A Revolution in R&D
A Revolution in R&D
A Revolution in R&D
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
A <strong>Revolution</strong> <strong>in</strong> R&D<br />
HOW GENOMICS AND GENETICS ARE TRANSFORMING<br />
THE BIOPHARMACEUTICAL INDUSTRY<br />
BCG REPORT
The Boston Consult<strong>in</strong>g Group is a general management consult<strong>in</strong>g firm<br />
that is a global leader <strong>in</strong> bus<strong>in</strong>ess strategy. BCG has helped companies<br />
<strong>in</strong> every major <strong>in</strong>dustry and market achieve a competitive advantage by<br />
develop<strong>in</strong>g and implement<strong>in</strong>g unique strategies. Founded <strong>in</strong> 1963, the<br />
firm now operates 51 offices <strong>in</strong> 34 countries. For further <strong>in</strong>formation,<br />
please visit our Web site at www.bcg.com.
A <strong>Revolution</strong> <strong>in</strong> R&D<br />
HOW GENOMICS AND GENETICS ARE TRANSFORMING<br />
THE BIOPHARMACEUTICAL INDUSTRY<br />
PETER TOLLMAN<br />
PHILIPPE GUY<br />
JILL ALTSHULER<br />
ALASTAIR FLANAGAN<br />
MICHAEL STEINER<br />
www.bcg.com<br />
NOVEMBER 2001
2<br />
© The Boston Consult<strong>in</strong>g Group, Inc. 2001. All rights reserved.<br />
For <strong>in</strong>formation or permission to repr<strong>in</strong>t, please contact BCG at:<br />
E-mail: imc-<strong>in</strong>fo@bcg.com<br />
Fax: 617-973-1339, attention IMC/Permissions<br />
Mail: IMC/Permissions<br />
The Boston Consult<strong>in</strong>g Group, Inc.<br />
Exchange Place<br />
Boston, MA 02109<br />
USA<br />
Credits: Left cover photo by Bob Waterston, Wash<strong>in</strong>gton University, St. Louis, Missouri. Used by permission.<br />
The photo shows a bird’s-eye view of one room <strong>in</strong> the DNA sequenc<strong>in</strong>g facility at the Whitehead Institute Center for<br />
Genome Research.
Table of Contents<br />
ABOUT THE AUTHORS 4<br />
FOREWORD 5<br />
EXECUTIVE SUMMARY 6<br />
INTRODUCTION 9<br />
CHAPTER 1: THE IMPACT OF GENOMICS 11<br />
Preface 11<br />
The Opportunities 12<br />
The Challenges 18<br />
A F<strong>in</strong>al Word 21<br />
CHAPTER 2: THE IMPACT OF GENETICS 24<br />
Preface 24<br />
Disease Genetics 27<br />
Pharmacogenetics 33<br />
A F<strong>in</strong>al Word 39<br />
CHAPTER 3: MANAGERIAL CHALLENGES 41<br />
Preface: Look<strong>in</strong>g Back and Look<strong>in</strong>g Forward 41<br />
Strategy—Search<strong>in</strong>g for Genomic Competitive Advantage 41<br />
Putt<strong>in</strong>g the Strategy <strong>in</strong>to Operation 49<br />
A F<strong>in</strong>al Word 56<br />
CONCLUSION 57<br />
METHODOLOGY 59<br />
3
4<br />
About the Authors<br />
About the Authors<br />
Peter Tollman is a vice president <strong>in</strong> the Boston office and leads BCG's biopharmaceutical R&D bus<strong>in</strong>ess.<br />
Philippe Guy is a senior vice president <strong>in</strong> the Paris office and leads the worldwide Health Care practice. Jill<br />
Altshuler is a manager <strong>in</strong> the Boston office and a key contributor to BCG’s genomics <strong>in</strong>itiative. Alastair<br />
Flanagan is a vice president <strong>in</strong> the London office and leads the U.K. Health Care practice. Michael Ste<strong>in</strong>er<br />
is a senior vice president <strong>in</strong> the Munich office and leads the German Health Care practice.<br />
Acknowledgments<br />
Sarah Cairns-Smith (Boston) pioneered BCG’s <strong>in</strong>vestigation of genomics. Samantha Gray (Boston) has made<br />
significant contributions throughout the research and writ<strong>in</strong>g phases of the report.<br />
The authors would like to thank the advisory team: Oliver Fetzer (Boston), Hamilton Moses (Wash<strong>in</strong>gton,<br />
D.C.), Niko Vrettos (Düsseldorf), and Craig Wheeler (Boston). The authors would also like to acknowledge<br />
the contributions of the project team: Dierk Beyer (Frankfurt), Markus Hild<strong>in</strong>ger (Boston), Raphael Lehrer<br />
(Wash<strong>in</strong>gton D.C.), Nancy Macmillan (Boston), Jonathan Montagu (London), and Joanne Smith-Farrell<br />
(Wash<strong>in</strong>gton, D.C.).<br />
For Further Contact<br />
The authors welcome your questions and comments. For <strong>in</strong>quiries about this report or BCG’s Health Care<br />
practice, please contact:<br />
Alastair Flanagan, London e-mail: flanagan.alastair@bcg.com<br />
Philippe Guy, Paris e-mail: guy.philippe@bcg.com<br />
Mark Lubkeman, Los Angeles e-mail: lubkeman.mark@bcg.com<br />
Michael Ste<strong>in</strong>er, Munich e-mail: ste<strong>in</strong>er.michael@bcg.com<br />
Mart<strong>in</strong> Reeves, Tokyo e-mail: reeves.mart<strong>in</strong>@bcg.com<br />
Peter Tollman, Boston e-mail: tollman.peter@bcg.com
Foreword<br />
To meet growth targets, pharmaceutical companies are go<strong>in</strong>g to have to <strong>in</strong>crease R&D productivity. By a fortunate<br />
co<strong>in</strong>cidence, that crisis <strong>in</strong> expectation is be<strong>in</strong>g counterbalanced by a surge of opportunity. Recent<br />
years have seen astonish<strong>in</strong>g advances <strong>in</strong> technology and explosions of data, which are driv<strong>in</strong>g two waves of<br />
change through the <strong>in</strong>dustry—a genomics wave and a genetics wave—and radically reshap<strong>in</strong>g R&D methods<br />
and economics <strong>in</strong> the process. Biopharmaceutical R&D is mov<strong>in</strong>g <strong>in</strong>to a new era: almost every l<strong>in</strong>k <strong>in</strong> the<br />
value cha<strong>in</strong> has the potential for tremendous boosts <strong>in</strong> efficiency or success.<br />
But these advances are not assured. Technological hurdles have yet to be overcome, particularly <strong>in</strong> the genetics<br />
wave. Moreover, because the productivity boosts are likely to be unequal and uncoord<strong>in</strong>ated, the value<br />
cha<strong>in</strong> itself will demand reconfigur<strong>in</strong>g. And so too, <strong>in</strong> consequence, will many traditional operational procedures<br />
and organizational structures. The repercussions of genomics, <strong>in</strong> other words, are go<strong>in</strong>g to reach the<br />
furthest recesses of corporate constitution and culture. A true revolution, <strong>in</strong> short—and one that is already<br />
well under way.<br />
BCG has evaluated deeply the economic and bus<strong>in</strong>ess implications of these disruptions. To bolster our <strong>in</strong>ternal<br />
understand<strong>in</strong>g, we gathered <strong>in</strong>formation and perspectives <strong>in</strong> an extensive program of <strong>in</strong>terviews with<br />
lead<strong>in</strong>g R&D scientists and executives. Our f<strong>in</strong>d<strong>in</strong>gs—based on the comb<strong>in</strong>ation of these <strong>in</strong>terviews, economic<br />
model<strong>in</strong>g, and client casework—form the substance of this report. Its three sections are devoted<br />
respectively to the impact of genomics, the impact of genetics, and some of the strategic and operational<br />
implications for biopharmaceutical firms.<br />
The first two sections have already been published separately. They generated considerable publicity, and—<br />
more important—considerable comment. We now look forward to your further responses to the report as a<br />
whole.<br />
Philippe Guy<br />
Senior Vice President<br />
5
6<br />
Executive Summary<br />
In the pharmaceutical <strong>in</strong>dustry’s struggle to reach<br />
the levels of growth expected of it, one of its key<br />
aims will be to <strong>in</strong>crease R&D productivity. And a key<br />
means of meet<strong>in</strong>g this challenge is to adopt some of<br />
the new technologies and approaches broadly<br />
def<strong>in</strong>ed as genomics. 1 That is bound to be a complicated,<br />
perilous, and often pa<strong>in</strong>ful process, but if<br />
companies get their strategy right and overcome<br />
the obstacles, they could, <strong>in</strong> the best case, as much<br />
as halve the cost of drug development.<br />
The report is divided <strong>in</strong>to three parts.<br />
The Impact of Genomics<br />
The first great advance of the genomics era is <strong>in</strong><br />
technology—above all, the <strong>in</strong>tegration of new highthroughput<br />
techniques with powerful new comput<strong>in</strong>g<br />
capabilities. The new technologies are active<br />
throughout R&D, most immediately at the drug discovery<br />
stage, and promise to enhance productivity<br />
by boost<strong>in</strong>g efficiency.<br />
The stagger<strong>in</strong>g <strong>in</strong>vestment needed to develop a<br />
drug—$880 million and 15 years is the pregenomics<br />
average—could be reduced by as much as<br />
$300 million and two years by apply<strong>in</strong>g genomics<br />
technologies. Productivity ga<strong>in</strong>s would be realized<br />
at every step <strong>in</strong> the value cha<strong>in</strong>. Potential obstacles<br />
abound, however. In particular, two broad challenges<br />
must be met to realize the sav<strong>in</strong>gs:<br />
•Target quality must be ma<strong>in</strong>ta<strong>in</strong>ed. Pursu<strong>in</strong>g new<br />
target classes could <strong>in</strong>volve unfamiliar costs <strong>in</strong>itially,<br />
and these could delay the rewards—though<br />
only temporarily. But to jeopardize target quality<br />
by withhold<strong>in</strong>g that early <strong>in</strong>vestment would be to<br />
risk higher failure rates downstream, and that<br />
would <strong>in</strong>volve far greater costs <strong>in</strong> the end.<br />
• Bottlenecks must be eased. Ow<strong>in</strong>g to the unevenness<br />
of the efficiency ga<strong>in</strong>s at different steps <strong>in</strong><br />
the value cha<strong>in</strong>, the pipel<strong>in</strong>e’s flow will be<br />
impeded at various chokepo<strong>in</strong>ts. If the requisite<br />
action is taken, an even flow should be restored<br />
and the promised rewards should be safeguarded.<br />
The Impact of Genetics<br />
The second great advance of the genomics era is <strong>in</strong><br />
the quantity and quality of data. From the data,<br />
<strong>in</strong>valuable <strong>in</strong>formation about <strong>in</strong>dividuals’ genetic<br />
variation can be extracted and exploited. In pharmaceutical<br />
R&D, genetics will be applied particularly<br />
to two tasks: identify<strong>in</strong>g genes whose carriers<br />
are susceptible to specific diseases (disease genetics);<br />
and subdivid<strong>in</strong>g patients <strong>in</strong> cl<strong>in</strong>ical trials<br />
accord<strong>in</strong>g to variations <strong>in</strong> drug response (pharma-<br />
1. Genomics <strong>in</strong> its narrow sense contrasts with genetics. Roughly, the former concerns itself with the common “standard” genetic makeup, the latter with<br />
the dist<strong>in</strong>ctive genetic makeup of <strong>in</strong>dividuals. But <strong>in</strong> its broader sense, genomics <strong>in</strong>cludes genetics. In this report, the context makes clear which sense is<br />
<strong>in</strong>tended.
cogenetics). The productivity ga<strong>in</strong>s will be realized<br />
mostly <strong>in</strong> later phases of the value cha<strong>in</strong>, through<br />
the boost<strong>in</strong>g of success rates.<br />
This genetics wave is still gather<strong>in</strong>g strength, but <strong>in</strong><br />
due course could make an even greater impact on<br />
R&D than the genomics wave. In an ideal scenario,<br />
the sav<strong>in</strong>gs would exceed half a billion dollars per<br />
drug. Several troubl<strong>in</strong>g hurdles would have to be<br />
negotiated first, however. These <strong>in</strong>clude:<br />
• Scientific and technical hurdles. For genetics<br />
approaches to work, the disease susceptibility or<br />
drug response has to be genetic <strong>in</strong> nature. The<br />
gene <strong>in</strong> question has to be identifiable and must<br />
lead to a drugable target and/or be found <strong>in</strong> time<br />
to streaml<strong>in</strong>e trials.<br />
• Economic and market hurdles. The cost of conduct<strong>in</strong>g<br />
genetics studies will need to drop, and<br />
the opportunity cost of a restricted label could<br />
offset the potential market upside of pharmacogenetics.<br />
Beyond these hurdles, other challenges will need to<br />
be addressed:<br />
• Difficult <strong>in</strong>vestment decisions will have to be<br />
made, weigh<strong>in</strong>g high risk aga<strong>in</strong>st potentially high<br />
rewards. Companies will need to decide exactly<br />
how to participate <strong>in</strong> genetics—whether to <strong>in</strong>vest<br />
<strong>in</strong> genetics approaches, and how deeply, consistent<br />
with their level of risk tolerance.<br />
• Unprecedented coord<strong>in</strong>ation between market<strong>in</strong>g<br />
and R&D will be necessary. Market<strong>in</strong>g will need to<br />
have a say <strong>in</strong> decid<strong>in</strong>g which markets and which<br />
genetic diseases R&D should concentrate on, and<br />
will need to become <strong>in</strong>volved earlier than ever.<br />
• Careful attention will need to be given to ethical<br />
considerations. Companies will have to ensure<br />
privacy of genetic material, and be prepared to<br />
address any concerns the public may have.<br />
Managerial Challenges<br />
With the new wealth of options and the <strong>in</strong>creased<br />
<strong>in</strong>terdependencies across the value cha<strong>in</strong>, strategic<br />
issues will prove more complex than <strong>in</strong> the past.<br />
Likewise operational issues: many traditional ways<br />
of do<strong>in</strong>g bus<strong>in</strong>ess will be disrupted by genomics<br />
technologies, and companies may need to restructure<br />
fairly drastically.<br />
The range of strategic options available to a<br />
company will be dictated by the company’s start<strong>in</strong>g<br />
position—its size, beliefs, aspirations, and capabilities.<br />
Given the magnitude of the opportunities and<br />
7
8<br />
the risks <strong>in</strong>volved, momentous <strong>in</strong>vestment decisions<br />
will need to be made, and at the very highest levels<br />
of the organization. And R&D executives will face a<br />
daunt<strong>in</strong>g new set of management responsibilities<br />
and challenges. These <strong>in</strong>clude:<br />
• Select<strong>in</strong>g an appropriate research focus—no<br />
longer just the therapeutic area or disease state of<br />
<strong>in</strong>terest, but also such dimensions as target class<br />
and treatment modality<br />
• Choos<strong>in</strong>g which technologies to implement and<br />
when and how to implement them—<strong>in</strong>-house, or<br />
through partner<strong>in</strong>g or licens<strong>in</strong>g<br />
• Rebalanc<strong>in</strong>g the value cha<strong>in</strong>—partly by reallocat<strong>in</strong>g<br />
resources but ma<strong>in</strong>ly by redesign<strong>in</strong>g processes<br />
and more actively plann<strong>in</strong>g and manag<strong>in</strong>g capacity<br />
• Establish<strong>in</strong>g a unified <strong>in</strong>formatics <strong>in</strong>frastructure—<strong>in</strong>clud<strong>in</strong>g<br />
a centralized knowledge management<br />
system<br />
• Establish<strong>in</strong>g the new organization—creat<strong>in</strong>g new<br />
<strong>in</strong>terfaces with<strong>in</strong> the R&D department, between<br />
departments, and even between corporations<br />
• Revis<strong>in</strong>g decision-mak<strong>in</strong>g procedures—fully<br />
exploit<strong>in</strong>g the latest data <strong>in</strong> order to select the<br />
most promis<strong>in</strong>g targets and compounds to move<br />
through the pipel<strong>in</strong>e and to optimize their relative<br />
resourc<strong>in</strong>g<br />
• Re<strong>in</strong>forc<strong>in</strong>g these various reforms by engag<strong>in</strong>g<br />
the emotional and behavioral issues as keenly as<br />
the operational ones<br />
All th<strong>in</strong>gs considered, companies cannot stand<br />
aside. Certa<strong>in</strong>ly there are risks <strong>in</strong> sign<strong>in</strong>g up for the<br />
revolution, but there is also a great risk <strong>in</strong> ignor<strong>in</strong>g<br />
it—the risk of becom<strong>in</strong>g uncompetitive. The revolution<br />
is real, and will leave no one untouched.
Introduction<br />
Throughout the pharmaceutical <strong>in</strong>dustry, executives<br />
are worried. They fear they will not be able to<br />
meet the double-digit annual growth expectations<br />
implied by high market capitalizations. The requisite<br />
new drugs will not be forthcom<strong>in</strong>g: R&D just<br />
cannot deliver them all.<br />
One standard response to this problem is to scale<br />
up—that has been the basis of many a recent<br />
merger—but while scale can pay off <strong>in</strong> commercialization,<br />
global development, market<strong>in</strong>g, and distribution,<br />
it is unlikely that scale alone can solve the<br />
R&D problem. Another standard response is to buy<br />
<strong>in</strong> drug candidates. Such a Band-Aid approach cannot<br />
work <strong>in</strong>def<strong>in</strong>itely, and is a risky one anyway,<br />
given that the price of these deals will cont<strong>in</strong>ue to<br />
rise as demand for them grows.<br />
The only sure way to address the problem is to<br />
<strong>in</strong>crease R&D productivity. And the way to ensure<br />
that is either to <strong>in</strong>crease efficiency (lower cost or<br />
higher speed) or reduce failure rates along the<br />
value cha<strong>in</strong>. Many companies have <strong>in</strong>creased productivity<br />
over the past decade, specifically by<br />
reeng<strong>in</strong>eer<strong>in</strong>g the development phase. That optimization<br />
may be reach<strong>in</strong>g its limits, however. As for<br />
the discovery phase, it has long been less amenable<br />
to such improvements. So the problem of produc-<br />
tivity persists. Traditional approaches cannot provide<br />
an answer, but genomics can. (See Exhibit 1.)<br />
It will not be easy, of course. There are some difficult<br />
obstacles en route—difficult, but not <strong>in</strong>surmountable.<br />
By mak<strong>in</strong>g <strong>in</strong>formed strategic choices,<br />
companies can overcome the obstacles and reap the<br />
productivity rewards. Those that embrace the revolution<br />
most boldly could potentially halve the cost<br />
and time it takes to develop a new drug—if they<br />
meet certa<strong>in</strong> challenges successfully.<br />
EXHIBIT 1<br />
GENOMICS IMPROVES R&D PRODUCTIVITY<br />
Cost (time)<br />
spent<br />
Failed targets/<br />
candidates<br />
Successful<br />
drug<br />
SOURCE: BCG analysis.<br />
Total cost to develop a drug<br />
Reduce<br />
failure<br />
Improve<br />
efficiency<br />
Total cost (time) per step<br />
9
Chapter 1: The Impact of Genomics<br />
Preface<br />
As the science of genomics has advanced, so has the<br />
def<strong>in</strong>ition. When the term was co<strong>in</strong>ed <strong>in</strong> 1986, it<br />
referred ma<strong>in</strong>ly to the study of the mammalian<br />
genome—specifically, the mapp<strong>in</strong>g, sequenc<strong>in</strong>g,<br />
and analyz<strong>in</strong>g of all its genes. The scope soon<br />
expanded, focus<strong>in</strong>g not just on the genes’ structure<br />
but on their function as well. More recently, the<br />
scope of the term has broadened further, focus<strong>in</strong>g<br />
no longer just on knowledge of the genome but also<br />
on the exploitation of that knowledge, especially<br />
for health care.<br />
Go<strong>in</strong>g beyond dictionary def<strong>in</strong>itions, our <strong>in</strong>terest is<br />
<strong>in</strong> what genomics means for the economics of pharmaceutical<br />
R&D. On the basis of our extensive<br />
research and many discussions with prom<strong>in</strong>ent people<br />
throughout the <strong>in</strong>dustry, we suggest characteriz<strong>in</strong>g<br />
genomics, for the purposes of this study, as the<br />
confluence of two <strong>in</strong>terdependent trends that are<br />
fundamentally chang<strong>in</strong>g the way R&D is conducted:<br />
<strong>in</strong>dustrialization (creat<strong>in</strong>g vastly higher throughputs,<br />
and hence a huge <strong>in</strong>crease <strong>in</strong> data), and <strong>in</strong>formatics<br />
(computerized techniques for manag<strong>in</strong>g and<br />
analyz<strong>in</strong>g those data). The surge of data—generated<br />
by the former, and processed by the latter—is<br />
of a different order from the data yields of the pregenomics<br />
era.<br />
To elaborate. The new high-tech <strong>in</strong>dustrialization<br />
has <strong>in</strong>creased the efficiency of certa<strong>in</strong> activities<br />
beyond recognition. Instead of assign<strong>in</strong>g <strong>in</strong>dividual<br />
scientists to work manually on modest <strong>in</strong>dividual<br />
experiments, companies now <strong>in</strong>voke automation<br />
and parallel process<strong>in</strong>g to conduct experiments<br />
much larger <strong>in</strong> scale and complexity, and at a much<br />
faster pace.<br />
Look around this lab—you have to search high<br />
and low to f<strong>in</strong>d a human heartbeat. Now robots<br />
can do the menial th<strong>in</strong>gs we did <strong>in</strong> grad school.<br />
—Research leader,<br />
lead<strong>in</strong>g biotech company<br />
The data that emerge are immensely greater both<br />
<strong>in</strong> quantity and <strong>in</strong> richness. Enormous databases—<br />
detail<strong>in</strong>g gene expression, for example, or homologous<br />
genes across species, or prote<strong>in</strong> structures—<br />
afford unprecedented comprehensive views of<br />
biological processes. Increas<strong>in</strong>gly, researchers can<br />
understand properties of the system rather than<br />
just <strong>in</strong>dividual parts, and that holds out the promise<br />
of a more rational approach to drug discovery.<br />
The new technology of <strong>in</strong>formatics serves to handle<br />
and process all these data. Without it, the data<br />
would rema<strong>in</strong> raw material. Informatics was nurtured<br />
by several co<strong>in</strong>cid<strong>in</strong>g factors: the ever-accelerat<strong>in</strong>g<br />
power of computers, ref<strong>in</strong>ed algorithms, the<br />
<strong>in</strong>tegration of data and technology platforms, and<br />
the versatility of the Internet. The effect is that<br />
overwhelm<strong>in</strong>g masses of <strong>in</strong>formation can now be<br />
marshaled, managed, and analyzed as never before.<br />
Data are transformed <strong>in</strong>to knowledge.<br />
We could never have achieved drug development<br />
that fast with traditional techniques. No way—<br />
without the computers we didn’t have a chance.<br />
—VP of chemistry,<br />
biotech company<br />
11
12<br />
The Opportunities<br />
What is the impact of genomics on the economics<br />
of R&D? To what extent will genomics improve productivity<br />
overall, and what will its effects be when<br />
applied at various po<strong>in</strong>ts of the value cha<strong>in</strong>? What<br />
other <strong>in</strong>cidental advantages might genomics br<strong>in</strong>g<br />
<strong>in</strong> its wake?<br />
These crucial questions have received a great deal<br />
of attention of late, and a wide variety of responses.<br />
To address the questions <strong>in</strong> a rigorous, fact-based<br />
way, we built an economic model of the entire R&D<br />
value cha<strong>in</strong>, grounded <strong>in</strong> a program of discussions<br />
with<strong>in</strong> the <strong>in</strong>dustry (more than 100 meet<strong>in</strong>gs with<br />
more than 60 scientists and executives from nearly<br />
50 companies and academic <strong>in</strong>stitutions.) (See the<br />
methodology section at the end of this report.)<br />
Realiz<strong>in</strong>g Sav<strong>in</strong>gs<br />
Before genomics technology, develop<strong>in</strong>g a new<br />
drug has cost companies on average $880 million,<br />
and has taken about 15 years from start to f<strong>in</strong>ish,<br />
that is, from target identification 2 through regulatory<br />
approval. (See Exhibit 2.) Of this cost, about 75<br />
percent can be attributed to failures along the way.<br />
By apply<strong>in</strong>g genomics technology, companies could<br />
on average realize sav<strong>in</strong>gs of nearly $300 million<br />
and two years per drug, largely as a result of efficiency<br />
ga<strong>in</strong>s. That represents a 35 percent cost and<br />
15 percent time sav<strong>in</strong>gs. (And those are the sav<strong>in</strong>gs<br />
possible with technologies that are available today;<br />
when new or improved genomics technologies<br />
emerge, the sav<strong>in</strong>gs will be even greater.) If companies<br />
wish to stay competitive, they have no choice:<br />
they must implement genomics technologies. (See<br />
Exhibit 3.)<br />
Do<strong>in</strong>g so, however, will hardly produce such huge<br />
sav<strong>in</strong>gs immediately, or automatically. It will take a<br />
few years, and many deft decisions, for the sav<strong>in</strong>gs<br />
to be realized. The early years of implementation<br />
may <strong>in</strong> fact <strong>in</strong>volve an <strong>in</strong>crease <strong>in</strong> costs as the learn<strong>in</strong>g<br />
curve is negotiated for novel targets—specifi-<br />
EXHIBIT 2<br />
DRUG R&D IS EXPENSIVE AND TIME-CONSUMING<br />
Cost: $880 million total<br />
Approximate cost ($M)<br />
165<br />
205<br />
Time: 14.7 years total<br />
Approximate time (yrs)<br />
1<br />
Biology<br />
2<br />
Target ID Target Validation<br />
40<br />
0.4<br />
Chemistry<br />
cally, as the necessary quality controls are established—and<br />
as major strategic decisions (about<br />
personnel and processes, for <strong>in</strong>stance) are confirmed<br />
or revised.<br />
More on these challenges later. But first, we will<br />
take a closer look at the long-term upside, detail<strong>in</strong>g<br />
the sav<strong>in</strong>gs at various steps along the value cha<strong>in</strong>.<br />
120<br />
2.7<br />
Screen<strong>in</strong>g Optimization<br />
90<br />
1.6<br />
Development<br />
260<br />
7<br />
Precl<strong>in</strong>ical Cl<strong>in</strong>ical<br />
SOURCES: BCG analysis; <strong>in</strong>dustry <strong>in</strong>terviews; scientific literature; public<br />
f<strong>in</strong>ancial data; Lehman Brothers; PAREXEL’S Pharmaceutical R&D<br />
Statistical Sourcebook 2000.<br />
NOTE: Cost to drug <strong>in</strong>cludes failures. Target identification <strong>in</strong>cludes <strong>in</strong>itial<br />
experiments that companies may have outsourced to academic research<br />
<strong>in</strong>stitutions.<br />
2. Includes <strong>in</strong>itial experiments to identify potential targets. Traditionally, companies have sourced much of this research from academia.
