Phenotypic vs. target-based drug discovery for first-in-class medicines

of dietary supplements labeled as ephe- 5. James, W.P. et al. Effect of sibutramine on as demonstrated by imatinib (Gleevec)1 cardiovascular outcomes in overweight and dra free is an example of important and gefitinib (Iressa),2 and it has raised obese subjects. N. Engl. J. Med. 363, 905–917
work that protects the public by helping expectations for the majority of drug rid the garden of pharmaceutical weeds. 6. Consumer Reports. What's behind our dietary discovery to follow the same path.
supplements cover If laws such as DSHEA were reversed, Before the advent of target-based drug less weeding would be necessary.
discovery, new medicines were discov- ered by evaluating different chemicals CONFLICT OF INTEREST
7. Sesso, H.D. et al. Multivitamins in the against phenotypes—an organism's The author declared no conflict of interest.
prevention of cardiovascular disease in men: the Physicians' Health Study II randomized observable characteristics—in authen- controlled trial. JAMA 308, 1751–1760 (2012).
tic biological systems, such as animals 8. Park, S.Y., Murphy, S.P., Wilkens, L.R., Henderson, or cells. Many factors influenced the 1. Foster, L. et al. Multiple dosing of ephedra- B.E. & Kolonel, L.N. Multivitamin use and the free dietary supplements: hemodynamic, risk of mortality and cancer incidence: the shift from a phenotypic approach to a electrocardiographic, and bacterial multiethnic cohort study. Am. J. Epidemiol. 173,
target-based approach, including the contamination effects. Clin. Pharmacol. Ther. 93,
906–914 (2011).
idea that a rational, measurable progres- 267–274 (2013).
9. US Food and Drug Administration. FDA recalls 2. Kernan, W.N. et al. Phenylpropanolamine and adulterated dietary supplemen sion from gene to clinic to registration the risk of hemorrhagic stroke. N. Engl. J. Med. would increase research and develop- 343, 1826–1832 (2000).
December 2012).
ment success and productivity. Rational, 3. Colman, E. Food and Drug Administration'sw 10. Qato, D.M., Alexander, G.C., Conti, R.M., Obesity Drug Guidance Document: a short Johnson, M., Schumm, P. & Lindau, S.T. Use of informed target-based approaches use history. Circulation 125, 2156–2164 (2012).
prescription and over-the-counter medications molecular tools of genetics, chemistry, 4. Pearce, N. et al. Case–control study of prescribed and dietary supplements among older adults and informatics to drive drug discovery fenoterol and death from asthma in New in the United States. JAMA 300, 2867–2878
Zealand, 1977–81. Thorax 45, 170–175 (1990).
and also provide criteria and boundaries for choosing patient populations, set- ting doses, and quantitatively measur- Phenotypic vs. Target-Based
ing efficacy and toxicity. Unfortunately, this shift in approach has not yet trans- formed the industry.
Drug Discovery for First-in-Class
To investigate whether some strategies have been more successful than others in the discovery of new drugs, my group analyzed the discovery strategies and the MMOA for new molecular entities DC Swinney1
and new biologics approved by the US Food and Drug Administration between Current drug discovery strategies include both molecular
1999 and 2008 (Figure 1).3 Of the 259
agents approved, 75 were first-in-class and empirical approaches. The molecular approaches are
drugs with new MMOAs. Of these, 50 predominantly hypothesis-driven and are referred to as target-
(67%) were small molecules and 25 based. The empirical approaches are referred to as phenotypic
(33%) were biologics. The results also because they rely on phenotypic measures of response. a recent
show that the contribution of pheno- typic screening to the discovery of first- analysis revealed the phenotypic approaches to be the more
in-class small-molecule drugs exceeded successful strategy for small-molecule, first-in-class medicines. The that of target-based approaches—with
rationalization for this success was the unbiased identification of
28 and 17 of these drugs coming from the molecular mechanism of action (mmoa).
the two approaches, respectively—in an era when the major focus was on target- based approaches.
Drug discovery and development in the at specific molecular locations in human The first-in-class medicines discov- past quarter century has focused on the DNA were found to be responsible ered by phenotypic screening included promise of molecular medicine to iden- for some cancers, raising the hope of those discovered using animal models tify medicines to treat unmet medical developing successful therapies tai- such as ezetimibe (Zetia) for reduc- need by targeting specific gene prod- lored to individual patients. The gene- ing levels of blood cholesterol; those ucts. For example, mutations, or defects, to-medicine approach has had success, discovered with cellular assays such as vorinostat (Zolinza), the first his- 1Institute for Rare and Neglected Diseases Drug Discovery, Mountain View, California, USA. Correspondence: tone deacetylase inhibitor, which was reported to come from the observa- tion that dimethyl sulfoxide had an CliniCal pharmaCology & TherapeuTiCs VOLUME 93 NUMBER 4 APRIL
link a specific molecular mechanism of action (MMOA) to the desired pheno- type, drug discovery can focus efforts toward addressing specific hypotheses (Figure 2).
