The Master said, “By nature, men are nearly alike; by practice, they get to be wide apart.”
The origin of cardiovascular disorders, as of all disease, lies in the duality of inborn predisposition, or susceptibility, and the exposure to external (environmental) influences. We apply this knowledge in our daily clinical practice as we consider anamnestic data from a patient’s family history as well as from his or her lifestyle and occupation in our assessment of cardiovascular risk. We also commonly draw a connection between these two etiologic domains, simplistic as it may be, by estimating a proportionately higher risk in the presence of “positive” findings from both categories. At the same time, we are keenly aware of how vague and inaccurate prognostications remain that are based on these recognized risk factors and how little specific guidance for the management of individual patients they offer. Is it possible that more detailed knowledge of the genetic component of a patient’s “risk profile” might raise the precision and specificity of our risk assessment and thus improve the success rate of intervention? To the geneticist, the specific, not simply additive, interaction between inherited characteristics and environmental factors is a well-appreciated phenomenon: neither exposure to dietary phenylalanine nor a defect in the gene encoding phenylalanine hydroxylase is by itself associated with disease, yet the combination of the two results in the severe syndrome of pronounced mental retardation and associated somatic abnormalities typical of phenylketonuria. We may expect that in other diseases that show the markings of both inherited and acquired disorders, similar specific “ecogenetic” interactions between external influences and genetic mutations exist that predict with a high degree of specificity an individual’s response or disease susceptibility to an environmental exposure. Complex diseases, which encompass most cardiovascular as well as other epidemiologically important disorders, are characterized by the concurrent interactive morbid effects of multiple genetic and environmental factors; although unraveling the precise network of gene-environment interactions that contribute to a given disease is a major task, the potential implications of such advances in medical knowledge are clearly considerable.
This issue of Circulation contains a remarkable account of such a gene-environment interaction. Amant and colleagues1 report a highly significant additive association between the D allele of the gene encoding ACE and the occurrence of restenosis after coronary stenting. Of note, this observation stands in contrast to their own and other investigators’ findings that this polymorphic marker shows no association with restenosis after conventional angioplasty. The authors speculate that this discrepancy may be related to perceived differences in the pathogenesis of restenosis after stenting compared with restenosis complicating regular angioplasty: preliminary findings indicate a more pronounced increase in neointimal mass after the former and a remodeling process that affects primarily the vascular smooth muscle cells after the latter. Seen in the context of gene-environment interaction, the proliferation-enhancing stimulus on neointimal growth exerted by the implant but not by the angioplasty procedure per se would be seen as representing the environmental component of a gene-environment interaction with the ACE gene. In this scheme, the number of D alleles an individual carries then determines, in an additive fashion, his or her susceptibility to a “foreign-body”–induced tissue reaction.
Conceptually, it is quite reasonable that heritable characteristics should govern or influence the response to injury, as exemplified by the spectrum of responses individuals show to simple lacerations, from uneventful healing to formation of severe scarring. In the arena of clinical genetics, the finding of Amant et al1 extends the well-recognized interaction between genetic makeup and response to pharmacological substances (often referred to as pharmacogenetics) to a similar interaction between heritable factors and interventional procedures and/or prosthetics. Thus, in time we may find that genetic variation explains differences in tolerance for orthopedic or cosmetic implants, influences the fate of implantable pacemakers or chronic indwelling catheters, or more generally contributes to the interindividual variability in response to surgical or minimally invasive intervention procedures. Knowledge of the genes and their molecular variants affecting these responses clearly has the potential to provide guidance for patient selection and risk assessment and to make procedures safer as well as possibly, by appropriate modification of preoperative and postoperative management, more effective. Selection of patients according to genetically defined risk profiles also carries important promise for the collection of clinically more homogeneous samples to evaluate new procedures, devices, or drugs, offering greater statistical power at significantly lower cost. Intriguingly, this approach may carry the potential of offsetting the financial disincentives of developing novel therapeutic approaches that target narrowly defined market sectors. It is of course ironic that we should recognize an example of gene-environment interaction so far downstream from where similar knowledge might have far greater impact, namely at the stage of prevention, not of costly repair of vascular lesions.
Finding the specific genetic and environmental components that are of relevance in such gene-environment interactions is not an easy task. As exemplified by the present study, gene variants that contribute to complex disease may generally be expected to act in concert with recognized environmental or biochemical risk factors by amplifying, as susceptibility genes or mutations, their morbid effects. Precise measurement of and careful accounting for all potentially important covariates will therefore be of paramount importance in performing, analyzing, and interpreting studies that focus on the genetics of complex disorders: as in the monogenic paradigm of phenylketonuria, gene variants may contribute to disease phenotype only in the presence (or absence) of certain external factors, and to the investigator, a pathogenetic contribution by a genetic variant may thus be apparent only if all relevant information is taken into consideration. The presence of gene-environment interactions also predicts that judicious choice of selection or stratification criteria will have the potential not only of significantly enhancing power but also of enabling the recognition of a genetic factor affecting a trait.