EXHIBIT 3<br />
GENOMICS CAN YIELD SIGNIFICANT SAVINGS<br />
Cost to drug<br />
Pre-genomics<br />
Post-genomics<br />
target ID<br />
Plus <strong>in</strong> silico<br />
chemistry<br />
Plus precl<strong>in</strong>ical and<br />
cl<strong>in</strong>ical advances 1<br />
Time to drug<br />
Pre-genomics<br />
Post-genomics<br />
target ID<br />
Plus <strong>in</strong> silico<br />
chemistry<br />
Plus precl<strong>in</strong>ical and<br />
cl<strong>in</strong>ical advances 1<br />
ID Biology<br />
Target ID Target Validation<br />
0<br />
0<br />
200<br />
Cost ($M)<br />
Target Discovery/Biology<br />
The identification of targets is be<strong>in</strong>g <strong>in</strong>dustrialized—through<br />
the use of technology such as gene<br />
chips to perform gene expression analysis, for<br />
example—and then further enhanced by bio<strong>in</strong>formatics.<br />
Scientists can now use a s<strong>in</strong>gle gene chip to<br />
compare the expression of thousands of genes, <strong>in</strong><br />
diseased and healthy tissue alike, all at once, and<br />
can then use <strong>in</strong>formatics technology to f<strong>in</strong>d follow-<br />
5<br />
Chemistry<br />
400<br />
Screen<strong>in</strong>g Optimization<br />
600<br />
10<br />
610<br />
590<br />
800<br />
740<br />
13.0<br />
12.7<br />
880<br />
13.8<br />
15<br />
Time (years)<br />
Development<br />
Precl<strong>in</strong>ical Cl<strong>in</strong>ical<br />
SOURCES: BCG analysis; <strong>in</strong>dustry <strong>in</strong>terviews; scientific literature; public<br />
f<strong>in</strong>ancial data; Lehman Brothers; PAREXEL’S Pharmaceutical R&D<br />
Statistical Sourcebook 2000.<br />
1,000<br />
14.7<br />
1Includes surrogate marker sav<strong>in</strong>gs from early elim<strong>in</strong>ation of unpromis<strong>in</strong>g<br />
candidates, not from early FDA approval; does not <strong>in</strong>clude potential sav<strong>in</strong>gs<br />
from pharmacogenetics.<br />
up <strong>in</strong>formation, on these or related genes, <strong>in</strong> databases<br />
around the world. (Target validation, however,<br />
seems difficult to <strong>in</strong>dustrialize, ow<strong>in</strong>g to the<br />
“slow” biology of whole-animal systems still<br />
<strong>in</strong>volved, and is not yet show<strong>in</strong>g significant productivity<br />
ga<strong>in</strong>s.)<br />
In all, the potential sav<strong>in</strong>gs per drug are on average<br />
about $140 million and just under one year of time<br />
to market, achieved entirely through improved efficiency.<br />
That would add about $100 million <strong>in</strong> value<br />
per drug (assum<strong>in</strong>g an “average” drug with peak<br />
annual sales of $500 million). So for this step <strong>in</strong> the<br />
value cha<strong>in</strong>, productivity would <strong>in</strong>crease vastly: it<br />
would be six times as high as before, assum<strong>in</strong>g the<br />
same level of <strong>in</strong>vestment. A sixfold <strong>in</strong>crease <strong>in</strong> the<br />
number of potential targets!<br />
Several companies have already benefited handsomely<br />
from this w<strong>in</strong>dfall. Take the case of<br />
Millennium, which was an early adopter of <strong>in</strong>dustrialized<br />
biology. The company, anticipat<strong>in</strong>g an overabundance<br />
of targets, established a bus<strong>in</strong>ess model<br />
<strong>in</strong> which it sells off much its output and uses that<br />
<strong>in</strong>come to fund <strong>in</strong>ternal research. Start<strong>in</strong>g from its<br />
early genomics platform, Millennium has strategically<br />
acquired or partnered with other platform<br />
companies to establish an <strong>in</strong>tegrated drug discovery<br />
value cha<strong>in</strong>. From the other perspective, pharmaceutical<br />
companies such as Bayer and Aventis<br />
have made deals with Millennium, <strong>in</strong> the expectation<br />
of profit<strong>in</strong>g from the new abundance of targets<br />
they can choose to pursue.<br />
Lead Discovery/Chemistry<br />
Chemistry is be<strong>in</strong>g revolutionized by <strong>in</strong> silico (that<br />
is, computer-aided) technology—specifically, virtual<br />
screen<strong>in</strong>g supported by chemo<strong>in</strong>formatics. In<br />
virtual screen<strong>in</strong>g, potential lead chemicals are<br />
assessed with computer algorithms to test how likely<br />
they are to <strong>in</strong>teract with a target. Chemo<strong>in</strong>formatics<br />
provides the necessary platform for virtual screen<strong>in</strong>g,<br />
us<strong>in</strong>g data and analysis from high-throughput<br />
screen<strong>in</strong>g (HTS) and other chemistry activities.<br />
This approach <strong>in</strong>creases efficiency by focus<strong>in</strong>g compound<br />
synthesis, reduc<strong>in</strong>g the number of assays,<br />
<strong>in</strong>creas<strong>in</strong>g the parallelization of screen<strong>in</strong>g steps,<br />
13
14<br />
and generally help<strong>in</strong>g to optimize screen<strong>in</strong>g. The<br />
power of this approach is expected to <strong>in</strong>crease dramatically<br />
with the availability of larger data sets for<br />
ref<strong>in</strong><strong>in</strong>g the predictive algorithms. (At the moment,<br />
however, <strong>in</strong> silico chemistry has one notable shortcom<strong>in</strong>g:<br />
it looks as if it will be suitable for only<br />
about 30 percent of targets—the rest fail to yield<br />
the requisite structural <strong>in</strong>formation—and even<br />
then might prove difficult to apply until lead optimization.<br />
Our sav<strong>in</strong>gs are calculated for those targets<br />
where <strong>in</strong> silico technology can be applied.)<br />
The potential sav<strong>in</strong>gs are on average about $130<br />
million and nearly one year per drug. That would<br />
add about $90 million <strong>in</strong> value per drug. For this<br />
step of the value cha<strong>in</strong>, then, productivity would<br />
double, assum<strong>in</strong>g the same level of <strong>in</strong>vestment.<br />
As a beneficiary of these advances, a good case <strong>in</strong><br />
po<strong>in</strong>t is Vertex. Start<strong>in</strong>g from an IDD (<strong>in</strong> silico drug<br />
design) platform <strong>in</strong> chemistry, the company has<br />
gone on to develop an <strong>in</strong>tegrated value cha<strong>in</strong> <strong>in</strong> its<br />
own right. In silico models have allowed more efficient<br />
design of small-molecule drugs than a purely<br />
traditional approach, and the company’s discovery<br />
focus has been on certa<strong>in</strong> target classes that benefit<br />
most from proprietary <strong>in</strong> silico technologies. This<br />
approach has met with considerable success, culm<strong>in</strong>at<strong>in</strong>g<br />
<strong>in</strong> one of the biggest biotech alliances so far<br />
(with Novartis, and worth $813 million). Vertex can<br />
fairly claim to have the strongest small-molecule<br />
drug pipel<strong>in</strong>e with<strong>in</strong> the biotech <strong>in</strong>dustry. With one<br />
drug on the market and twelve candidates <strong>in</strong> development,<br />
it compares favorably with some of the big<br />
pharmaceutical pipel<strong>in</strong>es.<br />
Serious money can be saved for the target classes<br />
where <strong>in</strong> silico chemistry works.<br />
—Director of chemistry,<br />
major pharmaceutical company<br />
Development<br />
Three key genomics advances look set to <strong>in</strong>crease<br />
capacity here. In silico ADME/tox (absorption, distribution,<br />
metabolism, and excretion/toxicity) and<br />
high-throughput <strong>in</strong> vitro toxicology are revolutioniz<strong>in</strong>g<br />
the precl<strong>in</strong>ical phase through their power to<br />
predict drug properties. And surrogate markers<br />
(physiological markers that correlate with elements<br />
of drug response), applied <strong>in</strong> both precl<strong>in</strong>ical and<br />
cl<strong>in</strong>ical trials, evaluate drug effects more efficiently<br />
than before: they are quick to identify fail<strong>in</strong>g compounds,<br />
and once regulatory approval is granted,<br />
will be used to identify pass<strong>in</strong>g compounds too.<br />
In comb<strong>in</strong>ation, the potential sav<strong>in</strong>gs available <strong>in</strong><br />
the short term are on the order of $20 million and<br />
0.3 years per drug. That would add about $15 million<br />
<strong>in</strong> value per drug. But these approaches will<br />
become even more valuable as cl<strong>in</strong>ical data on the<br />
relationship between genes, gene expression, and<br />
disease accumulate and regulatory agencies beg<strong>in</strong><br />
to accept cl<strong>in</strong>ical-marker data: the potential sav<strong>in</strong>gs<br />
could rise to $70 million.<br />
These technologies are be<strong>in</strong>g adopted by forwardlook<strong>in</strong>g<br />
chemistry companies, and are enabl<strong>in</strong>g<br />
them to pull certa<strong>in</strong> precl<strong>in</strong>ical activities <strong>in</strong>to the<br />
chemistry part of the value cha<strong>in</strong>. For example,<br />
ArQule has recently acquired Camitro to <strong>in</strong>corporate<br />
an <strong>in</strong>tegrated <strong>in</strong> vitro and <strong>in</strong> silico ADME/tox<br />
platform <strong>in</strong>to its own set of capabilities.<br />
These are not the only advances likely to transform<br />
productivity dur<strong>in</strong>g the development phase. Pharmacogenomics—through<br />
its power to identify subgroups<br />
of patients who respond differently to a<br />
drug under study—offers the promise of streaml<strong>in</strong><strong>in</strong>g<br />
cl<strong>in</strong>ical trials; we explore this topic <strong>in</strong> more<br />
detail later. Beyond genomics (and beyond the<br />
scope of the current report), “e-technologies,” such<br />
as electronic patient recruitment and monitor<strong>in</strong>g<br />
via the Internet, are expected to speed up the<br />
launch and completion of cl<strong>in</strong>ical trials.<br />
Beyond the Traditional Value Cha<strong>in</strong>:<br />
Chemical Genomics<br />
The various productivity ga<strong>in</strong>s just outl<strong>in</strong>ed occur<br />
with<strong>in</strong> specific steps of the value cha<strong>in</strong>. But suppose<br />
you could transcend the traditional value cha<strong>in</strong>, or<br />
refashion it to streaml<strong>in</strong>e R&D. That is one of the<br />
revolutionary prospects now open<strong>in</strong>g up. The key is<br />
chemical genomics, and the way it will dissolve the<br />
old boundaries is by <strong>in</strong>troduc<strong>in</strong>g <strong>in</strong>to the value<br />
cha<strong>in</strong> a k<strong>in</strong>d of parallel process<strong>in</strong>g. (See sidebar,<br />
“Chemical Genomics—Forward or Reverse.”)
CHEMICAL GENOMICS—FORWARD OR REVERSE<br />
When companies say they are pursu<strong>in</strong>g chemical<br />
genomics, they are usually referr<strong>in</strong>g to large-scale<br />
reverse chemical genetics. (That is how the term is<br />
used <strong>in</strong> our report.) This approach <strong>in</strong>volves f<strong>in</strong>d<strong>in</strong>g<br />
chemical compounds that b<strong>in</strong>d to a known target.<br />
Companies often perform this task for entire target<br />
classes; it is especially popular for prote<strong>in</strong> classes<br />
that are known to be highly drugable, such as Gprote<strong>in</strong><br />
coupled receptors (GPCRs). The assay for<br />
b<strong>in</strong>d<strong>in</strong>g does not need to provide functional <strong>in</strong>formation<br />
relevant to a specific disease state—biological<br />
function can be assessed <strong>in</strong> validation experiments.<br />
The alternative is forward chemical genetics. This<br />
approach beg<strong>in</strong>s with functional knowledge. A<br />
library of compounds is screened <strong>in</strong> an assay that<br />
tests for changes <strong>in</strong> a specific biological function.<br />
One immediate result would be to process the glut<br />
of identified targets more quickly: <strong>in</strong>stead of jo<strong>in</strong><strong>in</strong>g<br />
the logjam at the validation stage, a great many<br />
of them can now be diverted directly to screen<strong>in</strong>g.<br />
If they fail there, they can be discarded right away,<br />
and thus simply bypass most of the validation stage<br />
altogether. In other words, screen<strong>in</strong>g moves up the<br />
value cha<strong>in</strong> to rest alongside validation, <strong>in</strong> a parallel<br />
rather than consecutive position. By bracket<strong>in</strong>g<br />
the <strong>in</strong>dustrialized steps of target identification and<br />
chemical screen<strong>in</strong>g, chemical genomics has given<br />
the value cha<strong>in</strong> a remarkable makeover.<br />
The key is to move lengthy, messy biology far downstream<br />
where you know it’s worth pursu<strong>in</strong>g. Many<br />
targets aren’t drugable, so just validate the smaller<br />
drugable subset.<br />
—SVP of discovery,<br />
lead<strong>in</strong>g biotech company<br />
The effect of this new value cha<strong>in</strong> is dramatic: time<br />
to drug is cut by a further two years (that’s on top<br />
of the year already saved by us<strong>in</strong>g genomic target<br />
identification). On the other hand, there is a large<br />
3. In July 2001, Aurora Biosciences was acquired by Vertex Pharmaceuticals.<br />
The <strong>in</strong>tention is to screen a library aga<strong>in</strong>st all expressed<br />
genes <strong>in</strong> the system under <strong>in</strong>vestigation.<br />
This approach has the tremendous advantage of allow<strong>in</strong>g<br />
the identification of targets without any presumptions<br />
as to their function. Additionally, these<br />
targets can help to elucidate the mechanism of disease,<br />
thereby reveal<strong>in</strong>g other potential targets <strong>in</strong> relevant<br />
pathways. The drawback is that forward chemical<br />
genetics has not yet been <strong>in</strong>dustrialized, and<br />
throughput levels are therefore very low. Accord<strong>in</strong>g<br />
to our model, implement<strong>in</strong>g it today would <strong>in</strong>crease<br />
costs to more than $1 billion per drug, ow<strong>in</strong>g to the<br />
use of “slow” biology, which is needed to set up the<br />
screen<strong>in</strong>g assays <strong>in</strong> chemistry. The expert estimate is<br />
that forward chemical genetics is still as much as<br />
five years away from be<strong>in</strong>g economically feasible.<br />
<strong>in</strong>crease <strong>in</strong> cost, offsett<strong>in</strong>g all cost sav<strong>in</strong>gs from target<br />
identification. But the tradeoff is still positive.<br />
In a highly competitive market, where new entrants<br />
are cont<strong>in</strong>uously erod<strong>in</strong>g share, chemical genomics<br />
can add more than $200 million <strong>in</strong> value per drug.<br />
(In less competitive conditions, the value added<br />
may be as little as $20 million.)<br />
No doubt chemical genomics costs more—but you<br />
take the loss to ga<strong>in</strong> the speed. Time is money.<br />
—SVP of discovery and technology,<br />
major pharmaceutical company<br />
One important drawback of chemical genomics is<br />
this: it is limited ma<strong>in</strong>ly to known target classes.<br />
With targets of unknown function, results become<br />
very difficult to <strong>in</strong>terpret. The proxy assays used for<br />
screen<strong>in</strong>g—heat-stability assays, for <strong>in</strong>stance—tend<br />
to yield both false positives and false negatives.<br />
Nevertheless, chemical genomics is already be<strong>in</strong>g<br />
pursued throughout the <strong>in</strong>dustry. Several big pharmaceutical<br />
companies have adopted it, and genomics<br />
companies such as Aurora Biosciences 3 and<br />
15
16<br />
Cellomics are well positioned to exploit the expected<br />
result<strong>in</strong>g demand for screen<strong>in</strong>g resources.<br />
Aurora is a likely w<strong>in</strong>ner <strong>in</strong> the race to resolve chemical<br />
genomics-related bottlenecks, s<strong>in</strong>ce it boasts<br />
some of the most advanced screen<strong>in</strong>g and assay<br />
technologies <strong>in</strong> the <strong>in</strong>dustry. It has an unusual bus<strong>in</strong>ess<br />
model, <strong>in</strong> that it provides tools and discovery<br />
services but does not engage <strong>in</strong> any drug discovery<br />
of its own.<br />
* * *<br />
So much for the imm<strong>in</strong>ent efficiency sav<strong>in</strong>gs across<br />
the R&D value cha<strong>in</strong>. They are hardly the end of<br />
TECHNOLOGIES IN WAITING—OTHER TECHNOLOGIES EXAMINED,<br />
BUT OMITTED FROM OUR REPORT<br />
In this report we have focused on the technologies<br />
and approaches that are hav<strong>in</strong>g the greatest impact<br />
on R&D economics today. Several other excit<strong>in</strong>g advances<br />
appear likely to make a comparable impact<br />
beyond the next three to five years (too far ahead for<br />
<strong>in</strong>clusion <strong>in</strong> our analysis for this report), <strong>in</strong> particular,<br />
the use of proteomics <strong>in</strong> target identification,<br />
conditional gene <strong>in</strong>hibition <strong>in</strong> target validation, and<br />
<strong>in</strong>dustrialized structural biology <strong>in</strong> screen<strong>in</strong>g and<br />
drug design.<br />
Proteomics is the study of prote<strong>in</strong> expression and<br />
prote<strong>in</strong>-prote<strong>in</strong> <strong>in</strong>teractions. Its aim is an understand<strong>in</strong>g,<br />
and ultimately exploitation, of prote<strong>in</strong> function.<br />
Identify<strong>in</strong>g prote<strong>in</strong>s through sequence or structure<br />
homology has recently become much more efficient,<br />
thanks to bio<strong>in</strong>formatics’ role <strong>in</strong> analyz<strong>in</strong>g largescale<br />
experiments. One example of a genomics company<br />
apply<strong>in</strong>g proteomics is Oxford Glycosciences,<br />
which is engaged <strong>in</strong> identify<strong>in</strong>g targets and surrogate<br />
markers, both <strong>in</strong> collaboration with pharmaceutical<br />
companies and <strong>in</strong> an <strong>in</strong>dependent pipel<strong>in</strong>e. But proteomics<br />
is not really <strong>in</strong>dustrialized yet, and has high<br />
hurdles to overcome before it is.<br />
We exam<strong>in</strong>ed the economics of proteomic expression<br />
studies us<strong>in</strong>g two-dimensional gel analysis, followed<br />
the story, of course. Other technological advances<br />
are bound to improve R&D productivity further <strong>in</strong><br />
due course. Important emerg<strong>in</strong>g technologies<br />
<strong>in</strong>clude proteomics, partial target <strong>in</strong>hibition, and<br />
structural biology. (See sidebar, “Technologies <strong>in</strong><br />
Wait<strong>in</strong>g.”)<br />
Improv<strong>in</strong>g Decision Mak<strong>in</strong>g<br />
The economics of R&D h<strong>in</strong>ge on success rates, and<br />
success rates depend largely on a cascade of decisions<br />
that have to be made aga<strong>in</strong> and aga<strong>in</strong>:<br />
whether or not to pursue a target or lead, and if so,<br />
how—to what extent and with what approach.<br />
by identification of <strong>in</strong>terest<strong>in</strong>g prote<strong>in</strong>s through mass<br />
spectrometry.<br />
Under optimal conditions today, this approach has<br />
the potential to save about as much <strong>in</strong> cost as<br />
genomics-based approaches do, though not as much<br />
<strong>in</strong> time (about six months less). As the technology<br />
becomes <strong>in</strong>dustrialized, proteomics could well surpass<br />
genomics-based approaches, but that is still<br />
several years away.<br />
The aim of the second promis<strong>in</strong>g technology we<br />
<strong>in</strong>vestigated, conditional gene <strong>in</strong>hibition, is to overcome<br />
a common problem <strong>in</strong> target validation. Here<br />
is the background. A standard technique for target<br />
validation uses “target knockouts.” The potential target<br />
is removed, or “knocked out,” from an animal at<br />
conception; this results <strong>in</strong> the total <strong>in</strong>hibition of the<br />
target’s function from embryo to adult. The trouble is<br />
that drugs work differently. Very seldom do they<br />
<strong>in</strong>hibit target function fully, and they are taken only<br />
after genes have already fulfilled their developmental<br />
role <strong>in</strong> utero. So the use of target knockouts as a target<br />
validation technique does run the risk of creat<strong>in</strong>g<br />
false negatives (<strong>in</strong> some cases <strong>in</strong>dicated by death,<br />
because of the unnatural disruption of embryonic<br />
development). What is needed <strong>in</strong>stead of total gene
Genomics may offer an opportunity for companies<br />
to make the correct decision more often than<br />
before. For one th<strong>in</strong>g, genomics can ultimately provide<br />
more, better, and earlier <strong>in</strong>formation, and<br />
good <strong>in</strong>formation translates ultimately <strong>in</strong>to high<br />
success rates. For another, the implementation of<br />
genomics approaches will force companies to<br />
reth<strong>in</strong>k their <strong>in</strong>ternal decision-mak<strong>in</strong>g processes.<br />
Genomics-based <strong>in</strong>formation, together with the<br />
ability to m<strong>in</strong>e it productively, gives a company an<br />
enormous advantage. Such a company will now be<br />
able to make and execute decisions on targets and<br />
<strong>in</strong>hibition, therefore, is conditional gene <strong>in</strong>hibition,<br />
which mimics the partial <strong>in</strong>hibitory effect of a drug.<br />
Several promis<strong>in</strong>g approaches have emerged, <strong>in</strong>clud<strong>in</strong>g<br />
forward genetics, chemical genomics, and<br />
molecular switches that modulate gene expression,<br />
but their practicality has still to be proved.<br />
Examples abound of genomics companies engaged <strong>in</strong><br />
develop<strong>in</strong>g these target-validation techniques. Lexicon<br />
Genetics, Exelixis, and Ingenium, for <strong>in</strong>stance,<br />
are us<strong>in</strong>g mass mutagenesis on animals such as mice<br />
and zebrafish. In a more focused project, Hypnion is<br />
us<strong>in</strong>g forward genetics and other approaches to<br />
understand sleep-wake disorders <strong>in</strong> mammals.<br />
What benefits lie <strong>in</strong> wait? By elim<strong>in</strong>at<strong>in</strong>g the false<br />
negatives associated with the current knockout technique,<br />
these new technologies could double or even<br />
triple the number of validated targets, and <strong>in</strong> that<br />
way save up to $200 million per drug. At the<br />
moment, however, these new k<strong>in</strong>ds of validation<br />
(with the exception of certa<strong>in</strong> chemical genomics<br />
approaches, discussed <strong>in</strong> the ma<strong>in</strong> text) are still<br />
ma<strong>in</strong>ly limited to “slow” biology.<br />
F<strong>in</strong>ally, structural biology is used for generat<strong>in</strong>g and<br />
analyz<strong>in</strong>g the three-dimensional structure of targets<br />
leads with greater speed and consistency than<br />
before. Guided by more rigorous selection criteria,<br />
the company should go on to improve its success<br />
rates and hence its productivity.<br />
A mere 10 percent improvement <strong>in</strong> accuracy of<br />
decisions at any stage would confer disproportionately<br />
large benefits. Consider, for example, all the<br />
target/lead pairs that fail just before cl<strong>in</strong>ical trials:<br />
if a company were able to decide <strong>in</strong> just one out of<br />
ten such cases aga<strong>in</strong>st pursu<strong>in</strong>g the target <strong>in</strong> the<br />
first place, it would save as much as $100 million<br />
per drug on average. As for INDs that fail cl<strong>in</strong>ical<br />
for virtual screen<strong>in</strong>g, and is essential to <strong>in</strong> silico drug<br />
design. Unfortunately, it currently entails prote<strong>in</strong><br />
crystallization (to prepare the prote<strong>in</strong>s for visualization<br />
by X-ray diffraction), which is a difficult, labor<strong>in</strong>tensive<br />
manual process. Speedier alternatives,<br />
such as NMR spectroscopy, cannot predict overall<br />
prote<strong>in</strong> shape adequately, be<strong>in</strong>g restricted to prote<strong>in</strong><br />
subsegments. As a result, <strong>in</strong> silico model<strong>in</strong>g rema<strong>in</strong>s<br />
limited <strong>in</strong> its applicability: the algorithms cannot<br />
boast really high precision for target classes where<br />
no example structures are available.<br />
Several projects, both public and private, are under<br />
way to upgrade structural biology platforms to the<br />
po<strong>in</strong>t where they will achieve <strong>in</strong>dustrialized scale.<br />
Among the private endeavors is the Novartis Institute<br />
for Functional Genomics, founded by Novartis to<br />
identify and characterize targets us<strong>in</strong>g high-throughput<br />
technologies. In the biotech field, Structural<br />
Genomix aims to become a platform provider and<br />
generate revenues by sell<strong>in</strong>g prote<strong>in</strong> structures; the<br />
company may also decide to exploit its data <strong>in</strong>house,<br />
and extend <strong>in</strong>to <strong>in</strong> silico drug design. But it<br />
might be several years before technologies have<br />
developed far enough for the necessary scale effects<br />
to be realized.<br />
17
18<br />
trials, if the company were able to decide <strong>in</strong> just one<br />
out of ten such cases to abandon development earlier,<br />
it could save an additional $100 million per<br />
drug.<br />
Improv<strong>in</strong>g decision mak<strong>in</strong>g to that extent will take<br />
more than simply acquir<strong>in</strong>g and implement<strong>in</strong>g the<br />
new genomics technologies and approaches. It will<br />
take some serious strategic reth<strong>in</strong>k<strong>in</strong>g too, and possibly<br />
major organizational changes. Whether to keep<br />
all activities <strong>in</strong>-house, or seek partners, or buy <strong>in</strong> targets<br />
or leads. How to redistribute resources, reassign<br />
personnel, and revise l<strong>in</strong>es of communication and<br />
cha<strong>in</strong>s of command. Such operational and organizational<br />
quandaries will be addressed <strong>in</strong> detail <strong>in</strong> the<br />
f<strong>in</strong>al chapter of this report.<br />
We implemented a fast-<strong>in</strong>/fast-out decision policy<br />
about projects—if we didn’t have optimal conditions<br />
met <strong>in</strong> 18 months, we killed it. That made all<br />
the difference.<br />
—Former executive,<br />
lead<strong>in</strong>g pharmaceutical company<br />
Even the basic bus<strong>in</strong>ess skill of decision mak<strong>in</strong>g,<br />
then, is not immune to the <strong>in</strong>fluence of genomics<br />
technology. Whatever other benefits it br<strong>in</strong>gs,<br />
genomics serves as a wake-up call across the <strong>in</strong>dustry,<br />
even for companies try<strong>in</strong>g to shelter from the<br />
genomics revolution.<br />
The Challenges<br />
Although implement<strong>in</strong>g genomics offers companies<br />
great opportunities, it also presents them with<br />
formidable challenges. One of these is to ensure<br />
that the quality of the pipel<strong>in</strong>e rema<strong>in</strong>s uncompromised.<br />
Another is to put the new technologies <strong>in</strong>to<br />
efficient operation.<br />
Ma<strong>in</strong>ta<strong>in</strong><strong>in</strong>g Quality<br />
If the potential productivity ga<strong>in</strong>s are to be fully<br />
realized, the post-genomics R&D pipel<strong>in</strong>e will need<br />
to reta<strong>in</strong> or improve its pre-genomics quality. Any<br />
decl<strong>in</strong>e <strong>in</strong> quality—the quality of targets and<br />
leads—would obviously have an adverse effect on<br />
productivity. The ma<strong>in</strong> threat to quality derives<br />
from the unorthodoxy, the unfamiliar nature, of so<br />
many new targets. Entire target classes, previously<br />
unknown, will need <strong>in</strong>vestigat<strong>in</strong>g. The temptation<br />
to pursue leads prematurely is bound to arise, and<br />
quality control will need to be rigorously enforced<br />
to uphold the pipel<strong>in</strong>e’s usual success rates.<br />
In any given experiment, 70 percent of what I see is<br />
completely new. It could be a gold rush, or it could<br />
be junk—-there’s no way to tell until I sit at the<br />
bench and do more work.<br />
—Director of research,<br />
lead<strong>in</strong>g biotech company<br />
To appreciate the threat accurately, we need a<br />
proper def<strong>in</strong>ition of the term quality.<br />
The “<strong>in</strong>tr<strong>in</strong>sic quality” of a target or lead amounts<br />
to its likelihood of success, which is based on factors<br />
such as cl<strong>in</strong>ical relevance and drugability.<br />
Companies can do little to alter this type of quality.<br />
The “provisional quality” (or “<strong>in</strong>formational quality”)<br />
of a target or lead is based on the amount of<br />
data available on it at any given time—how much is<br />
known about its cl<strong>in</strong>ical relevance, drugability, and<br />
so on. (This <strong>in</strong>formational quality helps to predict<br />
success rates, but does not <strong>in</strong>fluence them.)<br />
Companies can alter this type of quality, by spend<strong>in</strong>g<br />
appropriately, and <strong>in</strong> that way can improve their<br />
ability to predict downstream success rates.<br />
This dist<strong>in</strong>ction is crucial. But it has at times been<br />
overlooked, result<strong>in</strong>g <strong>in</strong> some confusion <strong>in</strong> the<br />
<strong>in</strong>dustry. A widely publicized concern has been that<br />
novel targets identified through genomics would<br />
tend to be of <strong>in</strong>herently lower quality than pregenomics<br />
targets, and thus more likely to fail at<br />
some costly phase downstream. That <strong>in</strong>ference is an<br />
oversimplification, and is mislead<strong>in</strong>g.<br />
Certa<strong>in</strong>ly genomics proposes many more novel targets<br />
(as much as 60 to 70 percent of potential targets,<br />
<strong>in</strong> our <strong>in</strong>terviewees’ experience, may belong<br />
to previously unknown target classes), and their<br />
<strong>in</strong>formational quality at that early stage is duly modest.<br />
But that says noth<strong>in</strong>g about their <strong>in</strong>tr<strong>in</strong>sic quality.<br />
Any prudent company, no matter how bold, will<br />
strive to learn more about novel targets before<br />
decid<strong>in</strong>g to pursue them downstream. In our analysis,<br />
<strong>in</strong>vestments made to raise a novel target’s <strong>in</strong>for-
mational quality to the level of a known target’s<br />
would be more than recouped <strong>in</strong> due course.<br />
The overall cost of these novel targets—rais<strong>in</strong>g<br />
their <strong>in</strong>formational quality and then pursu<strong>in</strong>g them<br />
down the value cha<strong>in</strong>—is bound to rise <strong>in</strong>itially.<br />
However, with<strong>in</strong> three to five years from the <strong>in</strong>itial<br />
discovery of a target <strong>in</strong> a novel class, accord<strong>in</strong>g to<br />
our model, the overall cost <strong>in</strong>crease per novel-class<br />
drug could return to average.<br />
Where do the added costs come from? And what<br />
must happen to offset them?<br />
The Cost of Quality Control<br />
Our model predicts that the typical <strong>in</strong>crease will be<br />
about $200 million and more than one year per<br />
drug (that is, a total cost of $790 million versus<br />
$590 million, and a total time to drug of 13.8 years<br />
versus 12.7 years). The <strong>in</strong>crease is ma<strong>in</strong>ly attributable<br />
to the extra time needed to understand target<br />
function and develop appropriate assays <strong>in</strong> target<br />
validation and screen<strong>in</strong>g; also, to the need to screen<br />
a higher proportion of compounds, s<strong>in</strong>ce an appropriate<br />
subset of a larger library cannot be selected<br />
<strong>in</strong> advance.<br />
Chemical optimization costs would <strong>in</strong>crease only if<br />
the novel target required a novel compound (by no<br />
means a necessary requirement, though certa<strong>in</strong>ly a<br />
possible one occasionally). Our model exam<strong>in</strong>es<br />
this worst-case scenario explicitly. If a novel target<br />
does happen to require a novel compound, or a<br />
compound unfamiliar to the medic<strong>in</strong>al chemists,<br />
the potential efficiency loss causes a further<br />
<strong>in</strong>crease of $290 million and more than two years<br />
per drug (that is, a total cost of about $1.1 billion<br />
versus $590 million, and a total time to drug of 15<br />
years versus 12.7 years). The additional <strong>in</strong>creases<br />
here would be due to the extra time needed now for<br />
medic<strong>in</strong>al chemists to learn how to modify the compound<br />
and atta<strong>in</strong> specific properties through trial<br />
and error. But this worst-case scenario should not<br />
be very common.<br />
Mov<strong>in</strong>g further still down the value cha<strong>in</strong>, to the<br />
precl<strong>in</strong>ical and cl<strong>in</strong>ical phases, costs are not<br />
expected to <strong>in</strong>crease. The downstream success rate<br />
for novel compounds or targets should turn out to<br />
be much the same as that for known compounds or<br />
targets, as long as the same standards are applied.<br />
There should be no significant <strong>in</strong>crease <strong>in</strong> toxicity<br />
or decrease <strong>in</strong> efficacy, other than <strong>in</strong> very unlikely<br />
circumstances—for <strong>in</strong>stance, if exist<strong>in</strong>g animal<br />
models somehow proved less suitable, or if drugs<br />
for novel target classes were to <strong>in</strong>teract with metabolic<br />
pathways <strong>in</strong> utterly unfamiliar ways.<br />
Offsett<strong>in</strong>g the Costs<br />
Rais<strong>in</strong>g the <strong>in</strong>formational quality of novel targets<br />
<strong>in</strong>volves a heavy <strong>in</strong>vestment, but it is a wise <strong>in</strong>vestment.<br />
And a fairly quick one: knowledge about<br />
one novel target quickly elucidates other potential<br />
targets <strong>in</strong> the same class. Thanks to feedback<br />
loops, knowledge <strong>in</strong>creases geometrically. As more<br />
is learned, the level of <strong>in</strong>vestment can tail off<br />
accord<strong>in</strong>gly.<br />
In any case, the alternatives to mak<strong>in</strong>g that early<br />
<strong>in</strong>vestment <strong>in</strong> <strong>in</strong>formational quality are far from<br />
attractive. On the one hand, dropp<strong>in</strong>g the targets<br />
would be terribly short-sighted: companies would<br />
be forgo<strong>in</strong>g the opportunity to discover and exploit<br />
untapped sources of revenue. On the other hand,<br />
push<strong>in</strong>g novel targets onward without adequate<br />
<strong>in</strong>formation on them would almost certa<strong>in</strong>ly result<br />
<strong>in</strong> a higher failure rate downstream, with all the<br />
associated implications for cost. An <strong>in</strong>creased failure<br />
rate of just 10 percent across chemical optimization<br />
and all of development would on average<br />
<strong>in</strong>crease costs by about $200 million per drug.<br />
To sum up, then: costs <strong>in</strong>curred early <strong>in</strong> the value<br />
cha<strong>in</strong> (by <strong>in</strong>formation gather<strong>in</strong>g) look preferable<br />
to those that would otherwise be <strong>in</strong>curred later (as<br />
the result of a higher downstream failure rate). All<br />
the more so, given that the early costs should soon<br />
beg<strong>in</strong> fall<strong>in</strong>g (<strong>in</strong>vestment <strong>in</strong> <strong>in</strong>formation is almost<br />
always associated with an experience curve): as<br />
novel target classes become <strong>in</strong>creas<strong>in</strong>gly familiar, it<br />
will become <strong>in</strong>creas<strong>in</strong>gly efficient and economical<br />
to pursue new targets with<strong>in</strong> those classes. So with<br />
proper handl<strong>in</strong>g, the burden of that early cost<br />
<strong>in</strong>crease is just a short-term one, and the productivity<br />
of genomics-driven R&D should soon return<br />
19
20<br />
almost to that of more familiar target classes. We<br />
estimate the time required for this is about three to<br />
five years from the discovery of a novel target,<br />
which is the amount of time it should take to complete<br />
validation and early screen<strong>in</strong>g (assay development).<br />
(See Exhibit 4.)<br />
Putt<strong>in</strong>g the New Technology <strong>in</strong>to Operation<br />
It is one th<strong>in</strong>g to acquire and <strong>in</strong>stall new capabilities<br />
and another to get them to function as they are<br />
meant to. The challenge of mak<strong>in</strong>g genomics technologies<br />
operational has two major components:<br />
eas<strong>in</strong>g the bottlenecks that will develop, and resolv<strong>in</strong>g<br />
the personnel conundrums that are sure to<br />
arise.<br />
The Problem of Bottlenecks<br />
The bottlenecks result, <strong>in</strong> effect, from the unevenness<br />
of the efficiency ga<strong>in</strong>s at different po<strong>in</strong>ts <strong>in</strong> the<br />
value cha<strong>in</strong>. (See Exhibit 5.)<br />
Consider the sixfold <strong>in</strong>crease <strong>in</strong> target identification<br />
described above. This escalat<strong>in</strong>g quantity of<br />
targets could turn out to be not so much a glorious<br />
profusion as an exasperat<strong>in</strong>g glut. Unless there is<br />
some correspond<strong>in</strong>g <strong>in</strong>crease <strong>in</strong> the capacity to<br />
process them downstream, these targets will simply<br />
EXHIBIT 4<br />
IMPACT OF QUALITY ON COST TO DRUG<br />
Pre-genomics Post-genomics<br />
Mix of novel and<br />
known targets<br />
Novel targets only<br />
Novel targets and<br />
novel compounds only<br />
Benefits of experience<br />
over 3-5 years<br />
SOURCES: BCG analysis; <strong>in</strong>dustry <strong>in</strong>terviews.<br />
590<br />
590<br />
790<br />
880<br />
1,080<br />
Approximate cost ($M)<br />
EXHIBIT 5<br />
UNEVEN PRODUCTIVITY GAINS CREATE IMBALANCE<br />
Number today<br />
to get one drug<br />
Not to scale<br />
2,400 1<br />
Potential<br />
targets<br />
400<br />
Potential<br />
targets<br />
108<br />
Validated<br />
targets<br />
Increased<br />
productivity<br />
138<br />
Lead<br />
candidates<br />
loiter at their source <strong>in</strong> a wasteful logjam. Or consider<br />
chemical genomics. Implement<strong>in</strong>g this<br />
approach will build up huge pressure on screen<strong>in</strong>g<br />
resources: send<strong>in</strong>g unvalidated targets for screen<strong>in</strong>g<br />
could <strong>in</strong>volve a 120-fold <strong>in</strong>crease. So too with<br />
efficiency ga<strong>in</strong>s at other po<strong>in</strong>ts <strong>in</strong> the value cha<strong>in</strong>:<br />
without the necessary downstream adjustments,<br />
bottlenecks will <strong>in</strong>evitably develop.<br />
Our capacity to do functional experiments was<br />
completely choked by potential targets.<br />
—VP of discovery,<br />
major pharmaceutical company<br />
But the problem is a dynamic one, and accord<strong>in</strong>gly<br />
very awkward to deal with. Ease one bottleneck and<br />
you often create another downstream. Or ease it<br />
too much and you convert it <strong>in</strong>to a bulge—over-<br />
72<br />
Drug<br />
candidates<br />
Required<br />
productivity 2<br />
Targets Compounds<br />
30<br />
7<br />
INDs Drug<br />
ID Biology<br />
Chemistry<br />
Development<br />
Target ID Target Validation<br />
Screen<strong>in</strong>g Optimization<br />
Precl<strong>in</strong>ical Cl<strong>in</strong>ical<br />
SOURCES: BCG analysis; <strong>in</strong>dustry <strong>in</strong>terviews.<br />
NOTE: Does not <strong>in</strong>clude impact of pharmacogenetics, to be addressed <strong>in</strong><br />
next <strong>in</strong>stallment.<br />
1Number of targets identified by <strong>in</strong>vest<strong>in</strong>g same resources post-genomics as<br />