Each of these two approaches has its strengths and weaknesses, and advo- cates and detractors. Although phe- notypic approaches use semiempirical methods that do not require under- standing of the mechanism, they do require an understanding of biology to the extent that biomarkers that trans- late to human disease must have been identified. Additionally, it is difficult to accept the risk of moving a compound into development without some under- standing of mechanism to help evaluate dose–response relationships. Fortu- nately, there are many classic technolo- Figure 1 Distribution of new drugs discovered between 1999 and 2008, according to the discovery
gies that can aid in the identification of a strategy. The graph illustrates the number of new molecular entities (NMEs) in each category. Phenotypic screening was the most successful approach for first-in-class drugs, whereas target-based screening was new mechanism, including biochemical the most successful for follower drugs during the period of this analysis. The total number of medicines fractional isolation of activity and affin- discovered via phenotypic assays was similar for first-in-class and follower drugs—28 and 30, respectively. ity purification. This problem-solving The total number of medicines discovered via target-based screening was nearly five times higher for approach can also enable researchers follower drugs than for first-in-class drugs (83 vs. 17, respectively). Reprinted from ref. 3 with permission.
to utilize new molecular technologies of chemical biology, proteomics, and network biology. An example of a med- unexpected effect on cancer cells; and action" describes the way that biologi- icine whose predecessor was discov- those identified in bacterial assays cal parts collaborate to provide an effec- ered in a phenotypic assay is ezetimibe such as linezolid (Zyvox), an oxa- tive and safe medicine. Addressing the (Zetia), whose target was subsequently zolidinone antibiotic. Target-based MMOA would contribute to reversing identified using a genetic approach as successes included tyrosine kinase the low productivity of target-based the sterol transporter Niemann-Pick inhibitors for cancer, including gefi- discoveries because merely know- C1-Like 1 (NPC1L1).4 More recently, tinib (Iressa) (target, EGFR), imatinib ing the identity of a part involved in a Chung and co-workers from Glaxo- (Gleevec) (target, BCR-ABL), sorafenib defect may not be sufficient to repair a SmithKline demonstrated the use of (Nexavar) (target, Raf), and sunitinib malfunctioning machine. We postulate chemoproteomics to identify BET bro- (Sutent) (targets, VEGFR/PDGFR), and that a target-centric approach for first- modomains as the target for inhibitors antivirals, including maraviroc (Selzen- in-class drugs, without consideration of identified in phenotypic assays.5 try) (target, CCR5), raltegravir (Isen- an optimal MMOA, may contribute to An interesting question is whether tress) (target, HIV integrase), and the current high attrition rates and low more time and resources are required zanamivir (Relenza) (target, influenza productivity in pharmaceutical research to follow up empirical findings from and development.
phenotypic assays than to test multiple- Our previous paper3 proposed that These observations led to the proposal target hypotheses. In the phenotype lower productivity partly reflects target- that the progression of drug discovery approach, the early risk is decreased based discovery's lack of consideration from unmet medical need to best-in- as a result of the activity in a transla- of the molecular complexities of the class medicines is facilitated by the use tional phenotypic assay. Obviously, the drugs' action. Knowing the parts of an of phenotypic assays to identify first- predictive value of translational assays efficient machine—a watch, an automo- in-class medicines and their respec- for human biology must be tested and bile, or a computer—is not sufficient to tive MMOAs. Progression correlates validated for both approaches. The lack describe how it works. The parts must with an iterative increase in knowledge of understanding of mechanism may collaborate in precise ways to provide to specifically address a phenotype slow progression of the drug candidate accurate time, reliable transportation, related to the unmet medical need. because subsequent studies will need to or processed information.
Early in the progression, the knowl- be empirical. Perhaps more resources Biology is infinitely more complex. edge is achieved by empirical analysis. and time will be required earlier in order The phrase "molecular mechanism of Ideally, as more knowledge is gained to to understand the mechanism. With VOLUME 93 NUMBER 4 APRIL 2013 greater access to more reliable pheno- typic assays.
