Molecular Target Validation in preclinical drug discovery

A recent analysis of failures in Phase II and III trials in the past two years confirmed earlier reports that more than half of the drugs fail due to insufficient efficacy^1-3. In other words, the clinical target validation fails for about 50 per cent of therapeutic approaches. A retrospective analysis of drug development programs at Pfizer revealed some opportunities for optimisation of the drug development process⁴. Three knowledge pillars have been identified, which increase the likelihood of candidate survival in Phase II trials: deep understanding of the drug exposure at the site of action, target binding of the drug, and clear expression of functional pharmacological activity. The latest reached highest significance for prediction of success in clinical trials⁴. Hence, an in-depth biological understanding of a molecular target as one of the very early steps in the entire drug discovery and development process which can determine later success or failure of the emerging drug candidate is required.

Drug development often begins with identification of a novel molecular drug target candidate followed by detailed molecular target assessments consisting of experimental target validation and theoretical assessments of molecular druggability⁵. Preclinical drug target validation has the aim to increase confidence in a particular drug target. The process proves the initial hypothesis that a particular molecular target is key or even causative for pathogenic or symptomatic mechanisms in a disease. Several success factors seem to be of particular importance for the preclinical validation of a molecular target and are briefly discussed here.

Understanding the pathogenesis of the disease is key for successful target validation. Interdisciplinary research teams are working on breaking down a disease condition into its causative pathogenic events and to bridge the results from a variety of assays and models back to the clinical presentation.

The evolution of new technologies and data mining tools revealed the complexity and heterogeneity of previously thought monogenic disorders. In a brief letter, Kola & Bell provocatively called for reforming disease taxonomy⁶. The authors complained about the lack of recognition of disease heterogeneity in clinical development and medical practice. Indeed, clinical trials are becoming larger through increasing the number of patients enrolled in clinical trials to improve the ‘signal to noise’ ratio. This is supposed to counteract disease heterogeneity but certainly does not address the heterogeneity itself. A recent analysis of the European Federation of Pharmaceutical Industries and Associations confirmed that redefining diseases by their underlying molecular mechanisms is a top priority for successful drug development⁷. Clearly, proper stratification of patients based on their pathogenesis is quite powerful but requires deep disease understanding and a successful target validation followed by in-depth compound characterisation. Some cardiovascular and many inflammatory diseases are treated by a one-drug-one-target association, although there is increasing body of evidence that personalised medicine needs consideration⁸. For example, heart failure with preserved ejection fraction represents a major public health problem, but a specific treatment regimen has not yet been established. All patients are currently treated by therapeutic strategies established for patients who have heart failure with reduced ejection fraction⁸. Based on the underlying pathophysiological conditions, however, at least two distinct patient populations need to be considered when developing therapeutic approaches.

It is well accepted, though, that some cancer entities can only be successfully treated when multiple pathways and targets are addressed simultaneously as the cancer cell becomes successively addicted to growth stimulation and survival signals from multiple pathways, which can be brought about by one compound hitting multiple targets as the main aspect of successful poly-pharmacology, or by a combination of drugs hitting targets specifically in different pathways⁹.

There are examples where clinical observations led to new targets and even drugs subsequently. So the observation of individuals with CCR5 mutations, who are resistant against HIV, formed the bases for introducing CCR5 antagonists in that disease^10,11. Similarly, identification of the molecular basis of the Severe Combined Immunodeficiency Syndrome (SCID) validated Janus Kinase JAK3 as a promising drug target for immunosuppression¹².

Overall, basic research in the pathophysiology of diseases builds the basis for successful drug discovery. At the same time both clinical target validation as well as de-validation are important contributors to the growing pathophysiological understanding of diseases too. Examples are the application and neutralisation of cytokines in psoriasis, which led to new insights into the disease where TNF-alpha and IL12/23 seem to play a key role¹³.

Understanding the mechanism of action of a particular target and its network needs to be provided along with the preclinical target validation at the latest. Human data are crucial to gain confidence in the target by demonstrating pathway activity in diseased human tissue. So-called ‘experiments of nature’, which represent naturally occurring human conditions or states that modulate a biological target with a reproducible effect on human physiology, occupy a prominent position in the hierarchy of evidence to support the therapeutic hypothesis, with respect to both: potential beneficial and adverse events of a therapeutic targeting¹⁴. Again, interdisciplinary research involving basic and clinical scientists is a key prerequisite that generates best value and reflects the different perspectives on the molecular targets. Academic–industrial partnerships provide a unique opportunity to promote truly innovative projects¹⁵. The strategic alliance between the German Cancer Research Center and Bayer HealthCare is an example for an optimised drug discovery that facilitates deep knowledge on molecular mechanisms of cancer, innovative target ideas and high standard of technology platforms for target validation¹⁵. Some of the available tools are described below.