As intriguing as the results of the study by Amant and coworkers appear, they must be interpreted with greater caution than the results of conventional (nongenetic) risk-factor evaluations. Historically, epidemiological investigation has never had the opportunity to prove causation of disease, nor has it had the associated responsibility. As strong as the association between a disease and a suspected cause or risk factor may be, it can never distinguish between cause and effect and must therefore ultimately remain phenomenological in character. With the advent of techniques that allow us to directly assay DNA, the matrix of all life processes, a paradigm shift has occurred. It is now possible, on the basis of the association of specific molecular variants of genes with the absence or presence of disease, to discover and characterize pathogenic principles and mechanisms that are by definition primary and causative, because they are located upstream from all subsequent steps that govern biological function. Remarkable progress toward finding causative gene mutations has thus been made over the past few years in the area of single-gene disorders; as the next logical step, we are now beginning to apply molecular genetic methods to the broad realm of common, chronic, complex disease. Although less dramatic than in classic single-gene diseases, the contribution of genetic factors to the causation of common, complex diseases is likely to be of far greater public health impact than in the case of these rare disorders. Clearly, the major strength and claim of genetic studies, positive identification of primary, causal mechanisms, makes a cautious and circumspect approach toward such studies and toward the interpretation of their results particularly important.
This caveat may be particularly appropriate in the present case: the first candidate gene marker to be reported as being associated with differential risk for a common, complex disease, the ACE D/I polymorphism, has been studied in a host of pathological conditions and has spawned a large number of studies. There has been considerable controversy on the topic, no doubt reflecting the complexity of polygenic, multifactorial disease, with a number of negative studies failing to confirm earlier positive results and with a recent meta-analysis documenting an inverse correlation between study size and magnitude of relative risk associated with the polymorphism, a classic example of publication bias.2 Although it is unrealistic to expect that a study of the size published in this issue can provide the degree of statistical significance that some in the field have demanded (whose contention it is that any given candidate gene study is nothing less than part of a “virtual genome screen” and should thus be subject to the same stringent statistical thresholds as dictated by the large number of repetitive comparisons typical of genome screen3 ), it emphasizes the brittle character of the present results and the need to replicate these and similar findings in additional, independent, larger, more robust samples.
Much of the interest in the ACE D/I polymorphism can be attributed to the intriguing and well-documented correlation of the marker with plasma and tissue ACE levels that explains a sizable fraction of their variance and that would seem on the surface to tie in logically with an increased disease risk in carriers of the D allele: higher plasma ACE levels associated with the D allele are perceived as indicating an “activated” state of the renin-angiotensin system, which is viewed as pathogenic, mainly by inference based on beneficial effects of ACE inhibition in a number of disorders and from in vitro data demonstrating trophic effects of angiotensin II. Surprisingly, almost all investigations, including the present one, fail to take what would seem the logical next step, namely, to measure ACE activities and to carry out linear regression on this variable. Moreover, only one study4 so far has evaluated angiotensin levels—the ultimately relevant parameter, if this pathway is to be involved, with respect to the D/I polymorphism—and it found no correlation. (The reservation that measurements of plasma ACE may be irrelevant because they may not reflect tissue levels is a priori a valid one; however, in view of evidence for parallel gradients of ACE activity in tissues and plasma in association with the D/I genotype, this argument would seem less convincing.) Could it be that the ease with which PCR-based genotyping is accomplished today is luring clinical investigators into relying too much on a potentially powerful but equally dangerous strategy? Since basic biomedical research, with the impending completion of the human genome project, is swiftly reorienting its emphasis from genome to proteome and beyond, applied clinical-epidemiological genetic research must not forget its forte in protein biochemistry and integrative physiology to validate results from genetic studies.
If confirmed, the findings by Amant and coworkers are likely to attain direct practical importance in clinical medicine by providing a tool for risk assessment and prognostication. They would also most certainly be of paradigmatic character by providing an example for the degree of complexity we are to expect for gene-environment interactions that contribute to disease manifestations. It will be essential to demonstrate the functional substrate of what thus far is purely an observation before addressing the possibility of targeted intervention, as alluded to by the authors. Above all, we must recognize that complex disease genetics is still very much a developing discipline, that at the present time there are no patent answers to the many questions concerning investigative strategies and interpretation of results, and that we are all part of an ongoing learning process. Although we must expect that once the dust settles, much of what has been and is being published will appear obsolete, every piece of data represents a small step along this learning process that will, we hope, eventually allow us to apply the new tools to true and tangible progress in what has been termed “disease management,” that is, the individualized, risk factor profile–guided approach toward prevention and treatment of illness.
The opinions expressed in this editorial are not necessarily those of the editors or of the American Heart Association.
- Copyright © 1997 by American Heart Association
Amant C, Bauters C, Bodart J-C, Lablanche J-M, Grollier G, Danchin N, Hamon M, Richard F, Helbecque N, McFadden EP, Amouyel P, Bertrand ME. D allele of the angiotensin I–converting enzyme is a major risk factor for restenosis after coronary stenting. Circulation. 1997;96:56-60.
Samani NJ, Thompson JR, O’Toole L, Channer K, Woods KL. A meta-analysis of the association of the deletion allele of the angiotensin-converting enzyme gene with myocardial infarction. Circulation. 1996;94:708-712.
Lander ES, Kruglyak L. Genetic dissection of complex traits: guidelines for interpreting linkage results. Nat Genet. 1995;11:229-241.
Harrap SB, Davidson HR, Connor JM, Soubrier F, Corvol P, Fraser R, Foy CJ, Watt GC. The angiotensin I converting enzyme gene and predisposition to high blood pressure. Hypertension. 1993;21:455-460.