Vous êtes sur la page 1sur 3

Editorial

For reprint orders, please contact reprints@future-science.com

Network pharmacology for cancer drug


discovery: are we there yet?
“…network pharma­cology has allowed for a deeper understanding of these secondary
drug interactions, revealing that promiscuity often engages a synergistic combination of appropriate
high-value targets in cancer cells to produce treatment success.”

Keywords: bioinformatics n combination cancer drug discovery n de-risking drugs n network medicine n network pharmacology n synergistic drug
pairs n systems biology

In the era of targeted drug discovery, the major regimens [6] . The emergence of network biology
thrust has been on designing small-molecule has enhanced our knowledge of multi-pathway
drugs that show high binding affinity and speci- interactions in cancer and helped to make sense
ficity towards a druggable protein of interest [1] . of drug-response signature datasets that collec-
However, cancer is a complex disease arising tively decode the complex mechanisms of drug
from alterations in multiple pathways (biologi- action [7] . These holistic approaches have laid the
cal/protein networks) [2] that have evolved to be foundation for emerging concepts of ‘network
very robust [3] and designing drugs against single pharmacology’, which is solidifying its position
proteins has yet to yield optimal clinical success in cancer medicine. Unlike earlier reductionist
[4] . As they are fundamental organizing prin- ‘one drug, one target’ approaches, network phar-
ciples of life, the removal of individual compo- macology invokes the idea that a drug engages Asfar S Azmi
nents through single protein-targeted drugs from with multiple targets and, in fact, rarely with a Department of Pathology, Wayne
such networks has surprisingly little functional single protein in isolation. Lacking the proper State University School of Medicine,
consequence, even in pathological states of can- integrated tools, traditional molecular biology Karmanos Cancer Institute, 4100
John R, HWCRC Room 732,
cer. Such robustness is commonly observed with pooled these secondary interactions as off-target Detroit, MI, USA
many other types of highly optimized large-scale effects and many were categorized as toxic side Tel.: +1 3135768328
Fax: +1 3135768389
networks that have a general property to resist effects. Such a narrow view negatively impacted E-mail: azmia@karmanos.org
change. To have an impact, interventions within decisions regarding the clinical future of promis-
a biological network need to be multiple, but cuous drugs. However, network pharma­cology
highly selective. Most cancer drugs are designed has allowed for a deeper understanding of these
to target proteins having critical functions that secondary drug interactions, revealing that pro-
are structurally intricate and deeply seated in miscuity often engages a synergistic combination
complex networks. Predicting the functional out- of appropriate high-value targets in cancer cells to
come of interventions, such as those originating produce treatment success [8] .
from drug treatment, is not as straightforward as Networks are amenable to ana­lysis using several
originally thought. So-called ‘targeted drugs’, no branches of mathematics. The most simple way
matter how specific, are never chaste and usually of representing a biological network is through
show promiscuous behavior, targeting multiple graph points, more commonly termed nodes or
ligands at the same time [5] – hence, both disease vertices, which could be either genes, proteins or
understanding and drug discovery requires depar- even drugs connected by lines representing inter-
ture from a single protein or pathway-centric actions (called edges). Local and global proper-
approach to a more holistic multi-pathway or ties of this map can be evaluated and neighboring
network-centric approach [16] . substructures or the role of adjacent nodes can
Recent advancements in molecular biology and be inferred from information in patterns of con-
the emergence of integrated technologies, such nections/interactions. This information can be
as systems biology, have allowed rapid assess- used to identify sets of high-value nodes, some of
ment of large expression datasets from cancer which may serve as targets for drugs, depending
patients. These strategies have partially aided on the model of interest. Drugs generally exert
target-population stratification based on suscep- their effects through binding to proteins, thereby
tibility towards single or combination treatment modulating their activity. However, bioactive

10.4155/FMC.12.44 © 2012 Future Science Ltd Future Med. Chem. (2012) 4(8), 939–941 ISSN 1756-8919 939
Editorial | Azmi
compounds invariably influence more than one protein, pharmacology has also been utilized to discover anti-
either: as a consequence of structural similarities between cancer effects of drugs developed for other disease. For
the intended target and other proteins; through allosteric example, in a highly significant study, chemical systems
effects on other proteins; through pleiotropic mecha- biology demonstrated that pleiotropic effects against
nisms, where an interaction results in multiple down- cancer cells by Nelfinavir, a potent HIV protease inhibi-
stream effects on other proteins; or through multivalent tor, was due to weak inhibition of multi-kinase activity
target binding by different presentations of the active [15] . This example demonstrates the importance of this
molecule. Many prodrugs convert to active metabolites highly promising field, still in its infancy. Key points in
and these are subject to an exponential range of possible this area that need extensive evaluation include:
interactions within a target network or distantly unre- n Obtaining network insights on optimally synergistic

