In this issue of Hypertension, an article appears that involved considerable discussion among the reviewers, editors, and statistical consultants.1 . The article reports an association between the transforming growth factor-β1 gene and hypertension. The contentious issues result from a meaningful academic statistical argument. If multiple genetic loci are investigated for possible association with a phenotype without an a priori hypothesis, should a correction for repeated comparisons (à la Bonferroni) be carried out? If so, what is the correction? This issue then spilled over into a discussion of association studies as contrasted with linkage studies and the relative roles of each.
After a full exploration of this issue, the editors elected to publish the manuscript and use this opportunity to make the general readership aware of the fact that statistical inferences in this field must be interpreted carefully. Furthermore, the thresholds for statistical significance in genetic studies of polygenic human traits should ideally be agreed upon in advance by the experts in the field itself.
Readers should understand that association analysis and linkage analysis are complementary, each providing its own brand of information. A linkage analysis does not assume any causal association between a phenotype and marker gene, but such causal associations can be rigorously constructed by the study of related markers on the same chromosome. Once candidate genes are identified, it is appropriate to perform association analysis in an attempt to associate the gene and the trait. Each type of study provides different information, but together they can help make a compelling story. In any event, however, it must be noted that only when a specific disease-causing gene is identified, its effects quantified, and its mechanism of action elucidated can one be reasonably confident that a true understanding of the phenomenon has been achieved.
As an increasing number of polygenic human diseases come under genetic analysis, there are bound to be other disagreements among well-meaning and knowledgeable scientists. Indeed, disagreements of this sort have long been a part of classic genetics—as well as of other scientific disciplines. It is important that clinicians and scientists keep these honest contentions and disagreements in mind when interpreting results. It is even more important that these disagreements not be suppressed; rather, they must be permitted to evolve so that a deeper understanding and consensus regarding the underlying facts can be developed.