pre-genomics.<br />
2 Productivity required to exploit all potential targets from target identification.
esourced <strong>in</strong> relation to the flow from upstream,<br />
and hence wasteful once aga<strong>in</strong>. It will take some<br />
adroit adjustment of resources and processes along<br />
the value cha<strong>in</strong> to restore a smooth flow.<br />
This imbalance will affect <strong>in</strong>cumbents—<strong>in</strong>tegrated<br />
companies with established value cha<strong>in</strong>s—worst of<br />
all. They have resources and processes <strong>in</strong> place;<br />
changes are likely to be difficult and disruptive. To<br />
implement the new genomics technologies is troublesome<br />
enough, but then to have to redistribute<br />
resources along the entire value cha<strong>in</strong> will take real<br />
determ<strong>in</strong>ation. (To other companies, by contrast,<br />
bottlenecks might represent very favorable opportunities.<br />
In particular, genomics companies could<br />
benefit. (See sidebar, “Upstart Start-ups—the<br />
Competitors Classified.”)<br />
The Human Factor<br />
To flourish <strong>in</strong> the new genomics era, and possibly<br />
even to survive, companies are go<strong>in</strong>g to have to<br />
engage the new realities. It will not be easy. Some of<br />
the new technologies will tend to overstretch or<br />
even defy exist<strong>in</strong>g capabilities and organizational<br />
structures. All along the value cha<strong>in</strong>, processes and<br />
resources are go<strong>in</strong>g to have to be adjusted.<br />
The resources <strong>in</strong> question <strong>in</strong>clude human resources,<br />
and retrench<strong>in</strong>g, reassign<strong>in</strong>g, or supplement<strong>in</strong>g talented<br />
personnel is a far from straightforward procedure.<br />
But it will have to be done. Organizational<br />
restructur<strong>in</strong>g is likely to entail distress<strong>in</strong>g upheavals<br />
for corporate culture and personnel alike. The<br />
strategies adopted for manag<strong>in</strong>g it will require constant<br />
monitor<strong>in</strong>g and f<strong>in</strong>e-tun<strong>in</strong>g. New modes of<br />
cross-functional collaboration may need to be <strong>in</strong>stituted,<br />
new <strong>in</strong>centives offered, and so on.<br />
I spend half my time look<strong>in</strong>g for talent that isn’t<br />
out there, and the other half worry<strong>in</strong>g where they<br />
would fit if I found them.<br />
—Research director,<br />
lead<strong>in</strong>g biotech company<br />
* * *<br />
In sum, implement<strong>in</strong>g genomics technology will be<br />
very tricky. It will almost certa<strong>in</strong>ly require a holistic,<br />
cross-value-cha<strong>in</strong> perspective. We will discuss potential<br />
solutions to these operational challenges <strong>in</strong> the<br />
third chapter.<br />
A F<strong>in</strong>al Word<br />
By engag<strong>in</strong>g affirmatively with the brave new<br />
genomics world, companies are mak<strong>in</strong>g it possible<br />
to <strong>in</strong>crease R&D productivity substantially. They will<br />
br<strong>in</strong>g to bear an array of <strong>in</strong>dustrialized processes,<br />
<strong>in</strong>formatics, and rich data sets—a formidable comb<strong>in</strong>ation<br />
that promises to boost efficiency, and even<br />
improve success rates, all along the value cha<strong>in</strong>.<br />
Here we have discussed both the opportunities and<br />
the challenges that arise when a company adopts<br />
and implements genomics technologies that are<br />
available today.<br />
The opportunities add up to potential sav<strong>in</strong>gs of<br />
nearly $300 million per drug—about one-third of<br />
the cost—and the prospect of br<strong>in</strong>g<strong>in</strong>g each drug<br />
to market two years sooner. The challenges <strong>in</strong>clude<br />
manag<strong>in</strong>g quality control and deal<strong>in</strong>g with unfamiliar<br />
operational predicaments: bottlenecks along the<br />
pipel<strong>in</strong>e and a host of organizational difficulties.<br />
But for companies that choose not to meet the<br />
genomics revolution head on, the challenge is even<br />
greater: they will be unable to compete. These companies<br />
do more than leave money on the table.<br />
They face the <strong>in</strong>evitability of be<strong>in</strong>g left beh<strong>in</strong>d.<br />
To reap maximum benefit from the new technologies,<br />
companies will need to scrut<strong>in</strong>ize their resources,<br />
processes, and policies throughout the<br />
value cha<strong>in</strong>. Pharmaceutical and biotech managers<br />
will need to ask themselves some tax<strong>in</strong>g questions as<br />
they beg<strong>in</strong> to formulate their genomics strategy:<br />
• Which specific genomics technologies and<br />
approaches make the most sense for our company?<br />
What <strong>in</strong>vestments and capabilities would be<br />
needed to <strong>in</strong>tegrate these new technologies and<br />
approaches successfully?<br />
• What capabilities do we already have? What<br />
<strong>in</strong>vestment are we prepared to make to acquire<br />
those we lack?<br />
21
22<br />
UPSTART START-UPS—THE COMPETITORS CLASSIFIED<br />
As the e-commerce revolution has demonstrated,<br />
disruptive technologies tend to spawn start-ups that<br />
aim to exploit the disruption, either as suppliers to,<br />
or as replacements for, <strong>in</strong>cumbents disoriented by a<br />
chang<strong>in</strong>g world. In the case of genomics, the <strong>in</strong>cumbents<br />
are easy to identify: the traditional pharmaceutical<br />
companies and the larger fully <strong>in</strong>tegrated<br />
biotech companies. But who are the start-ups?<br />
Genomics companies can be classified <strong>in</strong> three<br />
broad groups, on the basis of how closely or distantly<br />
they are related to the actual develop<strong>in</strong>g and<br />
market<strong>in</strong>g of drugs.<br />
The group furthest away conta<strong>in</strong>s the companies<br />
that supply enabl<strong>in</strong>g technologies, <strong>in</strong> the form of<br />
hardware or software. These companies resemble<br />
the merchants of the California gold rush who sold<br />
pickaxes to the m<strong>in</strong>ers, or more recently, companies<br />
such as Cisco and Sun Microsystems that have been<br />
provid<strong>in</strong>g the necessary <strong>in</strong>frastructure for the multitude<br />
of e-commerce practitioners. Examples of such<br />
companies are PE Biosystems, the supplier of highthroughput<br />
sequenc<strong>in</strong>g mach<strong>in</strong>es, and Affymetrix,<br />
the preem<strong>in</strong>ent gene-chip manufacturer and supplier.<br />
The second broad group conta<strong>in</strong>s the companies<br />
supply<strong>in</strong>g <strong>in</strong>formation and knowledge, <strong>in</strong>clud<strong>in</strong>g<br />
those companies that generate proprietary databases<br />
and offer access to them through subscriptions or<br />
fee-per-use bus<strong>in</strong>ess models. One of the best-known<br />
examples is Celera, which sells subscription-based<br />
access to human and animal model-sequence data.<br />
The group also <strong>in</strong>cludes companies that are attempt<strong>in</strong>g<br />
to <strong>in</strong>tegrate and exploit those databases to conduct<br />
<strong>in</strong> silico R&D. An example is LION Bioscience,<br />
which <strong>in</strong>tegrates <strong>in</strong>formation from public and private<br />
sources <strong>in</strong>to a s<strong>in</strong>gle platform to make targets and<br />
leads easier to identify and analyze.<br />
F<strong>in</strong>ally, there is the group of companies that develop<br />
and sell more traditional “physical” drug <strong>in</strong>termediates—targets<br />
and lead compounds. We call these<br />
platform and orchestrator companies.<br />
Platform companies deploy proprietary technology <strong>in</strong><br />
the quest for promis<strong>in</strong>g targets and leads. One such<br />
company is Aurora Biosciences, which has developed<br />
proprietary high-throughput screen<strong>in</strong>g technology<br />
to exploit an opportunity <strong>in</strong> screen<strong>in</strong>g and chemical<br />
genomics. Another example is MorphoSys,<br />
which has developed a platform for rapid development<br />
of high-aff<strong>in</strong>ity antibodies, for use <strong>in</strong> target validation<br />
and therapeutic antibody discovery.<br />
Go<strong>in</strong>g one step further are orchestrator companies,<br />
which str<strong>in</strong>g together adjacent platforms to create<br />
optimized segments of the R&D value cha<strong>in</strong>. As the<br />
orchestrators extend their value cha<strong>in</strong>, they can sell<br />
drug candidates that have progressed further and<br />
further downstream—and have thus become more<br />
and more valuable. Although these companies are<br />
still sell<strong>in</strong>g only <strong>in</strong>termediates, they show every sign<br />
of graduat<strong>in</strong>g <strong>in</strong>to fully <strong>in</strong>tegrated drug companies.<br />
Millennium has already made that transition: concentrat<strong>in</strong>g<br />
<strong>in</strong>itially on genomics target discovery, it<br />
has subsequently developed a full R&D pipel<strong>in</strong>e <strong>in</strong><br />
its own right.<br />
What is the outlook of each of these groups? The<br />
first two (the pure suppliers, either of enabl<strong>in</strong>g technologies<br />
or of <strong>in</strong>formation and knowledge) would<br />
appear to be well positioned if they target areas of<br />
scarcity (that is, bottlenecks) with proprietary prod-
ucts, or if they have enough clients to achieve scale<br />
efficiencies reachable only by supply<strong>in</strong>g multiple<br />
companies. But so far, most of these companies<br />
have struggled to f<strong>in</strong>d a susta<strong>in</strong>able, profitable bus<strong>in</strong>ess<br />
model.<br />
Meanwhile, the third group seems <strong>in</strong> the most promis<strong>in</strong>g<br />
position. The pharmaceutical bus<strong>in</strong>ess rema<strong>in</strong>s<br />
attractive, with marg<strong>in</strong>s averag<strong>in</strong>g more than 80 percent,<br />
so it is easy to see why so many genomics<br />
companies aspire to become drug companies. But<br />
• Where is <strong>in</strong>dustrywide scale to be found rather<br />
than just company-level scale? Which capabilities<br />
should we therefore develop <strong>in</strong>-house, and which<br />
through partners?<br />
• How will any new technologies affect the rest of<br />
the value cha<strong>in</strong>? How can we optimize decision<br />
mak<strong>in</strong>g and <strong>in</strong>formation flow up and down the<br />
value cha<strong>in</strong>?<br />
•What are the implications for the organization of<br />
the changes we wish to make? How feasible is the<br />
necessary restructur<strong>in</strong>g? And what would be the<br />
most efficient way to carry it out?<br />
there are many hurdles en route, and overambitious<br />
companies risk tripp<strong>in</strong>g over them. Although many of<br />
these companies may fail, those that succeed will<br />
have a transformational impact on the <strong>in</strong>dustry.<br />
Moreover, the traditional drug companies seem to be<br />
mov<strong>in</strong>g <strong>in</strong> the opposite direction, <strong>in</strong>creas<strong>in</strong>gly outsourc<strong>in</strong>g<br />
portions of their R&D value cha<strong>in</strong>s. What is<br />
go<strong>in</strong>g on? We will try to answer that question <strong>in</strong> the<br />
third chapter, when we exam<strong>in</strong>e these trends <strong>in</strong><br />
more detail.<br />
These questions can be addressed by thorough,<br />
thoughtful analysis. Key <strong>in</strong>vestment decisions will<br />
be required, as well as a carefully planned implementation<br />
program to ensure that the value of<br />
those decisions is captured.<br />
In the next chapter, we turn to genetics and analyze<br />
its likely impact on R&D productivity. In the f<strong>in</strong>al<br />
chapter, we will exam<strong>in</strong>e more closely the strategic<br />
choices and operational implications of the various<br />
changes <strong>in</strong> prospect.<br />
23
24<br />
Chapter 2: The Impact of Genetics<br />
Preface<br />
Hav<strong>in</strong>g discussed the genomics wave <strong>in</strong> the previous<br />
chapter, and the way that it promises to enhance<br />
R&D productivity, we now turn to the genetics wave.<br />
Several broad differences suggest themselves immediately.<br />
Where the genomics wave is technologydriven,<br />
the genetics wave is better viewed as datadriven,<br />
exploit<strong>in</strong>g the known details of the human<br />
genome and <strong>in</strong>dividual variations with<strong>in</strong> it. Where<br />
the genomics wave br<strong>in</strong>gs benefits ma<strong>in</strong>ly at the<br />
drug-discovery and precl<strong>in</strong>ical phases, the genetics<br />
wave will prove its worth <strong>in</strong> both the earliest phase<br />
and the later phases of the value cha<strong>in</strong>—target discovery<br />
and the cl<strong>in</strong>ic. Where the genomics wave<br />
enhances R&D productivity ma<strong>in</strong>ly by secur<strong>in</strong>g<br />
great improvements <strong>in</strong> efficiency (with only modest<br />
improvements, if any, <strong>in</strong> success rates), the genetics<br />
wave could boost success rates dramatically as well.<br />
One further difference should be mentioned:<br />
where our model for the genomics wave was put forward<br />
with considerable confidence, our model for<br />
the genetics wave is more tentative. At this early<br />
stage, any assessment of genetics’ impact on the economics<br />
of R&D is bound to be provisional. Certa<strong>in</strong>ly<br />
genetics has huge potential: if all goes accord<strong>in</strong>g to<br />
plan, it will change R&D productivity beyond recognition.<br />
But between that potential and its full realization<br />
lie several years and many obstacles.<br />
The potential consists <strong>in</strong> tremendous sav<strong>in</strong>gs. First,<br />
genetics can br<strong>in</strong>g about great efficiency ga<strong>in</strong>s by<br />
mak<strong>in</strong>g it possible to shorten or even bypass various<br />
steps <strong>in</strong> the value cha<strong>in</strong>. Second, genetics holds the<br />
prospect of transform<strong>in</strong>g success rates: failures <strong>in</strong><br />
the R&D pipel<strong>in</strong>e currently account for 75 percent<br />
of the total cost to drug. But offsett<strong>in</strong>g such opportunities,<br />
dangers loom large. Rid<strong>in</strong>g the genetics<br />
wave <strong>in</strong>volves a greater risk than rid<strong>in</strong>g the<br />
genomics wave alone—though it is more exhilarat<strong>in</strong>g<br />
and, if the risks are successfully negotiated, ultimately<br />
more reward<strong>in</strong>g. How to choose between discretion<br />
and valor is a crucial strategic decision that<br />
companies will have to make.<br />
In analyz<strong>in</strong>g the economic implications of genetics,<br />
this chapter of our report considers the effect only<br />
on pharmaceutical R&D. But genetics is likely to<br />
affect health care far beyond R&D, <strong>in</strong> both the<br />
short and the long term. In the short term, new<br />
market opportunities should arise <strong>in</strong> the formerly<br />
sleepy diagnostics sector. (Drug companies may or<br />
may not be able to exploit these opportunities: see<br />
sidebar, “Diagnostics—an Opportunity Too Good to<br />
Miss…and Perhaps Too Good to Grasp.”) In the<br />
longer term, genetics is likely to transform the<br />
delivery of health care. Increas<strong>in</strong>gly, diseases will be<br />
redef<strong>in</strong>ed <strong>in</strong>to various subtypes—a ref<strong>in</strong>ement that<br />
should facilitate more appropriate care and more<br />
“rational” drug design. The comb<strong>in</strong>ation of new<br />
diagnostics, new disease def<strong>in</strong>itions, and new tailored<br />
drugs should prove a w<strong>in</strong>n<strong>in</strong>g one, and may<br />
well usher <strong>in</strong> an era of <strong>in</strong>dividualized medic<strong>in</strong>e.<br />
R&D rema<strong>in</strong>s the focus of our analysis here, however:<br />
specifically, the wide range of economic reactions<br />
that R&D might show under the impact of the<br />
new genetics <strong>in</strong>formation. We discuss the tremendous<br />
opportunities as well as the accompany<strong>in</strong>g
isks <strong>in</strong>herent <strong>in</strong> genetics-based R&D, and explore<br />
various ways of manag<strong>in</strong>g them.<br />
Two K<strong>in</strong>ds of Genetics Approaches<br />
There are two relevant approaches to consider<br />
when assess<strong>in</strong>g the economic impact of genetics on<br />
DIAGNOSTICS—AN OPPORTUNITY TOO GOOD TO MISS…<br />
AND PERHAPS TOO GOOD TO GRASP<br />
It will be several years before genetics fulfills its<br />
promise. In the meantime, however, companies<br />
might beg<strong>in</strong> to enjoy a prelim<strong>in</strong>ary reward, <strong>in</strong> the<br />
form of diagnostics—essentially a byproduct of their<br />
broader genomics research programs. Certa<strong>in</strong>ly diagnostics<br />
is the subject of great expectations, though<br />
whether and how soon it will meet them rema<strong>in</strong>s to<br />
be seen.<br />
Many research projects <strong>in</strong> genomics and genetics<br />
will devise diagnostic tests as a matter of course—<strong>in</strong><br />
parallel with research or simply as a prelim<strong>in</strong>ary<br />
step, perhaps—without portray<strong>in</strong>g them that way.<br />
Diagnostic tests can be understood <strong>in</strong> a fairly broad<br />
sense here. Disease genetics, for example, <strong>in</strong> identify<strong>in</strong>g<br />
a target, is <strong>in</strong> effect f<strong>in</strong>d<strong>in</strong>g a marker of disease<br />
susceptibility. Expression profil<strong>in</strong>g, <strong>in</strong> identify<strong>in</strong>g the<br />
molecular differences characteriz<strong>in</strong>g a disease’s different<br />
subtypes, is po<strong>in</strong>t<strong>in</strong>g the way to differentiated<br />
and f<strong>in</strong>e-tuned therapies. And pharmacogenetics, <strong>in</strong><br />
identify<strong>in</strong>g variations <strong>in</strong> drug response among various<br />
patients, could be help<strong>in</strong>g to suggest the most<br />
suitable drug for them.<br />
The opportunities <strong>in</strong>herent <strong>in</strong> diagnostics will appeal<br />
to drug companies at several levels. First, costs are<br />
low. The <strong>in</strong>tellectual capital needed to develop a diagnostic<br />
test comes free, courtesy of exist<strong>in</strong>g research<br />
<strong>in</strong> drug discovery and development; validation<br />
studies can be run <strong>in</strong> parallel with drug efficacy studies,<br />
or perhaps can even simply borrow their results<br />
and extrapolate from them; and as for safety studies,<br />
diagnostic tests don’t need any. All <strong>in</strong> all, then, the<br />
<strong>in</strong>cremental spend<strong>in</strong>g required to develop a marketable<br />
diagnostic test is, relatively speak<strong>in</strong>g, paltry.<br />
R&D: disease genetics and pharmacogenetics. They<br />
operate at different stages of the value cha<strong>in</strong>.<br />
Disease genetics is <strong>in</strong>voked earlier, dur<strong>in</strong>g the discovery<br />
phase: it <strong>in</strong>volves the search for genes that make<br />
people susceptible to particular diseases, with the<br />
aim of then f<strong>in</strong>d<strong>in</strong>g targets. Pharmacogenetics is the<br />
Second, rewards are prompt. Diagnostics, <strong>in</strong> bypass<strong>in</strong>g<br />
most of the traditional steps of pharmaceutical<br />
R&D, can be brought to market not only far more<br />
cheaply than drugs, but far more quickly too. Drug<br />
companies are thereby able to realize some unusually<br />
fast payback on their R&D spend<strong>in</strong>g.<br />
Third, the market outlook is favorable. As new therapies<br />
proliferate, more diagnostics will be demanded;<br />
and as technologies advance, new types of diagnostics<br />
will become available. The signs are good.<br />
These opportunities are to some extent offset, however,<br />
if not by risks, then at least by challenges.<br />
There is the challenge of novelty, for <strong>in</strong>stance: for<br />
many traditional companies, diagnostics would<br />
<strong>in</strong>volve manufactur<strong>in</strong>g an unfamiliar k<strong>in</strong>d of product—<br />
a kit—and that <strong>in</strong> turn would <strong>in</strong>volve develop<strong>in</strong>g new<br />
capabilities, or else partner<strong>in</strong>g with a dedicated diagnostics<br />
company. Companies that have an <strong>in</strong>-house<br />
diagnostics division, such as Hoffmann-La Roche,<br />
Abbott, and Bayer, will have an advantage here.<br />
Then there is the challenge of <strong>in</strong>tellectual-property<br />
rights: companies might f<strong>in</strong>d it more difficult to<br />
assert those rights over diagnostics than over their<br />
f<strong>in</strong>d<strong>in</strong>gs <strong>in</strong> pharmaceutical research.<br />
Perhaps the most daunt<strong>in</strong>g challenge is tim<strong>in</strong>g: diagnostic<br />
tests will tend to emerge too speedily, becom<strong>in</strong>g<br />
available sooner than the therapies they <strong>in</strong>dicate.<br />
So the chief appeal of <strong>in</strong>vest<strong>in</strong>g <strong>in</strong> diagnostics—its<br />
prompt availability—may be undercut. Drug companies<br />
may have to delay market<strong>in</strong>g their diagnostics<br />
(and thus delay capitaliz<strong>in</strong>g on the opportunities)<br />
until their drug R&D pipel<strong>in</strong>es catch up.<br />
25
26<br />
genetics-based form of pharmacogenomics (see<br />
sidebar, “Pharmacogenomics—Some Def<strong>in</strong>itions”),<br />
and comes <strong>in</strong>to play later, <strong>in</strong> the development<br />
phase: it <strong>in</strong>volves predict<strong>in</strong>g the efficacy and side<br />
effects of candidate drugs.<br />
The data explosion detonated by genomics technology<br />
has created vast amounts of genetic <strong>in</strong>formation,<br />
ready for sift<strong>in</strong>g. The f<strong>in</strong>d<strong>in</strong>gs of the<br />
Human Genome Project and related endeavors are<br />
merely the start<strong>in</strong>g po<strong>in</strong>t. The ultimate goal is to<br />
elucidate the genetic basis of human disease and<br />
drug response. In the short term, genetics research<br />
will enable scientists to predict disease susceptibility<br />
and likely drug response <strong>in</strong> <strong>in</strong>dividuals; <strong>in</strong> the<br />
longer term, it should help to improve the quality<br />
of pharmaceuticals and medical diagnoses.<br />
PHARMACOGENOMICS—SOME DEFINITIONS<br />
Pharmacogenomics is the use of genomics<br />
approaches to elucidate drug response. There are<br />
three relevant approaches: via DNA, via RNA, and<br />
via prote<strong>in</strong>s, and three correspond<strong>in</strong>g forms of pharmacogenomics:<br />
pharmacogenomics us<strong>in</strong>g genetic<br />
approaches (or pharmacogenetics), expression profil<strong>in</strong>g<br />
(or expression pharmacogenomics), and proteomics<br />
(or proteomic pharmacogenomics).<br />
Pharmacogenetics predicts patients’ drug response<br />
by analyz<strong>in</strong>g the genetic variations <strong>in</strong> their DNA. It is<br />
the form of pharmacogenomics discussed <strong>in</strong> the<br />
ma<strong>in</strong> text here.<br />
Expression pharmacogenomics predicts patients’<br />
drug response by analyz<strong>in</strong>g their RNA levels—specifically,<br />
by compar<strong>in</strong>g the amounts of RNA found <strong>in</strong><br />
different samples to determ<strong>in</strong>e which genes are<br />
expressed at different levels. An example: a research<br />
group at The Whitehead Institute study<strong>in</strong>g two very<br />
similar leukemias (AML and ALL) has observed a<br />
dist<strong>in</strong>ct difference <strong>in</strong> expression levels of specific<br />
genes, and thereby provided a quick and reliable<br />
method for differentiat<strong>in</strong>g them. Patients are now<br />
less at risk of be<strong>in</strong>g misdiagnosed and be<strong>in</strong>g given<br />
Atta<strong>in</strong><strong>in</strong>g the short-term goal is, conceptually, simple<br />
enough. The genetic codes of <strong>in</strong>dividuals differ<br />
<strong>in</strong> t<strong>in</strong>y, but sometimes decisive, details. By compar<strong>in</strong>g<br />
an <strong>in</strong>dividual’s genetic variations aga<strong>in</strong>st the<br />
“standard” genome, scientists should be able to predict<br />
whether that <strong>in</strong>dividual is at risk for a specific<br />
disease, and, if so, how well suited he or she is to a<br />
particular drug therapy—the work respectively of<br />
disease genetics and pharmacogenetics.<br />
The two approaches benefit R&D economics <strong>in</strong> different<br />
ways. Disease genetics will improve efficiency<br />
<strong>in</strong> target discovery and, by lead<strong>in</strong>g to the discovery<br />
of particularly high-quality targets, will br<strong>in</strong>g about<br />
improved success rates <strong>in</strong> validation and downstream.<br />
Pharmacogenetics, by enabl<strong>in</strong>g scientists to<br />
select patients more appropriately for cl<strong>in</strong>ical trials,<br />
an <strong>in</strong>correct, and possibly lethal, drug treatment: <strong>in</strong><br />
effect, the test screens for adverse drug response.<br />
Expression pharmacogenomics seems to be mov<strong>in</strong>g<br />
from academic studies and biotechs <strong>in</strong>to more ma<strong>in</strong>stream<br />
pharmaceutical R&D. Witness the recent purchase<br />
by Merck and Co. of Rosetta Inpharmatics, a<br />
biotech founded specifically to develop expression<br />
pharmacogenomics.<br />
F<strong>in</strong>ally, proteomic pharmacogenomics predicts<br />
patients’ drug response by analyz<strong>in</strong>g their prote<strong>in</strong><br />
levels—specifically, by compar<strong>in</strong>g prote<strong>in</strong> read<strong>in</strong>gs<br />
<strong>in</strong> different tissue samples to identify prote<strong>in</strong>s that<br />
differ either <strong>in</strong> structure or <strong>in</strong> expression levels.<br />
Consider the example of an aberrant fusion of two<br />
prote<strong>in</strong>s called Bcr and Abl, which occurs <strong>in</strong> more<br />
than 95 percent of patients with CML (chronic<br />
myeloid leukemia, which accounts for about 20 percent<br />
of all cases of adult leukemia). This aberrant<br />
fusion prote<strong>in</strong> is present only <strong>in</strong> cancer cells. It dist<strong>in</strong>guishes<br />
itself from its normal counterparts by its<br />
<strong>in</strong>creased size. It can be used not only to monitor the<br />
progression of the disease but also to test whether<br />
Gleevec, a revolutionary new drug, would provide an<br />
effective therapy.