lated network proteins. In whatever way these polyvalent combinations of targets, where optimality relates to
interactions occur, the end result is often unpredictable holistic network impact and not only to binding affin-
efficacy or reduction in toxicity, or both. It is impor- ity towards a single targeted protein;
tant to know that the more highly specific and the less
promiscuous a drug is for a particular cancer target, n Selection of drugs that intervene upon the function of
the more important that target must be in the cancer weak links in the targeted pathway to exert the desired
network for it to have a significant effect. effect through searching chemo­proteomic databases
for existing drugs with the desired binding and pleio-
tropic footprints, and use of reverse chemical engineer-
“It is important to know that the more highly ing (i.e., generating novel compounds with predictive
specific and the less promiscuous a drug is for a multi-targeted footprints);
particular cancer target, the more important that n Computation of the potential impact of proposed
target must be in the cancer network compounds on normal cellular networks and selection
for it to have a significant effect.” of most optimal drugs yielding maximum efficacy
with minimum effects on healthy cells.

Network pharmacology takes into account all of the


aforementioned principles to optimize the efficacy and “Network pharmacology is one such
safety of a candidate drug and their potent combinations. advancement that holds promise to
The first component of network pharmacology is the revolutionize our approach towards
selection of optimal intervention points in the network(s) next-generation cancer drug design
of relevance to cancer subtype. Having extracted/curated and development.”
cancer-specific network data, combinatorial network
impact algorithms can be applied to calculate the opti- Drug discovery in the last two decades has pursued
mal combination of proteins targeting, which should a ‘black box’ approach in which the pharmaceutical
maximally affect the network(s) while inducing mini- industry handpicked targets and focused on develop-
mal or acceptable modulation in healthy cell networks. ing small molecules with the highest binding affinity
Such combinatorial impact analyses can be quantified and specificity towards the identified protein of inter-
by many different integrated methods to evaluate how est. Such reductionist single protein/pathway-centric
much cancer network integrity can be changed by any approaches flooded the market with many promising
specific intervention. Since cancer-related networks typi- high-affinity drugs and these were rapidly pushed for
cally relate directly to systems-level function or dysfunc- clinical evaluation. The cost of bringing one cancer
tion, incorporating combinatorial impact in drug design drug to the market ranges from US$2 to $4 billion.
is predicted to influence the disease. These staggering numbers cover the cost starting from
Many examples in the literature exist where network high-throughput screening to identification of high-
pharmacology has been applied to cancer drug discov- affinity compounds, large-scale chemical synthesis of
ery. The many applications of this technology include candidate drugs, extensive pre­clinical evaluations in
identification of weak nodes in global cancer networks different models, multispecies pharmacokinetic/tox-
[9] , predicting drug toxicity [10] , drug repurposing [11] icity profiling and finally, the most expensive, clinical
and identifying multi-scale mechanisms of drug action testing. Although such candidate drugs have strong
[12] . In this direction, we were among the first to dem- preclinical data supporting their anticancer potential,
onstrate the application of network pharmacology in they rarely pass the ultimate test at Phase II and III
the rational design of potent anticancer drug combina- clinical trials. The majority of these drugs either do
tions [13] . Recently, other groups have also performed not translate their efficacy in patients or elicit seri-
large-scale evaluations on synergistic pairs [14] . Network ous and sometimes fatal adverse events, leading to

