Bad methods at the heart of drug discovery costs

The cost of newly licensed drugs are high and getting higher. Over the past 50 years the cost of cancer drugs has increased by a hundredfold and our spending on speciality drugs is expected to double again by 2020. Whilst R&D is expensive, human biology is complex and the regulatory hurdles the Pharma industry has to leap are significant, new research suggests that the real reason for the soaring prices of new medicines may be that the methods being used in drug discovery are wrong.

A paper published in PLOS ONE this week (link is external), says that drug discovery has focused on methods that are fashionable within the academic community or easy to industrialise, rather than whether the experimental results give an accurate reflection of a drugs effect in the patients for who they are designed. 

The rising cost of drug discovery may be due to poor choice of experimental methods
The rising cost of drug discovery may be due to poor choice of experimental methods

Dr Jack Scannell Associate Fellow of the Centre for the Advancement of Sustainable Medical Innovation (CASMI) at Oxford University and consultant Jim Bosley used a quantitative decision-theory model to show that the chance of discovering an effective drug is very sensitive to the patient predictive quality of the methods. Even small changes to the predictiveness of the methods can have a greater effect than running ten or even a hundred more experiments.

With the technological advancements made over the past few decades and the cost of this technology falling, theoretically the academic research being produced should be of a higher quality, providing a better platform for industrial development. But academia is suffering a reproducibility crisis (link is external) and analyses show that drugs are more likely to fail in the clinical development phase today than in the 1970s. The authors of this work demonstrate how screening and disease models that offer good predictability of patient responses correlate with reproducible methods, and suggest that there has been too much focus within industry on reductionist molecular models that provide poor predictiveness.

The pharmaceutical industry argues that increasing drug prices contain the cost of the increasing number of failed drugs that didn’t reach the market. However, Scannell and Bosley suggest in their work, that productivity within the pharmaceutical industry has declined because the most predictive methods lead to the discovery of good drugs, and consequently research in these areas stops. This leaves scientists working on as-yet-untreated diseases using less predictive technologies. Changing research fashions have exacerbated the problem by preventing focus on one set of methods and allowing their functionality and flaws to be fully exposed. This has meant that methods used in drug discovery have become less predictive of patient responses over time, worse decisions have been made in research programmes and the cost of drug discovery has soared.

Since 2012 the number of drug approvals has increased. According to Scannell and Bosley’s study this is likely due to the rising use of genetic information that offers good predictiveness in the discovery of treatments for rare diseases, but less so for common ones. Despite industrial advances in efficiency and the huge gains made in scientific knowledge the cost of drug development has continued to rise. The authors hypothesise that the rate of creation of predictive screening and disease models may be the major limiting factor in drug discovery and that more heed should be paid to this if effective drugs are to be developed at affordable cost.