To avoid becoming a commodities industry, combinatorial chemistry companies
are broadening their technology base.
Since Nature Biotechnology published the first-ever map of the combinatorial
chemistry universe based on Biovista data1, both the biotechnology
and pharmaceutical industries have continued to recognize combinatorial chemistry
as a must-have technology for the drug discovery business. But the way that
combinatorial chemistry is fitting into the drug discovery pipeline has changed.
The threat of becoming a commodities industry has heightened awareness of
the need to integrate practicing companies into the drug discovery process
as value-added strategic partners. This has resulted in several collaborations
between combinatorial chemistry companies and other biotechnology players
to provide a more robust array of technologies for drug development partnering.
Examples include deals between now defunct Cadus (Tarrytown, NY) and ArQule
for screening of leads in signal transduction, and Tripos (St. Louis, MO)
and MDL Information Systems (San Leandro, CA) in integrated combinatorial
software. Such developments in business-model approaches, coupled to significant
technical advances in the interim, are keeping this exciting field at the
forefront of drug discovery today.
Historical perspective Combinatorial chemistry, also known as test-tube evolution or applied molecular
evolution, owes its origins to the arrival of the principles of molecular
evolution almost 30 years ago, when RNA evolution experiments2
and theories on the self-organization of biopolymers3 suggested
how it was possible to "evolve" by iteration successive generations
of biopolymers from previous generations, with different properties. By the
mid-1980s, solid-phase synthesis of peptides was possible4,
setting the stage for automated large-scale synthesis of peptides of interest.
The development of PCR and nucleic acid synthesizers in the early 1990s extended
the possibilities further, and steady advances ever since in hardware, chemistries,
and software have created the vast platform technology which combinatorial
chemistry is acknowledged to be today.
Current state The pipeline for potential drug targets has been flooded through the parallel
public and private efforts to sequence both the human genome and the genomes
of single-cell pathogens. Advances in understanding molecular mechanisms such
as programmed cell death (apoptosis), signal transduction, telomere control
of chromosomes, cytoskeletal development, modulation of stress-related proteins,
and cell surface display of antigens by the major histocompatibility complex
molecules have provided a rational basis for sifting through these targets
to pick the most promising points of intervention. Based on these advances,
finding potential drug targets is no longer the bottleneck in drug development.
Developing lead candidates that can be used to rapidly validate these targets
and move them into clinical trials remains the black box in drug development.
Combinatorial chemistry is having its major impact in speeding this transition
from target identification to validated lead (see Lead validation, pp. 47−49). This is principally
due to its ability to generate leads at much higher magnitudes of scale. Using
traditional methods, it takes, on average, one medicinal chemist one month
to generate four compounds directed against a particular target—at a
total cost of about $30,000, or $7,500 per compound (see
Table 1). By comparison, combinatorial chemistry applied by one chemist
over one month can produce 3,300 compounds for $40,000, or about $12 per compound.
Although these data are from a 1996 survey, the differences between the two
approaches are even more stark today, as the economies of scale favor automation.
The efficiency of these methods is easily calculated: 1,000 times more compounds
than traditional methods for at least 600 times less cost per compound. What's
more, it is possible to buy ready-made combinatorial libraries built around
specific molecular themes and consisting of many thousands of compounds, and
test these libraries in high-throughput screening systems using automated,
off-the-shelf instrumentation and reagents5.
From a business perspective, depending on which side of the table you are
sitting on, perhaps the major negative consideration about this technology
is that barriers to entry to access this heightened efficiency is low. Because
it is available to anyone for a modest capital investment, it has come to
be regarded as a commodity. The associated low profit margins and high volumes
in this type of industry are forcing a gradual weeding out of companies that
adopt this model.
Industry challenges A key challenge to the industry has always been to convert the promise
of the field to commercial value. Combinatorial chemistry companies that are
healthy have sold themselves as a platform technology and have followed a
business model that generates value through strategic alliances with multiple
partners. They add value through generating not only molecules, but by validating
targets and optimizing leads. Recent estimates of the value of alliances of
this type for the 30 companies for which financial data are available exceeds
$2.6 billion. At present, more than 130 companies worldwide are engaged in
combinatorial chemistry-driven alliances, out of a total of about 180 companies
involved in the field6.
These alliances have as their objective the sourcing of compound libraries,
and/or the generation of new leads against specific targets, and/or the optimization
of already identified leads for particular targets. Examples of sourcing alliances
include those made by Axys, ArQule, Pharmacopeia, Dyax (Cambridge, MA), Tripos,
and CombiChem (San Diego, CA). Alliances that focus on generating leads against
known targets through combinatorial chemistry are based on the proprietary
capability to manipulate a molecular scaffold to develop new leads. Examples
of alliances in this category are those of Chiron (Emeryville, CA), Isis (Carlsbad,
CA), and Cadus. Finally, some alliances focus on lead optimization. Starting
with a lead that is known, such as a small peptide, the lead's activity
is optimized through combinatorial chemistry. In the case of peptides this
could be accomplished through phage display and/or peptidomimetic combinatorial
chemistry. Examples include alliances of Ontogen (Carlsbad, CA), Dyax, and
ArQule.
A review of the publicly available information on combinatorial chemistry
alliances reveals that there is a direct correlation between the number of
targets that the alliance is focusing on (see Table 2).
