Editorial Comment—Unraveling the Pagodian Architecture of Stroke as a Complex Disorder
The last decade has witnessed an explosion of hospital-based association studies aimed at deciphering the genetic architecture of complex disorders. Most of these studies attempt such major undertakings without taking into consideration the phenotypic heterogeneity of these disorders. The population-based meta-analyses presented by Schulz and colleagues1 in this issue of Stroke brings home a very significant point: a complex disorder is not a uniform entity, but a composite of several elemental pathologies affecting particular organ system(s).
Most genetic association studies on stroke, for example, pool its various subtypes, amalgamate several ethnicities for the sake of increasing sample size, and genotype only one or a few polymorphisms within candidate genes. In their endeavor to associate a single biallelic polymorphism with a complex disorder, such studies overlook the Factor(s) X which mediate the pathology. In essence, such an association is a jump across the Atlantic. The result is that the literature is flooded with a large number of published papers with extensive heterogeneity and publication bias.
Sporadic stroke, as are most complex disorders, is heterogeneous. It can be divided into two major categories, hemorrhagic and ischemic stroke, each having several subcategories which can be identified as distinct phenotypes with their own etiologies. In spite of the recognition of these concepts for a long time, Schulz and colleagues are among the very first to provide a proof. They present clear evidence that confirms prior notions about the genetic architecture of stroke: (1) there is greater genetic influence in pathogenesis if the disease develops at a younger age; (2) large vessel disease is predominantly due to atherosclerosis and shares common genetic denominators with hypertension and myocardial infarction; (3) cardioembolism is essentially a mechanical phenomenon and would unlikely be genetically determined. Investigators, in just over half of the studies reviewed recently, regarded ischemic stroke as a heterogeneous phenotype and more than a third of the studies made no attempt to address the clinical heterogeneity of the disease.2 Various systems have been used for subtyping ischemic stroke, including nonstandard ones, and there is little consistency among studies.2 Thus, there is a need for greater use of standardized and reliable systems for subtyping complex disorders in studies of genetic risk factors.
Complex disorders do not exhibit a flat architecture, but in fact possess a high degree of plasticity mediated by several nonlinear dynamic processes.3 A large number of these processes may be genetically determined and all would be partially influenced. The overall design is comparable to that of a pagoda with several independent units stacked on top of one another constituting the holistic phenotype. Yet, the units are partially interdependent as well. Rarely, we may stumble on polymorphisms which constitute the pillars of the pagoda and their influence transcends to the very top of the structure, such as the APOE ε4 in Alzheimer’s.4 Chance would certainly smile on us then, but only if we have been prepared enough to eliminate the confounding effect of ethnic and phenotypic heterogeneity.
In light of the same thought, it is not ostentatious to presume that some polymorphisms only influence the structure of a single pagodian unit; others shape several, yet the effect of some only manifests in the presence of polymorphisms which alter the structure of the preceding unit. This kind of dynamic interaction of multiple gene effects is responsible for shaping the overall architecture of a complex disorder. Thus, it would be difficult, if not impossible, to hit on absolute associations of a polymorphism within a candidate gene in a complex disorder. The polymorphism may certainly be involved, via several nonlinear dynamic processes and its influence may be significant, yet limited.
To tease out such effects, which indeed constitute the large majority, it is beneficial to look for associations on a limited scale, taking advantage of the fractal behavior of complex disorders. Genetic polymorphisms may alter the structure or expression of a protein, leading to a dysfunctional biologic pathway. Over time, this disturbance may manifest as an altered physiological characteristic of the organism. An example is ACE gene polymorphisms leading to renin-angiotensin system dysfunction which causes atherosclerosis, manifesting as increased intima media thickness (IMT) or systolic blood pressure variability due to rigid arteries.5 Indeed, these physiological characteristics can be easily measured and their development can be followed over time, since they are continuous in nature. Keeping in mind the fractal characteristics of complex disorders it is not presumptuous to infer that polymorphisms influencing measurable intermediate phenotypes certainly have a role to play in structuring the overall architecture of the complex disorder, even if no association is observed with the holistic phenotype. Additionally, polymorphisms influencing one complex disorder might be involved in the causation of another such disorder as well, as they might operate through the same intermediate phenotype(s). This is in line with the suggestion by Schulz and colleagues that the heritability of ischemic stroke may be partly conferred by an inherited tendency for hypertension.1
In view of the above, I would suggest a structured approach to genetic association studies of stroke. Polymorphisms in the exonic or promoter regions of candidate genes should be searched for associations with intermediate phenotypes, such as carotid IMT. The power of such studies can be enhanced by haplotyping and by using improvised study designs, such as sib-pair or TDT-trios, which strengthen the studies by tracking genetic transmission of haplotypic markers. The candidate genes yielding positive associations should be explored further for associations with stroke after accurate phenotyping. As suggested by Schulz et al, molecular genetic studies should target noncardioembolic stroke.1 Additionally, stroke association studies should focus separately on large vessel and small vessel disease, as well as independently evaluate hemorrhagic stroke. It is advantageous to look for and quantitate the combined effect of the polymorphisms in the various candidate genes on the stroke subtype6 and to verify the results in at least two ethnically distinct, homogenous populations of adequate sample sizes. Schulz et al demonstrate that hospital-based association studies do not suffer from the problem of inclusion-bias,1 which eases molecular genetic investigations for stroke. These might further benefit from recruiting younger diseased individuals (age <60 y) since the heritability is high, as suggested by the present study1 and the genetic effects of candidate polymorphisms is enhanced. Furthermore, unraveling complex disorders is a colossal task which will benefit greatly by the establishment of international multilateral collaborations and their organization into large consortiums.
Schulz UGR, Flossmann E, Rothwell PM. Heritability of ischemic stroke in relation to age, vascular risk factors and subtypes of incident stroke in population-based studies. Stroke. 2004; 35: 819–825.
Meschia JF. Addressing the heterogeneity of ischemic stroke phenotype in human genetics research. Stroke. 2002; 33: 2770–2774.
Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, Small GW, Roses AD, Haines JL, Pericak-Vance MA. Gene dose of Apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science. 1993; 261: 921–923.
Mancia G, Parati G, Castiglioni P, Tordi R, Tortorici E, Glavina F, Di Rienzo M. Daily life blood pressure changes are steeper in hypertensive than in normotensive subjects. Hypertension. 2003; 42: 277–282.
Naber CK, Husing J, Wolfhard U, Erbel R, Siffert W. Interaction of the ACE D allele and the GNB3 825T allele in myocardial infarction. Hypertension. 2000; 36: 986–989.