ing fun at the promise of the “beautiful game”
in the United States can seem uncomfortably
apt when applied to genomic medicine. It’s
now been 10 years since humans deciphered
the digital code that defi nes us as a species.
Although it may be hard to overestimate the
signifi cance of that achievement, it is easy to
misconstrue its meaning and promise. Peo-
ple argue about whether mapping the human
genome was worth the investment (1–3). With
global funding for genomics approaching $3
billion/year (4), some wonder what became
of all the genomic medicine we were prom-
ised (5). It thus seems an appropriate time to
take stock of whence the real benefi ts from
genomic research may come and how best to
attain a future in which genomics improves
Recent methodological progress in
genomics has been breathtaking. We now
regularly assay genomes at millions of loci
(6), and routine whole-genome sequencing
may soon be a reality (7). If this trajectory
continues, genomic research will illuminate
fundamental mechanisms of human disease
with a reasonable expectation of practical
results (8). But claims of near-term applica-
tions are too often unrealistic and ultimately
counterproductive. From the South Sea and
dot-com “bubbles” to the ongoing housing
market crisis, the world has seen its share of
infl ated expectations and attendant dangers.
Science is immune to neither.
If we fail to evaluate the considerable
promise of genomics through a realistic lens,
exaggerated expectations will undermine its
legitimacy (9), threaten its sustainability, and
result in misallocation of resources. Fueling
unrealistic expectations for predictive genetic
testing and uncritical translation of discov-
eries may also distract our gaze from other
promising approaches to preventing disease
and improving health.
occer is the sport of the future
in America … and it always will
be.” This oft-quoted epithet pok-
Impediments and Hyperbole
Substantial impediments to
realizing many of the claims
most frequently heard include
The problem of clinical utility
and relative risk. The numerous
genetic variants that mediate dis-
ease risk typically confer woe-
fully low relative risks (i.e., com-
pared with the much more mean-
ingful absolute risk) and are
thus meager in their predictive
power (10). Their applicability to
patient care shows little promise;
studies (11–14) demonstrate that
even combining dozens of risk
markers provides little clinically
meaningful information. In the
public health realm, the prospect
of effectively stratifying popula-
tions as high or low risk, thereby
guiding screening, is equally
dismal. Given the multifactorial
nature of common diseases and
the weak predictive properties of
genetic-risk alleles, the probabil-
ity of misclassifying individuals
as high or low risk is likely too
great to make such an approach
feasible in the general population
for guiding such things as mam-
mography or colorectal cancer
The illusion of parsing risk.
For common diseases, by defi -
nition, we are all at high levels
of absolute risk. In this setting,
defi ning precise relative risk on
the basis of individuals’ genetic
information is less meaningful;
interventions that lower risk will
be useful to everyone, regardless
of their relative risk. And for rare
diseases, shifting an individual’s
risk from an already low level
may not be very clinically mean-
ingful. For example, the life-
time risk for an individual in the
United States to develop Crohn’s
disease is about 1/1000. How
helpful is it for clinicians and
patients if that risk shifts to 1/500
The difficulty of changing behav-
iors. The idea that genetic information
will promote a healthy life-style has
emerged as a dominant claim by those
who promote genomic medicine (16,
17). However, there is little evidence
that simply telling someone they are
at a genetically increased risk for heart
disease or diabetes, for example, leads
to lasting benefi cial changes in diet
or exercise habits (18, 19).
Altering environments is
as a more effective way
of changing those coun-
that contribute most to poor
health in high-income coun-
tries—namely, diet, sedentary behavior,
smoking, and alcohol use (20).
The paradox of risk information.
Even if, despite evidence to the con-
trary, knowledge of one’s genetic status
drives behavior change, another prob-
lem emerges: for everyone identifi ed
at increased risk of a malady, there will
be an equal number at decreased risk.
Thus, if genetic information were actu-
ally found to be uniquely powerful in
changing behavior, it could well pro-
mote counterproductive behaviors.
The translation of science into the
clinic is inherently messy. The public,
researchers, and clinicians frequently
fail to appreciate that the history of med-
icine is strewn with ideas once thought
promising that did not pan out when
scrutinized through the lens of evidence-
based medicine (21). Hormone replace-
ment therapy, prostate-specifi c antigen
screening, peri–myocardial infarction
lidocaine, and many other good ideas,
when prematurely implemented, cre-
ated bubbles of expectation and invest-
ment, leaving sponsors disappointed and
patients ill-served when reality did not
live up to theoretical promise.