will not only help to make the trials faster and<br />
cheaper, but also allow some drugs to pass that<br />
would otherwise have failed ow<strong>in</strong>g to poor efficacy<br />
or side effects.<br />
High-Risk, High-Reward Research<br />
In the near-to-medium term, genetics promises to<br />
reduce R&D costs dramatically. Or does it? Our<br />
model shows that the application of disease genetics<br />
and pharmacogenetics together could, <strong>in</strong> the very<br />
best case, save as much as two-thirds of the current<br />
cost to develop a drug and nearly two years. Of<br />
these potential sav<strong>in</strong>gs, the vast majority comes<br />
from disease genetics. Of the rema<strong>in</strong>der, some cannot<br />
be clearly apportioned to either disease genetics<br />
or pharmacogenetics but must be credited to<br />
their jo<strong>in</strong>t efforts, be<strong>in</strong>g the product of synergies<br />
realized when disease genetics <strong>in</strong>formation is <strong>in</strong>corporated<br />
<strong>in</strong>to pharmacogenetics-driven cl<strong>in</strong>ical trials.<br />
(Although genetics can be comb<strong>in</strong>ed with the<br />
genomics technologies described <strong>in</strong> the previous<br />
chapter, the potential sav<strong>in</strong>gs are not additive. In<br />
the next chapter, we will assess the total sav<strong>in</strong>gs realistically<br />
achievable through the application of genomics<br />
and genetics.)<br />
Our model also shows that realiz<strong>in</strong>g this potential<br />
will be far from straightforward and is far from<br />
guaranteed. The high-end sav<strong>in</strong>gs estimate assumes<br />
a positive resolution of several scientific, technical,<br />
and market risks. With every setback, the sav<strong>in</strong>gs<br />
dim<strong>in</strong>ish. Too many setbacks, and the sav<strong>in</strong>gs fall to<br />
zero. Implement<strong>in</strong>g genetics could even turn out to<br />
have a negative impact on value, ow<strong>in</strong>g to adverse<br />
market dynamics.<br />
Such a double-edged sword is awkward to wield. If<br />
companies fail to grasp it at all, they forfeit the<br />
opportunity to reap enormous rewards—nearly<br />
double the sav<strong>in</strong>gs possible with genomics technologies.<br />
If they do grasp it, they put themselves <strong>in</strong><br />
some danger, and will need to develop a genetics<br />
strategy aligned with their risk profile and with specific<br />
market conditions. As an approach to research<br />
and development, genetics rema<strong>in</strong>s risky, but full of<br />
promise, too. This chapter assesses the promise and<br />
the risk alike.<br />
Disease Genetics<br />
Underly<strong>in</strong>g many diseases are genetic variants, or<br />
polymorphisms—alterations <strong>in</strong> the DNA sequence<br />
of a given gene that <strong>in</strong>fluence <strong>in</strong>dividual risk of disease.<br />
(The term polymorphism usually refers to a<br />
common variant—one found <strong>in</strong> more than 1 percent<br />
of the population.) Often the genetic difference<br />
consists of just a s<strong>in</strong>gle altered letter <strong>in</strong> the<br />
genetic code (the variant is then known as a s<strong>in</strong>gle<br />
nucleotide polymorphism, or SNP). Such a t<strong>in</strong>y alteration<br />
can have fatal consequences; for example, the<br />
substitution of an A for a T <strong>in</strong> the hemoglob<strong>in</strong> ß<br />
gene is responsible for sickle cell anemia. The goal<br />
of disease genetics is to identify such DNA alterations;<br />
so far, this goal has proved elusive for all but<br />
the most strongly <strong>in</strong>herited conditions. (See sidebar,<br />
“Disease Genetics—Various Approaches to<br />
Variants.”) But researchers persevere, s<strong>in</strong>ce f<strong>in</strong>d<strong>in</strong>g<br />
the causal variants can be a major step to f<strong>in</strong>d<strong>in</strong>g a<br />
treatment—and potentially a cure.<br />
Uncerta<strong>in</strong>ty persists: the technologies have yet to<br />
prove their full worth by produc<strong>in</strong>g the necessary<br />
quota of practical results. But if all goes accord<strong>in</strong>g<br />
to plan, the sav<strong>in</strong>gs realized by pharmaceutical<br />
companies will be enormous.<br />
The Potential<br />
The sav<strong>in</strong>gs would come partially from improved<br />
efficiency <strong>in</strong> discovery, but pr<strong>in</strong>cipally from<br />
improved success <strong>in</strong> target validation and cl<strong>in</strong>ical<br />
trials. The improvements <strong>in</strong> efficiency result from<br />
the consolidation of target discovery <strong>in</strong>to a s<strong>in</strong>gle<br />
step; the improved success rates result from the<br />
ref<strong>in</strong><strong>in</strong>g of target identification, mak<strong>in</strong>g it possible<br />
to p<strong>in</strong>po<strong>in</strong>t the targets associated with disease susceptibility<br />
<strong>in</strong> humans.<br />
Once the targets have been located, they can boast<br />
a collective superiority <strong>in</strong> three particular respects.<br />
First, their relevance to human disease is certa<strong>in</strong>.<br />
They have, after all, been validated <strong>in</strong> humans,<br />
show<strong>in</strong>g directly that modulation of gene activity<br />
leads to alteration <strong>in</strong> the <strong>in</strong>tensity or duration of<br />
disease. (By contrast, other targets are usually iden-<br />
27
28<br />
DISEASE GENETICS—VARIOUS APPROACHES TO VARIANTS<br />
When disease genetics is used to identify or test<br />
variant genes, the <strong>in</strong>vestigation can take a variety of<br />
forms. There are three key dimensions <strong>in</strong> the design<br />
of such an <strong>in</strong>vestigation: narrow/broad, l<strong>in</strong>kage/association,<br />
and direct/<strong>in</strong>direct. These dimensions all<br />
have a bear<strong>in</strong>g on cost and the chances of success.<br />
First, researchers can exam<strong>in</strong>e some of the genes (<strong>in</strong><br />
a narrow study, or candidate-gene study) or all of<br />
the genes (<strong>in</strong> a broad study, or genome-wide scan).<br />
Second, these genes can be exam<strong>in</strong>ed through <strong>in</strong>heritance<br />
patterns <strong>in</strong> families prone to the disease (<strong>in</strong> a<br />
l<strong>in</strong>kage study) or by compar<strong>in</strong>g <strong>in</strong>dividual patients<br />
with healthy <strong>in</strong>dividuals <strong>in</strong> the population at large (<strong>in</strong><br />
an association study).<br />
F<strong>in</strong>ally, with<strong>in</strong> association studies, each variant gene<br />
can be studied directly or <strong>in</strong>directly: researchers can<br />
test each variant <strong>in</strong>dividually for any <strong>in</strong>volvement <strong>in</strong><br />
the disease (<strong>in</strong> a direct study) or test clusters of<br />
closely positioned variants for the presence of a culprit<br />
among them (<strong>in</strong> an <strong>in</strong>direct study).<br />
The earliest disease genetics <strong>in</strong>vestigations, conducted<br />
prior to the 1980s, were association studies,<br />
and direct, and as narrow as could be—study<strong>in</strong>g just<br />
a s<strong>in</strong>gle gene, which had been selected on the basis<br />
of biological knowledge about disease mechanisms.<br />
The researcher would seek polymorphisms <strong>in</strong> the<br />
gene, and compare their frequencies <strong>in</strong> patients and<br />
controls. This early approach did achieve some<br />
notable successes, <strong>in</strong>clud<strong>in</strong>g, <strong>in</strong> 1956, the first discovery<br />
of an <strong>in</strong>herited genetic variation found to<br />
cause disease—the variant underly<strong>in</strong>g sickle cell<br />
anemia. The approach had a serious limitation, however:<br />
it allowed very few genes to be exam<strong>in</strong>ed, and<br />
it required a prior hypothesis.<br />
In the 1980s and 1990s, attention turned to families<br />
show<strong>in</strong>g an <strong>in</strong>herited pattern of disease. The<br />
<strong>in</strong>vestigations took the form of broad l<strong>in</strong>kage studies,<br />
and were tremendously successful <strong>in</strong> identify<strong>in</strong>g<br />
some genes responsible for s<strong>in</strong>gle-gene disorders,<br />
notably the cystic fibrosis gene <strong>in</strong> 1989. Such studies,<br />
be<strong>in</strong>g genome-wide, were now unbiased and<br />
comprehensive. But the actual identification of<br />
genes rema<strong>in</strong>ed a slow, pa<strong>in</strong>stak<strong>in</strong>g laboratory<br />
process. So the early versions of such studies were<br />
really suitable only for monogenic diseases. Hopes<br />
were raised <strong>in</strong> the early 1990s, when companies<br />
such as Millennium, Sequana, and Myriad were set<br />
up to develop and exploit these techniques <strong>in</strong> the<br />
quest to identify the genes implicated <strong>in</strong> common<br />
polygenic diseases. Their <strong>in</strong>itiative seems to be<br />
stalled at the moment: although the localiz<strong>in</strong>g of<br />
disease-related genes has become more efficient,<br />
the actual locat<strong>in</strong>g of them rema<strong>in</strong>s discourag<strong>in</strong>gly<br />
difficult. That task would be better suited to association<br />
studies.<br />
Given the limitations <strong>in</strong> genome-wide l<strong>in</strong>kage studies,<br />
association studies have recently come back <strong>in</strong>to<br />
fashion, fortified by the efforts of the Human<br />
Genome Project, Celera, and the SNP Consortium.<br />
These studies have high-throughput technologies to<br />
undergird them, as well as comprehensive databases<br />
of gene sequences and SNPs.<br />
Of the two possible approaches here, direct and <strong>in</strong>direct,<br />
the former looks like a very formidable task.<br />
The researcher conduct<strong>in</strong>g a direct association study,<br />
and aim<strong>in</strong>g to f<strong>in</strong>d the actual polymorphism underly<strong>in</strong>g<br />
the specified disease, is confronted by the entire<br />
set of common variants <strong>in</strong> the genome—expected to<br />
number some ten million. To exam<strong>in</strong>e such a horde<br />
of variants with current technology would be <strong>in</strong>ord<strong>in</strong>ately<br />
time-consum<strong>in</strong>g and expensive.<br />
Hence the hopes—and funds—now be<strong>in</strong>g <strong>in</strong>vested <strong>in</strong><br />
<strong>in</strong>direct studies. S<strong>in</strong>ce variants <strong>in</strong> close proximity<br />
tend to form clusters (known as haplotypes), it may<br />
be possible to track down the disease-related polymorphisms<br />
us<strong>in</strong>g only a small proportion of all<br />
SNPs. In light of recent research f<strong>in</strong>d<strong>in</strong>gs, <strong>in</strong>direct<br />
association studies of this k<strong>in</strong>d do look practicable,<br />
if not quite imm<strong>in</strong>ent.
tified through the use of animals and tissue cultures,<br />
and so their relevance to human disease is<br />
largely speculative.) In other words, there is no possibility<br />
of failure <strong>in</strong> target validation, because identified<br />
targets are ipso facto validated. (This is, of<br />
course, no guarantee of their drugability—their<br />
responsiveness to small-molecule <strong>in</strong>tervention.)<br />
It would not be possible to overstate the value of <strong>in</strong><br />
vivo human validation. Most of what passes for<br />
target validation today is largely conjectural <strong>in</strong><br />
relation to the disease <strong>in</strong> question.<br />
—Diabetes researcher,<br />
Harvard Medical School<br />
Second, the frequency of the causal polymorphisms<br />
is known at the outset. If a study identifies multiple<br />
genes associated with a particular disease, it will<br />
also reveal their relative culpability. Consider the<br />
example of Alzheimer’s disease, a heritable but<br />
genetically complex disorder. On the one hand,<br />
there are variants <strong>in</strong> three genes—PS1, PS2, and<br />
APP—that are very rare but very potent: if a person<br />
has any of them, he or she is almost certa<strong>in</strong> to<br />
develop Alzheimer’s. On the other hand, there is<br />
the ApoE4 polymorphism of the ApoE gene, which<br />
has a more modest effect on disease susceptibility<br />
but is much more common <strong>in</strong> the population at<br />
large, and among patients with Alzheimer’s.<br />
Information of this k<strong>in</strong>d can be useful for predict<strong>in</strong>g<br />
a drug’s potential marketability: although it<br />
might be equally feasible to develop a drug that<br />
<strong>in</strong>fluences the rarer variants, a drug target<strong>in</strong>g<br />
ApoE4 might expect broader effectiveness, and<br />
thus a larger market, and so might take precedence<br />
<strong>in</strong> further research. (The rarer variants may still be<br />
worth pursu<strong>in</strong>g, us<strong>in</strong>g pathway analysis.)<br />
F<strong>in</strong>ally, the nature of the relevant polymorphisms is<br />
known—different disease-<strong>in</strong>duc<strong>in</strong>g mechanisms<br />
among variant forms, for <strong>in</strong>stance. Such <strong>in</strong>formation<br />
may help to streaml<strong>in</strong>e cl<strong>in</strong>ical trials if used by<br />
efficacy-based pharmacogenetics to identify “nonresponders”—patients<br />
who lack the crucial DNA<br />
alteration, and hence are unlikely to experience the<br />
<strong>in</strong>tended effect of a candidate drug—and exclude<br />
them from the trials. (In model<strong>in</strong>g the potential of<br />
disease genetics, we have <strong>in</strong>cluded this effect of<br />
efficacy-based pharmacogenetics. See the section<br />
on pharmacogenetics below for further details.)<br />
Depend<strong>in</strong>g on the approach taken, cost sav<strong>in</strong>gs per<br />
drug could be as great as $420 million, with the<br />
potential time sav<strong>in</strong>gs rang<strong>in</strong>g from 0.7 to about 1.6<br />
years (produc<strong>in</strong>g an added $290 million of value<br />
per drug). (See Exhibit 6.) Of the cost sav<strong>in</strong>gs, the<br />
vast majority would be yielded by the improvements<br />
<strong>in</strong> success rates: $390 million, consist<strong>in</strong>g of $110<br />
million <strong>in</strong> validation and $280 million <strong>in</strong> the cl<strong>in</strong>ic.<br />
EXHIBIT 6<br />
DISEASE GENETICS OFFERS GREAT SAVINGS POTENTIAL<br />
Cost to drug<br />
Pre-genomics<br />
Candidate<br />
gene study<br />
Genome-wide scan<br />
Genome-wide scan<br />
plus pathway analysis<br />
Time to drug<br />
Pre-genomics<br />
Candidate<br />
gene study<br />
Genome-wide scan<br />
Genome-wide scan<br />
plus pathway analysis<br />
ID Biology<br />
Target ID Target Validation<br />
0<br />
0<br />
200<br />
5<br />
Chemistry<br />
400<br />
Screen<strong>in</strong>g Optimization<br />
10<br />
460<br />
485<br />
455<br />
600<br />
800<br />
13.1<br />
15<br />
14<br />
14.7<br />
Development<br />
16.5<br />
Precl<strong>in</strong>ical Cl<strong>in</strong>ical<br />
SOURCES: Industry <strong>in</strong>terviews; scientific literature; public f<strong>in</strong>ancial data;<br />
BCG analysis.<br />
880<br />
1,000<br />
$M<br />
20<br />
Years<br />
29
30<br />
Efficiency improvements <strong>in</strong> target discovery account<br />
for the rema<strong>in</strong><strong>in</strong>g sav<strong>in</strong>gs.<br />
The Uncerta<strong>in</strong>ty<br />
For these vast sav<strong>in</strong>gs to materialize, two requirements<br />
will have to be met. First, disease genetics<br />
must prove scientifically feasible for the relevant<br />
common diseases. Second, not only must studies <strong>in</strong><br />
humans work; <strong>in</strong> addition, the targets they identify<br />
must be drugable; fail<strong>in</strong>g that, identify<strong>in</strong>g the disease<br />
genes is po<strong>in</strong>tless, and all the effort that has<br />
gone <strong>in</strong>to f<strong>in</strong>d<strong>in</strong>g them will be wasted. (See sidebar,<br />
“Drug-Resistant?—Are Disease Genes Drugable<br />
Targets?”)<br />
Feasibility—the Limitations of Technology<br />
Fundamental technological concerns still hover over<br />
disease genetics. Can it actually be done? The results<br />
so far have been very modest. The bonanza of<br />
clearly documented disease-susceptibility genes for<br />
common multigenic diseases has yet to materialize.<br />
Candidate gene studies, for <strong>in</strong>stance, are by def<strong>in</strong>ition<br />
limit<strong>in</strong>g: they focus on a subset of genes def<strong>in</strong>ed<br />
by a prior hypothesis, and therefore risk exclud<strong>in</strong>g<br />
some crucial culprits. And genome-wide l<strong>in</strong>kage<br />
studies, although highly successful <strong>in</strong> address<strong>in</strong>g<br />
s<strong>in</strong>gle-gene diseases, have proved disappo<strong>in</strong>t<strong>in</strong>g for<br />
DRUG-RESISTANT?—ARE DISEASE GENES DRUGABLE TARGETS?<br />
The skeptics pose an awkward question: Will disease-related<br />
genes ever prove to be drugable <strong>in</strong> significant<br />
numbers? The record so far is hardly encourag<strong>in</strong>g.<br />
Some disease genes, such as CFTR <strong>in</strong> cystic<br />
fibrosis, were identified long ago, yet have failed to<br />
generate therapeutics. The <strong>in</strong>frequency of success<br />
stories, such as Ceredase—a drug for type I<br />
Gaucher’s disease that was essentially a creation of<br />
disease genetics—only highlights the general trend<br />
of failure.<br />
These long-identified disease genes tend to be for<br />
s<strong>in</strong>gle-gene disorders, however. And such disorders<br />
are by their nature difficult to cure. They are b<strong>in</strong>ary<br />
phenomena: the gene is broken, you get the disease.<br />
the more common multigenic k<strong>in</strong>d of disease: a disease-related<br />
gene might be accurately p<strong>in</strong>po<strong>in</strong>ted <strong>in</strong><br />
affected families (such as the BRCA1 breast cancer<br />
susceptibility gene), only for it then to show very low<br />
prevalence outside the families used <strong>in</strong> identify<strong>in</strong>g<br />
it. True, these two approaches might become more<br />
tractable now, <strong>in</strong> the wake of the sequenc<strong>in</strong>g of the<br />
human genome and the development of comprehensive<br />
SNP maps (catalogs of the characteristics<br />
and locations of SNPs <strong>in</strong> the genome).<br />
As for genome-wide association studies, considered<br />
by many experts to be the most promis<strong>in</strong>g of all,<br />
they have only recently became practicable: all the<br />
requisite tools (a full genome sequence with a SNP<br />
map to match, genotyp<strong>in</strong>g technologies, and so on)<br />
appear to be <strong>in</strong> place. But the approach rema<strong>in</strong>s virtually<br />
untested, ow<strong>in</strong>g to the still exorbitant cost of<br />
genotyp<strong>in</strong>g. The preferable form of genome-wide<br />
association studies would clearly be the <strong>in</strong>direct<br />
k<strong>in</strong>d—still cover<strong>in</strong>g the entire genome, but genotyp<strong>in</strong>g<br />
far fewer SNPs—yet even here the current<br />
cost is a prohibitive $400 million or so for each disease<br />
<strong>in</strong>vestigated. With<strong>in</strong> five years, however, genotyp<strong>in</strong>g<br />
costs are expected to fall to $20 million or<br />
less, and the essential proof-of-concept tests can<br />
then take place more rout<strong>in</strong>ely. (See sidebar,<br />
F<strong>in</strong>d<strong>in</strong>g a small-molecule therapeutic to repair a completely<br />
defective prote<strong>in</strong> is an extremely difficult challenge.<br />
(Indeed, Ceredase is a prote<strong>in</strong> therapeutic.)<br />
Most disorders, by contrast, are attributable not to a<br />
s<strong>in</strong>gle gene but to multiple genes, and perhaps other<br />
factors too. This means that the system as a whole<br />
can still function, just hampered to a greater or<br />
lesser degree. Such cases benefit from patch<strong>in</strong>g up,<br />
so drugs can be beneficial without actually constitut<strong>in</strong>g<br />
a cure. There is little reason to doubt that<br />
such palliative drugs will soon emerge <strong>in</strong> abundance,<br />
as disease genetics becomes ever faster at<br />
identify<strong>in</strong>g some of the genes implicated <strong>in</strong> multigenic<br />
disorders.