Smart usage of tools and models applicable for preclinical target validation together with qualitative research builds the technical requirement for success. Today, a variety of different methods, assays and models is being utilised for characterisation of disease mechanisms and potential drug target candidates. The techniques employ various -omics approaches, in vitro assays through the use of whole animal models to modulation of a desired target in diseased patients (Table 1). There is no one technique that would allow successful preclinical validation as each and every method has its limitations in terms of relevance for human and disease. Descriptive, often whole genome involving studies aim to identify molecular targets and networks regulated in diseased conditions. These data are often of human origin and provide a first valid hint but can bear targets with causative properties as well as biomarkers, which signal the response to a certain condition. Cell-based mechanistic studies in vitro reveal regulative characteristics of targets and pathways in which they are involved, while preclinical disease models are supposed to mimic the systemic implications of disease. Beside the various simplicity levels of assays and models, which have been summarised in Table 1, the importance of suitable tools needs to be highlighted. Preclinical target validation studies can employ either genetic knock-down or knock-out or by using target specific tools if either SMOL compounds or tool antibodies exist. It needs to be ascertained that whenever using SMOL compounds, those are administered at concentrations and doses selective for the molecular target of interest. Especially kinase inhibitors are unselective and prone to off-target effects. When dealing with BIOLs, neutralising antibodies are often species selective or need significant dose adjustment for use in mice.

Correspondingly, before using a certain preclinical model for target validation, the confidence in validity of the assay/model has to be established. This is particularly debatable when dealing with animal models, as numerous studies have addressed the translation ability of preclinical disease models^16-20. Assuming that functional orthology between rodents and man is given; knockout or transgenic animals can be used for target validation. Applying various methods, the researchers can limit the development of the target molecule in cellular or animal models (e.g. siRNA, shRNA, transgenic knockouts). If this has a positive effect on disease relevant parameters or the course of the disease, it can provide the first indication that inhibition of the target by drug therapy could also have a positive impact on the disease progression in humans. The difficulty lies in the complexity of the system and contributions from environmental conditions. Mouse studies often utilise in-bred strains of highly homogenous geno- and phenotype as opposed to the variability in genetic pool of higher laboratory species and finally in the patients. The housing factor is well controlled in animal husbandry as well. A recent study demonstrated that baseline tumour growth and immune control in laboratory mice were significantly influenced by subthermoneutral housing temperature¹⁹. These findings underline the fact that investigating mouse models under a single set of environmental conditions may lead to a misunderstanding of the antitumour immune potential.

Moreover, the time dependent course of disease needs to be carefully established. Chronic diseases especially evolve with significant consumption of time in humans and translation into murine life-span appears challenging. When modelling pathogenic conditions, caution towards applied methods needs further consideration. For example, a recent study by Seok et al.²⁰ questions the use mouse models for sepsis and inflammation and gained significant attention by a commenting article in the New York Times. Other researchers compared the inflammatory conditions applied by Seok et al. and found that the dose of the inflammatory stimulus was not appropriate to induce the conditions observed in humans and thereby limited the outcome of the comparison²¹. Once more, careful study design is needed for conclusive models. Finally, certain processes are primate specific and might be missed in rodent models although other pathogenic and mechanistic aspects of the disease can be simulated. To address the very high human specificity of certain pathways, genetically humanised mouse models can be applied replacing a mouse target gene by the human counterpart²². For example, MHC class I humanised mice have proved useful tools for investigating infectious and autoimmune diseases.

Improvements in animal modelling will hopefully facilitate preclinical target validation and enhance the quality and likely success of experimental drugs. Translational biomarker strategies bridging the gap between animal models and clinical studies have been proposed to increase the predictivity of disease models¹⁷. For example, urinary fibrinogen levels were found significantly up-regulated after kidney ischemia / reperfusion induced injury in rats, compared with controls, as well as in patients with clinically established multifactorial acute kidney injury, compared with healthy volunteers²³. Applying urinary fibrinogen as a biomarker for monitoring of disease and therapeutic efficacy seems predictive in preclinical models of acute kidney injury.

