Industry leaders say that the key issue is the low predictive power of existing pre-clinical models, with academic research supporting this hypothesis. We will address this by bringing more patient-derived approaches into the mainstream. These include advances in patient derived targets and biomarkers, and more complex models, such as organoids, for discovery and pre-clinical research. These technologies need to be woven throughout the discovery process to place the patient at the heart of the research process.
A new report from the UK has pointed to a productivity crisis in pharmaceutical research: new drugs have a high failure rate, the number of drugs launched per $1bn of research and development spending has fallen nearly thirty-fold over the last 40 years, leaving the pharmaceutical industry return on capital now at only 3.2%.
Let me briefly recap the phases of development that a new drug undergoes. Preclinical research is the stage of research that occurs before the drug is tested on humans. It usually involves in vitro and in vivo tests. In vitro (Latin: in glass) tests are sometimes called test-tube experiments, for example microorganisms in Petri dishes. More recently developed in vitro methods involve omics, such as genomics, proteomics or metabolomics. In vivo (Latin: within the living) tests are conducted on living organisms or cells. In biomedical research, in vivo methods generally involve animal experiments.
Preclinical research is followed by three stages of clinical trials on humans. Phase I is usually conducted for safety testing. If the drug is found to be safe, it is tested in Phase II to see whether it works as intended. Phases I and II involve small numbers of humans. If the drug is found to be safe and effective, it proceeds to Phase III, where it is tested on a larger number of people and compared to placebo or other treatments for the condition under study. Sometimes there is a fourth phase: after the drug has been marketed, further information is collected on effectiveness of the drug and side effects, or to investigate the effectiveness of the drug for a different condition or in combination with other drugs.
Three pharmaceutical industry groups collected data on clinical development success rates from 2006-2015 and found the following average rates:
NDA – New Drug Application to the FDA (US Food and Drug Administration)
BLA – Biologic License Application to the FDA
After in vitro and/or in vivo testing, on average only 9.6% of new drugs achieved approval from the FDA. Cancer drugs had the lowest approval rate (5.1%), haematology drugs the highest (26.1%). A dismal – and very expensive – failure rate.
Back to the UK report. Based on over 100 in-depth interviews with senior executives of UK drug discovery companies and electronic surveys of 250 experts, the authors summarised the problems as follows:
- Global R&D productivity is under unprecedented pressure
- The model of medicines R&D must be radically reshaped to meet patient needs
- A key problem is reliance on using inadequate models for human diseases
- Commercialising emerging technology will require new models of collaboration
- Data science is now indispensable to medicines R&D: research data is now generated in such high volumes that the ability to harness it has become a critical factor in developing new medicines
- It is imperative for the UK to provide industry with straightforward, well-governed access to consented patient data and human tissue samples – this is an acute problem for SMEs*
*SMEs – small and medium-sized enterprises
The authors of the report observed that too much of the preclinical research is patient-free and relies on animal models of disease and toxicology that are a poor approximation of humans. They wrote that drug discovery must be ‘humanised’:
Our interviews and surveys identified many emerging technologies that can ‘humanise’ the drug discovery process. These technologies make the early stages of research more predictive of how a drug will work in real life. They can generate a wealth of humanised in-vitro data, resulting in better drug candidates entering human trials. The benefit is lower attrition and therefore improved research productivity for industry.
… and pointed to new and emerging technologies that don’t involve animal research:
There are many emerging technologies that can make pre-clinical drug development more humanised. Most are derived from human stem cells and the resultant technologies that allow us to create and sustain human tissue in the laboratory. Just 20 years ago, keeping such tissue alive in the lab was a challenge. Now, thanks to pluripotent stem cells, advanced culture methods, microfluidics and precision gene editing we can manipulate the way such tissue grows and differentiates, even down to the substructures of cells and the stratum of the disease which the model reflects. When linked to large human cohorts, we can develop libraries of disease models that reflect the molecular spectrum of human disease, just as the Sanger Centre has done with their library of cancer cell lines. These complex predictive models, when used appropriately, have the potential to be much more discriminating in their ability to weed out the false positives in drug discovery i.e. those compounds that are too toxic, or insufficiently disease modifying.
The report also called for better collaboration between all stakeholders, the sharing of data and better access to consented patient data and human tissue samples.
Data from failed trials and failed pre-clinical projects could be transformative in reducing rework.
Further, a lack of validation efforts was noted. The experts that were interviewed said that ‘many potentially powerful human in vitro models remain in academia. There they have no obvious commercialisation path in the UK, given they often lack IP and so are hard to spin-out.’ Several people pointed out that validation is not a good fit for grant funding.
Many reasons tied up with their careers hold researchers in academic institutions back from leaving animal experiments behind:
It is important to recognise that researchers can be reluctant to invest time and money in implementing a new technique, or to replace an animal model that has served as the basis of their research for many years. … There may be concerns about a lack of historic data comparability, or invalidating past results. Setting up a new model can require additional technical expertise or development of new infrastructure. Referees are familiar with data from the ‘gold standard’ animal models, and may request additional in vivo data to be generated to support in vitro findings. These factors can delay publication in a highly competitive research environment and result in a lack of motivation to change models. (Jackson and Thomas 2017)
Researchers in the pharmaceutical industry are free of some of these constraints. The animal model research paradigm is truly outdated and better, human-relevant methods and technologies are available and are being further developed. This report by Medicines Discovery Catapult and the UK BioIndustry Association is a welcome guide to a future of biomedical research that serves patients, leaves behind cruel and unnecessary animal experiments, and promises a better return on investment for biomedical companies.
______________
Many thanks to Andrew Tilsley for his permission to use an image of his artwork ‘Cures for Diseases’.
ozsheba 