COMMENT Preclinical target validation using patient-derived cells Aled M. Edwards1, Cheryl H. Arrowsmith1, Chas Bountra2, Mark E. Bunnage3, Marc Feldmann4, Julian C. Knight5, Dhavalkumar D. Patel6, Panagiotis Prinos1, Michael D. Taylor7 and Michael Sundström8 on behalf of the SGC Open Source Target-Discovery Partnership*
The Structural Genomics Consortium (SGC) and its clinical, industry and disease-foundation partners are launching open-source preclinical translational medicine studies.
Structural Genomics Consortium (SGC), University of Toronto, 101 College Street, Toronto, Ontario M5G 1L7, Canada. 2 SGC, Nuffield Department of Clinical Medicine, University of Oxford, Old Road Campus Research Building, Roosevelt Drive, Oxford OX3 7LD, UK. 3 Worldwide Medicinal Chemistry, Pfizer Worldwide Research and Development, Cambridge, Massachusetts 02139, USA. 4 Kennedy Institute of Rheumatology, Roosevelt Drive, University of Oxford, Headington OX3 7FY, UK. 5 Wellcome Trust Centre for Human Genetics, University of Oxford, Roosevelt Drive, Oxford, OX3 7BN, UK. 6 Novartis Institutes for BioMedical Research, Basel CH-4002, Switzerland. 7 Division of Neurosurgery, Program in Developmental and Stem Cell Biology, Hospital for Sick Children, Suite 1504, 555 University Avenue,Toronto, Ontario M5G 1X8, Canada. 8 SGC, Department of Medicine, Karolinska University Hosptital and Karolinska Institutet, 171 76 Stockholm, Sweden. *See online for full author list. Correspondence to M.S. e-mail: [email protected] doi:10.1038/nrd4565 1
Although the annual number of new drug approvals is trending upwards, the number of ‘first-in-class’ therapies has remained relatively constant — often fewer than 10 per year. For such new medicines for ‘pioneer targets’, attrition in Phase II proof-of-concept clinical studies remains the biggest hurdle1, in large part because the target–disease associations derived from the currently dominant cell-line or animal preclinical models of disease often do not translate into clinical efficacy. It is increasingly appreciated that the use of disease models based on human samples is critical both to increase our understanding of pathophysiology and to reduce clinical attrition. However, securing regular access to well-annotated samples from patients is challenging to organize and raises ethical issues. Here, we propose that these issues can most easily be addressed by creating an open-source partnership.
methods were developed that enabled 90% of the total cells originating from the diseased joint to survive for 5–6 days was it possible to provide the first convincing evidence of the importance of TNF in joint inflammation, which was rapidly confirmed in animal models and then in proof-of-principle trials3. The discovery of anti-TNF therapy also provides two other lessons. First, success derived not only from the use of human samples by expert clinicians, but also from the use of excellent reagents — in this case, industry-derived antibodies specific for important mediators. Second, the speed by which the discoveries were translated into medicines was accelerated due to the openness of the scientists. This preclinical work was published immediately and, in addition (and unusually), the clinical finding that patients dramatically responded to a TNF-specific antibody was disclosed more than a year before publication.