Early safety de-risking of a novel therapeutic approach can and should be addressed during preclinical target validation. Due to pleiotropic effects, the same target may have different functions in various organ systems or at different time points during development and adulthood. It is therefore helpful to have a look at the expression level of the desired target throughout the human body. Although there are some exceptions, it can be assumed that the broader the expression, the higher the risk is for adverse events when the drug has to be administered systemically. If genetic deficiencies of the target of interest are known, phenotypic data provide an informative tool for assessment of potential target-mediated adverse events¹⁴. Differential expression in samples representing human disease versus healthy controls is an additional parameter contributing to an early assessment of putative target related adverse events. As an example, the extremely high tolerability of proton pump inhibitors for the treatment of gastric reflux diseases is largely explained by the near exclusive expression of their molecular target – the gastric H+/K+ ATPase – in gastric mucosa. When using genetically modified animals for target validation, further hints for potential adverse events can be obtained from phenotypic information on particular knockout mice. The relative importance of these descriptive criteria, however, varies between the individual indications. Obviously, the tolerability for adverse events is considerably higher in life threatening oncologic conditions than in less devastating diseases.

A reliable data package must be assembled using appropriate standards and controls by considering the needs of particular disease. Unfortunately, relying on pre-published data is often not sufficient to gain confidence in a molecular drug target candidate. An internal analysis of early research projects at Bayer revealed that literature data on potential drug targets should be viewed with caution, as we observed a general low rate of reproducibility across indications²⁴. Indeed only in around 20 – 25 per cent of our projects were the relevant published data completely in line with our in-house findings²⁴. Similar findings were reported by Amgen²⁵. These results underline the importance of confirmatory validation studies for pharmaceutical companies and academia before larger investments are made e.g. in assay development, high-throughput screening campaigns, lead optimisation and animal testing. The data further implies the need for high quality of preclinical research. Especially, using appropriate negative and positive controls increases confidence in data. An international consortium recently appealed for rigorous study design in preclinical research²⁶. The authors identified a core set of reporting standards to optimise the predictive value of preclinical research within the key parameters of randomisation, blinding, sample size and data handling²⁶.

Conclusion

Although final target validation usually occurs in Phase II clinical trials, a rigorous conducted preclinical target validation significantly increases the confidence in molecular drug target candidates and reduces the overall costs in drug discovery process, as the failure of potential targets can be shifted to the least expensive research phase before major investments are made in screening / panning and lead optimisation.

