Are you looking for a faster horse in drug discovery?

Posted 27th September 2017 by Jane Williams

Discovering and developing new drugs is difficult. Very difficult indeed. The cost of developing a new medicine, when you factor in the costs of all those projects which fail before getting to market, is increasing and is Very Large Indeed so what can be done to get us out of this situation?

Fortunately, biomedical science is not standing still in its pursuit of better human health. Knowledge is increasing rapidly, perhaps doubling every year or so and with many thousands of new publications appearing every single day there are lots of new ideas to help us – but therein also lies a problem. All of this new knowledge (only a fraction of which will directly allow breakthroughs and new therapies) is difficult for any one person to assimilate and even when firm conclusions can be drawn from it, they often then only signal the start of a drug discovery journey that takes 10 years or more to deliver a medicine for widespread clinical use.

In other words, if drug discovery were a maze, it is a maze where it’s extremely difficult to even to find the best way in and then, when you are in, well, you’re not getting out too quickly…

This is a classic example of a problem where incremental improvements are unlikely to fundamentally change the magnitude of the challenge and instead, something different is required. To paraphrase a great early 20^th century innovator (perhaps incorrectly attributed), “we really don’t need a faster horse.”

So, if we assume this particular horse is already galloping as fast as it possibly can, where do we go? Drug discoverers need to try some fundamentally different ways of working. The current, linear process (we all know the chevron diagrams) works of course. People and companies are very good at it so are very happy to continue doing it but remember, riding a horse will also get you from London to Edinburgh eventually, just not in a way which meets the needs of 21^st century society.

To break out of this, we need to rethink the process and use as many novel approaches as we can, challenging dogma as we go. Drug development is a highly regulated business (for good reason – patient safety should always remain the prime goal of drug developers), but there are still many ways of doing things differently which will also maintain, and even ultimately improve, the safety of new medicines. Here are some examples:

1. At the most basic level, small molecule drugs still need to be made in the chemistry lab, using linear sequences of, often labour-intensive chemical manipulations, frequently using equipment which would not seem out of place in a Victorian lab. In the modern lab, we should design and make more use of roboticized synthesisers to reduce the time taken to make molecules and then programme these synthesizers to make molecules which chemists have selected by making use of deep learning/AI assistance.
2. Of course, it’s only worth making molecules once you have a therapeutic hypothesis that you believe has relevance to clinical disease. As we highlighted earlier, finding such trustworthy hypotheses is not easy but by harnessing the rapidly growing power of artificial intelligence, the vastness of the biomedical literature can now be brought together comprehensively to highlight to drug discovery scientists larger numbers of potential hypotheses from which to pick the best ones. A new, but concordantly supported hypothesis can enable a new drug discovery project with more confidence or better, can find new uses for existing drug molecules.
3. Most drug molecules inhibit a specific protein target to elicit their pharmacological effect requiring the drug to stay in the body at sufficient concentration to maintain high target occupancy until the next dose is given. This frequently requires high drug doses which, in turn, increases the risk of unintended pharmacology (i.e. toxicity or adverse events). Instead of relying on this equilibrium occupancy-based efficacy, recently targeted protein degradation using untraditional molecules known as Protacs, has shown high levels of cellular and in vivo potency due to the unique, catalytic mode of action of these molecules. This has the genuine promise to translate to low clinical dose and thus better efficacy and therapeutic windows.
4. Once in the clinic, smart phones and devices are increasingly able to monitor patient health (and indeed influence it: have you completed your 10,000 steps today yet?). This provides opportunities for novel, and potentially more informative, clinical trial endpoints which can be monitored more cheaply. Drug compliance, so often a problem especially for chronic conditions leading to suboptimal clinical outcomes, can also be improved by use of electronic reminders or even monitoring to ensure correct dosing.

These are just a few examples of areas which, if successfully used together (and in concert with other new approaches also), could start to form a new drug discovery norm. Some approaches may prove to be transformational, others may disappear without trace after riding the hype curve for an all too short existence but one thing is for sure, if you don’t try some of these things in your drug discovery strategy you won’t find the ones which work but your competitors will, and they’ll happily leave you riding that old horse for a long while yet.

As we all work busily in our own corners of drug discovery, be it in Pharma, biotech or academia, ask yourself the question – are you just pushing that horse a little more or are you genuinely imagining and creating the model T of the 21^st century?

Ian Churcher is VP Drug Discovery & Preclinical Development at BenevolentBio.

The agenda has launched for the Global Medicinal Chemistry & GPCR Leaders Summit. View it here.

All thoughts belong to Ian Churcher and are not representative of BenevolentBio. 

Tags: artificial intelligence, disruption, drug discovery, innovation, Protac