Rationale for partnership The scientific need: predictive assays based on human tissue and primary cells. Preclinical studies with conventional proliferating cell lines and animal models of disease have usually proven unable to predict drug efficacy in human trials. In some cases, for example sepsis, hundreds of compounds are effective in animals but none have proven efficacious in humans2. The discovery of tumour necrosis factor (TNF) inhibitors for rheumatoid arthritis provides a classic example of where the use of human disease tissue helped define and validate a target, and also of the need to mimic the disease condition well. In the mid-1980s, synovial cultures from patients with rheumatoid arthritis were of the fibroblast-like synoviocytes, which are easy to culture but which comprise only 10% of the total cell population. The remaining inflammatory and immune cells, which need more specialized conditions, survive poorly in crude cell cultures and were routinely discarded. Only when culture
The organizational solution: an open partnership among academia, industry, hospitals and patient groups. Our aim is to identify and validate pioneer targets by profiling the highest-quality chemical and antibody tools in the most relevant, highest-quality assays based on patient cells or tissues. One main challenge is operational — selective, drug-like molecules or specific antibodies are challenging to discover and large well-annotated patient cohorts are rare and difficult to access. The other main challenge is organizational — the necessary ingredients are rarely all found in the same institution: industry usually has the most experience in the design and development of new chemicals or antibodies; the clinical academic community cares for the patients and provides deep disease expertise; and the academic research community customarily provides the molecular and technological insights necessary for mechanistic studies. The ideal way to discover better drug targets is to develop an organizational model that allows academics,
NATURE REVIEWS | DRUG DISCOVERY
VOLUME 14 | MARCH 2015 | 149 © 2015 Macmillan Publishers Limited. All rights reserved
COMMENT clinicians, foundations and drug hunters to collaborate seamlessly; however, this is not easily achieved in the current system. Accordingly, we are establishing an opensource target-discovery partnership in which pharmaceutical companies collaborate with academic scientists and clinicians to generate high-quality chemical and antibody probes (molecules that are potent, selective and active on cell systems) for unprecedented targets and to profile these probes in cells from patients with various diseases. The targets will be selected on the basis of novelty and tractability, with initial emphasis on epigenetics and protein kinases, and later emphasis on targets implicated by human genetic studies. The initial diseases and the participating centres will be selected based on the ongoing availability of blood and tissue, the degree of characterization of the patient cohort, the willingness to share reagents, data and knowledge without restriction, and the willingness to host a cell-assay team that works to industry-standard quality. The commitment to open-source and data sharing is a key feature of this plan, and is expected to accelerate the science, make data generation more transparent and thus more reproducible, reduce the costs and time associated with executing multi-institutional, multi-national and multi-sector collaborations, and alleviate ethical concerns that might arise when commercial and scientific interests are juxtaposed with patient samples.
Progress The partnership builds from the SGC, a public–private partnership that has proved able to generate and disseminate high-quality research tools. These and other high-quality tools will be distributed to a global network of hospital-affiliated laboratories, where they will be profiled using state-of-the-art, clinically relevant assays based on patient-derived cells. In Europe, as part of a new project supported by the Innovative Medicines Initiative, the partnership will focus on inflammatory and autoimmune diseases, including fibrosis, ankylosing spondylitis, lupus and myositis. Tissues and cells will be collected from well-characterized, managed patient cohorts in Oxford, UK, and at the Karolinska University Hospital in Sweden. Using cells from patients, we will generate robust, clinically meaningful assays to profile molecularly targeted probes using phenotypic and biomarker readouts (such as cytokine profiles), linking novel targets to new indications. Dedicated assay teams will ensure quality and reproducibility. In addition to the assays already established by our clinical partners, we will also develop innovative assays with potentially superior translational potential. This may involve morecomplex culture systems that more closely resemble human tissue — typically having more than one cell type and, in some instances, with cells embedded in matrix. In Canada, we will initially focus on oncology, with emphasis on cancers that can be cultured ex vivo, but that maintain key clinical disease features such as histo pathology, ability to engraft and clinical response to standard-of-care drugs. This approach is enabled by recent advances in the isolation and culture of cancer stem cells that can engraft, 3D cultures of organoids, and cultures grown on materials that mimic the extracellular
matrix, or with feeder cells that mimic elements of the tumour microenvironment. The use of these techniques makes it feasible to assay a wide panel of samples with the aim of covering, as much as is possible, the heterogeneity observed in the clinic. Responses to probes will be profiled, with monitored features including cell growth and viability, differentiation and senescence, response to standard-of-care drugs and epithelial–mesenchymal transition. Although still to be confirmed in systematic studies, we believe that this approach, which has worked in the past, may better predict efficacy in clinical trials, as primary patient cells better represent the genetic and clinical diversity of human disease. We also believe that, by comparing the same molecular tools across multiple diseases, we will gain tremendous insight into some of the shared molecular mechanisms of subtypes of disease, as well as for those common to cancer and inflammation.