GENOTYPING—HOPES RISE AS PRICES FALL<br />
For companies wish<strong>in</strong>g to pursue disease genetics,<br />
one of the major stumbl<strong>in</strong>g blocks has been the prohibitive<br />
cost of genotyp<strong>in</strong>g. The current average<br />
genotyp<strong>in</strong>g cost <strong>in</strong> the <strong>in</strong>dustry is about 50 cents per<br />
SNP. At that price, even narrow candidate-gene studies<br />
could cost as much as $15 million; a direct<br />
genome-wide association study could cost upwards<br />
of $5 billion.<br />
Matters are about to change, however. Costs are<br />
decl<strong>in</strong><strong>in</strong>g, and are expected to cont<strong>in</strong>ue to fall dramatically—as<br />
much as 50-fold—over the next few<br />
years, thanks to competition and customer demand<br />
on the one hand, and improved automation and<br />
“Genotyp<strong>in</strong>g—Hopes Rise as Prices Fall.”) And<br />
once that happens, the two major questions outstand<strong>in</strong>g<br />
can be resolved: the number of patients<br />
needed to atta<strong>in</strong> statistically significant results, and<br />
the utility of SNP maps.<br />
Regard<strong>in</strong>g patient populations, first: the more<br />
patients, of course, the easier it is to discern a true<br />
difference above randomly occurr<strong>in</strong>g fluctuations<br />
(noise)—and the more expensive. Size your sample<br />
too spar<strong>in</strong>gly, and you risk emerg<strong>in</strong>g empty-handed;<br />
too generously, and you overspend. Strik<strong>in</strong>g the<br />
right balance is tricky. It depends on how common<br />
the sought-for SNPs are, and that is very difficult to<br />
estimate. It also depends on the strength of the<br />
association between the disease and the suspect<br />
polymorphism. For common multigenic diseases,<br />
susceptibility depends on a specific comb<strong>in</strong>ation of<br />
several genetic changes, and is <strong>in</strong>fluenced by environmental<br />
factors as well, which weakens the association.<br />
The weaker the association, the larger the<br />
sample needed to detect the <strong>in</strong>fluence of a specific<br />
gene. Perhaps the issue will dissolve before be<strong>in</strong>g<br />
resolved. If prices plummet as expected, and if<br />
databases become as comprehensive as hoped, sufficient<br />
sample sizes will become easily affordable.<br />
SNP maps can help <strong>in</strong> the quest to identify a<br />
disease-susceptibility gene, but only if two condi-<br />
m<strong>in</strong>iaturization (which allows companies to reduce<br />
their consumption of expensive reagents) on the<br />
other.<br />
Our calculations are, accord<strong>in</strong>gly, based on a cost of<br />
one cent per SNP genotype—a likely price across the<br />
<strong>in</strong>dustry, accord<strong>in</strong>g to expert consensus, with<strong>in</strong> the<br />
next five years. (Some companies must already be<br />
benefit<strong>in</strong>g from genotyp<strong>in</strong>g costs considerably lower<br />
than the current <strong>in</strong>dustry average.) That said, the<br />
cost of conduct<strong>in</strong>g disease genetics studies will<br />
rema<strong>in</strong> far from negligible, and companies will need<br />
to cont<strong>in</strong>ue to take it <strong>in</strong>to account when assess<strong>in</strong>g<br />
risk.<br />
tions prevail. First, for any association studies<br />
(whether narrow or broad) to work, the genetic variants<br />
associated with the disease need to be fairly<br />
common—prevalent <strong>in</strong> more than 1 percent of the<br />
population at large—and that is far from guaranteed.<br />
Then, these polymorphisms need to be either<br />
recognizable (that is, they must produce discernible<br />
changes <strong>in</strong> a prote<strong>in</strong>) or at least detectable by an<br />
<strong>in</strong>direct measure (called l<strong>in</strong>kage disequilibrium;<br />
that is, the presence of a particular SNP cluster <strong>in</strong><br />
<strong>in</strong>dividuals with a given disease). Assum<strong>in</strong>g reasonable<br />
costs, <strong>in</strong>direct genome-wide association studies—of<br />
all the approaches, the one most likely to<br />
prevail—would, accord<strong>in</strong>g to our model, result <strong>in</strong> a<br />
total sav<strong>in</strong>gs of $395 million <strong>in</strong> cost per drug on average,<br />
with 0.7 years of time saved to market (produc<strong>in</strong>g<br />
an additional $260 million of value per drug).<br />
Of course, for that to happen, you need to do more<br />
than f<strong>in</strong>d disease genes—you have to turn them to<br />
good effect by ultimately produc<strong>in</strong>g a drug. And<br />
here disease genetics presents a further challenge:<br />
given the long odds <strong>in</strong>volved <strong>in</strong> pharmaceutical<br />
R&D, will the number of targets yielded by disease<br />
genetics be sufficient?<br />
Practicability—the Limitations of Human Studies<br />
The problem is that human studies—identify<strong>in</strong>g<br />
polymorphic genes <strong>in</strong> humans—will tend to pro-<br />
31
32<br />
duce targets high <strong>in</strong> quality but low <strong>in</strong> quantity, perhaps<br />
just one to five for an average disease. That<br />
would produce at best a 25–30 percent chance of<br />
yield<strong>in</strong>g a drug, given that chemistry and development<br />
rema<strong>in</strong> far from fail-safe. For disease genetics<br />
to live up to its promise, it will need to improve<br />
those odds considerably. And to do that, it will have<br />
to call on a supplementary technique: pathway<br />
analysis.<br />
Disease-susceptibility genes, if identified through<br />
disease genetics, serve not only as targets themselves,<br />
but also as guides to additional targets. The<br />
genes form part of broader disease pathways, and<br />
these pathways conta<strong>in</strong> other molecules that may<br />
serve as targets (perhaps 10 to 15 targets per pathway,<br />
accord<strong>in</strong>g to experts). These new targets are<br />
identified by pathway analysis, often tak<strong>in</strong>g the form<br />
of the study of “simple” experimental systems, such<br />
as those of Drosophila melanogaster or C. elegans (a.k.a.<br />
fruitflies and worms). When studied <strong>in</strong> the laboratory,<br />
these systems disclose the genetic components<br />
of a given pathway. (Other approaches to pathway<br />
analysis <strong>in</strong>clude expression profil<strong>in</strong>g of tissue samples<br />
and studies of prote<strong>in</strong>-prote<strong>in</strong> <strong>in</strong>teraction.)<br />
Implement<strong>in</strong>g pathway analysis will result <strong>in</strong> a lower<br />
cost per drug on average: it improves efficiency.<br />
Costs are reduced because pathway studies are relatively<br />
<strong>in</strong>expensive and fewer human-derived targets<br />
are required—pathway analysis expands the pool of<br />
potential targets tenfold or more. As targets, they<br />
prove to be of high quality, moreover (<strong>in</strong> keep<strong>in</strong>g<br />
with the disease-susceptibility gene that <strong>in</strong>spired<br />
their discovery), achiev<strong>in</strong>g good success rates <strong>in</strong><br />
cl<strong>in</strong>ical trials. For although they themselves are not<br />
yet validated <strong>in</strong> humans or clearly implicated <strong>in</strong> the<br />
disease, they participate <strong>in</strong> a pathway that is.<br />
So pathway analysis should give human studies the<br />
requisite boost, with enough targets emerg<strong>in</strong>g to<br />
yield a drug more often than not. On the down<br />
side, there are the time and cost of additional animal<br />
research and the loss of some advantages <strong>in</strong><br />
cl<strong>in</strong>ical trials. Add<strong>in</strong>g pathway analysis, via genetic<br />
studies of Drosophila melanogaster, to <strong>in</strong>direct<br />
genome-wide association studies would result <strong>in</strong> a<br />
total sav<strong>in</strong>gs of $425 million for each drug on average,<br />
regardless of the orig<strong>in</strong>al human study<br />
approach, though it would add nearly two years of<br />
additional work (produc<strong>in</strong>g an additional $255 million<br />
of value per drug).<br />
The first one to do genetic studies takes a huge risk.<br />
If it works, you’ll see everyone runn<strong>in</strong>g to jo<strong>in</strong> the<br />
crowd.<br />
—R&D executive,<br />
lead<strong>in</strong>g pharmaceutical company<br />
Implement<strong>in</strong>g Disease Genetics<br />
The sav<strong>in</strong>gs promised by disease genetics are enormous,<br />
and companies cannot ignore them. But they<br />
cannot ignore the risks either, and will need to exercise<br />
rigorous selectivity and discipl<strong>in</strong>e when it comes<br />
to pursu<strong>in</strong>g specific disease genetics studies—which<br />
approaches to adopt, for <strong>in</strong>stance, and which diseases<br />
to explore. Plac<strong>in</strong>g bets <strong>in</strong> this way is go<strong>in</strong>g to<br />
be nerve-rack<strong>in</strong>g enough. But companies face<br />
another difficult set of choices as well, <strong>in</strong> the various<br />
operational issues that need to be addressed.<br />
Plac<strong>in</strong>g Bets<br />
Some diseases have clear appeal as objects of<br />
genetic research: asthma, Alzheimer’s disease, and<br />
diabetes, for example, be<strong>in</strong>g complex diseases that<br />
afflict large populations and obviously conta<strong>in</strong> heritable<br />
factors. They have already become competitive<br />
areas of study. Although op<strong>in</strong>ion is still divided<br />
over the likely impact of disease genetics, substantial<br />
bets are be<strong>in</strong>g placed by various companies—<br />
emerg<strong>in</strong>g biotechs and established pharmaceutical<br />
companies alike. A few claim they are already see<strong>in</strong>g<br />
benefits from their <strong>in</strong>vestments: follow<strong>in</strong>g its<br />
alliance with Roche, deCODE genetics claims to<br />
have succeeded <strong>in</strong> identify<strong>in</strong>g a gene contribut<strong>in</strong>g<br />
to cerebrovascular disease; GlaxoSmithKl<strong>in</strong>e has<br />
announced f<strong>in</strong>d<strong>in</strong>g genes associated with migra<strong>in</strong>e,<br />
Type II diabetes, and psoriasis; and Genset, a<br />
French biotech company, is reported to have identified<br />
genes implicated <strong>in</strong> prostate cancer and<br />
schizophrenia.<br />
But all companies embarked on, or <strong>in</strong>tent on, pursu<strong>in</strong>g<br />
disease genetics must acknowledge that it is
still a scientifically risky endeavor. As they construct<br />
their customized portfolio of bets, they will have to<br />
keep review<strong>in</strong>g not just their <strong>in</strong>ternal capabilities but<br />
also the degree of risk they are prepared to accept.<br />
Putt<strong>in</strong>g Disease Genetics <strong>in</strong>to Operation<br />
If companies do opt to participate <strong>in</strong> disease genetics,<br />
they may still choose to ma<strong>in</strong>ta<strong>in</strong> some distance<br />
by outsourc<strong>in</strong>g the activity or licens<strong>in</strong>g <strong>in</strong> the<br />
results. Those that decide to launch a major disease<br />
genetics program <strong>in</strong>-house will confront a number<br />
of significant challenges as they put the program<br />
<strong>in</strong>to operation.<br />
Take the problem of obta<strong>in</strong><strong>in</strong>g the required samples,<br />
for <strong>in</strong>stance. From whom should samples be<br />
collected? And how are samples to be stored? The<br />
use of human tissue raises ethical considerations as<br />
well. What constitutes consent? How can privacy be<br />
protected? Who “owns” the tissue material? And<br />
who should profit from it? (Several companies,<br />
such as Genomics Collaborative and the not-forprofit<br />
First Genetic Trust, are emerg<strong>in</strong>g to address<br />
these conundrums.)<br />
And, of course, there are major organizational<br />
questions. What are the implications for human<br />
resources and labor relations? For big pharmaceutical<br />
companies, new capabilities will be demanded,<br />
and new skills will have to be acquired—statistical<br />
geneticists, for <strong>in</strong>stance, and experts on pathway<br />
genetic studies. And other capabilities may suddenly<br />
be less <strong>in</strong> demand—perhaps even obsolete.<br />
We will discuss implementation issues such as these<br />
<strong>in</strong> more detail <strong>in</strong> the f<strong>in</strong>al chapter of this report.<br />
Pharmacogenetics<br />
Just as some genetic variations among <strong>in</strong>dividuals<br />
may <strong>in</strong>fluence their susceptibility to diseases, so<br />
others may <strong>in</strong>fluence their responsiveness to drug<br />
treatments for those diseases. It is the goal of pharmacogenetics<br />
to seek out and characterize some of<br />
these latter variations.<br />
The sav<strong>in</strong>gs that pharmaceutical companies might<br />
hope to harvest are considerable, though noth<strong>in</strong>g<br />
like as high as those that disease genetics stands to<br />
achieve <strong>in</strong> ideal circumstances.<br />
The Potential<br />
The impact of pharmacogenetics on R&D productivity<br />
will derive from the <strong>in</strong>creased flexibility it<br />
<strong>in</strong>troduces <strong>in</strong>to cl<strong>in</strong>ical development. Currently,<br />
drug-development policy is dom<strong>in</strong>ated by a b<strong>in</strong>ary<br />
scenario <strong>in</strong> its later stages: either shepherd a compound<br />
through its cl<strong>in</strong>ical trials and out to market,<br />
or abandon it as unmarketable if it stumbles <strong>in</strong> the<br />
trials. Pharmacogenetics provides a more nuanced<br />
scenario, with an expanded range of possible outcomes,<br />
by allow<strong>in</strong>g the exclusion of patients genetically<br />
predisposed to respond poorly to the drug.<br />
Two particular benefits emerge: “standard” cl<strong>in</strong>ical<br />
trials can now be streaml<strong>in</strong>ed; and “fail<strong>in</strong>g” compounds<br />
can now be salvaged. (See Exhibit 7.)<br />
The streaml<strong>in</strong><strong>in</strong>g of trials would apply only to compounds<br />
dest<strong>in</strong>ed to proceed successfully through<br />
the cl<strong>in</strong>ical trial process anyway. That path can now<br />
be made smoother. The trials can be designed more<br />
subtly. They can be smaller and quicker than<br />
before, now that it is possible to preselect promis<strong>in</strong>g<br />
patients—that is, patients whose genetic<br />
makeup is likely to maximize the drug’s efficacy<br />
and m<strong>in</strong>imize its side effects. So, patients lack<strong>in</strong>g<br />
the drug-susceptible variation of the target gene<br />
EXHIBIT 7<br />
PHARMACOGENETICS EXPANDS DEVELOPMENT CHOICES<br />
Proportion of patients show<strong>in</strong>g<br />
poor or no response<br />
High<br />
Low<br />
SOURCE: BCG analysis.<br />
Current options<br />
Abandon drug<br />
before market<br />
Cont<strong>in</strong>ue cl<strong>in</strong>ical<br />
trials to market<br />
Options available with<br />
pharmacogenetics<br />
Cont<strong>in</strong>ue trials safely by<br />
exclud<strong>in</strong>g at-risk patients<br />
Optimize cl<strong>in</strong>ical trials,<br />
mak<strong>in</strong>g them<br />
smaller and shorter<br />
33
34<br />
would be excluded from the trial, <strong>in</strong> order to show<br />
high efficacy levels for the subset of patients who<br />
would eventually use the drug. Also excluded would<br />
be patients hav<strong>in</strong>g a specific genetic variation associated<br />
with side effects.<br />
To see the streaml<strong>in</strong><strong>in</strong>g effect of such exclusions,<br />
consider the case of Hercept<strong>in</strong>, a treatment for<br />
advanced breast cancer. It is effective only <strong>in</strong> a subset<br />
of patients—the 25–30 percent whose tumors<br />
overexpress the HER2/neu oncogene. It is this<br />
gene that serves as the drug target. By screen<strong>in</strong>g for<br />
HER2/neu expression, Genentech was able to<br />
exclude nonresponders—some two-thirds of the<br />
subjects orig<strong>in</strong>ally tested—early <strong>in</strong> the cl<strong>in</strong>ical trial.<br />
Without this prescreen<strong>in</strong>g, Genentech would have<br />
needed n<strong>in</strong>e times as many patients <strong>in</strong> phase III to<br />
achieve significant results. The cost of such a trial<br />
would have made Hercept<strong>in</strong> economically unviable.<br />
Turn<strong>in</strong>g to the second benefit, for fail<strong>in</strong>g compounds<br />
pharmacogenetics lowers the hurdle by eas<strong>in</strong>g<br />
the conditions for market viability. Consider<br />
specifically those candidate drugs that reveal serious<br />
side effects <strong>in</strong> a significant proportion of the subjects<br />
(such as the 7 percent of Caucasians with low<br />
levels of CYP2D6, an enzyme that helps metabolize<br />
some 25 percent of all drugs). Traditionally, any<br />
such drug would be perceived as too risky to market,<br />
and would be abandoned <strong>in</strong> precl<strong>in</strong>ical studies.<br />
Today, however, pharmacogenetics makes it possible<br />
to identify the at-risk patients, so the drug would not<br />
be disqualified right away, and could go on to prove<br />
marketable—patients would just need to be tested<br />
for vulnerability before be<strong>in</strong>g given prescriptions.<br />
These are the potential benefits to R&D, the primary<br />
focus of this report. Even greater potential,<br />
some observers believe, may lie <strong>in</strong> market advantages.<br />
Three such advantages are possible: price<br />
premium, share shift, and new patients. In other<br />
words, if a perception emerges that the pharmacogenetics-assisted<br />
drug is dist<strong>in</strong>ctly less risky or dist<strong>in</strong>ctly<br />
more efficacious for the (now restricted) target<br />
patient population, payers may tolerate a higher<br />
price for the drug, physicians may favor it when<br />
offer<strong>in</strong>g new patients a prescription, and patients<br />
who have shunned previous medications (ow<strong>in</strong>g to<br />
side effects, typically) may now choose to try it.<br />
Although it seems reasonable for pharmaceutical<br />
companies to expect some market upside from<br />
more efficacious, better tolerated therapies, it<br />
rema<strong>in</strong>s to be seen to what extent they will be able<br />
to reap these market rewards.<br />
Putt<strong>in</strong>g figures on the cost sav<strong>in</strong>gs pharmacogenetics<br />
benefits might achieve, our model estimates an<br />
average of $335 million <strong>in</strong> the cost to drug—if<br />
pharmacogenetics were to work every time it were<br />
applied. But pharmacogenetics won’t work every<br />
time. Given the set of cases where it is applied and<br />
succeeds, the expected sav<strong>in</strong>gs would average about<br />
$80 million, as discussed below. And of course, correspond<strong>in</strong>g<br />
to the potential market upside, there is<br />
the counterpart scenario—potential destruction of<br />
value <strong>in</strong> the market. Why the uncerta<strong>in</strong>ty? (See<br />
Exhibit 8.)<br />
EXHIBIT 8<br />
PHARMACOGENETICS’ POTENTIAL IS CONTINGENT<br />
Cost to drug<br />
Pre-genomics<br />
Pharmacogenetics:<br />
the promise 1<br />
Pharmacogenetics:<br />
expected sav<strong>in</strong>gs 2<br />
ID Biology<br />
Target ID Target Validation<br />
0<br />
200<br />
Chemistry<br />
400<br />
Screen<strong>in</strong>g Optimization<br />
600<br />
545<br />
800<br />
800<br />
Development<br />
Precl<strong>in</strong>ical Cl<strong>in</strong>ical<br />
SOURCES: Technical literature; <strong>in</strong>dustry <strong>in</strong>terviews; publicly available <strong>in</strong>formation;<br />
BCG analysis.<br />
1Sav<strong>in</strong>gs per drug assum<strong>in</strong>g pharmacogenetics can be applied across the<br />
R&D pipel<strong>in</strong>e.<br />
2 Average sav<strong>in</strong>gs across R&D pipel<strong>in</strong>e, given scientific and market limitations.<br />
880<br />
1,000<br />
$M
The Uncerta<strong>in</strong>ty<br />
The sizable potential sav<strong>in</strong>gs are mirrored by sizable<br />
risks—market and regulatory risks this time, as<br />
well as scientific and technical risks. In some circumstances,<br />
market dynamics might be so unfavorable<br />
that companies would be well advised to step<br />
back and forgo the potential sav<strong>in</strong>gs altogether.<br />
Once aga<strong>in</strong>, given the range of possible outcomes,<br />
the economic impact could ultimately be a negative<br />
one—far from result<strong>in</strong>g <strong>in</strong> sav<strong>in</strong>gs, apply<strong>in</strong>g pharmacogenetics<br />
could result <strong>in</strong> a net loss for R&D. To<br />
assess pharmacogenetics realistically, companies<br />
need to ask two questions: How feasible is it? And<br />
how desirable is it?<br />
Feasibility—the Technological Limitations<br />
Pharmacogenetics will not apply to all drugs. It will<br />
apply only where differ<strong>in</strong>g drug response is due<br />
entirely to genetic variation, and where that relationship<br />
can be elucidated.<br />
It is fairly rare for both of these conditions to prevail.<br />
For one th<strong>in</strong>g, biology-based variation <strong>in</strong> drug<br />
response can be due <strong>in</strong> part to environmental factors:<br />
grapefruit juice, for example, is known to<br />
modify the effect of certa<strong>in</strong> drugs <strong>in</strong> certa<strong>in</strong> <strong>in</strong>dividuals,<br />
sometimes rais<strong>in</strong>g the uptake to dangerous<br />
levels. For another, drug-response variation will<br />
often be the work of multiple genes, act<strong>in</strong>g severally<br />
or jo<strong>in</strong>tly, and that compounds the statistical and<br />
technological difficulties of the search.<br />
The bus<strong>in</strong>ess guys hear about this stuff, and are<br />
like, “Great! Make it happen!” We’re left scratch<strong>in</strong>g<br />
our heads, look<strong>in</strong>g like poor sports, because a lot of<br />
it just isn’t possible.<br />
—Senior scientist,<br />
major biotech company<br />
Even if the two conditions are fulfilled, a further<br />
challenge lies <strong>in</strong> wait—to f<strong>in</strong>d the relevant genes <strong>in</strong><br />
time to streaml<strong>in</strong>e the trial. For pharmacogenetics<br />
to effect this streaml<strong>in</strong><strong>in</strong>g, you need to be able to<br />
screen out nonresponders. And that means f<strong>in</strong>d<strong>in</strong>g<br />
those variants, or identify<strong>in</strong>g the nonresponder<br />
genotype before design<strong>in</strong>g any streaml<strong>in</strong>ed trial.<br />
Now, develop<strong>in</strong>g a robust pharmacogenetic test<br />
would generally require more than 1,000 patients.<br />
A phase I trial is far too small for that purpose, so<br />
no streaml<strong>in</strong><strong>in</strong>g would be possible for a phase II<br />
trial, despite the excitement surround<strong>in</strong>g that<br />
prospect. Streaml<strong>in</strong><strong>in</strong>g could be possible <strong>in</strong> time for<br />
phase III trials, but even that is far from assured.<br />
(See Exhibit 9.) (The other k<strong>in</strong>d of screen<strong>in</strong>g test,<br />
for side effects, is less problematic, s<strong>in</strong>ce the associated<br />
metabolic variations are often determ<strong>in</strong>ed <strong>in</strong><br />
precl<strong>in</strong>ical trials. But that k<strong>in</strong>d of test does not help<br />
to streaml<strong>in</strong>e cl<strong>in</strong>ical trials, which have to be sized<br />
to test for efficacy rather than for side effects.)<br />
It is hard to see how these phase II trials will be<br />
used for pharmacogenetics, because most of the<br />
variants are expected to be relatively <strong>in</strong>frequent.<br />
Certa<strong>in</strong>ly, 30 percent prevalence [which falls<br />
with<strong>in</strong> standard cl<strong>in</strong>ical trial sizes] would be relatively<br />
rare.<br />
—Geneticist,<br />
The Whitehead Institute<br />
EXHIBIT 9<br />
STATISTICAL REQUIREMENTS LIMIT PHARMACOGENETICS’<br />
POTENTIAL TO STREAMLINE TRIALS<br />
Number of patients required to develop a test<br />
5,000<br />
4,000<br />
3,000<br />
2,000<br />
1,000<br />
0<br />
10<br />
Typical phase II trial size<br />
Typical phase I trial size<br />
20<br />
30<br />
Prevalence >30% needed to<br />
develop a test prior to phase III trials<br />
40<br />
50<br />
60<br />
Frequency of responder genotype<br />
SOURCES: Technical literature; <strong>in</strong>dustry <strong>in</strong>terviews; publicly available <strong>in</strong>formation;<br />
BCG analysis.<br />
NOTE: Graph analyzes a scenario from Nature Biotechnology, vol. 18, May<br />
2000. Scenario uses efficacy pharmacogenetics to identify the ApoE<br />
responder genotype, a predictor of efficacy of Cognex (tacr<strong>in</strong>e) for<br />
Alzheimer’s. Strength of effect, a different variable, is not considered here.<br />
35
36<br />
Given these technological limitations, we estimate<br />
that less than 15 percent of drugs will be amenable<br />
to the application of pharmacogenetics.<br />
Desirability—Market Economics<br />
As already mentioned, there are circumstances <strong>in</strong><br />
which a company might have an <strong>in</strong>centive to shun<br />
pharmacogenetics entirely. After all, by exclud<strong>in</strong>g<br />
patients from trials, you are <strong>in</strong> effect giv<strong>in</strong>g the<br />
drug a restricted label when try<strong>in</strong>g to market it.<br />
Gaug<strong>in</strong>g the likely effect of a restricted label<br />
<strong>in</strong>volves some complex analysis. For a start, you<br />
need to consider two dist<strong>in</strong>ct groups of patients:<br />
those who take a prescription for the full course of<br />
treatment (which could last many years, or even the<br />
rema<strong>in</strong><strong>in</strong>g lifetime for those suffer<strong>in</strong>g from chronic<br />
diseases), and those who embark on a prescription<br />
but then discont<strong>in</strong>ue it for reasons of <strong>in</strong>efficacy or<br />
side effects.<br />
The pharmacogenetics test would shr<strong>in</strong>k these two<br />
potential patient groups <strong>in</strong> different ways. From the<br />
former, it would elim<strong>in</strong>ate the “placebo responders.”<br />
From the latter, it would elim<strong>in</strong>ate some of<br />
the nonresponders and negative responders.<br />
Market fragmentation has happened <strong>in</strong> many<br />
<strong>in</strong>dustries—the market<strong>in</strong>g group can’t put their<br />
heads <strong>in</strong> the sand. We have to figure out what to do<br />
about pharmacogenetics.<br />
—Genetics director,<br />
lead<strong>in</strong>g biotech company<br />
S<strong>in</strong>ce pharmacogenetics seems to be chipp<strong>in</strong>g away<br />
at a drug’s market base, why pursue it <strong>in</strong> the first<br />
place? The answer may lie, <strong>in</strong> part, <strong>in</strong> competitive<br />
dynamics and game theory: companies may have to<br />
embrace pharmacogenetics because their competitors<br />
are do<strong>in</strong>g so. Merck & Co., for example,<br />
accord<strong>in</strong>g to a recent Wall Street Journal article, is<br />
busy develop<strong>in</strong>g capabilities to reproduce pharmacogenetic<br />
analyses conducted by its competitors, if<br />
only to disprove any claims that a rival drug might<br />
be superior to its own.<br />
But the compensatory advantages can be more positive,<br />
too—the potential for market upside, once<br />
aga<strong>in</strong>: price premium, share shift, and new patients.<br />
What a company has to judge, before adopt<strong>in</strong>g<br />
pharmacogenetics for any drug <strong>in</strong> development, is<br />
the likely breakeven po<strong>in</strong>t—the po<strong>in</strong>t at which a<br />
price premium or <strong>in</strong>creased market share beg<strong>in</strong>s to<br />
offset the volume loss. Our model assumes a modest<br />
market premium of 20 percent, and calculates the<br />
breakeven po<strong>in</strong>t <strong>in</strong> various scenarios, based on four<br />
different approaches to pharmacogenetics. (See<br />
sidebar, “Pharmacogenetics—Four Applications,”<br />
and Exhibit 10.)<br />
Efficacy-based pharmacogenetics can reduce trial costs<br />
considerably. But the market dynamics could then<br />
cast a cloud over that economic picture. If the<br />
restricted label, by disqualify<strong>in</strong>g placebo responders<br />
and some nonresponders and negative responders,<br />
translates <strong>in</strong>to an overall revenue loss of just 2<br />
percent, that cancels out the sav<strong>in</strong>gs achieved <strong>in</strong> the<br />
cl<strong>in</strong>ical trials.<br />
EXHIBIT 10<br />
PHARMACOGENETICS’ VALUE DEPENDS ON MARKET<br />
DYNAMICS<br />
Patients lack<strong>in</strong>g good response (%) 1<br />
100<br />
80<br />
60<br />
40<br />
20<br />
0<br />
Conduct normal trials<br />
Abandon drug<br />
100<br />
SOURCES: Industry <strong>in</strong>terviews; BCG analysis.<br />
Pharmacogenetics can optimize cl<strong>in</strong>ical trials<br />
200<br />
Pharmacogenetics<br />
trials make drug viable<br />
300<br />
Revenue <strong>in</strong>crease required from market premium (%)<br />
1Example based on a scenario <strong>in</strong> Nature Biotechnology, vol. 18, May 2000<br />
of ApoE4 efficacy <strong>in</strong> tacr<strong>in</strong>e response; assumes response rate of 41 percent<br />
among patients with the SNP versus 20 percent among those without it;<br />
also assumes 50 percent of nonresponders discont<strong>in</strong>ue.
PHARMACOGENETICS—FOUR APPLICATIONS<br />
To evaluate pharmacogenetics properly, companies<br />
need to take an especially close look at the market<br />
dynamics. These dynamics vary accord<strong>in</strong>g to how<br />
pharmacogenetics is used. We have identified four<br />
such applications (each associated with a different<br />
category of patient respond<strong>in</strong>g to a given drug). The<br />
first three are used to exclude patients from trials;<br />
the fourth is used to expand the potential market for<br />
the drug.<br />
First, efficacy prediction identifies patients who will<br />
show no real or significant response to the drug—<br />
perhaps because they metabolize the drug <strong>in</strong> an<br />
unusual way, or have an unusual form or comb<strong>in</strong>ation<br />
of susceptibility genes. A typical drug produces<br />
this negligible response <strong>in</strong> about a third of patients,<br />
but sometimes the proportion is far higher. For<br />
example, Cognex (tacr<strong>in</strong>e), the first drug for<br />
Alzheimer’s, is <strong>in</strong>efficacious <strong>in</strong> more than 50 percent<br />
of patients. The vary<strong>in</strong>g response is associated with<br />
differ<strong>in</strong>g versions of the ApoE gene, and is therefore<br />
readily predictable by a pharmacogenetic test.<br />
Second, common-side-effect prediction identifies<br />
patients likely to experience familiar side effects, as<br />
a result of metabolic difficulties caused by wellknown<br />
enzymes. A test can screen out negative<br />
responders—“slow acetylators,” for <strong>in</strong>stance. The<br />
acetylation polymorphism <strong>in</strong> the NAT2 gene is one of<br />
the commonest genetic variations <strong>in</strong> drug metabolism;<br />
it has the effect of reduc<strong>in</strong>g the enzyme’s lifespan<br />
and thus reduc<strong>in</strong>g the effective amount of the<br />
enzyme <strong>in</strong> cells at any one time. This polymorphism<br />
is present <strong>in</strong> more than 50 percent of Caucasians,<br />
who are thus at greater risk of drug toxicity.<br />
Knowledge of this polymorphism could save a drug<br />
<strong>in</strong> cl<strong>in</strong>ical trials that would otherwise be abandoned.<br />
Third, very-rare-side-effect prediction identifies patients<br />
at risk for unconventional side effects, but<br />
comes <strong>in</strong>to play only after the drug is on the market.<br />
Unlike most of the common side effects, which are<br />
associated with metabolic pathways and usually<br />
emerge <strong>in</strong> precl<strong>in</strong>ical studies, these rare side effects<br />
tend to be provoked by nonmetabolic genes, and to<br />
be overlooked at first. They cannot easily be predicted,<br />
s<strong>in</strong>ce there are too many possible sources<br />
(modifications of the target or of the disease pathway,<br />
or unrelated pathways), and they may occur too<br />
rarely to show up <strong>in</strong> cl<strong>in</strong>ical trials.<br />
A case <strong>in</strong> po<strong>in</strong>t is Lotronex, a drug for irritable bowel<br />
syndrome, now withdrawn from the market. Only<br />
after its market launch, and 450,000 prescriptions,<br />
did its severe side effect (bowel impaction) become<br />
apparent. About one <strong>in</strong> 6,500 patients was<br />
affected—a frequency far too rare for a standard<br />
cl<strong>in</strong>ical trial to detect beforehand. (A typical trial<br />
<strong>in</strong>volves about 5,000 patients: for this side effect to<br />
have been manifest <strong>in</strong> a statistically significant way,<br />
a trial of nearly 100,000 patients would have been<br />
needed.) Pharmacogenetics could <strong>in</strong> certa<strong>in</strong> cases<br />
come to the rescue of such compromised drugs, by<br />
belatedly devis<strong>in</strong>g a screen<strong>in</strong>g test.<br />
F<strong>in</strong>ally, market expansion identifies patients who are<br />
currently unsuited to the drug but potentially responsive<br />
to it. S<strong>in</strong>ce f<strong>in</strong>e-tun<strong>in</strong>g of dosages or formulation<br />
can often reduce side effects and occasionally<br />
improve efficacy, pharmacogenetics could reassess<br />
and upgrade many of the supposedly <strong>in</strong>eligible<br />
patients. The market for the drug might expand considerably<br />
as a result.<br />
Take the case of cyclophosphamide, a chemotherapy<br />
drug, which works only when metabolized by the<br />
enzymes CYP3A4 and CYP3A5. Some patients<br />
appear underresponsive to it: a genetic variation<br />
suppresses the activity of the enzymes, thereby<br />
decreas<strong>in</strong>g the amount of active drug <strong>in</strong> the bloodstream.<br />
The best course is not to discont<strong>in</strong>ue the<br />
drug, but to compensate by tak<strong>in</strong>g a higher dosage.<br />
A pharmacogenetic test could identify the appropriate<br />
patients prior to treatment, and their consumption<br />
of the drug, <strong>in</strong>stead of decl<strong>in</strong><strong>in</strong>g to zero, would<br />
actually <strong>in</strong>crease.<br />
37
38<br />
Side-effect-based pharmacogenetics for common side effects<br />
can save a candidate drug that would otherwise fail.<br />
This form of pharmacogenetics does not streaml<strong>in</strong>e<br />
trials; <strong>in</strong> fact, it imposes a moderate <strong>in</strong>crease <strong>in</strong> costs,<br />
s<strong>in</strong>ce more patients have to be recruited <strong>in</strong>itially for<br />
the vett<strong>in</strong>g process. It requires a smaller upside to<br />
break even than efficacy-based pharmacogenetics,<br />
however, s<strong>in</strong>ce its powers of exclusion apply only to<br />
the second type of patient (those who would ord<strong>in</strong>arily<br />
try the drug and then discont<strong>in</strong>ue it). They do<br />
not apply to the first group (those who would take a<br />
full course of the drug), s<strong>in</strong>ce placebo responders<br />
do not suffer from side effects.<br />
Side-effect-based pharmacogenetics for very rare side effects<br />
is the type that would be applied for a drug already<br />
on the market. Once the number of adverse events<br />
(<strong>in</strong>stances of severe side effects) reaches a critical<br />
mass, the drug’s reputation suffers, and its cont<strong>in</strong>ued<br />
marketability is jeopardized. (In severe cases,<br />
regulatory agencies require the drug to be removed<br />
from the market.) To salvage it would <strong>in</strong>volve implement<strong>in</strong>g<br />
screen<strong>in</strong>g tests for all potential patients—<br />
a k<strong>in</strong>d of postmarket surveillance. This type of<br />
pharmacogenetics would <strong>in</strong>crease costs fairly<br />
steeply, yet it could still make economic sense if the<br />
drug were saved.<br />
The economics h<strong>in</strong>ge on a paradox: the fewer the<br />
adverse events, the harder it might be to save the<br />
drug. To identify the culprit genetic marker for use<br />
<strong>in</strong> the screen<strong>in</strong>g test, you need a certa<strong>in</strong> m<strong>in</strong>imum<br />
number of patients who have experienced the side<br />
effect. That could be as low as 20 (assum<strong>in</strong>g you<br />
achieved a 100 percent association with a s<strong>in</strong>gle<br />
SNP marker), and that would carry the modest<br />
price tag of $100,000 (assum<strong>in</strong>g the expected genotyp<strong>in</strong>g<br />
cost of one cent per SNP). But the required<br />
number could be 2,000 (assum<strong>in</strong>g you achieved<br />
only a 10 percent association), and the likelihood is<br />
that such a number of side-effect sufferers would<br />
simply never emerge.<br />
It’s a crime that a very small percentage of patients<br />
can sometimes elim<strong>in</strong>ate an otherwise highly beneficial<br />
drug from the market. Pharmacogenetics benefits<br />
everyone here.<br />
—Research executive,<br />
lead<strong>in</strong>g pharmaceutical company<br />
F<strong>in</strong>ally, market-expansion pharmacogenetics for the<br />
most part has highly favorable economics. S<strong>in</strong>ce its<br />
effect is to expand rather than contract the market,<br />
all it needs to ensure is that the expansion be large<br />
enough to cover the cost of the required trial.<br />
The prospects h<strong>in</strong>ge to some extent on the <strong>in</strong>cidence<br />
of the genetic variant. Consider two drugs,<br />
one produc<strong>in</strong>g side effects <strong>in</strong> poor CYP2D19<br />
metabolizers (<strong>in</strong>clud<strong>in</strong>g 20 percent of Asians) and<br />
the other <strong>in</strong> poor CYP2D6 metabolizers (<strong>in</strong>clud<strong>in</strong>g<br />
7 percent of Caucasians); and suppose that dosage<br />
adjustments could resolve the side effects. It might<br />
turn out that the former case warrants the <strong>in</strong>vestment<br />
and the latter does not, given the difference<br />
<strong>in</strong> their potential market expansions.<br />
All <strong>in</strong> all, these limitations reduce the expected sav<strong>in</strong>gs<br />
that pharmacogenetics would bestow on an<br />
average drug to about $75 million. But of course<br />
there is no such th<strong>in</strong>g as a true “average” drug. In<br />
some cases, pharmacogenetics could yield potential<br />
sav<strong>in</strong>gs as high as $335 million and potentially capture<br />
additional upside through price premiums or<br />
an <strong>in</strong>crease <strong>in</strong> market share. In other cases, it might<br />
save noth<strong>in</strong>g, or even destroy value <strong>in</strong> the market.<br />
The key to success is to be selective.<br />
Implement<strong>in</strong>g Pharmacogenetics<br />
Although the sav<strong>in</strong>gs atta<strong>in</strong>able through pharmacogenetics<br />
appear less dramatic than those atta<strong>in</strong>able<br />
through disease genetics, they are <strong>in</strong> the right circumstances<br />
quite substantial. And the total value<br />
added would be enormous if the hoped-for market<br />
advantages were realized.