940 Future Med. Chem. (2012) 4(8) future science group


Network pharmacology for cancer drug discovery: are we there yet? | Editorial
their withdrawal because of incomplete understand- more acceptable to molecular biologists and pharma-
ing of experimental evaluation or scientific explana- ceutical researchers will no doubt result in the use of
tion. Thus, thousands of candidate molecules in the network analyses as a priority tool in drug develop-
laboratory are reduced to just a couple for clinical use. ment. In conclusion, network pharmacology certainly
This consistent failure indicates that the field of drug has the potential to change our view of drug design
discovery needs to be totally revamped in its approach and understanding mechanisms of action and, if used
for designing newer molecules. Network pharmacology correctly, is predicted to significantly de-risk cancer
is one such advancement that holds promise to revolu- drug discovery.
tionize our approach towards next-generation cancer
drug design and development. Skeptics have argued Acknowledgements
that cancer is a heterogeneous biological process carry- The author wishes to give sincere thanks to FWJ Beck for proof
ing a number of variables, and drug behavior cannot be reading this article. This editorial is an overview of the theme dis-
predicted by mathematical or integrated network mod- cussed at the Molecular Medicine Tricon Conference 2012, Drug
eling alone. Recently, researchers have appreciated the Discovery and Development Channel, De-risking Drug Discovery.
power of integrated technologies, which has resulted A substantial proportion of the content of this article is taken ver-
in multidisciplinary interactions between molecular batim from the course materials prepared by Dr A V Whitmore,
biologist and bioinformaticists, albeit, still with some Dr J Wray and Professor M P Young of e-Therapeutics PLC and
skepticism. Therefore, proponents of this technology their contribution is gratefully acknowledged.
must be proactive in presenting convincing evidence
that promotes these emerging tools and rapidly merges Financial & competing interests disclosure
them into mainstream cancer drug discovery. These The author has no relevant affiliations or financial involvement
strategies may help revive some hastily discontinued with any organization or entity with a financial interest in or finan-
drugs and could also reduce the cost of new drugs cial conflict with the subject matter or materials discussed in the
entering the cancer pipeline. manuscript. This includes employment, consultancies, honoraria,
But are we there yet? Not quite; however, it is pre- stock ownership or options, expert t­estimony, grants or patents
dicted that within the next 10 years, newer technologi- received or pending, or royalties.
cal advances plus methodologies better developed to No writing assistance was utilized in the production of this
make ana­lysis software economical, easy-to-use and manuscript.

References 7 Schadt EE, Friend SH, Shaywitz DA. Pharmacol. Toxicol. 52, 505–521 (2012).
A network view of disease and compound 13 Azmi AS, Wang Z, Philip PA, Mohammad
1 Swinney DC, Anthony J. How were new
screening. Nat. Rev. Drug Discov. 8(4), RM, Sarkar FH. Proof of concept: network
medicines discovered? Nat. Rev. Drug. Discov.
286–295 (2009). and systems biology approaches aid in the
10(7), 507–519 (2011).
8 von Eichborn J, Murgueitio MS, Dunkel M, discovery of potent anticancer drug
2 Pache RA, Aloy P. A novel framework for the
Koerner S, Bourne PE, Preissner R. combinations. Mol. Cancer Ther. 9(12),
comparative ana­lysis of biological networks.
PROMISCUOUS: a database for network- 3137–3144 (2010).
PLoS ONE 7(2), e31220 (2012).
based drug-repositioning. Nucleic Acids Res. 14 Cokol M, Chua HN, Tasan M et al.
3 Larhlimi A, Blachon S, Selbig J, Nikoloski 39, D1060–D1066 (2011). Systematic exploration of synergistic drug
Z. Robustness of metabolic networks: a
9 Barabasi AL, Gulbahce N, Loscalzo pairs. Mol. Syst. Biol. 7, 544 (2011).
review of existing definitions. Biosystems
J. Network medicine: a network-based 15 Xie L, Evangelidis T, Xie L, Bourne PE. Drug
106(1), 1–8 (2011).
approach to human disease. Nat. Rev. Genet. discovery using chemical systems biology:
4 Gerlinger M, Rowan AJ, Horswell S et al. 12(1), 56–68 (2011). weak inhibition of multiple kinases may
Intratumor heterogeneity and branched
10 Wist AD, Berger SI, Iyengar R. Systems contribute to the anti-cancer effect of
evolution revealed by multiregion
pharmacology and genome medicine: a future nelfinavir. PLoS Comput. Biol. 7(4), e1002037
sequencing. N. Engl J. Med. 366(10),
perspective. Genome Med. 1(1), 11 (2009). (2011).
883–892 (2012).
11 Pujol A, Mosca R, Farres J, Aloy P. Unveiling 16 Young MP, Zimmer S, Whitmore AV. Drug
5 Peters JU, Hert J, Bissantz C et al. Can we
the role of network and systems biology in molecules and biology: network and systems
discover pharmacological promiscuity early in
drug discovery. Trends Pharmacol. Sci. 31(3), aspects. In: Designing Multi-target Drugs, RSC
the drug discovery process? Drug Discov.
115–123 (2009). Drug Discovery Series 21. Morphy JR, Harris
Today 17(17–18), 325–335 (2012).
12 Zhao S, Iyengar R. Systems pharmacology: CJ (Eds). Royal Soc. Chem. 32–49 (2012).
6 Anderson AR, Quaranta V. Integrative
network ana­lysis to identify multiscale
mathematical oncology. Nat. Rev. Cancer
mechanisms of drug action. Annu. Rev.
8(3), 227–234 (2008).

future science group www.future-science.com 941

Vous aimerez peut-être aussi