Transforming AstraZeneca's R&D productivity
Following a major review of our research and development strategy in 2010, we created a new 5R Framework to guide how we discover and develop new drug candidates. Looking at our productivity and success rates over the past 5 years we can now see a transformation in our productivity – enabling us to discover more innovative therapies for patients than ever before.
In a new report published in Nature Reviews Drug Discovery, Mene Pangalos, Executive Vice President, BioPharmaceuticals R&D and fellow scientists explain how our 5R framework (right target, right patient, right tissue, right safety, right commercial potential) has helped guide successful, efficient drug discovery and development. In just five years, we have achieved a five-fold improvement in the proportion of our pipeline molecules that have advanced from preclinical investigation to completion of Phase III clinical trials – from 4% to 19%. This improvement moves AstraZeneca well above the average success rates of 6% for small molecules in the 2013-2015 timeframe (Data sourced from CMR International’s 2016 Global R&D Performance Metrics Programme).
At the heart of this transformation has been a significant evolution of our culture. We have established an open, collaborative and ‘truth seeking’ culture where science thrives. Within this environment, we are never afraid to ask the ‘killer questions’ or to rigorously test our hypotheses. This has enabled us to improve the quality of the drug candidates we take forward into pre-clinical studies and subsequently into clinical trials.
Our success reflects an evolving approach in which we are integrating state-of-the-art technologies to identify new targets and the latest translational science to confirm proof-of-mechanism of our novel therapeutic modalities at safe therapeutic doses which can be taken into human studies.
We continually look to the future, giving our scientists time for curiosity and original thinking, enabling them to follow the science, further improve our R&D success rate and deliver transformative treatments for patients.
Selecting the right target remains the most important decision we make in the drug discovery process. We aim to identify molecular drivers of disease across our main therapy areas: oncology, cardiovascular, renal and metabolism, and respiratory and immunology, via our company-wide genomics initiative, launched in April 2016.
“A selective high-quality molecule will never become a medicine if it is modulating the wrong target. This is why target selection is the most important decision we make in research.” Mene Pangalos, Executive Vice President, BioPharmaceuticals R&D
Building on our growing knowledge of disease biology, we have expanded the classes of drug targets we investigate and we increasingly aim to identify compounds with novel mechanisms of action. Broadening our target class composition, significant investments in genomic data, next generation sequencing and CRISPR gene editing are enabling us to identify, interrogate and validate targets in a way not previously possible. This has translated to the increase in success rates from finding an active compound to an optimised molecule (‘hit to lead’) from 23 to 48%.
Integrating pre-clinical ‘absorption, distribution, metabolism, excretion’ (ADME) and safety data into our predictions of how drugs will act in patients is improving the quality of our drug candidates. We are advancing the best practice, predictive and translational science needed to show target engagement and proof of mechanism at safe, therapeutic doses that can be taken into human studies. As a result, we are delivering better quality candidates that are less likely to fail due to safety or other issues at later development stages.
To enhance our ability to match the right medicine to the right patient, we plan for biomarker-guided patient stratification at an early stage of our drug discovery projects. By establishing searchable, accessible human tissue biobanks for our scientists, we are supporting translation of biology from the clinic to the laboratory and this has contributed to improved screening assays and novel biomarkers. Consequently, approximately 80% of our 2012–2016 portfolio from Lead Optimisation onwards has a patient selection strategy compared to less than 50% in 2005-2010. Today, this figure is greater than 90% across our therapy areas.
Our focus on defining the ‘right patient’ has enabled us to launch nine companion diagnostics tests in 2012–2016, compared with just one in 2005–2010. We have already introduced biomarker-based diagnostic tests for EGFR, EGFR T790M, BRCA and PD-L1 to identify patients most likely to benefit from our targeted treatments for non-small cell lung cancer, ovarian cancer and bladder cancer. We also pioneered new tests using circulating tumour DNA as the biological sample type to broaden patient access.
By adapting and redesigning our preclinical and early phase clinical trials, we are setting rigorous standards so we rapidly and efficiently select only the best candidates for further development. Through the continued evolution and application of our 5R framework, we are converting our preclinical data into knowledge to be used to improve clinical study decision making, thereby narrowing the data-to-knowledge gap. AstraZeneca is already being recognised as leading the way in the application of artificial intelligence (AI) in early clinical trials through the iDecide research programme – an innovative five year collaboration between AstraZeneca, the University of Manchester Institute of Cancer Sciences, the Centre for Cancer Biomarker Sciences and the Christie NHS Foundation Trust (digitalECMT). As a result, we can make rigorous, quantitative decisions about the drug candidates we progress to the later stages of development.
“The Manchester collaboration is a tremendous opportunity for industry and academia to work hand in hand with patients. It brings together the AstraZeneca team, clinicians at Europe’s largest cancer hospital, University of Manchester scientists and Cancer Research UK. Together we can innovate in the conduct of clinical trials and the application of AI.” says Professor Andrew Hughes, Clinical Lead for Manchester Experimental Cancer Medicine.
As a result of these collaborations, we have published novel biological findings in several high impact journals together with our partners. In the past 18 months alone, this has included papers in Nature Medicine, Science Translational Medicine, Circulation and Science Advances. In oncology, through a highly successful collaboration with the Medical Research Council’s Laboratory of Molecular Biology, we used state-of-the-art cryo-electron microscopy (cryoEM), to describe, for the first time, the structure and activation mechanism of human ataxia-telangiectasia mutated (ATM) protein. This protein is a key trigger in the DNA damage response (DDR) and this collaborative work has resulted in insights that will help us uncover novel binding sites for future drug targeting.