Realiz<strong>in</strong>g this value is a matter not only of market<br />
dynamics, but also of various more speculative factors:<br />
how acceptable pharmacogenetics will prove<br />
to payers, patients, physicians, and regulatory agencies;<br />
how readily physicians and patients will<br />
embrace the screen<strong>in</strong>g tests to generate share shift;<br />
and so on. So the different types of pharmacogenetics<br />
will probably come <strong>in</strong>to effect at different<br />
times. Detection of rare side effects will probably be<br />
<strong>in</strong>troduced first, as pharmaceutical companies are<br />
highly motivated to save drugs from failure. Efficacy<br />
pharmacogenetics will probably progress on a<br />
slower timetable, ow<strong>in</strong>g to concerns about market<br />
fragmentation. It might even take FDA action to<br />
turn efficacy test<strong>in</strong>g <strong>in</strong>to a rout<strong>in</strong>e procedure.<br />
When contemplat<strong>in</strong>g their pharmacogenetics policy,<br />
companies will need to scrupulously analyze<br />
specific drugs and markets. Decid<strong>in</strong>g shrewdly just<br />
where and when to apply pharmacogenetics, for<br />
<strong>in</strong>stance, will mean assess<strong>in</strong>g market dynamics earlier<br />
than ever before <strong>in</strong> the cl<strong>in</strong>ical trials phase. And<br />
that <strong>in</strong> turn will demand new decision-mak<strong>in</strong>g<br />
processes and communication channels, <strong>in</strong>clud<strong>in</strong>g<br />
stronger ties between research and development,<br />
and between R&D and commercial activities. It is<br />
on operational and organizational issues of this<br />
k<strong>in</strong>d that the spotlight will fall <strong>in</strong> the next chapter<br />
of this report.<br />
A F<strong>in</strong>al Word<br />
If the new genetics can realize its full potential, the<br />
economics of pharmaceutical R&D will undergo a<br />
metamorphosis. Efficiency will improve handsomely<br />
and success rates will surge. The sums saved<br />
could exceed a half billion dollars per drug, more<br />
than halv<strong>in</strong>g the current cost.<br />
That prospect is far from assured. There are<br />
enough risks and uncerta<strong>in</strong>ties to temper excite-<br />
ment. The range of possible outcomes is wide, and<br />
companies will have to exam<strong>in</strong>e m<strong>in</strong>utely and apply<br />
selectively the various genetics opportunities.<br />
Contrast genomics technology: the productivity<br />
improvements promised by its implementation may<br />
be more modest, but they are clearly achievable,<br />
despite the operational challenges. With genetics,<br />
the operational challenges are formidable too, but<br />
they are compounded by less dist<strong>in</strong>ct and possibly<br />
more <strong>in</strong>tractable challenges: technological limitations,<br />
scientific unknowns, and (<strong>in</strong> the case of pharmacogenetics)<br />
the vagaries of the marketplace.<br />
So, companies determ<strong>in</strong>ed to acquire and exploit<br />
genetic <strong>in</strong>formation need to know what they are lett<strong>in</strong>g<br />
themselves <strong>in</strong> for. They need to consider how<br />
applicable genetics is to their current research<br />
strategy. They need to spell out the level of risk they<br />
are prepared to take on, and then plan how to manage<br />
that risk. In short, they need a genetics strategy.<br />
In the case of disease genetics, risk management<br />
would best beg<strong>in</strong> by contemplat<strong>in</strong>g the sheer magnitude<br />
of the undertak<strong>in</strong>g. Companies will be<br />
prompted to ask themselves questions such as these:<br />
How feasible is it for us to establish an extensive<br />
disease genetics program <strong>in</strong>-house? On which diseases<br />
should our program focus? Are there opportunities<br />
to share the risk, perhaps by jo<strong>in</strong><strong>in</strong>g a “precompetitive”<br />
<strong>in</strong>dustry consortium, along the l<strong>in</strong>es<br />
of the SNP Consortium? Or, should we adopt a waitand-see<br />
stance, and then hope to license <strong>in</strong> the<br />
fruits of others’ labor?<br />
In the case of pharmacogenetics, risk management<br />
beg<strong>in</strong>s by reevaluat<strong>in</strong>g the pipel<strong>in</strong>e. On that basis,<br />
companies will try to determ<strong>in</strong>e the drugs to which<br />
pharmacogenetics applications would add most<br />
value. Companies will also want to study <strong>in</strong>tently the<br />
market context and competitor landscape, th<strong>in</strong>k<strong>in</strong>g<br />
through potential competitor moves and counter-<br />
39
40<br />
moves, along with the relative scientific feasibility<br />
for each drug or therapeutic class. Such assessments<br />
will need to be revised cont<strong>in</strong>uously, as different<br />
drugs present themselves for consideration<br />
and perhaps suggest different approaches, and as<br />
the market and the regulatory environment cont<strong>in</strong>ue<br />
to evolve.<br />
A genetics strategy would encompass all of these<br />
issues, and would optimize any potential synergies<br />
among genetics approaches. If a company decides<br />
to implement both disease genetics and pharmacogenetics,<br />
it will need to decide how to <strong>in</strong>tegrate and<br />
harmonize them. Which diseases, for example,<br />
might be amenable to disease genetics on the one<br />
hand, and be likely to provide a market premium<br />
on the other? If genetic redef<strong>in</strong>ition of diseases<br />
makes it possible to develop suites of drugs and<br />
thereby address several smaller markets, how can<br />
research best collaborate with market<strong>in</strong>g to maximize<br />
the impact? The answers to these questions<br />
will generate still more questions: Do we have the<br />
requisite skills and capabilities to pursue the strategy?<br />
Do we have the right alliances <strong>in</strong> place, or the<br />
right alliance strategy?<br />
Genetics is a risky endeavor. Companies cannot<br />
avoid the risk—but they cannot lightly ignore the<br />
potential jackpot either. They need to be selective<br />
and smart <strong>in</strong> decid<strong>in</strong>g how and where to place their<br />
bets. With such vast w<strong>in</strong>n<strong>in</strong>gs at stake, it seems<br />
appropriate that the odds should be fairly long.
Chapter 3: Managerial Challenges<br />
Preface: Look<strong>in</strong>g Back and Look<strong>in</strong>g Forward<br />
The Story So Far<br />
The genomics revolution is poised to sweep aside<br />
the old economics of pharmaceutical R&D. The<br />
biotechnology and pharmaceutical <strong>in</strong>dustries—and<br />
perhaps health care delivery <strong>in</strong> general—are on the<br />
br<strong>in</strong>k of transformation, and companies that<br />
embrace the revolution <strong>in</strong> the right way stand to<br />
reap enormous benefits. Develop<strong>in</strong>g a new drug<br />
should become considerably less unpredictable and<br />
much less expensive. Companies will record<br />
improvements both <strong>in</strong> efficiency and <strong>in</strong> success<br />
rates all along the value cha<strong>in</strong>, and the average cost<br />
and time needed to br<strong>in</strong>g a new drug to market will<br />
fall correspond<strong>in</strong>gly. (See sidebar, “Potential<br />
Sav<strong>in</strong>gs—From Theoretical to Practical.”)<br />
But this benign prospect is clouded by some warn<strong>in</strong>gs:<br />
great rewards will require comparably great<br />
efforts; a new paradigm <strong>in</strong> R&D economics may<br />
necessitate paradigm shifts <strong>in</strong> R&D management;<br />
above all, the great promise is offset by great risks—<br />
though, as <strong>in</strong> any revolution, the risks of stand<strong>in</strong>g<br />
aside may be greater than those of gett<strong>in</strong>g <strong>in</strong>volved.<br />
Ensur<strong>in</strong>g Your Future<br />
All biopharmaceutical companies are, or should be,<br />
actively decid<strong>in</strong>g how best to engage <strong>in</strong> the revolution.<br />
Mak<strong>in</strong>g such decisions is no easy matter. The<br />
familiar bear<strong>in</strong>gs are no longer there, s<strong>in</strong>ce the<br />
competitive and regulatory landscapes have<br />
changed so much—and cont<strong>in</strong>ue to change—<strong>in</strong> response<br />
to the promise that genomics offers. Compa-<br />
nies have been rush<strong>in</strong>g to claim <strong>in</strong>tellectual property<br />
rights (<strong>in</strong> the so-called IP land grab), now that<br />
the sequenc<strong>in</strong>g of the human genome has been<br />
completed. Statutes and court decisions regulat<strong>in</strong>g<br />
those IP rights keep emerg<strong>in</strong>g and modify<strong>in</strong>g the<br />
picture. (See sidebar, “Intellectual Property—Lost<br />
and Found.”) And the corporate map is be<strong>in</strong>g redrawn:<br />
the major mergers of recent years have created<br />
<strong>in</strong>dustry superpowers, and the pace of acquisitions<br />
and alliances is set to quicken, if anyth<strong>in</strong>g.<br />
(See sidebar, “Industry Changes.”)<br />
With so much change occurr<strong>in</strong>g, there are bound<br />
to be w<strong>in</strong>ners and losers. Although the decisions<br />
will be unfamiliar and difficult, success will <strong>in</strong> the<br />
end be determ<strong>in</strong>ed by traditional criteria. The w<strong>in</strong>ners<br />
will be those who make optimal strategic<br />
choices and then implement them <strong>in</strong> an optimal<br />
way. The two components of the w<strong>in</strong>n<strong>in</strong>g comb<strong>in</strong>ation<br />
will differ from company to company, accord<strong>in</strong>g<br />
to each company’s size, aspirations, f<strong>in</strong>ancial<br />
power, capabilities, and so on. In this f<strong>in</strong>al chapter<br />
of our report, we identify the strategic and operational<br />
issues and exam<strong>in</strong>e the various options that<br />
different companies might exercise.<br />
To beg<strong>in</strong> with the strategic issues, then—specifically,<br />
the challenge of def<strong>in</strong><strong>in</strong>g a strategy <strong>in</strong> the<br />
genomics era.<br />
Strategy—Search<strong>in</strong>g for Genomic<br />
Competitive Advantage<br />
Before genomics, biopharmaceutical companies<br />
used two basic tools—chemistry and molecular biol-<br />
41
42<br />
POTENTIAL SAVINGS—FROM THEORETICAL TO PRACTICAL<br />
In the first two chapters of this report, we assessed<br />
the potential sav<strong>in</strong>gs for the two waves of the<br />
genomics revolution: first, the substantial sav<strong>in</strong>gs<br />
atta<strong>in</strong>able through genomics technologies; second, the<br />
greater but far less certa<strong>in</strong> sav<strong>in</strong>gs atta<strong>in</strong>able through<br />
genetics approaches, notably disease genetics and<br />
pharmacogenetics. Those assessments show the high<br />
end of the achievable range, and they view the two<br />
waves s<strong>in</strong>gly rather than jo<strong>in</strong>tly; that is, they <strong>in</strong>dicate<br />
discrete and best-case scenarios, which will be difficult<br />
for companies to realize <strong>in</strong> practice and impossible<br />
to comb<strong>in</strong>e.<br />
A more <strong>in</strong>tegrated assessment needs to average out<br />
the achievable range—to take account of worst-case<br />
scenarios too. And it has to analyze the various likely<br />
comb<strong>in</strong>ations of approaches from the two waves,<br />
rather than treat<strong>in</strong>g genomics and genetics<br />
approaches <strong>in</strong> isolation.<br />
Accord<strong>in</strong>g to the comb<strong>in</strong>ation selected, the R&D<br />
value cha<strong>in</strong> as a whole will assume a particular new<br />
form and favor a particular subset of potential drugs.<br />
That is a crucial consideration for a company engaged<br />
<strong>in</strong> build<strong>in</strong>g a portfolio of technologies: the more com-<br />
DRUG R&D VALUE CHAIN ACTIVITIES<br />
Biology<br />
Target ID Target Optimization Validation<br />
Genomic target discovery<br />
Chemical genomics target discovery<br />
Genetic target discovery with pathway analysis<br />
Chemistry<br />
Screen<strong>in</strong>g Optimization<br />
In silico<br />
Traditional<br />
b<strong>in</strong>ations of approaches, the greater the company’s<br />
versatility <strong>in</strong> pursu<strong>in</strong>g different drug subsets.<br />
Draw<strong>in</strong>g once aga<strong>in</strong> on our economic model, we<br />
have exam<strong>in</strong>ed the impact of each feasible<br />
genomics-based approach to target discovery, augmented<br />
by pharmacogenetics and genomics technology<br />
whenever possible. And we have estimated the<br />
realizable value <strong>in</strong> each case: first, by calculat<strong>in</strong>g the<br />
cost, time, and added value for each possible comb<strong>in</strong>ation<br />
of approaches, and then—adjust<strong>in</strong>g for the<br />
percentage of targets each approach is able to<br />
process—by calculat<strong>in</strong>g a weighted average per<br />
drug. The result is three broad scenarios:<br />
• A genomics-based approach: <strong>in</strong>dustrialized target<br />
identification, supplemented where applicable by<br />
downstream genomics technologies (<strong>in</strong> silico<br />
chemistry, <strong>in</strong> silico toxicology, <strong>in</strong> vitro ADME, surrogate<br />
markers, and pharmacogenetics)<br />
• Chemical genomics: <strong>in</strong>dustrialized target identification,<br />
followed by chemistry and traditional validation<br />
conducted <strong>in</strong> parallel, and supplemented<br />
where possible by the downstream technologies<br />
just listed<br />
30%<br />
70%<br />
Development<br />
Precl<strong>in</strong>ical Cl<strong>in</strong>ical<br />
Genomics:<br />
In silico/<strong>in</strong> vitro tests, surrogate markers<br />
Pharmacogenetics<br />
Genomics and pharmacogenetics<br />
Traditional<br />
73%<br />
• Disease genetics supported by pathway analysis,<br />
comb<strong>in</strong>ed where possible with the various downstream<br />
technologies<br />
The chart on page 42 shows the three scenarios,<br />
and the weight<strong>in</strong>g applied to calculate the average<br />
cost per drug. Here, by way of example, are realistic<br />
sav<strong>in</strong>gs estimates generated by our economic model<br />
for these three approaches.<br />
First, a genomics-based approach <strong>in</strong> a traditional<br />
value cha<strong>in</strong> structure. This provides an ideal foundation<br />
for a portfolio perspective on drug discovery. It<br />
applies to both known and unknown target classes,<br />
and offers impressive potential sav<strong>in</strong>gs of $200 million<br />
and 1.5 years per drug.<br />
Next, chemical genomics is probably the most competitive<br />
approach for targets where there is an established<br />
high-throughput chemical screen<strong>in</strong>g assay.<br />
The time sav<strong>in</strong>gs are particularly important—nearly<br />
3.5 years—boost<strong>in</strong>g drug revenues by means of<br />
<strong>in</strong>tellectual property rights and first-to-market<br />
advantages. When comb<strong>in</strong>ed with other genomics<br />
approaches, chemical genomics offers potential sav<strong>in</strong>gs<br />
of about $100 million per drug. The approach<br />
lends itself particularly well to certa<strong>in</strong> targets, such<br />
as GPCRs, so a company elect<strong>in</strong>g not to pursue<br />
chemical genomics would be at a disadvantage if it<br />
reta<strong>in</strong>ed such targets on its wish list.<br />
F<strong>in</strong>ally, disease genetics supported by pathway<br />
analysis is, theoretically, the most direct route to<br />
understand<strong>in</strong>g human disease. This approach is<br />
applicable to known and unknown target classes and<br />
to targets overlooked <strong>in</strong> animal-based studies. And<br />
on the face of it, it is the most attractive approach<br />
f<strong>in</strong>ancially, with cost sav<strong>in</strong>gs of $400 million. There<br />
is an extended time to drug, however, amount<strong>in</strong>g to<br />
about one year, though that drawback would often<br />
be offset by the early secur<strong>in</strong>g of <strong>in</strong>tellectual property<br />
rights. Unfortunately, this <strong>in</strong>vit<strong>in</strong>g approach is<br />
still not affordable, and <strong>in</strong> fact its scientific feasibility<br />
rema<strong>in</strong>s unproven. (See the graphs below for a<br />
summary of expected sav<strong>in</strong>gs.)<br />
REALIZABLE RESULTS BASED ON DISCOVERY APPROACH<br />
Weighted average of approaches across value cha<strong>in</strong><br />
Cost to drug<br />
Pre-genomics<br />
Genomics-based<br />
approach<br />
Chemical genomicsbased<br />
approach<br />
Genetics-based<br />
approach 1<br />
Time to drug<br />
Pre-genomics<br />
Genomics-based<br />
approach<br />
Chemical genomicsbased<br />
approach<br />
Genetics-based<br />
approach 1<br />
0<br />
0<br />
200<br />
5<br />
400<br />
10<br />
475<br />
600<br />
11.3<br />
675<br />
800<br />
13.3<br />
15<br />
780<br />
14.7<br />
880<br />
Cost ($M)<br />
15.8<br />
1,000<br />
20<br />
Time (years)<br />
1With pathway analysis, 50% through genetic simple systems, 50% through<br />
genomics expression profil<strong>in</strong>g. Assumes resolution of scientific and technological<br />
questions.<br />
43
44<br />
ogy—to discover new drugs. Broadly speak<strong>in</strong>g, the<br />
drugs that emerged were much <strong>in</strong>debted to<br />
serendipity. Research strategy consisted ma<strong>in</strong>ly of<br />
choos<strong>in</strong>g which therapeutic areas to <strong>in</strong>vestigate, and<br />
discovery efforts focused on <strong>in</strong>dividual drug targets.<br />
Development provided even fewer strategic choices:<br />
a promis<strong>in</strong>g compound emerg<strong>in</strong>g from chemistry<br />
would be tested on animals and humans <strong>in</strong> large<br />
and <strong>in</strong>efficient trials (<strong>in</strong>efficient because there was<br />
no means of identify<strong>in</strong>g <strong>in</strong> advance likely responders<br />
or nonresponders). With the rise of genomics,<br />
there have come new technologies, new approaches,<br />
new <strong>in</strong>formation, and new ways of th<strong>in</strong>k<strong>in</strong>g about<br />
INTELLECTUAL PROPERTY—LOST AND FOUND<br />
Gene patent applications are flourish<strong>in</strong>g: <strong>in</strong> 2000<br />
alone, more than 20,000 were submitted to the<br />
United States Patent and Trademark Office. Despite<br />
the large number of applications, two central questions<br />
have yet to be answered. What exactly can be<br />
patented? And what rights does a patent actually<br />
confer on its holder?<br />
For a gene or gene fragment (an expressed sequence<br />
tag, or EST) to secure a patent, its “utility” has to be<br />
established. In January 2001, the United States<br />
Patent and Trademark Office issued Utility Exam<strong>in</strong>ation<br />
Guidel<strong>in</strong>es to clarify the standard used. (That<br />
<strong>in</strong> itself was encourag<strong>in</strong>g to those <strong>in</strong> favor of gene<br />
patent<strong>in</strong>g, re<strong>in</strong>forc<strong>in</strong>g the view that genetic material<br />
can <strong>in</strong>deed be patented.) Follow<strong>in</strong>g on the <strong>in</strong>terim<br />
guidel<strong>in</strong>es released <strong>in</strong> 1999, the new guidel<strong>in</strong>es<br />
advise patent seekers to provide at least one “specific,<br />
substantial, and credible” use for the gene or<br />
gene fragment <strong>in</strong> question. This restatement effectively<br />
fixes the height of the hurdle for applicants<br />
and disqualifies undersubstantiated applications<br />
from the outset. Some uncerta<strong>in</strong>ty rema<strong>in</strong>s, however:<br />
whether it is necessary when present<strong>in</strong>g evidence<br />
of utility to understand the actual biological<br />
function of the genetic material, and whether gene<br />
research and development. These have brought<br />
with them a new opportunity, or imperative, to turn<br />
research to competitive advantage.<br />
So companies now have weighty strategic issues to<br />
address. At the corporate level, the question is how<br />
much to <strong>in</strong>vest, given the current environment. For<br />
R&D leadership, the question tends to be where to<br />
focus those <strong>in</strong>vestments—<strong>in</strong> what therapy areas, on<br />
what target classes, and so on—as well as which<br />
technologies to adopt and how to adopt them (<strong>in</strong>house<br />
or externally, for example), and how to mitigate<br />
the associated risks.<br />
fragments, as dist<strong>in</strong>ct from full-length genes, are eligible<br />
for a patent.<br />
If gett<strong>in</strong>g a patent approved seems daunt<strong>in</strong>g, all the<br />
more so is enforc<strong>in</strong>g it, or shelter<strong>in</strong>g confidently <strong>in</strong><br />
its protective embrace. The strength of protection<br />
afforded by a gene patent is still a develop<strong>in</strong>g legal<br />
issue. One recent court decision, though, clearly<br />
marks a setback for patent holders—Festo Corp. v<br />
Shoketsu K<strong>in</strong>zoku Kogyo Kabushiki, decided by the<br />
federal circuit court <strong>in</strong> November 2000. It appears<br />
to weaken many patents by preclud<strong>in</strong>g a broad <strong>in</strong>terpretation<br />
of most patent claims; it does so by virtually<br />
exclud<strong>in</strong>g the “doctr<strong>in</strong>e of equivalents.” This is a<br />
doctr<strong>in</strong>e that generally allows extension of a patent’s<br />
claim beyond its literal language, so that a would-be<br />
<strong>in</strong>fr<strong>in</strong>ger who makes trivial changes to the patented<br />
product is not thereby exempt from the patent’s constra<strong>in</strong>ts.<br />
Accord<strong>in</strong>g to the Festo rul<strong>in</strong>g, if the language<br />
of the legal claim diverges from that of the<br />
patent itself, the doctr<strong>in</strong>e no longer applies. Many<br />
patents now look very narrow and vulnerable, and<br />
companies will have to plan their patent submissions<br />
even more carefully <strong>in</strong> the future. Or hope for<br />
a change of fortune: the U.S. Supreme Court is<br />
expected to review Festo <strong>in</strong> the 2001–2002 term.
The Start<strong>in</strong>g Position<br />
Although these same broad questions will apply<br />
equally to all companies, there can be no standard<br />
answers. The actual options available to any company<br />
will depend on its start<strong>in</strong>g position.<br />
Company Size<br />
A key constra<strong>in</strong>t on a company’s strategic options is<br />
size. The largest pharmaceutical companies boast<br />
capabilities and f<strong>in</strong>ances on a scale that allows full<br />
participation <strong>in</strong> the new technologies, even when<br />
the risk is high. Not that this exempts them from<br />
hav<strong>in</strong>g to make choices. In fact, s<strong>in</strong>ce scale gives<br />
them so many options, they arguably carry a greater<br />
burden of strategic decision mak<strong>in</strong>g. How to select<br />
from such an embarrassment of riches? In addition,<br />
they face the challenge of manag<strong>in</strong>g complexity. If<br />
they are not selective enough, and embrace too<br />
many options, the operational problems could<br />
prove overwhelm<strong>in</strong>g.<br />
The narrower capabilities and lesser scale of smallto-medium-sized<br />
pharmaceutical companies and the<br />
larger biotech companies could represent either a<br />
severe drawback or a dist<strong>in</strong>ct advantage. On the one<br />
hand, there are reduced opportunities and even the<br />
prospect of be<strong>in</strong>g locked out by the big pharmaceutical<br />
firms: with disease genetics, for <strong>in</strong>stance, a company<br />
with <strong>in</strong>sufficient scale to build an <strong>in</strong>-house<br />
capability would risk forfeit<strong>in</strong>g potentially lucrative<br />
<strong>in</strong>tellectual property rights. On the other hand,<br />
s<strong>in</strong>ce lesser scale often means lesser complexity,<br />
these modest-sized companies can compete more<br />
flexibly, chang<strong>in</strong>g their tactics quickly <strong>in</strong> response to<br />
technological advances or competitor moves.<br />
To see how scale can affect a company’s options,<br />
consider the differ<strong>in</strong>g ways <strong>in</strong> which large and midsize<br />
companies approach the target land grab. The<br />
larger companies have been able to take very<br />
aggressive approaches—scal<strong>in</strong>g up or pursu<strong>in</strong>g big<br />
deals to secure <strong>in</strong>tellectual property rights to targets.<br />
The smaller companies, lack<strong>in</strong>g <strong>in</strong> resources,<br />
have been unable to follow suit, but some of them<br />
have compensated by choos<strong>in</strong>g very focused strategies,<br />
concentrat<strong>in</strong>g on their special competencies<br />
and impos<strong>in</strong>g higher quality standards.<br />
Build<strong>in</strong>g the Fact Base<br />
Apart from company size, the two most important<br />
facets of a company’s start<strong>in</strong>g po<strong>in</strong>t are the beliefs<br />
and hypotheses held by its leadership team (roughly,<br />
its corporate culture) and its current R&D capabilities.<br />
Companies need to scrut<strong>in</strong>ize both.<br />
It is crucial to understand and shape the beliefs and<br />
hypotheses of leaders throughout the organization,<br />
especially s<strong>in</strong>ce, with genomics and genetics, the<br />
contributions and effects are cross-functional—that<br />
is, the managers or sections that contribute most<br />
are not necessarily those that benefit most. All<br />
those affected need to articulate their perceptions<br />
of the value and applicability of genomics and<br />
genetics to the company. Once tested, these perceptions<br />
should be given considerable weight when<br />
it comes to def<strong>in</strong><strong>in</strong>g company strategy.<br />
An equally thorough assessment needs to be made<br />
of the company’s relevant R&D capabilities—its<br />
technologies, skills, specific knowledge of diseases<br />
and disease mechanisms, and so on. Ideally, this will<br />
<strong>in</strong>clude an audit of current R&D productivity at<br />
every step <strong>in</strong> the value cha<strong>in</strong>, identify<strong>in</strong>g bottlenecks<br />
and other constra<strong>in</strong>ts. The more accurate<br />
and detailed the assessment, the more effectively<br />
the company can address the strategic questions as<br />
they perta<strong>in</strong> to its specific situation.<br />
Corporate Decisions:<br />
How Much to Invest and Where<br />
As suggested above, even the largest pharmaceutical<br />
companies will have to make choices. Consider<br />
some of the huge deals of recent times: the $500<br />
million deal between Bayer and Millennium for targets,<br />
the $800 million deal between Novartis and<br />
Vertex for <strong>in</strong> silico chemistry, the $500 million deal<br />
between Roche and deCODE for disease genes.<br />
Note that these deals concern discrete steps of the<br />
value cha<strong>in</strong>: <strong>in</strong> each case, it appears likely that the<br />
companies concerned were act<strong>in</strong>g on an explicit<br />
preference—an established strategic preference.<br />
After all, given the magnitude of these deals, it<br />
seems unlikely that any one company would have<br />
placed all three bets. More to the po<strong>in</strong>t, such large<br />
deals, although essentially R&D ventures, are not<br />
45
46<br />
INDUSTRY CHANGES<br />
The genomics landscape features many small startups<br />
amid the larger genomics companies and the<br />
genomics divisions of big pharmaceutical corporations.<br />
But that landscape is chang<strong>in</strong>g. The number<br />
of deals—of genomics companies comb<strong>in</strong><strong>in</strong>g with<br />
each other or be<strong>in</strong>g taken over by big pharmaceutical<br />
firms—has been grow<strong>in</strong>g steadily. What is driv<strong>in</strong>g<br />
this tendency toward consolidation?<br />
The Pressure to Extend Scope<br />
Increas<strong>in</strong>gly, genomics companies are aspir<strong>in</strong>g to<br />
become full-fledged drug companies. No specialized<br />
company, it seems, has yet succeeded <strong>in</strong> build<strong>in</strong>g a<br />
truly stable competitive position as a drug-<strong>in</strong>dustry<br />
supplier; <strong>in</strong>tellectual-property statutes do not<br />
appear to be enough to guarantee long-term protection;<br />
and the chances of proprietary advantage are<br />
be<strong>in</strong>g nullified by the trend toward public-private<br />
partnerships or consortia, underwritten by big pharmaceutical<br />
companies.<br />
Wall Street appears to place a far higher value on<br />
<strong>in</strong>tegrated drug producers than on pure technology<br />
companies (if only because the drug sector has traditionally<br />
enjoyed such high profits and such high<br />
regard among <strong>in</strong>vestors). Accord<strong>in</strong>g to a recent USB<br />
Warburg study, the average <strong>in</strong>tegrated drug company<br />
has been able to raise $870 million, as aga<strong>in</strong>st<br />
a mere $330 million for the average technology<br />
company. (The study noted a further <strong>in</strong>terest<strong>in</strong>g<br />
divergence among technology companies themselves:<br />
biology companies—those focused on target<br />
identification and validation—raised $480 million<br />
on average, whereas companies <strong>in</strong> the chemistry<br />
area—those focused on screen<strong>in</strong>g and lead optimization—raised<br />
on average only $170 million.)<br />
In keep<strong>in</strong>g with this expansionist aspiration, most of<br />
the recent deals have consisted of acquisitions of<br />
downstream drug-development capabilities. Witness<br />
LION’s acquisition of Trega (for $35 million),<br />
Celera’s acquisition of AxyS (for $173 million), and<br />
Lexicon’s acquisition of Coelacanth (for $32 million).<br />
The Pressure to Achieve Scale<br />
As sections of the value cha<strong>in</strong> have become <strong>in</strong>dustrialized,<br />
the value of scale <strong>in</strong> R&D has ga<strong>in</strong>ed<br />
prom<strong>in</strong>ence. And for genomics platform companies<br />
and pharmaceutical companies alike, it may appear<br />
quicker and neater to achieve scale through a merger<br />
than through pa<strong>in</strong>stak<strong>in</strong>g <strong>in</strong>-house upscal<strong>in</strong>g. (Of<br />
course, pharmaceutical companies might have other<br />
reasons to acquire genomics companies: to jumpstart<br />
their genomics efforts, for <strong>in</strong>stance, or to<br />
acquire otherwise rare capabilities.)<br />
Sure enough, most of the recent mergers and acquisitions<br />
have clearly been <strong>in</strong>itiated for the sake of<br />
<strong>in</strong>creas<strong>in</strong>g scale: Sequenom’s acquisition of Gem<strong>in</strong>i<br />
Genomics (for $238 million), for example, or<br />
Sangamo’s acquisition of Gendaq (for $40 million).<br />
These scale deals have been primarily <strong>in</strong> target identification<br />
and validation rather than <strong>in</strong> chemistry—a<br />
reflection of the urgency of the land grab.<br />
The Pressure to Spend<br />
Whatever the <strong>in</strong>ducements to merge, there is a traditional<br />
impediment—lack of wherewithal. The spirit<br />
is will<strong>in</strong>g but the purse is weak. That is certa<strong>in</strong>ly not<br />
a constra<strong>in</strong>t, however, on some of the large<br />
genomics companies at the moment. In mid-2001,<br />
three top companies—Human Genome Sciences,<br />
Celera, and Millennium—boasted over $4 billion <strong>in</strong><br />
cash between them, represent<strong>in</strong>g about 25 percent<br />
of their comb<strong>in</strong>ed market capitalizations. Idle money<br />
cries out to be spent—probably, for these companies,<br />
on diversification more than on scal<strong>in</strong>g.<br />
Market expectations, a sense of urgency, an abundance<br />
of funds: all signs po<strong>in</strong>t<strong>in</strong>g to cont<strong>in</strong>ued consolidation<br />
<strong>in</strong> the near term.
