Editorial Comment—The Pendulum’s Swing: The Way Forward in the Genetics of Stroke
In a field that has been paved with frustration because of inconsistent reports, the review article by Rosand and Altshuler1 comes at the right time. It is refreshing to see that the authors rightly adopt an optimistic view on the prospects of molecular genetic studies in unraveling the molecular and genetic architectures of stroke.
In a nutshell, the goal of molecular genetic investigations is to identify genetic mutations that confer an individual’s genetic susceptibility to the disorder. The identification of causative mutations forms the basis for diagnostic and prognostic tests. It also allows us to comprehend the molecular etiology and pathophysiology of the disorder, which in turn serves as a springboard for the development of therapeutic modalities that are tailored to the yet-to-be-discovered molecular abnormalities. So much for the easy part. Stroke indeed, as all other complex disorders, is underscored by the combined effects of several to many genes with reduced penetrance. Furthermore, it is likely that different sets of deleterious genes contribute to the disease in different populations or families. How we should go about identifying susceptibility genes remains a burning issue.
Twenty years ago, the general opinion was that, if researchers wanted to characterize genetic effects of diseases (of monogenic types), they had to strictly adhere to linkage analysis procedures. One of the revolutions of molecular geneticists in the 1980s was to show that modern tools of recombinant DNA technologies could be applied to the unraveling of complex (then called multifactorial) disorders, whereby polygenic factors interplay with environmental and epigenetic factors. Those common disorders, which are chronic and degenerative in nature, include atherosclerotic diseases, hypertension, diabetes, allergies, cancers, Alzheimer disease, and stroke, among others. The higher degree of complexity detracted many classical geneticists, and association studies in particular were largely frowned upon. In the mid-1990s, benefits and shortcomings of different strategies were reasonably assessed.2,3 The result was an explosion of association studies to the extent that the pendulum seemed to have swung from “purely linkage” to “exclusively association” lately.
In fact, both general strategies, together with variations on their themes, are complementary, and investigators should be very much aware of advantages but also of limitations and pitfalls of each of them. Rosand and Altshuler’s balanced view on the topic is even more so remarkable that, compared with other complex clinical entities, stroke is clearly more amenable to case-control types of investigations because of both late age of onset and associated mortality. And although both linkage4 and case-control studies5 have so far yielded positive results, they still fall short of giving definite answers.1 Clearly, there is a need for better diagnostic categorization, larger samples sizes, and more powerful methods.1 The review gives practical ways of avoiding some of the pitfalls of association study designs.
For example, usually accepted P values of 0.05 are clearly no longer good enough, and sample sizes should be sufficiently high to reach statistical significance down to an order of 10−6 instead. Another key element is the replication of data and, even better, to carry out meta-analyses. Adequate samples sizes can better be achieved through multicenter study designs, with obvious implications on how to share and assign due credit to all collaborators. Then, while the selection of patients with well-defined clinical criteria is a critical issue, the major difficulty resides in the recruitment of controls. These are usually age- and sex-matched individuals who are at best disease-free at the time of sampling, although they may very well be presymptomatic already. Collectively, they therefore should be more appropriately referred to as comparison (rather than control) groups. While focusing on end-point clinical phenotypes (such as stroke), probing for associations with endo- (intermediate) phenotypes will yield critical information to explain underlying pathophysiological mechanisms. Two other topics would also deserve special attention: (1) Besides drastically increasing sample sizes, another way to improve the power of analyses is to construct haplotypes (combinations of markers on the same chromosomal region) as opposed to using single markers in order to define alleles on which to then look for causative variants.6 At several loci indeed, linkage disequilibrium has been shown to be lost after 3.5 kilobase pairs. (2) The identification of culprit genes in chronic conditions relies on 2 basic assumptions, those of clinical and genetic homogeneities. While the idea of defining strict yardsticks for inclusion of cases and controls is well accepted, genetic make-ups of sample populations under investigation are rarely questioned. Yet genetic isolates or specific ethnicities would clearly be more appropriate at first than populations of mixed, recent genetic origins, to tease out major genetic effects.
Clearly the time is ripe to establish proper sets of guidelines that will arm investigators with convincing criteria to identify disease-susceptibility genes. We have now learned enough lessons, and there has to be law and order to prevent wastage of time and resources.
The review even touches upon evolutionary considerations. The authors give preference to the model stating that “ … mutations causing common, late-onset diseases are likely to be evolutionarily neutral …,” although the usual view is that we have inherited those genes that allowed our ancestors to survive under harsh conditions. Examples include the “thrifty gene” hypothesis for diabetes, and the fight or flight response leading to increased blood pressure, and through the same molecular pathways, to essential hypertension. Similarly, the mechanisms developed throughout evolution for storing and retaining as much energy as possible from very limited food supplies have turned against individuals of affluent societies, in whom relatively recently improved environmental conditions lead to early development of atherosclerosis. In the case of stroke, which is the fourth leading cause of death in the world (and the third one in developed countries), Rosand and Altshuler1 reconcile the 2 views in arguing that the molecular pathways leading to cerebrovascular accidents may have exerted pleiotropic effects on survival. Selective advantages could find their roots in mechanisms similar to those leading to atherosclerosis.
No review could possibly cover all important aspects of the field of genetics of complex disorders. Bearing this reality in mind, Rosand and Altshuler have done an excellent job at producing a balanced, synthetic overview of genetics and stroke. They are encouraging readers to raise their awareness of underlying concepts and methods, thereby alerting them that efficiently generated results are quickly crossing the doors of research laboratories to enter clinical practice.
Rosand J, Altshuler D. Human genome sequence variation and the search for genes influencing stroke. Stroke. 2003; 34: 2512–2517.
Lander ES, Schork NJ. Genetic dissection of complex traits. Science. 1994; 265: 2037–2048.
Risch N, Merikangas K. The future of genetic studies of complex human diseases. Science. 1996; 273: 1516–1517.
Hassan A, Markus HS. Genetics and ischaemic stroke. Brain. 2000; 123: 1784–1812.