References

1. Arrowsmith, J. and P. Miller, Trial watch: phase II and phase III attrition rates 2011-2012. Nat Rev Drug Discov, 2013. 12(8): p. 569
2. Arrowsmith, J., Trial watch: Phase II failures: 2008-2010. Nat Rev Drug Discov, 2011. 10(5): p. 328-9
3. Arrowsmith, J., Trial watch: phase III and submission failures: 2007-2010. Nat Rev Drug Discov, 2011. 10(2): p. 87
4. Morgan, P., et al., Can the flow of medicines be improved? Fundamental pharmacokinetic and pharmacological principles toward improving Phase II survival. Drug Discov Today, 2012. 17(9-10): p. 419-24
5. Gashaw, I., et al., What makes a good drug target? Drug Discov Today, 2011. 16(23-24): p. 1037-43
6. Kola, I. and J. Bell, A call to reform the taxonomy of human disease. Nat Rev Drug Discov, 2011. 10(9): p. 641-2
7. Forda, S.R., et al., Priorities for improving drug research, development and regulation. Nat Rev Drug Discov, 2013. 12(4): p. 247-8
8. Volzke, H., et al., Personalized cardiovascular medicine: concepts and methodological considerations. Nat Rev Cardiol, 2013. 10(6): p. 308-16
9. Knight, Z.A., H. Lin, and K.M. Shokat, Targeting the cancer kinome through polypharmacology. Nat Rev Cancer, 2010. 10(2): p. 130-7
10. Hendel, H., et al., Contribution of cohort studies in understanding HIV pathogenesis: introduction of the GRIV cohort and preliminary results. Biomed Pharmacother, 1996. 50(10): p. 480-7
11. Dhami, H., et al., The chemokine system and CCR5 antagonists: potential in HIV treatment and other novel therapies. J Clin Pharm Ther, 2009. 34(2): p. 147-60
12. Russell, S.M., et al., Mutation of Jak3 in a patient with SCID: essential role of Jak3 in lymphoid development. Science, 1995. 270(5237): p. 797-800
13. Laws, P.M. and H.S. Young, Current and emerging systemic treatment strategies for psoriasis. Drugs, 2012. 72(14): p. 1867-80.
14. Plenge, R.M., E.M. Scolnick, and D. Altshuler, Validating therapeutic targets through human genetics. Nat Rev Drug Discov, 2013. 12(8): p. 581-94
15. Wellenreuther, R., et al., Promoting drug discovery by collaborative innovation: a novel risk- and reward-sharing partnership between the German Cancer Research Center and Bayer HealthCare. Drug Discov Today, 2012. 17(21-22): p. 1242-8
16. Dolgin, E., Animalgesic effects. Nat Med, 2010. 16(11): p. 1237-40
17. Wendler, A. and M. Wehling, The translatability of animal models for clinical development: biomarkers and disease models. Curr Opin Pharmacol, 2010. 10(5): p. 601-6
18. Cook, N., D.I. Jodrell, and D.A. Tuveson, Predictive in vivo animal models and translation to clinical trials. Drug Discov Today, 2012. 17(5-6): p. 253-60
19. Kokolus, K.M., et al., Baseline tumor growth and immune control in laboratory mice are significantly influenced by subthermoneutral housing temperature. Proc Natl Acad Sci U S A, 2013. 110(50): p. 20176-81
20. Seok, J., et al., Genomic responses in mouse models poorly mimic human inflammatory diseases. Proc Natl Acad Sci U S A, 2013. 110(9): p. 3507-12
21. Cauwels, A., B. Vandendriessche, and P. Brouckaert, Of mice, men, and inflammation. Proc Natl Acad Sci U S A, 2013. 110(34): p. E3150
22. Scheer, N., et al., Generation and utility of genetically humanized mouse models. Drug Discov Today, 2013. 18(23-24): p. 1200-11
23. Krishnamoorthy, A., et al., Fibrinogen beta-derived Bbeta(15-42) peptide protects against kidney ischemia/reperfusion injury. Blood, 2011. 118(7): p. 1934-42.
24. Prinz, F., T. Schlange, and K. Asadullah, Believe it or not: how much can we rely on published data on potential drug targets? Nat Rev Drug Discov, 2011. 10(9): p. 712
25. Begley, C.G. and L.M. Ellis, Drug development: Raise standards for preclinical cancer research. Nature, 2012. 483(7391): p. 531-3
26. Landis, S.C., et al., A call for transparent reporting to optimize the predictive value of preclinical research. Nature, 2012. 490(7419): p. 187-91

Biographies

Isabella Gashaw is Senior Scientist / Group Leader at Bayer HealthCare’s Hematology Research Department based in San Francisco. She leads and supports hematology projects on rare blood and bleeding disorders by pre-clinical pharmacology at different phases of the drug discovery process. Before this position, she was involved in preclinical drug discovery for gynecologic diseases at Bayer. Isabella received her PhD in pharmacology from the Humboldt University in Berlin. Prior to joining Bayer in 2008, she researched different aspects of reproductive health disorders at the University Hospital in Essen (Germany).

Khusru Asadullah is Vice President and Head of Global Biomarkers at Bayer HealthCare Pharmaceuticals. His group is responsible for the discovery and development of biomarkers and according assays, including companion diagnostics, for all indication areas of Bayer. Before this, he headed Target Discovery and Inflammation and Immunology research at Bayer Schering Pharma. He is a Professor of Medicine at the University Hospital Berlin. He graduated from the Medical School Charité, Berlin and the Harvard Business School, Boston. He is a board certified Dermatologist and Immunologist.

Martin Bechem is Senior Vice President heading the Cardiology & Hematology Therapeutic Research Group at Bayer, which initiates and realises all cardiology and hematology research projects in close cooperation with other research functions. Martin joined Bayer in 1984 and was directing Central Nervous System research before he has headed the Molecular Screening Technology Department at Bayer’s research center in Wuppertal (Germany). He studied biology at Bochum University, and wrote his PhD thesis on single-channel investigations of the heart’s calcium channels.

Karl Ziegelbauer is Senior Vice President of Global Drug Discovery in the Therapeutic Research Group Oncology / Gynecological Therapy at Bayer. He is responsible for research activities and the early portfolio in oncology and gynecological therapy. Working together with other research functions, his team identifies targets and disease mechanisms, and drug candidates are then optimised, pharmacologically characterised and made available for further development. Starting at Bayer as a research scientist in 1993, he held positions with increasing responsibility in several different areas and locations including anti-infectives, immunology in Japan, as well as alliance management and supporting product related research for marketed products.