Outlook and call for collaborators The two main objectives of our initiative are to increase our understanding of the molecular bases of cancer and inflammatory diseases (as well as other diseases as our network grows), and to identify specific targets for which pharmacological modulation ameliorates key disease phenotypes. For some diseases and cell types, such efforts may include the use of induced pluripotent stem cells; for example to generate neuronal cells from somatic cells of patients with neurological disease. Pilot projects are already underway with cells from patients with Rett syndrome. Importantly, all projects within the partnership must have a commitment to open access. No participant will file for patent protection on any result from the cellbased assays and all reagents will be made available to the research community without restriction. For assay results derived from clinical biospecimens, the public will receive aggregated, quality-assured and analysed data. Data and cells from individual patients, although anonymized, cannot be distributed freely owing to ethical considerations and associated legislation. The partnership currently comprises ten pharmaceutical companies and has research sites at the University of Oxford, the University of Toronto (Canada), the Karolinska Institute, ETH Zürich (Switzerland) and the University of Campinas (Brazil). Clinical nodes have been established at the Karolinska University Hospital, the Kennedy Institute in the UK, as well as at the Hospital for Sick Children, the University Health Network and the Montreal Neurological Institute in Canada. We expect the number of partners to increase as the project progresses, and that patient organizations will be interested in participating. Indeed, the CHDI Foundation has already joined the partnership to mount an open-source search for new drug targets for Huntington disease. 1. 2. 3.
Bunnage, M. E. Getting pharmaceutical R&D back on target. Nature Chem. Biol. 7, 335–339 (2011). Marshall, J. C. Why have clinical trials in sepsis failed? Trends Mol. Med. 20, 195–203 (2014). Brennan, F. M. et al. Inhibitory effect of TNFα antibodies on synovial cell interleukin-1 production in rheumatoid arthritis. Lancet 2, 244–247 (1989).
Competing interests statement
The authors declare no competing interests.
150 | MARCH 2015 | VOLUME 14
www.nature.com/reviews/drugdisc © 2015 Macmillan Publishers Limited. All rights reserved
The Structural Genomics Consortium (SGC) and its clinical, industry and disease-foundation partners are launching open-source preclinical translational medicine studies.
Structural Genomics Consortium (SGC), University of Toronto, 101 College Street, Toronto, Ontario M5G 1L7, Canada. 2 SGC, Nuffield Department of Clinical Medicine, University of Oxford, Old Road Campus Research Building, Roosevelt Drive, Oxford OX3 7LD, UK. 3 Worldwide Medicinal Chemistry, Pfizer Worldwide Research and Development, Cambridge, Massachusetts 02139, USA. 4 Kennedy Institute of Rheumatology, Roosevelt Drive, University of Oxford, Headington OX3 7FY, UK. 5 Wellcome Trust Centre for Human Genetics, University of Oxford, Roosevelt Drive, Oxford, OX3 7BN, UK. 6 Novartis Institutes for BioMedical Research, Basel CH-4002, Switzerland. 7 Division of Neurosurgery, Program in Developmental and Stem Cell Biology, Hospital for Sick Children, Suite 1504, 555 University Avenue,Toronto, Ontario M5G 1X8, Canada. 8 SGC, Department of Medicine, Karolinska University Hosptital and Karolinska Institutet, 171 76 Stockholm, Sweden. *See online for full author list. Correspondence to M.S. e-mail: [email protected] doi:10.1038/nrd4565 1
Although the annual number of new drug approvals is trending upwards, the number of ‘first-in-class’ therapies has remained relatively constant — often fewer than 10 per year. For such new medicines for ‘pioneer targets’, attrition in Phase II proof-of-concept clinical studies remains the biggest hurdle1, in large part because the target–disease associations derived from the currently dominant cell-line or animal preclinical models of disease often do not translate into clinical efficacy. It is increasingly appreciated that the use of disease models based on human samples is critical both to increase our understanding of pathophysiology and to reduce clinical attrition. However, securing regular access to well-annotated samples from patients is challenging to organize and raises ethical issues. Here, we propose that these issues can most easily be addressed by creating an open-source partnership.