With our respiratory and immunology research, we have identified the sub-set of dendritic cells, called cDC2 cells, that are essential for priming the immune response against invaders. Identifying how and where cDC2 cells fit into the complex process of immunity enables us to examine new targets in diseases driven by inappropriate antibody responses such as asthma, chronic obstructive pulmonary disease and autoimmunity.
In cardiovascular, renal and metabolic diseases, our extensive programme of cardiac regeneration research is identifying new targets and pathways that may play a role in repairing damaged heart muscle in people with heart failure. In recent studies with Professor Bin Zhou at the University of Chinese Academy of Sciences in Shanghai, we highlighted the importance of paracrine factors in cardiac regeneration.
With our scientists at the AstraZeneca Integrated Cardio Metabolic Centre (AZ-ICMC) at the Karolinska Institute, we have contributed two seminal pieces of research to the understanding of diabetes. By identifying a new role for insulin-like growth factor 1 (IGF1) in driving the formation of harmful epicardial fat tissue in the heart we have opened up fresh opportunities for tackling the growing burden of heart disease and obesity. In addition, by elucidating the functionality of sub-sets of islet cells in the pancreas, we have improved understanding of their importance in diabetes.
In neuroscience, our scientists at the AstraZeneca-Tufts Neuroscience Laboratory at Tufts University are bringing new insights into control mechanisms for nerve excitation in the brain, including increasingly detailed knowledge of the essential role of neuronal potassium-chloride transporter protein (KCC2). We are collaborating with researchers at the University of Cambridge to study the cellular processes for degrading unwanted proteins, with the aim of activating these mechanisms to degrade misfolded proteins such as huntingtin in Huntington’s disease.
“We are interested in partners who are curiosity based, who believe that understanding the science will shorten the route to clinic. Our partnership with AstraZeneca is a natural fit because AstraZeneca is truly focused on the science and on ‘blue-sky’ thinking that’s informed by current state-of-the-art knowledge.” Professor Iain McInnes, Director, Institute of Infection, Immunity and Inflammation, University of Glasgow
As we apply our knowledge of disease biology, we are diversifying our chemical toolbox to develop new therapies against an array of drug target classes. We no longer focus on small molecules alone, with around 30% of our programmes now exploring new modalities and drug delivery devices as part of our effort to make every target druggable.
Our everyday use and development of CRISPR/Cas 9 provides fast, precise, efficient gene editing to help us discover new drug targets and create more relevant cell lines and animal models. Through world-leading partnerships and our in-house team of experts, we continually push the technology to improve screening and efficiency. In two recent high impact publications, we showed how it is possible to barcode guide-RNA to improve screening and develop novel hybrid DNA:RNA guides to improve binding efficiency.
Speeding up the automated testing of thousands of potential new molecules is NiCoLA-B – the world’s most advanced drug discovery robot. NiCoLA-B uses sound waves to move tiny droplets of potential drugs from storage tubes into miniature ‘wells’ on assay plates – billionths of a litre at a time.
Our investment in multimodal molecular imaging is enabling us to uncover new insights into our drug targets, and to see the impact of our drug candidates on molecular and cellular pathways in ways that were previously impossible. With mass spectrometry, we have created detailed images of the deposition of asthma drugs in multiple structures of the lung over time. We have also mapped drug and metabolite distribution for combinations of targeted cancer therapies to evaluate their impact on the tumour microenvironment.
To further improve our ability to predict the effects of our drug candidates on humans, we are collaborating with world-leading experts in ‘organ-on-a-chip’ design, technology and biology to develop microphysiological systems (MPS).
Rapid progress with our genomics initiative and next-generation sequencing (NGS) are enabling us to identify novel targets and pathways within large patient populations. Indeed, we have already analysed more than 200,000 of the 2 million genomes we plan to explore by 2026, including 500,000 from our own clinical trials. This is supported by on our recent partnership with UK Biobank and Regeneron to sequence the genomes of 500,000 UK Biobank samples, to accelerate the largest widely-available ‘big data’ human sequencing resource.
AI is already helping us address our biggest challenges in chemistry and AI-based informatics are starting to convert ‘big data’ into valuable knowledge. The development of novel AI techniques are currently being applied in ongoing clinical trials to improve the identification and prediction of safety and tolerability signals. WATCHER, is an AI-driven system that notifies clinicians and study teams involved in clinical trials of potential safety issues. The system can ‘reason’ over clinical trial data to assess the risk of specific clinical events enabling the team to take appropriate action early.
“WATCHER is a decision support system being developed to continuously monitor specific patient risk and alert the study team early. Our aim is that Watcher will ultimately identify those patients at potential risk before they materialize and ameliorate this same risk.” says Dr Dónal Landers, Senior Director Physician OncTMU Early Clinical Development Director – iDecide Programme, CRUK Manchester Institute
Across R&D we will never be complacent; we will continually look beyond the way we develop drug candidates today and explore how we can best use emerging technologies to accelerate the design and testing of tomorrow’s medicines. With our novel drug discovery platforms, we are moving towards multiple classes of medicines that target the biology of disease in totally new ways. By replacing today’s conventions with tomorrow’s innovations, we are turning science fiction into science fact.