Drug discovery challenges now the low hanging fruit has been harvested

Published: 07 Sep 2018

Up until the late 70s, pharmacological discovery was a process exemplified by increasing technical expertise in surgery and bioassay, and was associated with a series of important discoveries. However we tend to view this ‘golden era’ through rose tinted spectacles. The reality is that the vast majority of research, as will be evident from a perusal of old volumes of pharmacology journals, was unimportant and anodyne. We remember only the ‘good bits’.

Moreover, the great discoveries such as the mechanism of action of aspirin were, to a large extent, the plucking of low hanging fruit. Much of the discovery, such as the antiarrhythmic action of amiodarone, was by chance. And some, such as the discovery of the mechanism of action of digitalis, was far from immediate, taking (in this example) several hundred lugubrious years in a process littered with wishful thinking and the invention of catch-all concepts such as ‘multifactorial mechanisms’ (a euphemism for ‘it does lots of things but we don’t know how this leads to the primary beneficial effect’).

The main purpose of pharmacology is drug discovery. One of the problems in pharmacology is that when the low hanging fruit were plucked, or serendipity delivered a new medicine, we took inordinate encouragement from these great successes, and have tended to imagine this reflects the excellence of our knowhow and the sophistication of our discipline. Unfortunately, in more recent years, the truth is beginning to dawn on some of us: progress is slow and translation commonly fails, and the drug discovery process is more haphazard than one would expect if driven by genuine expertise and knowhow. Consider the post-Viagra success rate of Pfizer, Sandwich, for example. There is even a perception that we are getting worse at drug discovery.

The reality may be that we are neither better nor worse, but appear worse because the low hanging fruit are now largely gone and the hit rate has consequently declined. This means that drug discovery is, and has always been, intrinsically a low yield activity, characterized by false discovery and failed translation.

The emergent ‘dry pipeline’ has allowed those opposed to animal research to argue that animal models are misleading and outcomes will not accurately predict the human response. This may be true in cases where there are no drugs effective in humans (positive controls) with which to validate a model, meaning the model is unvalidated and potentially invalid. However, it cannot be the main reason for failed translation of treatment of conditions where it is easy to validate a model with positive and negative controls. Take this explanation for failed translation away and we start to wander into murkier areas.

For example, part of the explanation for the ‘dry pipeline’ of failing translation, perhaps the larger part, may be that we are not (and never were) very good at experimental design, with nonblinded non-randomized studies yielding false positives. Early legendary observations, such as when ACh is injected into a dog it had profound and obvious effects, did not require a great deal of experimental design to reveal themselves.

These early observations, also, came about from the pursuit of curiosity, often by comfortably-off men (in the main) not reliant on their research for their primary income. There was also no need to publish regularly in high journal impact factor (JIF) journals, no ‘publish or perish’ tyranny, and no need to put a positive spin on every trivial finding. As time progressed, the workplace environment has changed, becoming highly competitive. The primary measure of success is no longer an impactful discovery and a new medicine, but a ‘high impact’ paper and citations.

With rewards based on ‘impactfulness’, findings need only be positive and exciting. Reproducible (i.e., correct) findings are no longer necessary for a ‘successful’ career in preclinical (especially academia-based) research. Moreover, all the while the work stays preclinical, if the findings are incorrect this is likely to go unchallenged, partly because people work in silos, with replication of other people’s work regarded as an unoriginal (and largely unfundable) pursuit.

In the meantime, knowhow about experimental design has not improved. If I say that most if not all pharmacology journals publish in almost every issue papers with n=3/group, and report ‘highly significant’ effects (P<0.001) even with small samples, the tragedy is that many reading this statement will think ‘so what?’. Indeed, far too many pharmacologists (sometimes cheerfully) admit they know nothing about design and statistical analysis.

Unfortunately this has always been so. But today, with the low hanging fruit gone, and the pressure to publish so great, the effect is that we are polluting the literature with an increasing preponderance of findings that are likely to be false. Indeed, there are some individuals who are doing this knowingly. How many? It is hard to tell, and it is widely regarded as distasteful to broach the topic. Regardless of the relative contribution of ignorance versus deliberate fraud, unless things change, eventually the funders will begin to realise that the ‘breakthrough’ promises are not forthcoming and start to speculate that their investment may not be good value for money.

That said, given that members of the public are often naïve, and far too willing to put hope before evidence (witness the numbers who believe in space aliens, the investment value of a lottery ticket, ghosts, and god), the processes leading to our being found out will likely be slow. That is no excuse for letting the situation drift, however.

I would like to see pharmacologists applying more rigour to their experimental design. Sadly I suspect that the ‘publish or perish’ mentality is now so pervasive that attempts to improve standards will be resisted (‘don’t rock the boatism’), and that without a better coordinated effort, with unswerving leadership, nothing much will change. We shall see.


Blog post currently doesn't have any comments.

If you are a British Pharmacological Society member, please sign in to post comments.

About the author

Mike graduated with a BSc in Pharmacology from Chelsea College in 1979, and a PhD from University of British Columbia in 1986 (under the supervision of Michael Walker). After three years’ postdoctoral training at the Rayne Institute (under the supervision of David Hearse), Mike became a lecturer in Pharmacology at King’s College London in 1989, and reader in 1996. His research is cardiovascular and his main interest is antiarrhythmic and proarrhythmic drugs. He has published over 100 papers (cited over 5000 times) and has an h index of 34. Mike has supervised 12 PhD students, two of whom were AJ Clark scholars. He has a keen interest in teaching and training, and has published several research guidance articles including the British Journal of Pharmacology’s design and analysis guidance (2015) and the Lambeth Conventions arrhythmia guidance (2013). Mike has held several editorial roles including reviews and themed issues editor for the British Journal of Pharmacology (to finish a 17 year run on its editorial board) and has been editor in chief of J Pharm Tox Methods since 2001. He served on the executive committee of the Society for three years, and that of the British Society for Cardiovascular research for 17 years (15 as treasurer).

Post archive

No data found