An example of a significant recent commitment is the $47 million paid by American
Home Products (Madison, NJ) to Affymax (Santa Clara, CA) to develop leads
against 11 targets. R&D investment in the rest of the field ranged from
$1 million to this amount or more. While this may not be surprising, the data
suggest that developing the critical mass to offer one-stop shopping for a
major partner confers significant leverage in the field (see
Table 3).
Table 2. Financial characteristics of combinatorial chemistry alliances
There also seems to be a high confidence factor across the board associated
with this field based on an average of $12 million in up-front financial commitments
on each deal through a combination of equity and up-front payment. This is
not the norm, given that typical biotech alliances now are not "front-loaded",
but have payment terms that are contingent on the achievement of defined goals
and milestones. This suggests that there is significant confidence on the
part of the sponsoring party toward the combinatorial chemistry company, and
the sponsor is therefore prepared to pay up-front for the lead discovery program.
Backing up this hypothesis is the fact that investments made in the form of
equity in this field are done at a significant premium over the market price
for equity that was raised in the last round of financing. On average, a 26%
premium was paid, demonstrating again that sponsor companies acknowledged
the value of the alliance and the potential of the companies they were investing
in.
Clinical progress The fact remains that despite its almost decade-long commercial development,
to date no drugs on the market have been discovered by the exclusive application
of combinatorial chemistry. There are, however, a large number of leads—over
350—in various stages of pre-clinical development from the 180+ companies
that are active in the field, as well as leads being developed directly through
the application of combinatorial chemistry. Thus, it is only a matter of time
before a drug developed from combinatorial beginnings enters the market.
Because the information is proprietary, it is difficult to know exactly
which compounds now in clinical trials were derived exclusively from combinatorial
chemistry programs unless the company publicly discloses the information.
Some examples of these disclosures include Eli Lilly's small molecule
agent for the central nervous system in human phase II/III trials, which took
less than two years to identify and enter into trials. In addition, Pfizer
also has a compound generated through library generation and screening in
phase I/II trials, and Magainin Pharmaceuticals (Exton, PA) has developed
a novel antibacterial drug lead through the application of combinatorial chemistry
that is also in phase I/II trials. Finally, Trega has a lead candidate for
pain and asthma in early trials derived from its combinatorial chemistry program.
Future directions It is widely perceived that the field is at the leading edge of innovation.
On the hardware front, there are numerous players and instruments to fit every
need and budget. Examples include Hewlett Packard's (Meriden, CT) HP
7686 Synthesizer, a synthesizer based on a collaboration between Perkin-Elmer
(Norwalk, CT) and Tecan (Switzerland), Advanced ChemTech's (Louisville,
KY) ReacTech semiautomated synthesizer, Argonaut's (San Carlos, CA) Quest
system, Charybdis' (San Diego, CA) Calypso system, Robbins Scientific
(Sunnyvale, CA) FlexChem synthesizer, and many others. These systems, all
well within reach of most biotechnology companies today, ensure that the practice
of basic combinatorial chemistry will continue to be cost-effective.
On the chemistry and analytical side, there are numerous developments to
watch for. The future will see new chemistries based on solution reactions
that use photogenerated reagents now being developed, and these may turn out
to be quite efficient for the synthesis of organic compound libraries that
are easy to deconvolute7. Furthermore, a recent report describes
the use of combinatorial chemistry to generate mimetics of peptide secondary
structure8, as opposed to simple primary structure variants.
In addition, platforms exist today to perform the screening of multi-million
member combinatorial libraries in 1,536-well assay formats, illustrating the
parallel development of the combinatorial chemistry and high-throughput screening
fields9.
In addition, the use of computers to carry out virtual combinatorial chemistry
will become increasingly common, validating approaches, scaffolds, and linking
chemistries before more expensive "wet lab" experiments are undertaken10. This will also link this field to the emerging area of cheminformatics,
which annotates small molecules and also libraries with structure−function,
synthesis, and all other relevant data used to design and develop better drugs.
In addition to the development of new library chemistries, computer approaches,
and assay formats, the deconvolution of libraries will benefit from advances
in analytical capabilities, such as mass spectrometry11. This
is important if the field is to continue to rely on more and more compounds
being made and tested in parallel, which will require the analytical methods
to keep pace.
Finally, interesting new ways to generate or analyze libraries will constantly
appear. An interesting recent approach is that of using polymeric matrices
that swell in organic solvents and keep specified geometric shapes12.
The discrete shapes themselves of these miniature "monoliths",
as they are called, enable the rapid and easy deconvolution of libraries essentially
by shape discrimination. Because it is easy to load these structures with
high amounts of reactants, it is envisaged that the future of this approach
holds significant promise.
Conclusions Combinatorial chemistry companies have sought to overcome the "commodities"
business model by integrating themselves into the drug discovery pipeline
as a value-added partner. Although combinatorial chemistry has yet to boast
a specific drug on the market, the pervasiveness of the technology in all
major pharmaceutical and biotechnology companies suggests that it is only
a matter of time before the current pipeline delivers on that promise13. It is likely that as combinatorial chemistry instrumentation becomes
commonplace, those companies that are the first to deliver drugs to market
based on their proprietary abilities to validate targets and optimize leads,
will capture significant market share. For those companies that are not part
of the team that brings leads to market quickly, their best opportunities
may lie in their ability to gather critical mass through biotech-to-biotech
collaborations so that they can offer a more robust "second-wave"
technology to the drug development community.
