Correspondence
Will Genetics Revolutionize Medicine?
N Engl J Med 2000; 343:1496-1498November 16, 2000
- Article
To the Editor:
We find two major flaws in the Sounding Board article by Holtzman and Marteau (July 13 issue).1 First, if the authors find little clinical value in the genetics revolution, then they must find no value in taking a family history — a conclusion surely at odds with the experience of Hippocrates2 and 95 subsequent generations of physicians. Holtzman and Marteau rightly focus on the genetics of common diseases, but they restrict their analysis to the diagnostic use of isolated genetic information. They do not consider the full Bayesian framework on which modern methods of diagnosis are based. Genetic information will be an improvement over the family history and, like the family history, will be used largely to set prior probabilities, not to make definitive diagnoses.
Second, Holtzman and Marteau would have us ignore another Hippocratic precept: making a precise diagnosis.2 A precise diagnosis, whether genetic or microbial, is essential to precise treatment. The authors' claim that “no interventions based on the identification of disease-related genes have yet proved safe and effective” is premature. This would be akin to saying, in 1900, that no interventions based on the identification of disease-related microbes have yet proved safe and effective. It takes time for clinical trials to catch up with basic science, and the revolution has only just begun.
2 ReferencesJohn G. Sotos, M.D.
Hugh Y. Rienhoff, Jr., M.D.
DNA Sciences, Mountain View, CA 940431
Holtzman NA ,Marteau TM . Will genetics revolutionize medicine? N Engl J Med 2000;343:141-144
Full Text | Web of Science | Medline2
Hippocrates. The genuine works of Hippocrates: translated from the Greek by Francis Adams. Adams F, trans. Huntington, N.Y.: R.E. Krieger, 1972.
To the Editor:
You are to be commended for inviting discussion of whether progress in genetics will constitute a revolution in medicine. Although we applaud the dialogue, we take issue with the view expressed by Holtzman and Marteau that genetics will not revolutionize medicine, for several reasons.
First, the traditional concept of penetrance, which is at the heart of the authors' argument, is at home only in an impoverished context in which detailed knowledge of genomics and its environmental influences is lacking. It is this dearth of insight that evolutionary progress in functional genomics and bioinformatics promises to ameliorate.
Second, the Human Genome Project, even at this relatively early juncture, has already yielded some arguably revolutionary benefits. Hereditary hemochromatosis, which the authors mention but do not discuss, is a prime example. In the past, those who had this disease would most often be given the diagnosis, if at all, only after severe secondary complications, such as end-stage liver disease, had developed. Thus, simple treatment through phlebotomy, though completely effective if started before the onset of tissue injury, was often instituted only after patients with hereditary hemochromatosis had advanced well beyond the fully treatable stage. Although itself an example of a polygenic disease with a critical environmental interaction, and once thought to be rare, hereditary hemochromatosis is now known to be quite common. The most common forms of the disease can now be identified by means of genetic testing and effectively managed decades before the onset of irreversible tissue injury.
Although we understand that Holtzman and Marteau's purpose is to temper irrational exuberance about the promise of genetic medicine, this very exuberance may itself be evidence that the revolution has already begun.
Geoffrey D. Block, M.D.
University of Pittsburgh School of Medicine, Pittsburgh, PA 15213Mark P. Aulisio, Ph.D.
Case Western Reserve University Center for Biomedical Ethics, Cleveland, OH 44109-1998To the Editor:
Holtzman and Marteau1 use three arguments against popular and scientific claims that genetics will revolutionize medicine: the incomplete penetrance of genotypes for common diseases; the low proportion of common diseases attributed to genotypes; and our limited ability to tailor interventions to genotypes. Because most common diseases, if not all, result from the interaction among several genes and environmental factors (such as diet, drugs, and infectious agents), the same arguments can be used to show how interacting factors can increase the penetrance of genotypes at single loci, refine our concept of attributable risk, and target medical and behavioral interventions.