methods were developed that enabled 90% of the total cells originating from the diseased joint to survive for 5–6 days was it possible to provide the first convincing evidence of the importance of TNF in joint inflammation, which was rapidly confirmed in animal models and then in proof-of-principle trials3. The discovery of anti-TNF therapy also provides two other lessons. First, success derived not only from the use of human samples by expert clinicians, but also from the use of excellent reagents — in this case, industry-derived antibodies specific for important mediators. Second, the speed by which the discoveries were translated into medicines was accelerated due to the openness of the scientists. This preclinical work was published immediately and, in addition (and unusually), the clinical finding that patients dramatically responded to a TNF-specific antibody was disclosed more than a year before publication.
Rationale for partnership The scientific need: predictive assays based on human tissue and primary cells. Preclinical studies with conventional proliferating cell lines and animal models of disease have usually proven unable to predict drug efficacy in human trials. In some cases, for example sepsis, hundreds of compounds are effective in animals but none have proven efficacious in humans2. The discovery of tumour necrosis factor (TNF) inhibitors for rheumatoid arthritis provides a classic example of where the use of human disease tissue helped define and validate a target, and also of the need to mimic the disease condition well. In the mid-1980s, synovial cultures from patients with rheumatoid arthritis were of the fibroblast-like synoviocytes, which are easy to culture but which comprise only 10% of the total cell population. The remaining inflammatory and immune cells, which need more specialized conditions, survive poorly in crude cell cultures and were routinely discarded. Only when culture
The organizational solution: an open partnership among academia, industry, hospitals and patient groups. Our aim is to identify and validate pioneer targets by profiling the highest-quality chemical and antibody tools in the most relevant, highest-quality assays based on patient cells or tissues. One main challenge is operational — selective, drug-like molecules or specific antibodies are challenging to discover and large well-annotated patient cohorts are rare and difficult to access. The other main challenge is organizational — the necessary ingredients are rarely all found in the same institution: industry usually has the most experience in the design and development of new chemicals or antibodies; the clinical academic community cares for the patients and provides deep disease expertise; and the academic research community customarily provides the molecular and technological insights necessary for mechanistic studies. The ideal way to discover better drug targets is to develop an organizational model that allows academics,
NATURE REVIEWS | DRUG DISCOVERY
VOLUME 14 | MARCH 2015 | 149 © 2015 Macmillan Publishers Limited. All rights reserved
COMMENT clinicians, foundations and drug hunters to collaborate seamlessly; however, this is not easily achieved in the current system. Accordingly, we are establishing an opensource target-discovery partnership in which pharmaceutical companies collaborate with academic scientists and clinicians to generate high-quality chemical and antibody probes (molecules that are potent, selective and active on cell systems) for unprecedented targets and to profile these probes in cells from patients with various diseases. The targets will be selected on the basis of novelty and tractability, with initial emphasis on epigenetics and protein kinases, and later emphasis on targets implicated by human genetic studies. The initial diseases and the participating centres will be selected based on the ongoing availability of blood and tissue, the degree of characterization of the patient cohort, the willingness to share reagents, data and knowledge without restriction, and the willingness to host a cell-assay team that works to industry-standard quality. The commitment to open-source and data sharing is a key feature of this plan, and is expected to accelerate the science, make data generation more transparent and thus more reproducible, reduce the costs and time associated with executing multi-institutional, multi-national and multi-sector collaborations, and alleviate ethical concerns that might arise when commercial and scientific interests are juxtaposed with patient samples.
