The moving target

A past experience with target selection reminded me of the uncertainty and messiness of drug discovery.

One of the most confounding realities of drug discovery is that it rarely offers clear-cut “right” or “wrong” answers. Instead, researchers navigate a murky landscape shaped by the staggering complexity of human biology and the unforgiving pressures of the marketplace. This makes the field challenging, even if exciting. Here's a concrete example that I have been thinking about.

A few years back I was part of a team effort to pick targets at a small company. Good target validation has the rather unfortunate distinction of being both a very tricky exercise and a very important one; wrong targets can take you down very expensive rabbit holes and are often the number one reason for drugs failing because of lack of efficacy. Now, you can always draw up a list of criteria for picking a good target - clinical validation, structural data, disease mutations, existing chemical matter, market size etc. but this does not mean everyone will agree on the end result. In fact the same criteria that one person uses for nominating a target will be considered by another person as criteria that should disqualify that target.

For instance, you may pick a target on the basis of novelty because there is no existing chemical matter for it, but someone else may disqualify that target as being too intractable for the same reason. You may pick a target because it's targeted by blockbuster drugs and therefore is clearly clinically validated, but someone else may disqualify it for the exact same reason, because it is too well-treaded upon. Someone can pick a target because there is a lot of structural data for it, but another might think that all that structural data is irrelevant because the clinical validation is not quite there. Target selection can start feeling like a Rashomon test, with everyone weighing the importance of a target based on personal preferences, biases and experience.

Factors like resources, feasibility of a novel technology platform, investor interest and even fads and fashions in the field can heavily influence final target choice. Even targets like KRAS G12C, Myc or p53 which have been favorites for years may not be suitable in the context of a specific organization with its own specific set of constraints. Sometimes a target looks attractive, only to present insurmountable downstream challenges. When that happens, the sunk cost fallacy can prevent researchers and management from jettisoning the target. Ultimately, what's a "good" or "bad" target becomes "obvious" only in retrospect. In my own example, predictably nobody was entirely happy with the final list of targets, and everyone found something to love and hate about it.

Now apply this uncertainty in target selection and validation to almost any other aspect of drug discovery and you can understand why the process feels so fraught. Trying out a synthetic route for quick hit and lead synthesis and validation? You never know what’s going to be the success rate or how scalable it is. Testing different preclinical assays for a challenging target? You never know what the false positive or negative rate is and whether, crucially, they will translate to in vivo data. Deciding on the right animal model and wondering whether it will translate to human disease? Good luck with that! And it goes on and on.

This is a very different situation from basic scientific research where, as hard as the problems can be, there is usually a right answer. Either DNA is a double helix or it is not. Either spacetime is curved or it is not. Either energy comes in little packets or it does not. Scientific truths are not subject to the whims of investors and the market; a field like drug discovery is subject to both. But what really confounds the field is biological complexity and ignorance. There can be little consensus on picking a target because nobody really knows how druggable the target is or how important it will turn out to be in the grand scheme of things, especially when very large patient populations are involved. It is a messy universe if there was one, one that makes drug discovery subject to a constantly shifting landscape of opinions fueled by old and new data. It is what makes us feel frustrated, but also keeps us on our toes every day. It ensures that we will never run out of ideas and that there will always be opportunities to explore and discover. It is what makes our world an open world, a world without end.