Author + information
- Calum A. MacRae, MB, ChB, PhD* ( and )
- Patrick T. Ellinor, MD, PhD
- ↵*Reprint requests and correspondence:
Dr. Calum A. MacRae, Cardiovascular Research Center, Massachusetts General Hospital, 149 13th Street, Charlestown, Massachusetts 02129
Inherited cardiovascular diseases in which sudden death may be the initial clinical presentation pose two basic questions for the cardiologist. First, how should we screen family members for disease? Second, once someone has been identified through screening, how should they be treated and, specifically, what is their risk of sudden death? These questions have proven difficult to answer, despite the remarkable advances in our understanding of the molecular basis for cardiovascular disease (1).
Intrinsic biases in inherited disease
In large measure, the difficulty in addressing such clinical questions is a result of several intrinsic features of these Mendelian disorders. The high incidence of premature sudden death, particularly if it occurs before reproductive age, is a major cause of the various forms of heterogeneity seen in inherited cardiac disease. The intense selective pressure of such a phenotype cannot be underestimated. Multiple genes may result in very similar phenotypes, possibly because the pathways affected are uniquely sensitive to perturbation. This genetic heterogeneity is well documented in hypertrophic cardiomyopathy (HCM), the long-QT syndrome, and arrhythmogenic right ventricular dysplasia (1). As the most malignant mutations will rapidly be eliminated from the population, these disorders are predicted to have a high frequency of de novo mutations and therefore are likely to be relatively uncommon. For any given gene, each affected family is likely to exhibit a different mutation (allelic heterogeneity), which also may be a source of phenotypic variation. Highly morbid genes paradoxically may affect multiple generations only when the effects of a specific mutation are modified, as more penetrant mutations will result in significant lethality. Thus, kindreds in which there is a family history may already be preselected for relatively “mild” mutations and more likely to display a broad range of different phenotypes, or pleiotropy, exhibiting exquisite sensitivity to genetic or environmental modifiers. This is perhaps most extreme in congenital heart disease, where the pleiotropy is such that often only the directed evaluation of apparently normal family members uncovers any evidence of an inherited familial predisposition (2).
The net effect of these biases is that standard epidemiologic approaches to defining diagnostic criteria and assessing risk are extremely difficult to interpret. The conflicting published data, both clinical and genetic, in syndromes such as HCM to a large extent reflect real differences in study cohorts. There is substantive variation between studies in the mode of subject recruitment or the representation of individual families, genes, and alleles, each of which are fundamental confounders.
The identification of disease genes, through genetic mapping and positional cloning, requires large extended families, which often may not perfectly reflect the disease seen in individual patients as a result of biologic or referral biases. However, these fundamental studies are critical to establish the causality of mutations and to enable the generation of mechanistically faithful disease models (1). Once the causal genes have been defined, investigators have gone on to study collections of families that map to these loci. Documentation of the spectrum of mutations has also led to a greater understanding of the role of specific protein domains in the pathogenesis of the respective syndromes. These studies also offer initial estimates of the proportion of disease attributable to each locus, the disease penetrance, and potential genotype-phenotype correlations (3).
In HCM such studies have defined some significant differences between the phenotypes seen at various loci. However, taken as a whole, genotype-phenotype correlations have produced rather disparate results. The unavoidable inclusion of multiple subjects from small numbers of large families has compounded the biases outlined earlier. When combined with the confounding effects of shared genetic background or environmental exposures, the lack of consistency is not surprising (4).
The ideal study design for genotype-phenotype correlation would include large numbers of unrelated subjects (both clinically affected and silent) with identical mutations. Although such a study is not feasible in HCM (several groups have begun to screen large series of patients for mutations), as sequencing costs have fallen and other cheaper screening technologies have become more established (5,6). These studies, although still limited, offer a complementary perspective to that obtained from earlier family studies. The emerging picture is of a lack of predictive utility for genotyping in patients with HCM.
In this issue of the Journal, Mogensen et al. (7) present the first results from a hybrid design that combines serial probands and their families, known as a “kin-cohort” study. This design collects multiple subjects with each mutation, and importantly, includes the two clinically relevant populations: those with clinical disease and their relatives. The size and scope of this effort, the first systematic study of cardiac troponin I mutations in HCM, are commendable, involving a thorough cardiovascular evaluation of the probands as well as affected and unaffected family members in nearly 750 families over a 10-year period.
What has emerged? The study reinforces the findings of genetic and allelic heterogeneity seen in other large studies of this form of cardiomyopathy. Comprehensive screening for cardiac troponin I mutations revealed abnormalities in only 3.1% of families. Thirteen distinct cardiac troponin I mutations in 23 probands and 77 family members were identified. Widely varying clinical presentations were observed not only between families with identical mutations, but also within families. The investigators concluded that there were no discernible patterns of a gene- or mutation-specific phenotype. Of note, large families are the exception and not the rule in HCM, precluding an initial mapping stage in the routine evaluation for mutations.