Given these hurdles to practical
application, why has genomics been the
recipient of such hyperbole? Impatience
for practical applications from genetic
advances is understandable. To be sure,
there is much room for improvement in
modern medicine: Screening programs
Defl ating the Genomic Bubble
James P. Evans, 1 * Eric M. Meslin, 2 Theresa M. Marteau, 3 Timothy Caulfi eld 4
CREDIT: P. HUEY/SCIENCE
*Author for correspondence. E-mail: email@example.com
Unrealistic expectations and uncritical
translation of genetic discoveries may
undermine other promising approaches to
preventing disease and improving health.
1Departments of Genetics and Medicine, University of
North Carolina at Chapel Hill, Chapel Hill, NC 27599, USA.
2Center for Bioethics, Indiana University, Indianapolis, IN
46202, USA. 3Health Psychology Section, King’s College
London, London SE1 9RT, UK. 4Faculty of Law and School
of Public Health, Health Law Institute, University of Alberta,
Edmonton, Alberta T6G 2H5, Canada.
www.sciencemag.org SCIENCE VOL 331 18 FEBRUARY 2011
Published by AAAS
on February 18, 2011
18 FEBRUARY 2011 VOL 331 SCIENCE www.sciencemag.org Download full-text
are inherently ineffi cient given the need to
test entire populations, drugs have widely
variable effi cacy, and diseases strike capri-
ciously. But there are other drivers of infl ated
expectations ( 22). Researchers gain fund-
ing, jobs, and fame, while pressure to com-
mercialize their work adds fuel to the ten-
dency to oversell ( 23, 24). In a distressing
way, biomedical research is often viewed by
governments as primarily an engine of eco-
nomic growth ( 25) and, only secondarily, as
an engine of scientifi c and medical progress.
Further pressure results from retail market-
ing of genomic information, such as direct-
to-consumer genetic testing. As boundar-
ies between private and public efforts erode,
academic endeavors are increasingly sub-
ject to market forces that demand quick pay-
offs. Finally, the press plays an obvious role
in creating unrealistic hopes ( 24). Collec-
tively, these factors contribute to heightened
expectations; left unchallenged, they take on
a momentum that is hard to unseat.
Harold Varmus observed that the full poten-
tial of a DNA-based transformation of med-
icine will be realized only over the course
of decades ( 8). We agree; the true promise
of genomics is to help lay bare the mecha-
nisms of human disease. Genes responsible
for most Mendelian disorders will soon be
identifi ed. Genome-wide association studies
are illuminating loci that contribute to com-
mon disease, and novel drug targets are being
identified that will ultimately lead to new
therapies. But the timeline for translation of
such discoveries will be long.
Pharmacogenomics (PGx; the study
of influence of genetic variation on drug
response) may represent a near-term payoff
of genomic research for carefully selected
treatments and could enhance the safety and
utility of treatments used for serious disor-
ders ( 26, 27). But it is unrealistic to expect
PGx to revolutionize the use of all (or perhaps
even most) drugs, given that much variability
in effi cacy is not genetically determined ( 28).
Indeed, the most powerful predictor of drug
effi cacy is whether a patient takes the drug,
highlighting the importance of human behav-
ior in health outcomes.
If properly harnessed and based on evi-
dence, appropriate risk assessment could aid
in clinical decision-making ( 29). The ability
to make diagnoses, especially for disorders
that result from disruption of a single gene,
will provide tangible benefi t in the near term.
Enhanced diagnostic capacity promises to
spare both anxiety and money, ending the
cruel “diagnostic odyssey” of families who
go for years without a defi nitive diagnosis.
But we should not overestimate the value of
diagnosis or risk stratifi cation. Without effec-
tive interventions, a diagnosis is only a dimly
realized, partially fulfi lled hope.
Couples will be empowered to make
informed reproductive decisions, as pre-
conceptual screening, augmented by robust
genomic analysis, allows them to learn
whether they are carriers of disease-related
genes. Newborn screening will also benefi t
as medically actionable conditions are iden-
tifi ed. Such advances hold great promise if
the information so gathered is useful, cost-
effective, and welcome (since not all parents
may welcome such information).
So how do we avoid infl ating an unsus-
tainable genomic bubble but still realize the
true—and considerable—promise of the
“genomic revolution”? Solutions range from
the political to the personal, from short term
to long term. We offer a short list of recom-
mendations as a starting point for debate,
aimed at defl ating the genomic bubble and
realizing the fi eld’s long-term promise:
1. Reevaluate funding priorities. A sober
assessment of disease etiology suggests that
funding priorities may be mismatched to the
potential for practical benefi t. Much morbid-
ity and premature mortality in high-income
countries results from smoking, sedentary
behavior, and excessive food and alcohol
consumption ( 30, 31). It is likely that com-
mon diseases arising from these behaviors
can be reduced by behavioral change ( 32,
33), but our knowledge of how to effect
such change across populations is limited.
