, and the cost will be seven figures if not eight. It’s not surprising – the fee-for-service provider had to invest heavily to create and maintain that platform, just as the pharma companies do.
Yet the arms-length nature of this relationship yields a critical, but often overlooked, benefit: the cost of accessing the platform is completely transparent to the project leader. Upfront, he knows the nature of the information he will receive from applying the platform to his asset, and he knows the price he will have to pay. While its impossible to know the value of the information once its delivered, at least he has a fighting chance of making a sound judgment knowing the cost so clearly.
Virtual drug development is, then, nothing more than a series of these cost:benefit judgments by the drug developer.
Contrast that with the situation inside any large pharma company. Most of the processes the project manager responsible for any given drug candidate asset needs to access are available in-house. Better still, the platform technologies available to him are likely the best that money can buy.
But his judgment as to which one to apply, and in what order, to de-risk the asset he is working on, and move it towards the clinic, or towards approval, is hampered by one critical omission: the price of accessing each. Since the platform is internal, the task of accessing the tool is not formulated as a transaction. Critically, now neither the benefit of the information nor the cost of obtaining it are transparent to the decision-maker.
Of course, its difficult to estimate the “internal transfer pricing” in any case. The marginal cost of using the platform for each asset should be accessible (but, actually, is rarely known), but the cost of creating the platform in the first place – usually the largest component of the cost – is difficult to apportion among the subsequent users over time.
The same pharma company with 6,000 preclinical efficacy models also told DrugBaron that they do not know the marginal cost of running any given one. They know very well the aggregate cost of the 250 staff running the models, and the total consumables bill for the unit – but could not even estimate the marginal cost of running one individual study. Any attempt to guess the cost of developing and validating any one model, so that could be added to the marginal cost of running it, was beyond the realms of possibility.
The contrast is stark then: a researcher in a virtual asset-centric biotech with a candidate drug to treat lupus can access several different preclinical models of the disease from a supplier such as RxCelerate. As he works out his ideal study design, he knows quickly the cost implication – should he try his drug candidate in two models in parallel? Should he add an extra dose group? Or a range of different end-points? With each option, comes a transparent cost directly attached to the different information he will receive. That cost:benefit judgment can then be compared with other similar judgments he must make in parallel to have the best chance of delivering a clinical candidate for the budget he has available.
In the pharma company, by contrast, the costs are opaque, so the researcher – not unreasonably – simply designs the best study design he is able to. Cost implications are only indirectly impacting him: if senior management cut the total capacity of the unit delivering preclinical models, they may ask him to refine (trim) his study design to fit with the new capacity constraint. But by and large, the available capacity is allocated without any consideration for where it will be most useful.
That is not to say that either one of them, the virtual developer or the big pharma scientist, is better at making the right decision about the size and design of his preclinical study – but only the virtual developer has the cost implications of each design change in mind to assist him.
This dichotomy is the single biggest reason why pharma cannot copy asset-centric drug development. Such a large fraction of their R&D costs are consumed by the platforms, that even as they squeeze the direct costs of the project teams working on an asset the impact on the overall budget is negligible.
And the consequences of this opaque cost problem go much wider. How do you decide which platforms were worth their investment, given the acknowledged problem of quantifying benefit? If the platform is set up by an independent service company, the answer to that is easy: if they can turn a profit selling access to the platform then the investment in developing and maintaining it was worthwhile. But if the price they have to charge to recoup the investment exceeds the price clients are willing to pay for the information the platform delivers, then arguably developing that particular platform was a drain on R&D productivity.
With access to platforms not treated as a financial transaction internally within pharma, there is no such guidance as to which platforms are “worth” the cost of having them. Furthermore, the decision to create a new platform is a competition between ideas for a fixed budget for technological innovation. In the small company universe, the decision for a service provider to invest in creating and maintaining a new service is a very transparent return-on-investment calculation.
Is there a solution for pharma?
Out-sourcing solves the problem for generic capabilities – there is no reason why pharma cannot access all the same processes as the virtual asset-centric biotech, and access them from the sample suppliers at similar prices. Doing so, may (or may not) cost more than delivering the same service internally – although it will be difficult to make that comparison of course, since the true cost of delivering the same service internally is almost never actually known. But the advantage of the cost transparency, and the improved judgment of the cost:benefit ratio of his decision-making, more than outweighs the profit the service-provider needs to earn on his capital.
But out-sourcing is not an option for the platinum-grade proprietary tools that arguably define pharma companies. The point of these technologies is that are supposed to be delivering competitive advantage. So how can they simultaneously be proprietary and yet have a transparent cost associated with them? In principle, it should be possible to create a reasonably accurate “internal transfer price” by accounting differently for the cost centres (on a much more granular basis than at present). But given the size and complexity of a very large pharma company that may be harder to achieve in practice than it sounds.
A better solution may be to spin out the platform technologies into a separate company. Such a move will force the calculation of an appropriate costed transfer price for each service. Some of the new companies technologies may be generic enough to sell to third party customers, like any other out-sourced drug development house, but others may be restricted in their sale to the originating pharma company, preserving those proprietary cutting-edge technologies perceived to be essential for competitive advantage. The strategic partnership between the two would also allow a discussion about where the platforms company should be making new investment to create improved tools – tools that the pharma would be prepared to pay a certain price to access.
Such a move may sound bizarre – how could such a re-organization do anything other than increase administration costs? Does it not run counter to the entire theory of economies of scale that underpinned the M&A trail which created the global pharma companies in the first place?
The proposition for consideration by the readers of this article is that such a move would improve R&D productivity to such a large degree that it would leave any extra costs visible in only the last decimal place. The inability to properly assess the cost associated with each new process undertaken on a project is a daily handicap across hundreds of multi-million dollar projects inside each company. If it is not addressed somehow, the assumption that platform technologies boost output by more than their cost will be tested in a very painful experiment – one which kills the companies.
Pharma companies are just beginning to recognize that they are too large. Some are even splitting, trying to disentangle parts of the business that show no synergy (such as Abbott’s split into a drug and a healthcare business, or mooted splits for Pfizer’s animal health division or GSK’s consumer products division). These “de-mergers” do little to address the operating problems that come from scale – the businesses being separated are operationally independent anyway. Splitting Abbott in two provided more flexibility for the investors to decide which part of the business might offer the higher return on capital in the future (which is a good thing) but it did nothing to make either business unit operate better or more efficiently, or to improve its fundamental return on capital.
A braver de-merger would separate platform technologies from drug candidate assets. Who will be the first to try? And if you are not quite brave enough to consider this surgical option, then a healthy dose of pricing transparency for creation and use of internal platform technologies is a must.
DrugBaron is the blog from David Grainger, PhD, partner at Index, biopharma consultant, and interim CEO of XO1 Ltd.