R&D decisions alone. Almost certa<strong>in</strong>ly, the decisions<br />
were thrashed out at the corporate level.<br />
For more modest-sized companies, strategic choices<br />
often go beyond matters of preference or emphasis.<br />
The question might be whether to concentrate all<br />
their efforts on some value cha<strong>in</strong> steps and forgo<br />
others altogether. Certa<strong>in</strong>ly it no longer makes<br />
sense for even midsized pharmaceutical companies<br />
to compete <strong>in</strong> target identification. And at the<br />
smaller end of the scale, companies with less than<br />
$400 million <strong>in</strong> R&D, say, may f<strong>in</strong>d themselves ask<strong>in</strong>g<br />
even more radical questions: Can we afford<br />
research at all? Should we not focus exclusively on<br />
licens<strong>in</strong>g <strong>in</strong>stead? Aga<strong>in</strong>, it is at the corporate level,<br />
rather than with<strong>in</strong> R&D alone, that such questions<br />
will eventually be settled.<br />
It is not just through major partnerships and<br />
<strong>in</strong>vestment decisions, however, that the corporate<br />
level is imp<strong>in</strong>g<strong>in</strong>g on R&D strategy. More and more,<br />
specific R&D activities are hav<strong>in</strong>g ramifications<br />
beyond R&D itself, and <strong>in</strong>vok<strong>in</strong>g corporate-level<br />
participation. Pharmacogenetics, for <strong>in</strong>stance,<br />
often touches on corporate strategy as much as on<br />
R&D strategy. Should the company cont<strong>in</strong>ue to pursue<br />
a promis<strong>in</strong>g compound, say, when the risk of<br />
market fragmentation might outweigh the positive<br />
market effects? Should the company attempt to resurrect<br />
candidate drugs previously killed because of<br />
rare side effects? And so on.<br />
R&D Leadership Decisions:<br />
Where and How to Compete<br />
With genomics and genetics now part of the landscape,<br />
R&D decision mak<strong>in</strong>g has become more<br />
complex. The options are far more numerous:<br />
there are more ways of ga<strong>in</strong><strong>in</strong>g access to capabilities,<br />
more technologies to choose among, and even<br />
new dimensions <strong>in</strong> which to compete. R&D executives<br />
must select a comb<strong>in</strong>ation of options that not<br />
only dovetail with the company’s start<strong>in</strong>g position<br />
and aspirations but can also be <strong>in</strong>tegrated smoothly<br />
with one another.<br />
Choos<strong>in</strong>g a Research Focus<br />
The dimensions of competition <strong>in</strong>clude:<br />
• Disease states. Some disease states have become<br />
more tractable, thanks to genomics approaches,<br />
and any company cont<strong>in</strong>u<strong>in</strong>g to <strong>in</strong>vestigate them<br />
will have to deploy genomics if it is to rema<strong>in</strong><br />
competitive. Just which therapeutic areas or disease<br />
states are most amenable to genomics is<br />
determ<strong>in</strong>ed by several factors: the degree to<br />
which the disease is genetic <strong>in</strong> nature, the current<br />
understand<strong>in</strong>g of disease processes at a molecular<br />
or genetic level, and so on.<br />
• Target class. Some genomics approaches are at<br />
odds with traditional therapeutic-area borders,<br />
and favor a broader deployment—around target<br />
class—rather than the old focus on disease state.<br />
(The targets with<strong>in</strong> a class are usually similar <strong>in</strong><br />
structure and biochemical function.)<br />
• Therapeutic modalities. Small-molecule drugs<br />
still dom<strong>in</strong>ate the market, but they no longer<br />
monopolize it. Some new therapeutic modalities<br />
have already established a foothold—<strong>in</strong>jectible<br />
prote<strong>in</strong> therapeutics, for <strong>in</strong>stance, based on<br />
secreted factors and antibodies. Others rema<strong>in</strong><br />
very much <strong>in</strong> the experimental stage—gene therapy<br />
and anti-sense techology, for example—though<br />
adventurous companies are pursu<strong>in</strong>g them<br />
undaunted (as exemplified by Lilly’s recent $200<br />
million deal with Isis to ga<strong>in</strong> access to anti-sense<br />
capabilities).<br />
These dimensions are <strong>in</strong>terconnected, of course,<br />
and even <strong>in</strong>terdependent. Take Novartis’s <strong>in</strong>terest<br />
<strong>in</strong> oncology, for example—a broad disease state.<br />
Given that <strong>in</strong>terest, it made sense for the company<br />
to focus on k<strong>in</strong>ases, a key target class <strong>in</strong> oncology.<br />
K<strong>in</strong>ases constitute one of the few target classes that<br />
are amenable to a particular genomics approach, <strong>in</strong><br />
silico drug design. Novartis has duly set about augment<strong>in</strong>g<br />
its expertise with the appropriate<br />
genomics technology, form<strong>in</strong>g an alliance with<br />
Vertex to that end.<br />
47
48<br />
Select<strong>in</strong>g Technologies<br />
Accord<strong>in</strong>g to the research focus adopted by the<br />
company, certa<strong>in</strong> technologies will press their<br />
claims immediately. An oncology program, for<br />
<strong>in</strong>stance, would certa<strong>in</strong>ly argue for the <strong>in</strong>corporation<br />
of a transcription profil<strong>in</strong>g approach, as more<br />
and more cancers are be<strong>in</strong>g redef<strong>in</strong>ed at the level<br />
of RNA expression. But each claim would have to be<br />
assessed by reference to the company’s aspirations<br />
and current capabilities. How comfortably would a<br />
candidate technology fit <strong>in</strong> with the company’s risk<br />
profile or exist<strong>in</strong>g skills mix, for example?<br />
In addition, companies will need to consider the<br />
current stage of development of genomics technologies.<br />
When is the best time to buy <strong>in</strong>to the<br />
favored technology? As noted throughout this<br />
report, although some genomics approaches are<br />
practicable today—<strong>in</strong> the early steps of the value<br />
cha<strong>in</strong>, notably—others rema<strong>in</strong> speculative:<br />
genome-wide association studies, for <strong>in</strong>stance. A<br />
company’s risk profile will determ<strong>in</strong>e whether it<br />
wishes to be on the “bleed<strong>in</strong>g edge” or to be a technology<br />
follower. Either way, the company will want<br />
to chart the evolution of genomics technologies<br />
and approaches, and adjust its own strategy accord<strong>in</strong>gly.<br />
A technology scout<strong>in</strong>g function is <strong>in</strong>dispensable,<br />
now more than ever.<br />
Whether the technology is proven or unproven,<br />
companies will need to decide not just whether and<br />
when to <strong>in</strong>vest, but also how—how to keep a sharp<br />
focus and mitigate the risks <strong>in</strong>volved. The options<br />
vary from company to company, aga<strong>in</strong> accord<strong>in</strong>g to<br />
company size. With disease genetics, say, a large<br />
pharmaceutical company that chose to pursue the<br />
technology <strong>in</strong>-house would face the question of how<br />
to apply it—to which therapeutic areas, for example.<br />
A smaller company, by contrast, unable to build<br />
a program <strong>in</strong>-house, and obliged to take a different<br />
approach, would face such questions as what k<strong>in</strong>d<br />
of jo<strong>in</strong>t ventures to pursue and what focus to apply.<br />
Decid<strong>in</strong>g How to Acquire or<br />
Ga<strong>in</strong> Access to Capabilities<br />
In general, there are several ways to atta<strong>in</strong> a desired<br />
capability, but <strong>in</strong> some cases the options are limited.<br />
When the item is a proprietary database or tool, for<br />
<strong>in</strong>stance, the company will have to license it <strong>in</strong> (or<br />
pay a provider for service) rather than buy it outright;<br />
or when a company views its own <strong>in</strong>formation<br />
as too confidential to outsource, it will be forced to<br />
implement the related technology <strong>in</strong>-house. In<br />
many cases, though, a company will face the choice<br />
between build<strong>in</strong>g <strong>in</strong>-house capabilities and outsourc<strong>in</strong>g.<br />
The <strong>in</strong>-house option, to justify itself,<br />
would have to confer some significant strategic or<br />
cost advantage. A company could have a cost advantage<br />
if it had developed a proprietary method, for<br />
example, or if it could boast greater scale or experience<br />
<strong>in</strong> a given approach.<br />
Some though not all of the new technologies show<br />
clear scale benefits, thanks to <strong>in</strong>dustrialized processes<br />
and <strong>in</strong>formatics. (Among the most oblig<strong>in</strong>g<br />
technologies <strong>in</strong> this regard are expression profil<strong>in</strong>g,<br />
traditional HTS and µHTS, and exploitation of<br />
<strong>in</strong>formatics-based analysis. The least oblig<strong>in</strong>g are<br />
medic<strong>in</strong>al chemistry and animal models, and somewhere<br />
<strong>in</strong> between are compound synthesis and<br />
management, proteomic expression analysis, structural<br />
biology, and <strong>in</strong> silico chemistry.) Unfortunately,<br />
build<strong>in</strong>g scale <strong>in</strong>-house could be disproportionately<br />
costly for small-to-midsized pharmaceutical<br />
companies, even for the most scale-friendly<br />
technologies. These companies are unlikely to realize<br />
cost advantages; they risk spread<strong>in</strong>g their technology<br />
dollars too th<strong>in</strong>. The wiser option would be<br />
partner<strong>in</strong>g or licens<strong>in</strong>g.<br />
If a company decides to develop a given technology<br />
<strong>in</strong>-house, it should review that decision regularly.<br />
What is today a strategically advantageous capability<br />
may be commoditized tomorrow. The perception of<br />
sequenc<strong>in</strong>g, for <strong>in</strong>stance, seems to be shift<strong>in</strong>g, from<br />
a need-to-have technology to someth<strong>in</strong>g that can<br />
readily be outsourced.<br />
If a company decides to outsource a given technology,<br />
it will have to decide further on a prospective<br />
partner or partners. It might even opt to jo<strong>in</strong> forces<br />
with competitors. A model partnership of this k<strong>in</strong>d<br />
has been the SNP Consortium. A group of pharmaceutical<br />
companies, helped by various academic
<strong>in</strong>stitutions, banded together to identify 300,000<br />
SNPs (<strong>in</strong> the end, the total was about one million)<br />
and put them <strong>in</strong>to the public doma<strong>in</strong>. This jo<strong>in</strong>t<br />
effort had two very beneficial effects for its participants.<br />
First, it enabled the companies to concentrate<br />
more on their core <strong>in</strong>terest, f<strong>in</strong>d<strong>in</strong>g drugs;<br />
second, it forestalled the efforts of genomics companies,<br />
which would have sought to patent and<br />
extract rents from these SNPs. Other candidates for<br />
“coopetition” of this k<strong>in</strong>d <strong>in</strong>clude prote<strong>in</strong> structure<br />
model<strong>in</strong>g and broad-scale sample collection for disease<br />
association studies.<br />
CASE STUDY: BRINGING RESEARCH INTO FOCUS<br />
A midsized pharmaceutical company <strong>in</strong>itiated an<br />
R&D strategy overhaul. With the broad goal of <strong>in</strong>creas<strong>in</strong>g<br />
productivity, the effort was directed at reduc<strong>in</strong>g<br />
the number of and concentrat<strong>in</strong>g the areas of<br />
<strong>in</strong>vestigation and boost<strong>in</strong>g the company’s access to<br />
relevant technologies.<br />
The CEO and R&D director commissioned a review<br />
of the company’s research and development capabilities.<br />
A cross-functional project team then set about<br />
def<strong>in</strong><strong>in</strong>g those capabilities precisely at all steps of<br />
the value cha<strong>in</strong>, rat<strong>in</strong>g their quality, align<strong>in</strong>g them<br />
with the diseases and markets of <strong>in</strong>terest, and identify<strong>in</strong>g<br />
gaps and synergies.<br />
With specific therapeutic areas <strong>in</strong> m<strong>in</strong>d, the company<br />
next turned to optimiz<strong>in</strong>g its genomics technology<br />
portfolio. The project team embarked on a threestage<br />
assessment of the various strategic options<br />
and their correspond<strong>in</strong>g technologies.<br />
First, hav<strong>in</strong>g audited the company’s capabilities and<br />
research <strong>in</strong>terests, the team identified relevant<br />
<strong>in</strong>dustry trends, and from these generated a list of<br />
the strategic options. (Genetics was listed, for example,<br />
as a major target-discovery trend <strong>in</strong> a favored<br />
research area—a therapeutic area with many heritable<br />
diseases.) The team then compiled a list of<br />
Putt<strong>in</strong>g the Strategy <strong>in</strong>to Operation<br />
Def<strong>in</strong><strong>in</strong>g a genomics strategy is a good start, but<br />
even the most brilliant strategy is futile if it rema<strong>in</strong>s<br />
def<strong>in</strong>ed on paper only. The po<strong>in</strong>t is to put it <strong>in</strong>to<br />
operation. Putt<strong>in</strong>g a strategy <strong>in</strong>to operation consists<br />
essentially of mak<strong>in</strong>g changes and manag<strong>in</strong>g them<br />
effectively. In the case of genomics and genetics,<br />
the changes that need to be made are profound,<br />
affect<strong>in</strong>g all aspects of the R&D organization and,<br />
by extension, the corporation as a whole—core<br />
processes, organizational structure, job descrip-<br />
match<strong>in</strong>g technologies, whether currently owned or<br />
not—those that would enhance the options’ chance<br />
of success. (Aga<strong>in</strong>st the genetics option, for<br />
<strong>in</strong>stance, were listed such matches as SNP maps<br />
and genotyp<strong>in</strong>g technologies.) F<strong>in</strong>ally, the team evaluated<br />
each technology’s likely impact on productivity,<br />
us<strong>in</strong>g a sophisticated productivity model the<br />
company had established.<br />
What emerged was a rank<strong>in</strong>g of various strategic<br />
options and their required technologies—<strong>in</strong> effect,<br />
the basis for a new, <strong>in</strong>tegrated technology strategy<br />
and the bluepr<strong>in</strong>t for an optimized technology portfolio.<br />
Company executives were now <strong>in</strong> a position to<br />
ponder access arrangements—whether upgrad<strong>in</strong>g,<br />
licens<strong>in</strong>g, or partner<strong>in</strong>g.<br />
More broadly, the result of this effort has been a dramatic<br />
focus<strong>in</strong>g and align<strong>in</strong>g of research activities,<br />
directed by a well-articulated strategy. The R&D<br />
managers have been able to commit to specific productivity<br />
metrics and time frames. They have been<br />
able to agree on a dist<strong>in</strong>ct research focus: three key<br />
therapeutic areas and limited modalities. And they<br />
now have a specific technology strategy, with a clear<br />
plan for acquir<strong>in</strong>g the necessary components and an<br />
<strong>in</strong>vestment program to make it happen.<br />
49
50<br />
tions, <strong>in</strong>terfaces, and so on. The necessary work can<br />
be divided <strong>in</strong>to three broad areas:<br />
•Rebalanc<strong>in</strong>g the value cha<strong>in</strong><br />
• Establish<strong>in</strong>g the new organization and its<br />
governance<br />
• Manag<strong>in</strong>g organizational change<br />
Rebalanc<strong>in</strong>g the Value Cha<strong>in</strong><br />
The old ways of conduct<strong>in</strong>g R&D are often unsuited<br />
to the new era. As the first chapter showed, the traditional<br />
R&D value cha<strong>in</strong> no longer works. For one<br />
th<strong>in</strong>g, its smooth flow quickly becomes disrupted by<br />
a series of bottlenecks, <strong>in</strong>duced by the different<br />
productivity ga<strong>in</strong>s at different phases. For another,<br />
its sequence is unsusta<strong>in</strong>able, s<strong>in</strong>ce the new technologies<br />
dance to a different schedule: much of the<br />
chemistry phase might now take place simultaneously<br />
with target validation, for example. To re<strong>in</strong>state<br />
a smooth flow (while enjoy<strong>in</strong>g the new, much<br />
accelerated rate of throughput), R&D needs to ease<br />
the bottlenecks and adjust to a reconfigured value<br />
cha<strong>in</strong>. And that means redistribut<strong>in</strong>g resources<br />
and, more importantly, redesign<strong>in</strong>g processes, as<br />
well as keep<strong>in</strong>g the new value cha<strong>in</strong> <strong>in</strong> balance.<br />
Restor<strong>in</strong>g Balance: Reallocation versus Redesign<br />
At first sight, the bottleneck problem would seem<br />
fairly simple to resolve: scale up downstream steps<br />
to meet the <strong>in</strong>creased demand. But how feasible<br />
would that be? The number of targets identified<br />
could <strong>in</strong>crease sixfold or more. To scale up to meet<br />
that <strong>in</strong>crease, a company accustomed to spend<strong>in</strong>g<br />
$1 billion on all of R&D would now have to spend<br />
more than $1.5 billion on target validation alone.<br />
Another simple approach suggests itself: adjust<br />
resources along the value cha<strong>in</strong> <strong>in</strong> order to br<strong>in</strong>g<br />
the uneven phases back <strong>in</strong>to balance, shift<strong>in</strong>g funds<br />
from more efficient phases (notably target discovery)<br />
to less efficient downstream phases (such as<br />
precl<strong>in</strong>ical). But such reallocation of resources is,<br />
on its own, an overcautious measure, and will not<br />
have a really dramatic impact on R&D economics. It<br />
neglects, or even distracts from, the central opportunity<br />
that genomics offers: the opportunity to<br />
“raise the game” by chang<strong>in</strong>g fundamentally the<br />
way R&D is conducted.<br />
This transformation of R&D will derive above all<br />
from the bold reconfigur<strong>in</strong>g of processes, for the<br />
sake of both physical process flow and <strong>in</strong>formation<br />
flow. For the former, new technology platforms<br />
need to be <strong>in</strong>tegrated and optimized, both with<strong>in</strong><br />
value cha<strong>in</strong> steps and across the value cha<strong>in</strong>. (In<br />
some cases, this may require a discipl<strong>in</strong>e and a<br />
rearrangement comparable to the mov<strong>in</strong>g assembly<br />
l<strong>in</strong>e <strong>in</strong>troduced by Henry Ford <strong>in</strong> 1913.) As for<br />
<strong>in</strong>formation flow, the tremendous amount of data<br />
generated by the new technologies rema<strong>in</strong>s worthless<br />
unless translated <strong>in</strong>to functional <strong>in</strong>formation<br />
and supplied punctually—that is, <strong>in</strong> time to <strong>in</strong>fluence<br />
the decisions be<strong>in</strong>g made. (See sidebar,<br />
“Establish<strong>in</strong>g a Unified Informatics Structure.”)<br />
The extent of the redesign, and the particular<br />
shape the new flows take, depend very much on the<br />
company’s strategy choices. Processes that are newly<br />
<strong>in</strong>dustrialized, but that still follow a traditional<br />
R&D sequence, need to be systematized. In some<br />
cases, however, the traditional R&D value cha<strong>in</strong> will<br />
need to be disrupted. To <strong>in</strong>tegrate chemical genomics<br />
and genetics, for <strong>in</strong>stance, would necessitate a<br />
major restructur<strong>in</strong>g of the value cha<strong>in</strong>. Chemical<br />
genomics <strong>in</strong>troduces a new parallelism, as target<br />
validation and chemistry activities are conducted<br />
simultaneously; the two processes now <strong>in</strong>teract<br />
rather than just <strong>in</strong>terface. And genetics <strong>in</strong>troduces<br />
feedback loops, where late-phase f<strong>in</strong>d<strong>in</strong>gs (such as<br />
genetic <strong>in</strong>formation from the cl<strong>in</strong>ic) feed back <strong>in</strong>to<br />
earlier steps of the value cha<strong>in</strong> (such as diseasegenetics-based<br />
target discovery).<br />
In anticipation of any process redesign, <strong>in</strong>dividual<br />
function heads should be ponder<strong>in</strong>g the cont<strong>in</strong>gencies:<br />
how and when genomics might affect<br />
them, and what actions to take when it does. As one<br />
bottleneck is relieved, another is created: When will<br />
the bottleneck reach their step <strong>in</strong> the cha<strong>in</strong>, and<br />
what will its impact be? What new technologies and
ESTABLISHING A UNIFIED INFORMATICS INFRASTRUCTURE<br />
For a company to extract full value from the new<br />
technologies and the copious data that will emerge<br />
from them—that is, to transform data <strong>in</strong>to knowledge—it<br />
will have to devise a comprehensive <strong>in</strong>formatics<br />
vision and architecture. The vision will articulate<br />
the role of <strong>in</strong>formatics as a potential source of<br />
competitive advantage. The architecture will have<br />
three essential components<br />
First, there must be an optimized <strong>in</strong>formation flow<br />
across the newly <strong>in</strong>dustrialized research and development<br />
value cha<strong>in</strong>. Standards will need to be<br />
established to ensure that data are formatted, organized,<br />
and def<strong>in</strong>ed consistently. Hardware and<br />
applications will have to be l<strong>in</strong>ked and networked<br />
appropriately so that <strong>in</strong>formation can move where it<br />
needs to, feed<strong>in</strong>g subsequent steps <strong>in</strong> the process.<br />
Second, a centralized knowledge management system<br />
is required, to capture and store the data, <strong>in</strong>tegrate<br />
it with external data, and make it available<br />
throughout the company. F<strong>in</strong>ally, powerful analytical<br />
tools will be required, to m<strong>in</strong>e and make sense of the<br />
data—sophisticated algorithms, visualization tools,<br />
and so on.<br />
To develop and <strong>in</strong>tegrate these components, most<br />
companies will have to <strong>in</strong>vest heavily. The costs may<br />
look particularly high <strong>in</strong> relation to traditional costs,<br />
but that is partly because the <strong>in</strong>dustry has generally<br />
under<strong>in</strong>vested <strong>in</strong> IT. (One large biotech company, follow<strong>in</strong>g<br />
its <strong>in</strong>formatics upgrade, reports a threefold<br />
<strong>in</strong>crease <strong>in</strong> its annual IT budget.) To focus the <strong>in</strong>vestment<br />
accurately, a coherent plan is once aga<strong>in</strong><br />
essential. A critical decision is whether to develop<br />
the capabilities <strong>in</strong>-house, outsource to solutions<br />
providers, or purchase and <strong>in</strong>tegrate <strong>in</strong>formatics<br />
packages. The choice or choices made will depend<br />
on such factors as available <strong>in</strong>ternal expertise, the<br />
amount of <strong>in</strong>tegration with legacy <strong>in</strong>formation sys-<br />
tems required, and the availability of reliable <strong>in</strong>tegration<br />
vendors, package suppliers, or solutions<br />
providers.<br />
That last factor may prove particularly difficult to<br />
assess. Who would provide the most reliable and<br />
suitable assistance? Today, no s<strong>in</strong>gle provider of<br />
application software provides all the functionality<br />
needed. The <strong>in</strong>formatics <strong>in</strong>dustry is crowded with<br />
small start-ups offer<strong>in</strong>g niche products; the cumulative<br />
market capitalization of all publicly listed bio<strong>in</strong>formatics<br />
companies scarcely amounts to one-eighth<br />
of GlaxoSmithKl<strong>in</strong>e’s annual R&D budget. And while<br />
larger IT solutions providers are gear<strong>in</strong>g up to serve<br />
the burgeon<strong>in</strong>g needs of this market, they are still <strong>in</strong><br />
the process of develop<strong>in</strong>g <strong>in</strong>ternal life-sciences capabilities.<br />
Given that each prospective solutions<br />
provider is likely to try to make its offer<strong>in</strong>g the centerpiece<br />
of the company's <strong>in</strong>formatics architecture,<br />
and given that multiple solutions will need to be<br />
knitted together, companies would be wise to solicit<br />
<strong>in</strong>dependent, unbiased advice before decid<strong>in</strong>g on<br />
particular vendors.<br />
In addition to manag<strong>in</strong>g all this <strong>in</strong>formatics complexity,<br />
companies will have to deal with a further<br />
challenge if they are to implement the new <strong>in</strong>formation<br />
regime successfully. They will have to f<strong>in</strong>d a way<br />
to resolve the human-resources and organizational<br />
issues that are bound to arise. Talented, experienced<br />
<strong>in</strong>formatics personnel are difficult to come by: how<br />
to f<strong>in</strong>d and keep the right people and how to fit them<br />
<strong>in</strong>to the organizational structure are questions that<br />
companies will need to address more actively and<br />
imag<strong>in</strong>atively than ever. So too the question of how<br />
to change processes and behaviors generally,<br />
throughout the organization, to ensure fullest use of<br />
the new <strong>in</strong>formatics tools—a question exam<strong>in</strong>ed <strong>in</strong><br />
some detail elsewhere <strong>in</strong> this chapter.<br />
51
52<br />
CASE STUDY: REDISCOVERING DISCOVERY RESEARCH<br />
In a global pharmaceutical company, the discovery<br />
division was slowly undergo<strong>in</strong>g a change of character,<br />
from a pre-genomics one of small, <strong>in</strong>dependent<br />
efforts to an up-to-date one of highly def<strong>in</strong>ed and<br />
sequenced processes. The company was anxious to<br />
speed up and optimize this <strong>in</strong>evitable transition. It<br />
had recently created several centers of excellence <strong>in</strong><br />
research, each conta<strong>in</strong><strong>in</strong>g a particular comb<strong>in</strong>ation<br />
of technologies, resources, and expertise, but these<br />
new group<strong>in</strong>gs rema<strong>in</strong>ed <strong>in</strong> need of improved<br />
process flow, both with<strong>in</strong> and between them.<br />
An “<strong>in</strong>dustrializ<strong>in</strong>g” approach was proposed: why<br />
not treat each center of excellence as a factory, and<br />
<strong>in</strong> that way reth<strong>in</strong>k or refashion the discovery program<br />
as a whole?<br />
A factory-based structure imposes a strict discipl<strong>in</strong>e.<br />
Typically, factories have clear, measurable objectives,<br />
with def<strong>in</strong>ed <strong>in</strong>puts and outputs, and specified<br />
resources and roles. Efficiency is closely regulated:<br />
<strong>in</strong>ternal processes and <strong>in</strong>terfaces are optimized for<br />
approaches will be available, and how effectively<br />
will they relieve the bottleneck? The head of development,<br />
for example, should already be contemplat<strong>in</strong>g<br />
the <strong>in</strong>evitable <strong>in</strong>crease <strong>in</strong> demand for cl<strong>in</strong>ical<br />
trial capacity and weigh<strong>in</strong>g the various options,<br />
such as pharmacogenetics, for meet<strong>in</strong>g it.<br />
Reta<strong>in</strong><strong>in</strong>g Balance:<br />
Capacity Plann<strong>in</strong>g and Management (CPM)<br />
After the <strong>in</strong>itial jolt of genomics, supply and demand<br />
should get back <strong>in</strong>to alignment, thanks to the comb<strong>in</strong>ed<br />
forces of resource reallocation and process<br />
redesign. But this restored balance is a precarious<br />
one, and needs careful and regular ma<strong>in</strong>tenance.<br />
That is where capacity plann<strong>in</strong>g and management,<br />
or CPM, can play an <strong>in</strong>valuable role. By enabl<strong>in</strong>g an<br />
organization to keep supply and demand aligned,<br />
CPM also enables it to make rational plans, l<strong>in</strong>ked to<br />
capacities and resources, and thereby to manage<br />
projects with optimal efficiency.<br />
scale, quality, and productivity; external <strong>in</strong>teractions<br />
are monitored regularly for compatibility and cost<br />
effectiveness.<br />
In keep<strong>in</strong>g with this ethos, the company set about<br />
def<strong>in</strong><strong>in</strong>g processes with<strong>in</strong> each potential factory as<br />
clearly as possible. The project team set targets for<br />
<strong>in</strong>puts, outputs, and quality standards; it identified<br />
activities that could be completed <strong>in</strong>side the factory,<br />
as dist<strong>in</strong>ct from those supplied as support from outside;<br />
and it itemized l<strong>in</strong>ks between factories themselves,<br />
between factory and nonfactory research,<br />
and between research units and units outside research<br />
and development.<br />
The result has been a subtly redesigned discovery<br />
division. Processes and functions are now clearly<br />
assigned to specific factories, expectations and<br />
achievements are more transparent than before, and<br />
the <strong>in</strong>teractions throughout discovery research are<br />
now easily tracked from factory to factory, with the<br />
map be<strong>in</strong>g constantly ref<strong>in</strong>ed.<br />
Though well established <strong>in</strong> high-profile corporations<br />
such as General Electric, Hewlett-Packard,<br />
and Cisco, CPM is conspicuously rare <strong>in</strong> biotech<br />
and pharmaceutical companies. For the genomics<br />
revolution to realize anyth<strong>in</strong>g like its full productivity<br />
potential, efficient CPM will be immensely beneficial<br />
if not imperative.<br />
Establish<strong>in</strong>g the New Organization and its<br />
Governance<br />
Implement<strong>in</strong>g the process changes just mentioned<br />
will entail a thorough review of a company’s exist<strong>in</strong>g<br />
hierarchies and procedures. For the process<br />
changes to yield optimal value, changes also need<br />
to be made <strong>in</strong> traditional decision-mak<strong>in</strong>g methods<br />
and <strong>in</strong> organizational structures.<br />
New L<strong>in</strong>kages and Interfaces<br />
To beg<strong>in</strong> with organizational changes. With the<br />
value cha<strong>in</strong> so much altered <strong>in</strong> appearance, and
processes now so different from before, many<br />
disparities and stresses will <strong>in</strong>evitably develop <strong>in</strong><br />
an unaltered managerial system. The old structures<br />
will creak and stra<strong>in</strong> under the unfamiliar<br />
new pressures. To restore congruence, a company<br />
may need to undertake some bold organizational<br />
reshap<strong>in</strong>g—shift<strong>in</strong>g or remov<strong>in</strong>g divisional<br />
borders, reassign<strong>in</strong>g personnel, redistribut<strong>in</strong>g<br />
areas of responsibility, and so on—not just with<strong>in</strong><br />
R&D, but also with<strong>in</strong> the company as a whole and<br />
even beyond, <strong>in</strong> the alliances the company might<br />
enter <strong>in</strong>to.<br />
The R&D Department. Incorporat<strong>in</strong>g the requisite<br />
new capabilities, it goes without say<strong>in</strong>g, represents a<br />
formidable organizational challenge: not only do<br />
they have to mesh with exist<strong>in</strong>g capabilities, they<br />
need to coord<strong>in</strong>ate with one another as well. To<br />
implement <strong>in</strong> silico drug design, for example, it<br />
would be almost essential to provide an <strong>in</strong>formatics<br />
<strong>in</strong>terface between structural biology and chemistry<br />
data. Meanwhile, a comparable reorganization of<br />
personnel has to be undertaken. Biologists and<br />
chemists, for example, can no longer proceed <strong>in</strong><br />
isolation, but must now work alongside each<br />
CASE STUDY: MANAGING CAPACITY—EMPOWERMENT THROUGH CPM<br />
A large pharmaceutical company was fac<strong>in</strong>g a capacity<br />
crisis. Both development and staff<strong>in</strong>g levels were<br />
under pressure, ma<strong>in</strong>ly as a result of productivity<br />
improvements <strong>in</strong> basic research and competition for<br />
scientific talent. As a key part of the remedy, the company<br />
undertook worldwide implementation of capacity<br />
plann<strong>in</strong>g and management—a considerable challenge<br />
for such a complex organization, where demand<br />
was uncerta<strong>in</strong> and resources were not fungible.<br />
Development of CPM had four ma<strong>in</strong> components:<br />
• Quantify<strong>in</strong>g capacity and demand. Appropriate<br />
units of capacity and demand were def<strong>in</strong>ed for<br />
each function. In cl<strong>in</strong>ical departments, for<br />
<strong>in</strong>stance, the typical unit of capacity was def<strong>in</strong>ed<br />
as a team of monitors, coord<strong>in</strong>ators, and support<br />
personnel, and the unit of demand was a study.<br />
• Bus<strong>in</strong>ess processes. To exploit CPM fully and foster<br />
cooperation among departments and project<br />
teams, various new bus<strong>in</strong>ess processes were <strong>in</strong>itiated—most<br />
importantly, the track<strong>in</strong>g and <strong>in</strong>terpretation<br />
of demand and capacity <strong>in</strong>formation, and<br />
the consequent adjustment of timel<strong>in</strong>es and<br />
resource allocation. Appropriate l<strong>in</strong>kages needed<br />
to be made to related functions such as facilities<br />
plann<strong>in</strong>g and human resources.<br />
• Change management. S<strong>in</strong>ce CPM tends to affect<br />
deeply the way an organization operates—publi-<br />
ciz<strong>in</strong>g the relative productivity and workload of<br />
different departments, for example—some managers<br />
react more negatively than others. The company<br />
took steps, both before and dur<strong>in</strong>g the<br />
implementation of CPM, to ease the transition.<br />
The message was constantly re<strong>in</strong>forced—that the<br />
changed regimen was beneficial, essential, and<br />
permanent.<br />
• IT support. With CPM quickly generat<strong>in</strong>g a wealth<br />
of <strong>in</strong>formation, some centralized and some requir<strong>in</strong>g<br />
broad dissem<strong>in</strong>ation, the company recognized<br />
that its CPM <strong>in</strong>itiative needed extra IT support. It<br />
identified suitable vendors with the requisite flexibility<br />
and pharmaceutical experience.<br />
The CPM endeavor has been widely hailed. No<br />
longer is the question “Do we have the capacity to<br />
do these projects?” met with silence. Nowadays, the<br />
CPM team can provide a detailed, graphical depiction<br />
of capacity and demand <strong>in</strong> each department and<br />
overall, and an analysis of the capacity impact of<br />
each project.<br />
Look<strong>in</strong>g ahead, the company expects CPM to contribute<br />
to enhanced revenues, by speed<strong>in</strong>g time to<br />
market and <strong>in</strong>creas<strong>in</strong>g the number of <strong>in</strong>dications per<br />
compound. It also expects to use CPM to improve<br />
portfolio management, and to save costs through<br />
more rational <strong>in</strong>vestment <strong>in</strong> hir<strong>in</strong>g and facilities.<br />
53
54<br />
other—often literally—on collaborative projects or<br />
<strong>in</strong> formal discovery partnerships. And genetics<br />
requires far closer collaboration between basic<br />
research and development than ever before.<br />
One excellent example of reth<strong>in</strong>k<strong>in</strong>g traditional<br />
organizational structure and boundaries is<br />
GlaxoSmithKl<strong>in</strong>e. Alert to the impact of scale, the<br />
company has on the one hand consolidated functions<br />
where scale and coord<strong>in</strong>ation provide a clear<br />
advantage, and on the other, engaged <strong>in</strong> decentraliz<strong>in</strong>g<br />
where size and complexity could prove a drawback.<br />
Specifically, prompted by the scale benefits,<br />
the company decided to organize centrally both the<br />
front end and the back end of R&D (that is, target<br />
discovery and full development). For the steps <strong>in</strong><br />
between, conversely, where the company’s enormous<br />
scale would risk encumber<strong>in</strong>g <strong>in</strong>novation, it<br />
has established smaller, more autonomous centers<br />
of excellence (based on different therapeutic<br />
areas), which attempt to simulate the feel of smaller<br />
biotech companies.<br />
The Entire Corporation. So, enhanced control of data<br />
and <strong>in</strong>creased cross-functionality of personnel are<br />
set to change the structure and tone of the R&D<br />
department. But their sphere of operation is<br />
broader than that. As with the strategic issues discussed<br />
earlier, the company as a whole is implicated.<br />
New l<strong>in</strong>es of communication, and possibly<br />
new cha<strong>in</strong>s of command, will need to be extended<br />
between R&D and other units. In particular, the<br />
relationship between R&D and market<strong>in</strong>g will be<br />
fundamentally transformed: with R&D fac<strong>in</strong>g<br />
greater choice and plac<strong>in</strong>g bigger bets earlier than<br />
ever, commercial <strong>in</strong>put will be crucial. And pharmacogenetics<br />
will require new ways of th<strong>in</strong>k<strong>in</strong>g<br />
about markets, competitors, and customers. (Pharmacogenetics<br />
may also <strong>in</strong>spire new l<strong>in</strong>kages<br />
between pharmaceutical and diagnostic units for<br />
corporations that have both).<br />
Coord<strong>in</strong>at<strong>in</strong>g the commercialization process<br />
between R&D and market<strong>in</strong>g has always been a delicate<br />
balanc<strong>in</strong>g act. Most biopharmaceutical companies<br />
have established product development project<br />
teams to drive the process. These cross-<br />
functional teams are charged with develop<strong>in</strong>g product<br />
strategy and coord<strong>in</strong>at<strong>in</strong>g the various functions<br />
as products progress from R&D <strong>in</strong>to the market.<br />
The job has now become even more complex and<br />
tricky, ow<strong>in</strong>g to larger global efforts, greater <strong>in</strong>formation<br />
flow, more specialized functions, and<br />
<strong>in</strong>creased liaison with global strategic market<strong>in</strong>g<br />
(especially when companies consider the options<br />
for apply<strong>in</strong>g pharmacogenetics to molecules <strong>in</strong><br />
development).<br />
Beyond the Corporation. F<strong>in</strong>ally, new partnership<br />
models need to be considered. Although traditionally<br />
organized partnerships are still appropriate <strong>in</strong><br />
many cases, new and more flexible forms of alliance<br />
will sometimes be required, notably when it comes<br />
to collaborat<strong>in</strong>g with academic or not-for-profit<br />
<strong>in</strong>stitutions and to jo<strong>in</strong><strong>in</strong>g horizontal networks or<br />
consortia.<br />
R&D Governance<br />
One potential source of ga<strong>in</strong> <strong>in</strong> R&D is improved<br />
decision mak<strong>in</strong>g. Consider aga<strong>in</strong> the example at the<br />
start of the value cha<strong>in</strong>—the glut of identified targets<br />
and the need to decide which ones should proceed<br />
to the next phase. Genomics technologies<br />
have created this quandary, but they have also provided<br />
the means for solv<strong>in</strong>g it. Us<strong>in</strong>g new genomic<br />
methods of “aptitude-test<strong>in</strong>g,” decision makers can<br />
confidently preselect the most promis<strong>in</strong>g targets<br />
and forward them downstream.<br />
Even decisions unrelated to genomics technologies<br />
stand to improve, s<strong>in</strong>ce the new genomics regimen<br />
fosters a culture of rigorous selection criteria. In<br />
fact, one of the most important, though perhaps<br />
least noted, benefits of genomics is the way it<br />
encourages a thorough reth<strong>in</strong>k<strong>in</strong>g of decisionmak<strong>in</strong>g<br />
processes. New k<strong>in</strong>ds of data now present<br />
themselves for <strong>in</strong>terpretation, and they enter the<br />
calculations earlier and <strong>in</strong> greater abundance than<br />
the old k<strong>in</strong>ds did. And R&D decision makers have<br />
to take <strong>in</strong>to account a new set of factors too, beyond<br />
the conf<strong>in</strong>es of R&D, <strong>in</strong> order to maximize<br />
value—factors such as market<strong>in</strong>g and IP implications,<br />
for example.
