e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 5 ( 2 0 0 8 ) 1–4
available at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/ejps
Commentary
Integrating systems approaches into pharmaceutical sciences Hans V. Westerhoff a , Erik Mosekilde b , Christian R. Noe c , Anne Marie Clemensen b,∗ a b c
Department of Molecular Cell Physiology BioCentrum, Amsterdam Faculty of Biology, Vrije Universiteit, The Netherlands Biophysics Group, Department of Physics, The Technical University of Denmark, 2800 Lyngby, Denmark Institut für Pharmazeutische Chemie, der Universität Wien, Austria
a r t i c l e
i n f o
a b s t r a c t
Article history:
During the first week of December 2007, the European Federation for Pharmaceutical Sci-
Received 28 May 2008
ences (EUFEPS) and BioSim, the major European Network of Excellence on Systems Biology,
Accepted 28 May 2008
held a challenging conference on the use of mathematical models in the drug development
Published on line 14 June 2008
process. More precisely, the purpose of the conference was to promote the ‘Integration of Systems Approaches into Pharmaceutical Sciences’ in view of optimising the development
Keywords:
of new effective drugs. And a challenge this is, considering both the high attrition rates in
Modelling and simulation
the pharmaceutical industry and the failure of finding definitive drug solutions for many of
Systems biology
the diseases that plague mankind today. The conference was co-sponsored by the Ameri-
Drug development
can College of Clinical Pharmacology, the European Center for Pharmaceutical Medicine, and the Swiss Society of Pharmaceutical Sciences and, besides representatives from the European Regulatory Agencies and FDA, the meeting was attended by 75 industrial and some 45 academic participants. © 2008 Elsevier B.V. All rights reserved.
During the first week of December 2007, the European Federation for Pharmaceutical Sciences (EUFEPS) and BioSim, the major European Network of Excellence on Systems Biology, held a challenging conference on the use of mathematical models in the drug development process. More precisely, the purpose of the conference was to promote the ‘Integration of Systems Approaches into Pharmaceutical Sciences’ in view of optimising the development of new effective drugs. And a challenge this is, considering both the high attrition rates in the pharmaceutical industry and the failure of finding definitive drug solutions for many of the diseases that plague mankind today. The conference was co-sponsored by the American College of Clinical Pharmacology, the European Center for Pharmaceutical Medicine, and the Swiss Society
∗
Corresponding author. Tel.: +45 45253246; fax: +45 45931669. E-mail address:
[email protected] (A.M. Clemensen). 0928-0987/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.ejps.2008.05.011
of Pharmaceutical Sciences and, besides representatives from the European Regulatory Agencies and FDA, the meeting was attended by 75 industrial and some 45 academic participants. The conference provided the forum for an intensive and lively discussion on the position of modelling and simulation (M&S) in the pharmaceutical industries, now and in the future. Many scientific groups are already deeply engaged in the use of detailed mechanism-based models to better understand the biological processes underlying their experimental data, and participants of the conference were provoked with a number of paradigm shifting visions. Some saw the drug company of the future as being centered around an extensive activity which would drive drug discovery, development and trials. This was contrasted with descriptions of today’s com-
2
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 5 ( 2 0 0 8 ) 1–4
panies where M&S was largely isolated from the rest of the organization, and where there was little, if any, personal union between discovery and development. Refreshing was the view from Thomas Senderovitz, UCB Pharma that the industry should serve the patient before the share holder. Dr Senderovitz went on to explain how UCB Pharma involves patient groups in its drug development programme. The issue turned less controversial when the industry’s dependence on publicly financed training of its human capital and the importance of substantial publicprivate partnerships for the scientific and technological bases of new drugs, were highlighted. Examples of such partnerships are the 2 GD Innovative Medicines Initiative (IMI) of the European Commission and the 0.3 GD Top Institute Pharma of the Dutch Government, both involving many, though interestingly not all, leading pharma multinationals. Other examples are AstraZeneca’s Systems Biology Chair at the University of Manchester, supporting both research and training, and Novo Nordisk’s