First, although incomplete penetrance of a single gene leads to a low positive predictive value, we have shown previously that the combination of a genotype and an interacting cofactor (another gene or exposure) leads to higher penetrance than either one alone.2 The positive predictive value in the subgroup with even one interacting cofactor can markedly increase (Table 1Table 1
Positive Predictive Value of Tests for Susceptibility-Conferring Genotypes in the Presence of an Interacting Cofactor, According to the Frequency of the Cofactor and Relative Risk Associated with It, for a Genotype Relative Risk of 2 and a Disease with a Lifetime Risk of 5 Percent.). The overall dilutional effect on penetrance arises because not everyone with a genotype has one or more interacting cofactors. Incomplete penetrance of single genes could increase if multiple gene panels were used or if subjects were stratified according to exposure status.Second, whereas genotypes at single loci have low attributable fraction, common diseases are etiologically heterogeneous. A genotype that accounts for 5 percent of a disease with 5 percent lifetime risk translates to 2.5 per 1000 persons affected by a common condition due to a genotype. Such a disease is as common as hereditary hemochromatosis3 and more frequent than all known mendelian conditions. Such a calculation helps us view a common disease as made up of multiple conditions with different sets of genetic–environmental interactions.
Third, the authors argue that there are only a few conditions for which interventions could prevent disease in affected people after genetic testing. More information will emerge on how to tailor environmental interventions (with drugs or behavioral and dietary interventions) to particular genotypes. Although it is prudent to caution against exaggerated claims about genetic technology, Holtzman and Marteau evoke outdated concepts of nature versus nurture by stating that “differences in . . . environment account for much larger proportions of disease than genetic differences.” Coincidentally, in the same issue of the Journal, a heritability study concluded that most cancers are “environmental” rather than “genetic.”4 The accompanying editorial had a simple message that is applicable to all diseases: “it is time to drop . . . debate over nature versus nurture in favor of efforts to exploit every opportunity to identify and manipulate both environmental and genetic risk factors to improve the control of cancer.”5
5 ReferencesMuin J. Khoury, M.D., Ph.D.
Centers for Disease Control and Prevention, Atlanta, GA 303411
Holtzman NA ,Marteau TM . Will genetics revolutionize medicine? N Engl J Med 2000;343:141-144
Full Text | Web of Science | Medline2
Khoury MJ ,Wagener DK . An epidemiological evaluation of the use of genetics to improve the predictive value of disease risk factors. Am J Hum Genet 1995;56:835-844[Erratum, J Hum Genet 1996;58:253.]
Web of Science | Medline3
Cogswell ME ,McDonnell SM ,Khoury MJ ,Franks AL ,Burke W ,Brittenham G . Iron overload, public health, and genetics: evaluating the evidence for hemochromatosis screening. Ann Intern Med 1998;129:971-979
Web of Science | Medline4
Lichtenstein P ,Holm NV ,Verkasalo PK , et al. Environmental and heritable factors in the causation of cancer: analyses of cohorts of twins from Sweden, Denmark, and Finland. N Engl J Med 2000;343:78-85
Full Text | Web of Science | Medline5
Hoover RN . Cancer -- nature, nurture, or both. N Engl J Med 2000;343:135-136
Full Text | Web of Science | Medline
To the Editor:
Using simple epidemiologic principles, Holtzman and Marteau demonstrate that under the vast majority of conditions, common genotypes associated with common human diseases will have little predictive power and that individual genotypes will account for only a small fraction of the associated disease within a population.
Despite these findings, however, we disagree with their conclusion that common genotypes with low penetrance will not have an important role in clinical medicine. Although their conclusion seems to be supported by conventional measures of effect, such as positive predictive value and population attributable risk, we would like to point out a simple new index for use in interpreting complex genetic associations, which clearly reveals how these genetic associations can be effectively incorporated into clinical medicine.
The number needed for genetic effect is the number of persons with a disease-associated genotype among whom one event can be expected to occur as the result of a genetic effect.1 It is calculated as the reciprocal of the increased absolute risk of disease that results from having the disease-associated genotype. For example, if a genotype triples the lifetime risk of colon cancer from 5 percent to 15 percent, the number needed for genetic effect is 10, or 1÷(0.15–0.05), indicating that only 1 of every 10 persons with this genotype will develop colon cancer as the result of a genetic effect.