CASE STUDY: REDESIGNING R&D GOVERNANCE<br />
A large drug company recently completed an <strong>in</strong>tensive<br />
three-month project to redesign discovery governance<br />
and is already reap<strong>in</strong>g the benefits. The<br />
company had always placed a high value on the<br />
quality of its scientists and their entrepreneurial<br />
drive. Now, however, it was grow<strong>in</strong>g <strong>in</strong>creas<strong>in</strong>gly dissatisfied<br />
with its exist<strong>in</strong>g system of allocat<strong>in</strong>g<br />
resources: the decision-mak<strong>in</strong>g procedures were<br />
prov<strong>in</strong>g very troublesome to navigate, decision makers<br />
were difficult to identify, communication was<br />
poor, and the decisions themselves often seemed<br />
politically motivated rather than guided by scientific<br />
and commercial promise.<br />
In redesign<strong>in</strong>g the decision-mak<strong>in</strong>g procedures, the<br />
company began with a thorough review of its current<br />
governance process, both as espoused and as practiced<br />
(the two were remarkably dist<strong>in</strong>ct <strong>in</strong> certa<strong>in</strong><br />
<strong>in</strong>stances). Various root causes of undesirable outcomes<br />
were identified: these <strong>in</strong>cluded perverse<br />
<strong>in</strong>centives (that is, <strong>in</strong>centives encourag<strong>in</strong>g behavior<br />
Manag<strong>in</strong>g Organizational Change<br />
With the advent of genomics, R&D personnel suddenly<br />
f<strong>in</strong>d themselves <strong>in</strong> alien territory. As the scientific<br />
methods change, the old <strong>in</strong>st<strong>in</strong>ctive<br />
approaches and behaviors need to change as well.<br />
Among the greatest challenges fac<strong>in</strong>g R&D executives<br />
is manag<strong>in</strong>g the human side of change.<br />
How the Scientist’s Job is Chang<strong>in</strong>g<br />
R&D science is shift<strong>in</strong>g from an arena of experimentation<br />
to one <strong>in</strong>creas<strong>in</strong>gly concerned with theoretical<br />
biology. The challenge is now less how to<br />
get the data than what to do with the data collected.<br />
Scientists who formerly could do their jobs virtually<br />
on their own—conduct their own experiments, and<br />
generate and analyze the data themselves—now<br />
f<strong>in</strong>d they need to collaborate with others who have<br />
more specialized technological skills, <strong>in</strong> areas such<br />
as <strong>in</strong>formatics, robotics, or microfabrication.<br />
Indeed, the scientists of the pre-genomics era are<br />
dest<strong>in</strong>ed to evolve <strong>in</strong>to two k<strong>in</strong>ds of successors:<br />
at odds with company strategy); unclear criteria,<br />
which project champions were dis<strong>in</strong>cl<strong>in</strong>ed to clarify,<br />
let alone follow; and <strong>in</strong>adequate allocation of decision<br />
rights (that is, too vague a def<strong>in</strong>ition of who was<br />
entitled to make which decisions), which often<br />
meant that no decision was made at all.<br />
From the lessons learned, a new governance process<br />
was devised. Not just devised, but activated: by<br />
modify<strong>in</strong>g <strong>in</strong>centives, the company ensured that<br />
practice was now properly aligned with espousal.<br />
The new process is work<strong>in</strong>g well: R&D managers<br />
navigate it easily, and decisions are be<strong>in</strong>g made and<br />
communicated clearly and consistently. It allows scientists<br />
more time to focus on their projects, and it<br />
gives those projects appropriate fund<strong>in</strong>g and management<br />
<strong>in</strong>volvement. It has accord<strong>in</strong>gly won the<br />
confidence of those affected by it, and can claim a<br />
considerable contribution to the marked improvement<br />
<strong>in</strong> productivity that has followed its adoption.<br />
those who <strong>in</strong>terpret the data and devise plans for<br />
exploit<strong>in</strong>g it, and those who cont<strong>in</strong>ue to develop<br />
and optimize the technologies required for generat<strong>in</strong>g<br />
the data. (Companies should be sure to recognize<br />
and reward the latter group for its contributions,<br />
and not relegate it to second-class status.)<br />
All scientists will need to become comfortable with<br />
new ways of work<strong>in</strong>g together—more shar<strong>in</strong>g or collectivist<br />
now, less conducive to solitary <strong>in</strong>itiative.<br />
The scientists of the future will still take responsibility<br />
for their own work, but perhaps will no longer<br />
take the credit for it: that will be ascribed to team<br />
effort.<br />
Manag<strong>in</strong>g the Transition<br />
Chang<strong>in</strong>g from bench-based to <strong>in</strong>formation-based<br />
work <strong>in</strong> this way, and from favor<strong>in</strong>g fairly <strong>in</strong>dependent<br />
endeavors to promot<strong>in</strong>g a more collaborative<br />
ethos, is bound to be awkward or even pa<strong>in</strong>ful for<br />
most of those <strong>in</strong>volved, scientists and managers alike.<br />
55
56<br />
The formidable operational and organizational<br />
changes will entail cultural changes too: <strong>in</strong> fact, the<br />
new processes and structures may prove far less difficult<br />
to establish than new habits and attitudes.<br />
Consider <strong>in</strong>formatics. It is not enough simply to<br />
<strong>in</strong>troduce powerful new IT tools with<strong>in</strong> traditional<br />
silos—with<strong>in</strong> chemistry, for example, where <strong>in</strong> silico<br />
approaches would boost the efficiency of screen<strong>in</strong>g<br />
and optimization. To achieve their full impact,<br />
these IT tools need to be deployed across functions:<br />
to br<strong>in</strong>g biologists and chemists together, to <strong>in</strong>corporate<br />
data from the cl<strong>in</strong>ic <strong>in</strong>to discovery, and so<br />
on. And that will require not just new software, or<br />
even new managerial positions, but new ways of<br />
th<strong>in</strong>k<strong>in</strong>g and of relat<strong>in</strong>g to colleagues.<br />
Some idea of what lies <strong>in</strong> store can be gleaned from<br />
the history of another transformational technology—CAD/CAM<br />
for airplane design. Like<br />
genomics, it promised to transform a costly and<br />
labor-<strong>in</strong>tensive R&D process <strong>in</strong>to a highly automated<br />
and efficient one. After languish<strong>in</strong>g <strong>in</strong> niche<br />
applications <strong>in</strong> the 1970s and ’80s, it f<strong>in</strong>ally proved<br />
its worth <strong>in</strong> the 1990s, when Boe<strong>in</strong>g used it <strong>in</strong><br />
design<strong>in</strong>g the first “paperless” airplane, the Boe<strong>in</strong>g<br />
777. To exploit the technology fully, the company<br />
had to break down departmental barriers and<br />
encourage collaboration across the full range of<br />
functions. Jobs and job responsibilities had to<br />
change. Cherished traditions were called <strong>in</strong>to question.<br />
The company held quarterly meet<strong>in</strong>gs at<br />
which employees could ask questions and voice<br />
their concerns. The transformation was a struggle,<br />
but ultimately a great success: Boe<strong>in</strong>g cont<strong>in</strong>ues to<br />
push the envelope <strong>in</strong> “<strong>in</strong> silico” airplane design.<br />
When pharmaceutical companies convert to genomics,<br />
they will have to temper the discomforts of<br />
transition <strong>in</strong> their turn. And that means engag<strong>in</strong>g<br />
the emotional and behavioral issues—the human<br />
issues—as deeply as the operational ones. 4 Attentive<br />
management of the human issues, which has played<br />
such a prom<strong>in</strong>ent role <strong>in</strong> so many <strong>in</strong>dustries <strong>in</strong> the<br />
throes of reform, is go<strong>in</strong>g to be particularly crucial<br />
when it comes to the massive <strong>in</strong>stitutional changes<br />
demanded by the genomics revolution.<br />
A F<strong>in</strong>al Word<br />
To stake a claim <strong>in</strong> the chang<strong>in</strong>g biopharmaceutical<br />
landscape, let alone feature prom<strong>in</strong>ently with<strong>in</strong> it, a<br />
company will have to make itself radically amenable<br />
to change. Def<strong>in</strong><strong>in</strong>g a strategy is certa<strong>in</strong>ly a step <strong>in</strong><br />
that direction, and <strong>in</strong>itiat<strong>in</strong>g that strategy is certa<strong>in</strong>ly<br />
a gesture of commitment. But wholehearted<br />
commitment is evidenced not by <strong>in</strong>itiat<strong>in</strong>g the strategy<br />
but rather by ma<strong>in</strong>ta<strong>in</strong><strong>in</strong>g it—that is, monitor<strong>in</strong>g<br />
the new structures and procedures constantly,<br />
respond<strong>in</strong>g to shifts <strong>in</strong> external and <strong>in</strong>ternal circumstances,<br />
and <strong>in</strong>troduc<strong>in</strong>g further changes<br />
repeatedly, aggressive or defensive, as new opportunities<br />
or new challenges arise, though always <strong>in</strong> l<strong>in</strong>e<br />
with the controll<strong>in</strong>g wisdom of the strategy itself.<br />
If the unfamiliar outer landscape provokes feel<strong>in</strong>gs<br />
of unease, so too will a company’s <strong>in</strong>ner landscape,<br />
once all the requisite operational and organizational<br />
changes are <strong>in</strong> place. In particular, the <strong>in</strong>crease<br />
<strong>in</strong> cross-functional activity may be disorient<strong>in</strong>g<br />
for some executives of the old school. Many of<br />
the ancient landmarks, tidy borders, and familiar<br />
categories will no longer be there to give them their<br />
bear<strong>in</strong>gs. Short of attempt<strong>in</strong>g a counterrevolution<br />
or withdraw<strong>in</strong>g <strong>in</strong>to obscurity, they will need to<br />
familiarize themselves with the new terra<strong>in</strong> fairly<br />
promptly—and accept it affirmatively, not grudg<strong>in</strong>gly.<br />
Changes <strong>in</strong> attitude will perhaps prove the<br />
most difficult changes of all to br<strong>in</strong>g about, and a<br />
company’s prosperity could be <strong>in</strong> jeopardy if they<br />
fail to take effect.<br />
4. For a fuller discussion of the emotional aspects of change, read The Change Monster, by Jeanie Daniel Duck, published by Crown <strong>in</strong> 2001.
Conclusion<br />
The <strong>in</strong>ternational pharmaceutical <strong>in</strong>dustry is press<strong>in</strong>g<br />
ahead <strong>in</strong> an unexpectedly difficult environment.<br />
Drug companies face unfamiliar frustrations.<br />
On one side, pric<strong>in</strong>g policies are com<strong>in</strong>g <strong>in</strong>creas<strong>in</strong>gly<br />
under threat (witness the recent moves <strong>in</strong> various<br />
U.S. states to restrict access to costlier drugs<br />
for Medicaid patients). From the other side, the<br />
pressure of expectation <strong>in</strong>creases too, with f<strong>in</strong>ancial<br />
analysts cont<strong>in</strong>u<strong>in</strong>g to count on triumphant product<br />
launches and enormous growth. In such an<br />
environment, corporate well-be<strong>in</strong>g, or even survival,<br />
depends on boost<strong>in</strong>g productivity.<br />
It is aga<strong>in</strong>st this background that the genomics revolution<br />
is unfold<strong>in</strong>g. In their quest for improved<br />
productivity, companies should welcome the new<br />
technologies and approaches. Genomics promises<br />
prodigious benefits: it will unlock storehouses of<br />
<strong>in</strong>formation about the work<strong>in</strong>gs of human disease,<br />
and greatly ref<strong>in</strong>e—perhaps even personalize—<br />
health care. More to the po<strong>in</strong>t, it promises to transform<br />
how pharmaceutical research is conducted.<br />
The paradigm will shift from small-scale and<br />
serendipitous to global, <strong>in</strong>dustrialized, and systematic;<br />
and from methodical and compartmentalized<br />
to fluid and cross-functional. The impact on R&D<br />
economics is likely to be tremendous: <strong>in</strong> the best<br />
case, productivity could as much as double.<br />
Look<strong>in</strong>g beyond R&D, genomics and genetics also<br />
promise to transform the way pharmaceutical companies<br />
conduct their bus<strong>in</strong>ess <strong>in</strong> the com<strong>in</strong>g years.<br />
If genetics realizes its potential, for example, treatments<br />
will become more sophisticated, markets may<br />
fragment, and the shape and value of market<strong>in</strong>g<br />
and sales organizations will change dramatically.<br />
The entire system of health care delivery, already <strong>in</strong><br />
flux, will complete its metamorphosis.<br />
The offer that genomics and genetics are hold<strong>in</strong>g<br />
out is really an offer that companies cannot refuse.<br />
Companies that fail to accept the offer adequately<br />
will f<strong>in</strong>d themselves not simply uncompetitive but<br />
possibly right out of contention. There is nowhere<br />
to hide, and certa<strong>in</strong>ly no safety <strong>in</strong> <strong>in</strong>action. To shun<br />
the promise of pharmacogenomics out of a fear of<br />
market fragmentation, for <strong>in</strong>stance, is not to avert<br />
the fragmentation but simply to cede the market to<br />
one’s rivals.<br />
Embrac<strong>in</strong>g the revolution appropriately will require<br />
both boldness and f<strong>in</strong>esse: managers will have to<br />
make major strategic decisions, and to implement<br />
them will have to radically reconfigure operations.<br />
The decisions take careful analysis to get right, and<br />
the operational hurdles need nimble negotiation to<br />
surmount. It all adds up to a formidable but by no<br />
means impossible task. And for companies that do<br />
it well, the rewards will be handsome.<br />
The opportunities are unprecedented. So are the<br />
challenges. The shrewd company will be one that<br />
rema<strong>in</strong>s responsive to both, as it tries to keep its<br />
head and to prosper <strong>in</strong> these revolutionary times.<br />
57
Methodology<br />
The many diverse studies on drug development<br />
have reached diverse conclusions. They have put<br />
the current price tag at anywhere between $350 million<br />
and $800 million per drug (all underestimates,<br />
<strong>in</strong> our view: our own calculation is $880 million).<br />
Not surpris<strong>in</strong>gly, when it comes to the likely impact<br />
of genomics and genetics on the economics of drug<br />
development, op<strong>in</strong>ions diverge aga<strong>in</strong>.<br />
For our study, we conducted an extensive program<br />
of discussions <strong>in</strong> an effort to compile accurate figures<br />
for all the ma<strong>in</strong> activities <strong>in</strong> the R&D process,<br />
both pre- and post-genomics. The result is a robust<br />
bottom-up model of R&D, based on the time, cost,<br />
and likely success rate for each step of the value<br />
cha<strong>in</strong>.<br />
Our model goes beyond exist<strong>in</strong>g models of R&D <strong>in</strong><br />
three important ways:<br />
• It is the product of primary research. Other estimates<br />
have tended to build on the f<strong>in</strong>d<strong>in</strong>gs of earlier<br />
work; our model also draws on more than 100<br />
discussions at nearly 50 companies and academic<br />
<strong>in</strong>stitutions.<br />
• It analyzes the discovery phase more closely than<br />
has previously been possible. Earlier models typically<br />
assigned a conjectural figure to represent<br />
the sunk cost of discovery, but the art of discovery<br />
is becom<strong>in</strong>g <strong>in</strong>dustrialized, and we have duly<br />
been able to model its activities more scientifically.<br />
A more detailed understand<strong>in</strong>g of the economics<br />
of discovery has resulted, and that <strong>in</strong> turn<br />
has allowed us to more accurately quantify the<br />
impact of genomics on R&D. (See the chart on<br />
page 60 for technologies modeled.)<br />
• It is activity-based and flexible. The numbers<br />
cited <strong>in</strong> this report represent an average drug,<br />
unless noted otherwise, but <strong>in</strong> our research we<br />
ranged far wider than that, and assessed each step<br />
of the value cha<strong>in</strong> under a range of circumstances.<br />
So the model allows for scenario build<strong>in</strong>g<br />
and sensitivity analysis, as well as enables us to<br />
tailor <strong>in</strong>puts to match the unique circumstances<br />
of <strong>in</strong>dividual companies.<br />
As already mentioned, all numbers cited <strong>in</strong> the text<br />
are for an average drug. In any <strong>in</strong>dividual case, cost<br />
and time will vary accord<strong>in</strong>g to factors such as therapeutic<br />
area and target class.<br />
All numbers cited <strong>in</strong> the text are for a relevant<br />
drug, that is, one to which the technology under<br />
discussion could be applied. Various technologies<br />
may not apply to all targets or drugs. Where specific<br />
limitations are likely to be a significant factor, that<br />
is po<strong>in</strong>ted out.<br />
When we discuss the “value added” to a drug, we<br />
are referr<strong>in</strong>g to its net present value: the current<br />
59
60<br />
value of expected profits, discounted by a representative<br />
hurdle rate, less the cost of develop<strong>in</strong>g the<br />
drug. For the average drug, we assume peak annual<br />
sales of $500 million and 11 years to patent expiration.<br />
Also, our numbers reflect the fact that R&D<br />
Post-genomics Pre-genomics<br />
GENOMICS TECHNOLOGY ASSUMPTIONS<br />
Biology<br />
Target ID Target Validation<br />
Target identification<br />
• Limited numbers of genes<br />
• Molecular biology and biochemistry<br />
techniques<br />
Target validation<br />
• Cell and tissue studies<br />
• Mouse knockouts<br />
Target identification<br />
• Large numbers of genes<br />
• Industrialized techniques<br />
(e.g., gene chip expression)<br />
• Bio<strong>in</strong>formatics<br />
(e.g., database searches for homologies)<br />
Target validation<br />
• Cell and tissue studies<br />
• Mouse knockouts<br />
SOURCES: BCG analysis; <strong>in</strong>dustry <strong>in</strong>terviews; scientific literature.<br />
dollars saved are pretax dollars; these “saved” dollars<br />
are subject to taxation, which expla<strong>in</strong>s <strong>in</strong> part<br />
why the expressed NPV calculations can be lower<br />
than the R&D sav<strong>in</strong>gs.<br />
Chemistry<br />
Screen<strong>in</strong>g Optimization<br />
Screen<strong>in</strong>g<br />
• Parallel synthesis for library design<br />
• Assay development for high-throughput<br />
screen<strong>in</strong>g (HTS)<br />
• HTS<br />
Chemical optimization<br />
• Bench synthesis<br />
• Parallel synthesis<br />
Screen<strong>in</strong>g<br />
• Structural biology (target structure)<br />
• SAR profil<strong>in</strong>g of library<br />
• Assay development for LTS1 • Virtual screen<strong>in</strong>g and LTS1 Chemical optimization<br />
• In silico-supported bench synthesis<br />
• In silico early ADME/tox<br />
1 LTS = LOw-throughput screen<strong>in</strong>g; generally more <strong>in</strong>formation-rich, but less standardized, assays that cannot be used <strong>in</strong> HTS.<br />
Development<br />
Pre-cl<strong>in</strong>ical Cl<strong>in</strong>ical<br />
Precl<strong>in</strong>ical (ADME/tox)<br />
• Animal test<strong>in</strong>g<br />
Cl<strong>in</strong>ical<br />
• Patient trials<br />
Precl<strong>in</strong>ical (ADME/tox)<br />
• Animal test<strong>in</strong>g<br />
• In silico ADME/tox<br />
• In vitro toxicology<br />
• Surrogate markers<br />
Cl<strong>in</strong>ical<br />
• Patient trials<br />
• Surrogate markers
The Boston Consult<strong>in</strong>g Group publishes other reports<br />
that may be of <strong>in</strong>terest to senior health care executives.<br />
Recent examples <strong>in</strong>clude:<br />
The Pharmaceutical Industry <strong>in</strong>to Its Second Century:<br />
From Serendipity to Strategy<br />
A report by The Boston Consult<strong>in</strong>g Group, January 1999<br />
Ensur<strong>in</strong>g Cost-Effective Access to Innovative Pharmaceuticals:<br />
Do Market Interventions Work?<br />
A report by The Boston Consult<strong>in</strong>g Group and Warner-Lambert,<br />
April 1999<br />
Patients, Physicians, and the Internet: Myth, Reality, and Implications<br />
A report by The Boston Consult<strong>in</strong>g Group, January 2001<br />
For a complete list of BCG publications and <strong>in</strong>formation about<br />
how to obta<strong>in</strong> copies, please visit our Web site at www.bcg.com.<br />
Vital Signs: The Impact of E-Health on Patients and Physicians<br />
A report by The Boston Consult<strong>in</strong>g Group, February 2001<br />
Vital Signs Update: The E-Health Patient Paradox<br />
A BCG Focus by The Boston Consult<strong>in</strong>g Group, May 2001<br />
Vital Signs Update: Doctors Say E-Health Delivers<br />
A BCG Focus by The Boston Consult<strong>in</strong>g Group, September 2001<br />
In addition, BCG’s Health Care practice publishes Opportunities for Action<br />
<strong>in</strong> Health Care, essays on topical issues for senior executives.
Amsterdam<br />
Athens<br />
Atlanta<br />
Auckland<br />
Bangkok<br />
Berl<strong>in</strong><br />
Boston<br />
Brussels<br />
Budapest<br />
Buenos Aires<br />
Chicago<br />
Cologne<br />
Copenhagen<br />
Dallas<br />
Düsseldorf<br />
Frankfurt<br />
Hamburg<br />
Hels<strong>in</strong>ki<br />
BCG<br />
www.bcg.com<br />
Hong Kong<br />
Istanbul<br />
Jakarta<br />
Kuala Lumpur<br />
Lisbon<br />
London<br />
Los Angeles<br />
Madrid<br />
Melbourne<br />
Mexico City<br />
Milan<br />
Monterrey<br />
Moscow<br />
Mumbai<br />
Munich<br />
New Delhi<br />
New York<br />
Oslo<br />
Paris<br />
San Francisco<br />
São Paulo<br />
Seoul<br />
Shanghai<br />
S<strong>in</strong>gapore<br />
Stockholm<br />
Stuttgart<br />
Sydney<br />
Tokyo<br />
Toronto<br />
Vienna<br />
Warsaw<br />
Wash<strong>in</strong>gton<br />
Zürich