center for diabetes research and treatment in Oxford. It was broadly accepted that a basic shift in methodology is needed for the pharmaceutical industry to survive the challenges of increasing attrition rates and global competition. It was also noted that progress on finding drugs that cure present-day diseases has been extraordinarily slow, to the extent that one must worry whether the applied methods will be effective at all for future progress. The diseases of tomorrow were characterized as network diseases, requiring a shift from the reductionist approach to a systems oriented understanding of how function derives from interactions, and not from isolated components. New approaches will be necessary to empower functional genomics with modelling, simulation and systems concepts and to enable us to better understand the enormous complexity and the fantastic range of temporal and spatial scales involved in normal biological function, drug action and disease development. By virtue of the interconnectedness of the regulatory mechanisms in the living organism and because of the need of contributions from many different disciplines, Systems Biology may be too much of a challenge for individual pharmaceutical industries or academic institutions. A joint, pre-competitive initiative accomplishing the equivalent of ‘man on the moon’, or ‘sequencing of the human genome’ is, therefore, timely for Systems Biology, one of the important scientific challenges of the coming years. These ideas were met with due scepticism by some, finding the Genomics/Systems Biology band wagon premature and too focused on signal transduction. Genomics had not delivered as promised, and it might be better to start with disease modelling than with the study of intra- and intercellular processes. Admittedly, drug development for some diseases had been outright sloppy. However, attrition rates could be greatly reduced simply by better implementation of existing PK/PD modelling. Others suggested that a diversity of approaches will be important based, perhaps, on various forms of integration of disease modelling and PK/PD modelling with Systems Biology at the molecular, cellular, and tissue levels. The recently developed physiological and mechanismbased PK/PD methods may be seen as successful attempts of such a ‘middle-out’ integration.
Fig. 1 – Achievement of some of the aims of integrating systems biology with pharmaceutical sciences. Assessed was the percentage of the invited talks at the conference that met with the following criteria: (1) drug/therapy discovery and development, (2) modelling, (3) mathematical data analysis, (4) experimental, (5) bridging physical chemistry and biology, (6) enjoying industrial/user interest, (7) enjoying academic interest, (8) cross involvement of academia and industry/user, (9) involving a network of collaborators and (10) training.
Also otherwise the meeting was dynamic: 13 posters, illustrating simulation and modelling in the context of pharmaceutical research, were presented at a speaker’s corner during an informal gettogether of the participants. This was followed by poster tours and individual poster discussions. This arrangement affected a much better ‘take’ of the poster information than traditional poster sessions. The conference’s eight break-out sessions were devoted to: Education and training, Harmonising and sharing of highquality data, Software/libraries/systems for M&S, M&S for risk mitigation, A global regulatory strategy for model-based drug development, Translational pharmaceutical sciences, Integration of company research: What and when, and R&D Organisation: How and when. Each of these sessions led to concrete conclusions and recommendations, which were reported and discussed in pleno at the end of the conference. The success of the conference reaching its goal of ‘Integrating Systems Biology into the Pharmaceutical Sciences’, may be assessed by various criteria. Fig. 1 shows that many of these criteria, in our view, were met by most of the presentations. The critical issues of training, networking of researchers, and mathematical analysis of the data were among the least addressed. With respect to the issue of training, however, this was compensated by significant attention in the break-out sessions. The invited speakers at the conference represented a unique mix of heads of M&S departments in multinational pharmaceutical industries, scientists from regulatory agencies, and academic researchers. A couple of the speakers were personally involved in the experimental treatment of patients, based on insights gained from detailed mathematical models. The presentations were organised in a series of sessions covering various aspects of the interface between Systems Biology and Pharmaceutical Sciences.