Ultimately, these investigators will extend their screening to the other genes implicated in HCM. The challenge remains how to extract meaningful information from such kin-cohorts, where the biological influences on proband presentation and family size remain potent sources of bias. Optimally, the ascertainment of the proband's relatives should be blinded to the affection status of each family member. Novel statistical or biologic strategies for extracting clinically useful information from monogenic families are needed. Concepts such as transmissible lifetime risk may be useful, but it could take several generations for a study such as this to collect sufficient data.
What then are the implications of the genotype-phenotype studies published to date for both screening and risk assessment?
At the present time, if an inexpensive, sensitive, and specific genetic test for HCM existed, the benefits would include the identification of asymptomatic carriers at risk for sudden death and obviation of the need for conventional screening in genotypically unaffected family members. Eventually, the identification of asymptomatic carriers may be useful for implementing validated preventive strategies.
Unfortunately, all the available data suggest that such a test simply does not exist. Hypertrophic cardiomyopathy displays such genetic and allelic heterogeneity that only discovery technologies offer the ability to detect mutations even in the known genes. The sensitivity and specificity of these techniques are not known for most of the relevant genes, and more than 20% of HCM is not explained by mutations in these genes. It is far from clear how isolated sequence abnormalities in affected subjects, which may be rare polymorphisms of no consequence, can be interpreted without functional assays for each gene. This is particularly important in the interpretation of genotype in those patients who have potential mutations in more than one gene, an issue in a significant number of those screened (5,8). The psychological impact on those who are genotype-positive but not yet affected must be balanced with the benefit for genotype-negative family members, and careful counseling will be essential both before and after testing (9).
Objective data on those who are genotype-negative, observed rates of “progression” to disease, the risks of events before diagnosis, and comparison of genotyping with state-of-the-art clinical diagnostic techniques will be required before any of these questions can be answered. At present, clinical and genetic screening can be recommended only in the context of ongoing evaluation at a research center (4).
Characterization of risk
If genetic testing is to offer any benefit in the assessment of risk, then individual genotypes must be useful over and above the simple diagnosis of HCM. The data presented in this issue of the Journalsuggest that, at least for troponin I-associated disease, genotypic information allows one only to discriminate those who have inherited any risk (and then only in those in whom a mutation can be detected using current technology), but does not allow gradation of that risk. In this study, all those family members who went on to have a cardiac arrest were detectable by conventional electrocardiographic or echocardiographic screening. Mogensen et al. (7) reasonably conclude that genetic testing offers no obvious benefit to those who are genotype positive.
Similar conclusions have emerged from the other large studies of genotype-phenotype correlation irrespective of their design. The available data do not yet support genotyping as a tool in risk stratification (4), but once again, it is vital that such data continue to be collected in the context of active research.
There is so much variation in conventional phenotypes and outcomes in many Mendelian disorders that current approaches to genotype-phenotype correlation may be incapable of generating clinically useful information. A substantial change in our understanding of phenotypic variation with genotype and its transmission through families will be required before the promised genetic revolution can take place. There is in vitro evidence of mechanistic heterogeneity even with different mutations in the same gene, and genetic or environmental modifiers may contribute to profound differences between subjects in the same family. The use of animal models to map and clone modifier genes is already underway. However, if we are to establish robust genotype-phenotype correlations, we need to identify and refine new phenotypes at the same time as we improve our genotyping capabilities. A systematic search for novel phenotypes closer to the proximate cause of HCM may help. More precise understanding of functional groupings may be feasible using in vitro analyses or even patient-based cellular profiling. Functional assays may be better predictors of outcome, allowing the integration of multiple factors, including genotypes at all the sarcomeric gene loci, remote modifiers, environmental effects, and even potential therapies.
↵* Editorials published in the Journal of the American College of Cardiologyreflect the views of the authors and do not necessarily represent the views of JACCor the American College of Cardiology.
- American College of Cardiology Foundation
- Watkins H.,
- Seidman J.G.,
- Seidman C.E.
- Maron B.J.,
- McKenna W.J.,
- Danielson G.K.,
- et al.
- Richard P.,
- Charron P.,
- Carrier L.,
- et al.
- Van Driest S.L.,
- Ellsworth E.G.,
- Ommen S.R.,
- Tajik A.J.,
- Gersh B.J.,
- Ackerman M.J.
- ↵Mogensen J, Murphy RT, Kubo T, et al. Frequency and clinical expression of cardiac troponin I mutations in 748 consecutive families with hypertrophic cardiomyopathy. J Am Coll Cardiol 2004;44:2315–25.
- Blair E.,
- Price S.J.,
- Baty C.J.,
- Ostman-Smith I.,
- Watkins H.
- Charron P.,
- Heron D.,
- Gargiulo M.,
- et al.