When the number needed for genetic effect is used, it becomes clear that common genotypes associated with common human diseases will have a quantitatively large and clinically meaningful effect under two conditions (Table 1Table 1
Number Needed for Genetic Effect for Genotypes of Various Frequencies, Conferring Various Relative Risks for Common Diseases of Various Prevalences.). The first condition is a very rare exposure, such as a certain combination of genotypes and environmental factors. Identifying and eliminating the environmental factors that interact with various genotypes to produce a substantially increased risk of disease can result in individualized primary-prevention strategies based on a person's genotype.The second condition is a very prevalent outcome, such as a response (or susceptibility) to a particular medication or intervention. Thus, it may be possible to choose medications to treat certain disorders on the basis of the genotype, leading to individualized secondary-prevention strategies.
Unlike Holtzman and Marteau, we believe that genetics can revolutionize clinical medicine by allowing us to choose individualized primary- and secondary-prevention strategies according to a patient's genotype. The quantitative rationale for this revolution can be expressed as the number needed for genetic effect.
1 ReferencesBrian A. Ference, M.D.
Manish S. Chauhan, M.D.
Seigo Izumo, M.D.
Beth Israel Deaconess Medical Center, Boston, MA 02215Author/Editor Response
The authors reply:
To the Editor: Neither we nor our critics defined a revolution in medicine. We mean a paradigm shift in theory or practice. Sotos and Rienhoff's plea for “precise diagnosis” epitomizes the current paradigm. In most of those who will have common disorders, the interaction of genetic, environmental, and behavioral factors makes the quest for precise diagnosis illusive. (Khoury's charge that we “evoke outdated concepts of nature versus nurture” is simply wrong. We did not suggest the environment acts independently of inherited factors.)
Khoury implies, and Block and Aulisio say directly, that determining the genotype in asymptomatic persons with respect to hereditary hemochromatosis would improve the outcome of the disease. Screening young men for elevated transferrin saturation, rather than genotyping, will identify most, if not all, cases of incipient hemochromatosis in time to lead to cure by periodic phlebotomy, possibly with less mislabeling and unnecessary treatment than with genotyping.
Sotos and Rienhoff maintain that we have not considered “the full Bayesian framework”; Khoury and Ference et al. take Bayes's theorem a few steps further. Khoury's table showing that “the combination of a genotype and an interacting cofactor . . . leads to higher penetrance than either one alone” is indisputable. What Khoury fails to recognize is biomedical researchers' inability to identify the various genotypes and cofactors when more than two or three are involved, as is often the case.
Nor do we dispute the calculations of Ference et al., just their relevance. The first condition for which they point out that the number needed for genetic effect will be low, “very rare exposure, such as a certain combination of genotypes and environmental factors,” is just that — very rare. Genotype frequencies of 0.2 or more that confer relative risks of more than 5 for common diseases also yield low numbers needed for genetic effect but are unlikely to be encountered because of selection against them. The second condition of Ference et al. is “a very prevalent outcome, such as a response . . . to a particular medication or intervention.” If a good response is “a very prevalent outcome,” why identify the genotypes in the first place? Only if the intervention would be selectively harmful to people without the genotype would this be necessary.
The revolution in medicine will come with the recognition, based in part on genetic research, that the quest for single causes for common diseases will seldom be fruitful and that a new paradigm of a causal web must be adopted. Interventions must be directed at the most vulnerable points in the web. Sometimes this will involve biomedical interventions. At other times, it will involve modifying aspects of our social structure, lifestyle, or environment that increase the risk of disease.
Neil A. Holtzman, M.D., M.P.H.
Johns Hopkins Medical Institutions, Baltimore, MD 21205-2004Theresa M. Marteau, Ph.D.
Guy's, King's and St. Thomas' Medical School, London SE1 9RT, United Kingdom- Citing Articles (1)
Citing Articles
1
Norbert W. Paul. (2010) Medizinische Prädiktion, Prävention und Gerechtigkeit: Anmerkungen zu ethischen Dimensionen eines biomedizinischen Ideals. Ethik in der Medizin 22:3, 191-205
CrossRef