Our analysis found high success of phe- notypic approaches to small-molecule first-in-class drug discovery in an era when the majority of efforts were focused on molecular target–based approaches. This finding is surprising because mod- ern medical research is based on the assumption that a clearer understand- regardless of the ing of the molecular mechanism of dis- ease, enabled by genetic and molecular Iterative increase in knowledge
advances, would lead to an increase in new medicines. In drug discovery the preferred scenario has been that molecu- Progression from empirical to hypothesis testing
lar mechanisms associated with disease are represented by targets and that quan- titation of target modulation facilitates Figure 2 Progression of drug discovery from unmet medical need to best-in-class medicines. This
a more rational development. However, simplified schematic highlights the contribution of empirical approaches to first-in-class medicines, the mechanistic details to enable this hypothesis-driven approaches for best-in-class medicines, and the role of mechanistic understanding. Progression correlates with an iterative increase in knowledge to specifically address a phenotype approach are not always available, vali- related to the unmet medical need. Early in the progression, the knowledge is achieved by empirical dated, or sufficient for the specific medi- analysis. Ideally, as more knowledge is gained to link a specific mechanism of modulation to the desired cal need. It is unrealistic to assume that phenotype, drug discovery can focus efforts to address specific hypotheses. The relative timing of we can know the exact molecular and employing empirical vs. hypothesis-driven approaches is influenced by the validation of mechanistic mechanistic details of complex human understanding. MMOA, molecular mechanism of action.
diseases. Empirical analyses, including phenotypic assay, have been successful in the past and require fewer mechanistic target-based approaches, the mecha- that the vast majority of academic biol- assumptions. The challenge is to use an nistic hypothesis should enable rapid, ogy is also hypothesis-driven. At issue appropriate combination of empirical and measured progress to clinical proof- is the value of hypothesis-driven medi- mechanistic research and development of-concept studies, although it may be cal research for new discoveries. Is the to enable good ideas to successfully necessary to evaluate more than one hypothesis overvalued at the expense of move forward.
candidate target and MMOA to find a traditional empirical evaluation? It can CONFLICT OF INTEREST
winner. It is therefore possible that a tar- be argued that in seeking the best path to The author declared no conflict of interest.
get-based approach will add to the cost new medicines, academic science should of development because of the need to be focusing not on gene-based, hypoth- evaluate multiple hypotheses. Interest- esis-driven research but on translating 1. Capdeville, R., Buchdunger, E., Zimmermann, ingly, in current drug discovery discus- disease knowledge into disease-relevant J. & Matter, A. Glivec (ST571, imatinib), a sions the central feature of any approach phenotypic assays for screening and rationally developed, targeted anticancer is the level of mechanistic understanding chemical biology approaches to screen- drug. Nat. Rev. Drug Discov. 1, 493–502 (2002).
2. Barker, A.J. et al. Studies leading to the required to move a compound forward. ing and target identification as well as on identification of ZD1830 (Iressa): an orally However, an understanding of mecha- systematic approaches to understanding active, selective epidermal growth factor nism is not required for regulatory the MMOA. Even with the many new receptor tyrosine kinase inhibitor targeted to the treatment of cancer. Bioorg. Med. Chem. approval; the regulatory agencies are less technologies that are now available for Lett. 11, 1911–1914 (2001).
concerned with the MMOA of a com- phenotypic assays—e.g., high-content 3. Swinney, D.C. & Anthony, J. How were new pound than with whether it is effective.
screening with stem cells, primary medicines discovered? Nat. Rev. Drug Discov.
10, 507–519 (2011).
Both Paul Janssen and Sir James Black human cells, zebrafish, and Caeno- 4. Garcia-Calvo, M. et al. The target of ezetimibe emphasized the importance of using rhabditis elegans—in many therapeutic is Niemann-Pick C1-Like 1 (NPC1L1). Proc. Natl. assays that translate to the human dis- areas it is not routine to establish and Acad. Sci. USA 102, 8132–8137 (2005).
5. Chung, C. et al. Discovery and characterization ease.6,7 The decline in productivity in validate phenotypic assays that trans- of small molecule inhibitors of the BET family relation to research investment in the late effectively to human disease. This is bromodomains. J. Med. Chem. 54, 3827–3838
pharmaceutical industry has also been particularly evident with animal models 6. Black, J. Personal perspective on Dr. Paul matched by a similar decline in trans- whose predictability for human diseases Janssen. J. Med. Chem. 48, 1687–1688 (2005).
lational research in academia. A very is not always reliable. Greater focus on 7. Black, J. Learning by doing [interview]. Mol. important aspect of the debate is the fact translational research should lead to Interv. 4, 139–142 (2004).
CliniCal pharmaCology & TherapeuTiCs VOLUME 93 NUMBER 4 APRIL

Source: http://www.plengegen.com/wp-content/uploads/Swinney_ClinPharmTher_2013_phenotypic-screens.pdf