Yet, U.S. National Institutes of Health and
Department of Energy spending on genom-
ics vastly exceeds the budget for behavioral
and social science research ( 4, 34). Given
that even a small improvement in our ability
to alter behaviors could yield major benefi ts,
we suggest a reappraisal of the apportioning
of funds to promote the promise of improved
2. Foster a realistic understanding among
the scientific community, the media, and
the public of the incremental nature of sci-
ence and need for statistical rigor. Scientists
can start this process by making responsible
claims and by advocating that reporters and
editors do the same.
3. Maintain focus on developing high-
quality evidence before integrating good
ideas into medical practice. Develop novel
ways of assessing evidence so as not to delay
implementing promising modalities.
We believe that genomic discovery and
resultant applications will provide great
benefi ts to human health. Ours is not a call
to gut existing research or too rigidly tie
funding to the degree of disease burden.
Indeed, the nature of scientifi c progress is
arguably not optimized by a rigid alloca-
tion of resources to purely practical need.
But failing a (desirable but unlikely) mas-
sive expansion of total funding for all types
of research, a realistic view of the promise
of genomics and an appropriate prioritiza-
tion of research funding are vital to realiz-
ing that future. The pursuit of our common
goal—improved human health—demands
that we take a hard look at disease causation
and order our priorities accordingly.
References and Notes
1. A. Pollack, New York Times, 14 June 2010, p. B5.
2. P. Kraft, D. J. Hunter, N. Engl. J. Med. 360, 1701 (2009).
3. J. N. Hirschhorn, N. Engl. J. Med. 360, 1699 (2009).
4. J. R. Pohlhaus, R. M. Cook-Deegan, BMC Genomics 9,
5. F. S. Collins, N. Engl. J. Med. 341, 28 (1999).
6. Editors, Nature 464, 649 (2010).
7. E. Pettersson, J. Lundeberg, A. Ahmadian, Genomics 93,
8. H. Varmus, The Art and Politics of Science (Norton,
New York, 2009).
9. A. Petersen, Monash Bioeth. Rev. 28, 5.1 (2009).
10. L. A. Hindorff et al., Proc. Natl. Acad. Sci. U.S.A. 106,
11. S. Wacholder et al., N. Engl. J. Med. 362, 986 (2010).
12. S. Kathiresan et al., N. Engl. J. Med. 358, 1240 (2008).
13. V. Lyssenko et al., N. Engl. J. Med. 359, 2220 (2008).
14. N. L. Pedersen, JAMA 303, 1864 (2010).
15. M. S. Pepe, H. Janes, G. Longton, W. Leisenring,
P. Newcomb, Am. J. Epidemiol. 159, 882 (2004).
16. F. Collins, Nature 464, 674 (2010).
17. Navigenics Inc., Genetics and health; http://www.
18. J. B. McClure, Am. J. Prev. Med. 22, 200 (2002).
19. T. M. Marteau et al., Cochrane Database Syst. Rev. 10,
20. WHO Regional Committee for Europe, “Resolution:
Behaviour change strategies & health: The role of health
systems” (World Health Organization, Geneva, 2008).
21. K. Goodman, Ethics and Evidence-Based Medicine:
Fallibility and Responsibility in Clinical Science
(Cambridge Univ. Press, Cambridge, UK, 2003).
22. J. Kimmelman, Gene Transfer and the Ethics of First-in-
Human Research: Lost in Translation (Cambridge Univ.
Press, Cambridge, UK, 2009).
23. L. P. Garrison Jr. et al., Drug Metab. Rev. 40, 377 (2008).
24. T. Caulfi eld, PLoS. Med. 1, e38 (2004).
25. B. Carlsson, Z. J. Acs, D. B. Audretsch, P. Braunerhjelm,
Ind. Corp. Change 18, 1193 (2009).
26. S. Mallal et al., N. Engl. J. Med. 358, 568 (2008).
27. M. P. Goetz et al., Breast Cancer Res. Treat. 101, 113
28. G. Tucker, BMJ 329, 4 (2004).
29. A. J. Vickers, E. Basch, M. W. Kattan, Ann. Intern. Med.
149, 200 (2008).
30. R. L. Keeney, Oper. Res. 56, 1335 (2008).
31. W. C. Willett, Science 296, 695 (2002).
32. W. C. Knowler et al., Lancet 374, 1677 (2009).
33. R. Doll, R. Peto, J. Boreham, I. Sutherland, BMJ 328,
34. H. J. Silver, A. L. Sharpe, H. Kelly, P. Kobor, G. Sroufe, in
AAAS Report 33: Research and Development 2009
(AAAS, Washington, DC, 2008), chap. 19; www.aaas.org/
35. The authors thank R. Whiting and Genome Alberta.
Published by AAAS
on February 18, 2011