Progress The partnership builds from the SGC, a public–private partnership that has proved able to generate and disseminate high-quality research tools. These and other high-quality tools will be distributed to a global network of hospital-affiliated laboratories, where they will be profiled using state-of-the-art, clinically relevant assays based on patient-derived cells. In Europe, as part of a new project supported by the Innovative Medicines Initiative, the partnership will focus on inflammatory and autoimmune diseases, including fibrosis, ankylosing spondylitis, lupus and myositis. Tissues and cells will be collected from well-characterized, managed patient cohorts in Oxford, UK, and at the Karolinska University Hospital in Sweden. Using cells from patients, we will generate robust, clinically meaningful assays to profile molecularly targeted probes using phenotypic and biomarker readouts (such as cytokine profiles), linking novel targets to new indications. Dedicated assay teams will ensure quality and reproducibility. In addition to the assays already established by our clinical partners, we will also develop innovative assays with potentially superior translational potential. This may involve morecomplex culture systems that more closely resemble human tissue — typically having more than one cell type and, in some instances, with cells embedded in matrix. In Canada, we will initially focus on oncology, with emphasis on cancers that can be cultured ex vivo, but that maintain key clinical disease features such as histo pathology, ability to engraft and clinical response to standard-of-care drugs. This approach is enabled by recent advances in the isolation and culture of cancer stem cells that can engraft, 3D cultures of organoids, and cultures grown on materials that mimic the extracellular
matrix, or with feeder cells that mimic elements of the tumour microenvironment. The use of these techniques makes it feasible to assay a wide panel of samples with the aim of covering, as much as is possible, the heterogeneity observed in the clinic. Responses to probes will be profiled, with monitored features including cell growth and viability, differentiation and senescence, response to standard-of-care drugs and epithelial–mesenchymal transition. Although still to be confirmed in systematic studies, we believe that this approach, which has worked in the past, may better predict efficacy in clinical trials, as primary patient cells better represent the genetic and clinical diversity of human disease. We also believe that, by comparing the same molecular tools across multiple diseases, we will gain tremendous insight into some of the shared molecular mechanisms of subtypes of disease, as well as for those common to cancer and inflammation.
Outlook and call for collaborators The two main objectives of our initiative are to increase our understanding of the molecular bases of cancer and inflammatory diseases (as well as other diseases as our network grows), and to identify specific targets for which pharmacological modulation ameliorates key disease phenotypes. For some diseases and cell types, such efforts may include the use of induced pluripotent stem cells; for example to generate neuronal cells from somatic cells of patients with neurological disease. Pilot projects are already underway with cells from patients with Rett syndrome. Importantly, all projects within the partnership must have a commitment to open access. No participant will file for patent protection on any result from the cellbased assays and all reagents will be made available to the research community without restriction. For assay results derived from clinical biospecimens, the public will receive aggregated, quality-assured and analysed data. Data and cells from individual patients, although anonymized, cannot be distributed freely owing to ethical considerations and associated legislation. The partnership currently comprises ten pharmaceutical companies and has research sites at the University of Oxford, the University of Toronto (Canada), the Karolinska Institute, ETH Zürich (Switzerland) and the University of Campinas (Brazil). Clinical nodes have been established at the Karolinska University Hospital, the Kennedy Institute in the UK, as well as at the Hospital for Sick Children, the University Health Network and the Montreal Neurological Institute in Canada. We expect the number of partners to increase as the project progresses, and that patient organizations will be interested in participating. Indeed, the CHDI Foundation has already joined the partnership to mount an open-source search for new drug targets for Huntington disease. 1. 2. 3.
Bunnage, M. E. Getting pharmaceutical R&D back on target. Nature Chem. Biol. 7, 335–339 (2011). Marshall, J. C. Why have clinical trials in sepsis failed? Trends Mol. Med. 20, 195–203 (2014). Brennan, F. M. et al. Inhibitory effect of TNFα antibodies on synovial cell interleukin-1 production in rheumatoid arthritis. Lancet 2, 244–247 (1989).
Competing interests statement
The authors declare no competing interests.
150 | MARCH 2015 | VOLUME 14
www.nature.com/reviews/drugdisc © 2015 Macmillan Publishers Limited. All rights reserved
