<img alt="" src="https://secure.lote1otto.com/219869.png" style="display:none;">
Skip to content

Maximizing Preclinical Knowledge for Optimal R&D



By Esha Senchaudhuri

In response to its R&D productivity from 2005 – 2010, AstraZeneca took the initiative in 2011 to implement what it has called the 5R Framework to strengthen its capabilities. In a Perspectives article from Nature Reviews Drug Discovery [1], Paul Morgan and his team provided complex details about the success of this framework from the perspective of every stage of drug development. Between 2005 and 2010, AstraZeneca was behind industry averages in every phase of clinical development except Phase 1.  After the implementation of 5R, AstraZeneca success rates improved substantially. Indeed, it was announced by IDEA Pharma in March 2018 that AstraZeneca had topped its 2018 innovation index.  Here we examine to what Morgan et al, attribute AstraZeneca’s  success.  

Key to identifying whether a clinical program will be successful is the use of a variety of predictive analytics ranging from predictive biomarkers assessed when choosing targets, to the uses of PK/PD to identify the right tissue, to maximize chances of success and make each phase of a trial more efficient.
AstraZeneca has initiated numerous ways to leverage preclinical data to improve clinical trial design and optimizing success has meant significant improvements in both pipeline production and safety trials.  Below we examine the five R’s of the 5R Framework – Right target, Right tissue, Right safety, Right patient, and Right commercial potential. We also examine how getting these five parameters right requires improved modeling, forecasting and portfolio construction tools.

Right target:
287 molecule-discovery programs began between 2005 and 2010, with 28% of the development portfolio consisting of backup programs. This number was reduced to 76 new discovery programs between 2012 and 2016, with fewer than 7% backup. In addition, the number of annual high-throughput screens – the primary strategy for lead development – reduced from 45 per year to 20 per year during this time.
Driving these reductions was the use of genomics, phenotypic screening and other methods of biological knowledge, (aided by requisite investments in chemo-pharmacological technologies,) to focus the composition of pipelines while also seeking to develop compounds with “novel [causal or operative] mechanisms.” Genomic editing, applied early in the development process, has also proven useful in choosing the right target.

Right tissue:
A central component of improving molecule-discovery programs has been AstraZeneca’s extensive use of PK/PD analysis for proof-of-mechanism. Pre-clinically, PK analysis is helpful in understanding a number of factors that shape molecule design such as molecular absorption, distribution, metabolism and excretion. Although obtained pre-clinically, this knowledge can then be scaled to provide critical insights about dosage and stratification. In particular, scientists at AZ note that if more than one compound has been developed that successfully engages a target, early PK/PD analysis can ensure that the compound chosen for testing is the one best positioned to pass safety and toxicity tests. Further, Phase 2 decision-making can take advantage of pre-clinical data to obtain the right dose. From an R&D standpoint this means that fewer compounds are tested in Phase 2; those which are tested are better suited to succeed in Phase 2; and that these successes pass on to Phase 3 testing with superior dosage-models to inform Phase 3 decision-making. By investing in the right modeling up front, every phase of the trial is more rigorously informed as well as considerably shortened.

Right Safety:
The AstraZeneca 5R framework also integrates preclinical data with Phase 1 and Phase 2 safety testing. This is accomplished by greater reliance on early testing using a combination of in silico, in vitro and in vivo methods, which generate data that can positively impact early phase safety testing. Once again, moving from preclinical to clinical settings requires numerous translation capabilities, not least of which is dependable modeling of preclinical toxicity.

Right Patient:
Biomarker-driven patient selection has become a popular tool for clinical development. Unsurprisingly, its use by AstraZeneca has generated immense benefits for its R&D, with 62% of trials with prospective patient selection yielding results, when compared to 44% success rates of trials without prospective patient selection. This is not merely the result of a change in methodology but also a change in culture or attitude: scientists are beginning to think about patient selection much earlier in the development process, so that suitable diagnostic exams can also be developed early enough to leverage enrichment for patient selection. Early planning across all phases has yielded an 18% increase in success rates.


Digital target over computing matrix on black backgroundRight Commercial Potential:
AstraZeneca has had to make a separate cultural decision in its focus on preclinical and early phase development, which is to substitute governance decisions based on commercial potential to those based on scientific potential. Given the numerous uncertainties in the beginning of program development the early phase focus has elicited concerns that commercial decisions cannot be accurately calculated. Scientific focus, though, can give a drug the best possible chance it has for commercial success.
In this regard, Cytel believes there is reason for optimism. Through forecasting and careful portfolio optimization, there are numerous ways to translate uncertainties into commercial potential without sacrificing scientific rigor.

 Cytel Comment

In the process of implementing the 5R framework, Go/No-Go decision-making became a central feature of drug development. Promising new medicines needed to be identified quickly, and compared to other promising new medicines across several parameters. Cytel’s OK GO ( currently in beta testing)  software was developed to facilitate such assessment, originally developed with AstraZeneca scientists as dECiDe.

In addition to refined decision-rules, Morgan et al., point out that once Go/No-Go decision rules have been established, interim analyses in early phases may also be used to evaluate the likelihood of a ‘go’ decision at the end of the study. Such analyses can stop for futility as well as accelerate development given the predicted likelihood of a go decision. In the AstraZeneca 5R framework, such a quantitative approach is applied throughout early phases, beginning with preclinical studies and ending with Phase 2.

Click the button below to learn more about OK GO. 


Further Reading and Resources

Morgan, P., Brown, D., Lennard, S., Anderton, M., Barrett, J., Eriksson, U., Fidock, M., Hamrén, B., Johnson, A., March, R., Matcham, J., Mettetal, J., Nicholls, D., Platz, S., Rees, S., Snowden, M. and Pangalos, M. (2018). Impact of a five-dimensional framework on R&D productivity at AstraZeneca. Nature Reviews Drug Discovery, 17(3), pp.167-181.

Blog: Decision-Making in Early Clinical Development

Poster: Model-based predictions of pharmacodynamic responses

Webinar Replay: Dual Target Methods for Go/No-Go Decision Making

Frewer, P., Mitchell, P., Watkins, C. and Matcham, J. . Decision-making in early clinical drug development. Pharmaceutical Statistics, 15(3), (2016) pp.255-263.


contact iconSubscribe back to top