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 5 ( 2 0 0 8 ) 1–4
The vice rector of the University of Vienna, the mathematician Heinz Engl, director of the Radon Institute of the Austrian Academy of Sciences, started the first session on ‘Approaching biosystems: How and when’ from the extreme end of the spectrum: He had earlier been engaged in modelling of iron production in a blast furnace, so hot that little could be measured inside. In his talk he showed how the mathematical techniques he developed for that system can now be used to deduce internal mechanistic aspects of the cell cycle and circadian rhythms by only measuring the external properties of the cell. The assumption of sparsity of connections applied in the solution of such inverse problems, was challenged in the subsequent discussion. Morten Colding-Jörgensen of NovoNordisk showed how he used relatively simple models, tailored to answer specific questions, to reach subtle, yet surprising conclusions. He particularly pointed to the possible role of neural control in the overall metabolic regulation of the body: There may be a direct neural control not only of the secretion of insulin, but also of the cellular response to insulin. Robert McBurney demonstrated the value of a molecular systems biology approach to solving problems in drug discovery and development through a study conducted at BG Medicine Inc. He identified networks of properties that correlate in the natural variability of mouse adipose tissue. These networks were different between diabetic and non-diabetic cases, and the challenge was to see if drugs such as rosiglutazone could redress these. The subsequent session dealt with virtual organs, possibly en route to the ‘virtual physiological human’. Blanca Rodrigez from the University of Oxford presented her work on ‘multiscale modelling of drug-induced effects on cardiac activity’. She showed how spatial heterogeneity of parameters caused the contraction waves in her model heart ventricle to break up into independent oscillators. The calculated patterns were very similar to what was observed experimentally through surface inspection with membrane potential probes. In line with the results of his simulations of ensembles of interacting nonlinear oscillators, Peter Tass from the Research Centre Jülich thereafter showed how synchronous oscillations in brain cells, mediating Parkinson disease, could be effectively desynchronized by means of a novel phase resetting technique, involving three phase shifted pulses of oscillatory stimulation. Compared with standard approaches to deep brain stimulation, the required power was significantly reduced, and preliminary results pointed to the possibility of long-lasting beneficial effects. Samsun Lampotang of the University of Florida then demonstrated, how in operating theatres used for training, his impressive ‘virtual patient’ technology is used to show what happens in patients undergoing anaesthesia. A similar approach was used to demonstrate to patients the importance of compliance with their prescriptions. The next session featured the disease component of model building, where Carolyn Cho of Pfizer in her talk on ‘a pathway-based strategy for discovering oncology therapeutics’ discussed how she correlated cues (external signals), signals and outputs at various molecular levels, using text mining and most recently also rule-based modelling. Igor Goryanin from the Edinburgh Centre for Bioinformatics, described his complete metabolic map for the human genome and showed how this facilitated understanding of the effect of
3
cyclooxygenase inhibitors on aspirin action. In his talk on ‘virtual populations and biomarker patterns in the cardiovascular physiology lab’, Entelos’ Thomas Paterson thereafter showed how physiology can be placed in the central position between genomics and disease, leading to a virtual-patient platform that can integrate large numbers of genetic and imaging markers and may provide the basis for a useful patient stratification. In the session on the population aspects of model building, Francis Lévi from INSERM in Paris, showed how orchestrating drug dosing both with the day–night rhythm and with appropriate gender modifications markedly improved certain forms of cancer therapy. This presentation sparked an intense discussion of the role of circadian effects in the treatment of other diseases. It was concluded that this problem deserves further attention. Roberto Gomeni of GlaxoSmithKline showed how the rather dramatic placebo response in trials for antidepressants could be recognized by Magnetic Resonance Imaging of the brain and managed by stratifying the population. Illustrating a well-functioning collaboration between academics, industry and regulatory agencies, Amin Rostami-Hodjegan of the University of Sheffield and Simcyp Ltd. and Steffen Thirstrup of the Danish Medicines Agency held a twin presentation. They showed how physiology- and mechanism-based PK/PD modelling could be used to implement all available information, for instance to support the requested renal co-morbidity investigation when the drug is secreted through the kidneys. The conference thereafter focused on the exposure-response component in pharmacometrics. Meindert Danhof of Leiden University started the presentations in this area with a talk on ‘state-of-the-art PK/PD modelling’ in which he outlined the main challenges relating to the use of models in drug discovery and development. In this connection he emphasized the significance of distinguishing between drug specific and system specific parameters and he also provided an example of how drug response mechanisms can be represented in PK/PD modelling. This was continued by Mats Karlsson from Uppsala University showing how similar principles enabled the pooling of data concerning agonists and antagonists in the test of drugs for prostate cancer. His way of incorporating prior information into his models provided a handle to mechanism-based modelling. Thorsten Lehr of Boehringer-Ingelheim Pharmaceuticals Inc. emphasized that a pharmacometrics data analysis environment is badly needed. He discussed how so-called ‘what-if’ scenarios can be used to optimize the design of trials and showed how neural net methodologies may be used, useful particularly when modelling needs to be fast. The subsequent session focused on the simulation of clinical trials. Carl Peck from the Center for Drug Development Science in Washington read a set of slides sent by Joga Gobburu of the US Food and Drug Administration. These showed that the FDA is critically engaged in M&S, for instance when investigating whether sleep maintenance rather than sleep initiation matters for the treatment of insomnia. As part of the documentation for a new drug, the FDA appears ready to accept and critically evaluate a broad range of different models. In his talk on ‘trial simulation to facilitate decision making in drug development’, Marc Pfister of the Bristol Myers Squibb Research Institute showed examples of model-based trial simulations that had been developed in his company. He also emphasized
4
e u r o p e a n j o u r n a l o f p h a r m a c e u t i c a l s c i e n c e s 3 5 ( 2 0 0 8 ) 1–4
the need for establishing a model-based drug development paradigm (MBDD) that could ensure a centralized platform for the systematic integration of information from different disciplines and sources. Johan Mouton of the Canisius-Wilhelmina Hospital explained how clinical trial simulations are different for antimicrobial drugs, because the drug’s receptor is associated with the microbe and does not sit in the human body itself. The activity of the antimicrobial drug can therefore be examined both in vitro and in animal models. This philosophy enabled the establishment of separately assessed ‘areas under the curve’ and ‘minimal inhibitory doses’. The last morning of the conference was devoted to presentation of some of the recent initiatives stimulating R & D in drug development and systems biology, and to an intensive wrap up session. Ole Bjerrum of the University of Copenhagen read slides sent by Arnd Hoeveler, European Commission on the Innovative Medicines Initiative, which will have an independent governing board and a stakeholders forum. Foci will include Safety, Knowledge Management, Education & Training, Efficacy, with a 0.25 GD first call defined at http://Ec.europa.eu/research/health/imi/index en.html. Daan Crommelin of the Top InstitutePharma in Leiden described this operative model for a pre-competitive private-public partnership bridging the gap between industry and academia, with 50% bonus funding from the Dutch government. Only proposals described on less than 5 pages are considered for funding. Adriano Henney then described how AstraZeneca was implementing Systems Biology, for instance leading to the discontinuation of a project modelled to be unsuccessful, thereby saving the company 20 M$. He referred to the Pharma 2020 vision of PriceWaterhouseCooper and the extensive report from UK’s Royal Society of Engineering and Academy of Medical Sciences, both concluding that a paradigm shift from reductionist and holist approaches to a systems oriented approach is needed. Zili Li of Merck & Co described the special and rapidly altering situation for the pharmaceutical industry in the World’s most populous country. The Chinese regulatory agencies are looking at new ways of interacting with drug companies in order to deal with the slow acceptance of drugs in China, caused by the required prior acceptance of the same drug abroad. Carl Peck of the US Center for Drug Development Science, with a substantial track record within the FDA, then highlighted the US critical path initiative, which is somewhat lame because of the absence of dedicated funding. Its in sil-
ico liver injury group is quite relevant, however, in view of liver systems biology activities in Germany and China. He also mentioned the FDA-UCSF joint course on drug development and regulatory sciences. FDA’s European sister organization EMEA was represented by Hans-Georg Eichler, who described the strongly increasing activities of the agency in the assessment of new technologies, and in post marketing Risk/Benefit assessment. M&S is mentioned in EMEA documents as important, but is not yet considered as candidate substitute for clinical trials. Erik Mosekilde of the Technical University of Denmark and Chair of BioSim led an extensive overview and discussion of what had been attained in the breakout sessions, and Hans Westerhoff of the Free University of Amsterdam and Manchester Centre for Integrative Systems Biology in his wrap-up showed how the conference had led to the recognition of some of the 10 commandments of systems-based drug discovery and development: It had dared to climb its Mount Sinai from the other side. He concluded that a Grand Action on Systems Biology for drug design is now called for. ESF, IMI-EC, EUFEPS, EFPIA, FEBS and BioSim may all be asked to get the ball rolling. The aims of the meeting, as clearly expressed by its chair and its co-chairs Hartmut Derendorf, University of Florida, Erik Mosekilde, Technical University of Denmark, and Christian R Noe, University of Vienna, have been amply met: To describe a visionary and effective integration of the new technologies “modelling”, “biomarkers”, “biosimulation” and “systems biology” into drug development activities, and to develop a foundation for further research, development and utilisation of these methodologies. When the conference broke up, it was evident that the pharmaceutical companies and academia present at the meeting intended to put the concepts emerging from the conference into some common practice. After all, both Systems Biology/Genomics and Pharmaceutical Sciences, as intriguing as they have been, have remained superficial: They have not yet succeeded in penetrating sufficiently into the heart of the matter, which is an engineering and systems level understanding of drug action on patient populations, informed by relevant knowledge from the chemical, physical and biological sciences. By starting from the interface between these approaches, the pharmaceutical industry may stand a fair chance of reversing the increase in the cost/benefit ratio of bringing effective, safe and affordable drugs to the market.
