Dilated cardiomyopathy and the desmin gene.

To the Editor: Li et al1 report a mutation screening study of actin and desmin genes in 44 probands with familial dilated cardiomyopathy (DCM) and conclude that a mutation of the desmin gene can cause DCM. We believe that the genetic data are not yet sufficient to justify this conclusion. Attempts to find genes for autosomal dominant DCM by linkage analysis have been frustrating; although 6 chromosomal loci have been mapped, none of these linkages have yet led to disease-gene identification. Not surprisingly, investigators have turned to direct candidate gene analysis. However, in forgoing linkage analysis, one of the pillars of evidence supporting a real association is removed. Other criteria for assessing a causal effect are that the DNA variant in the affected individuals is not present in a large number of normal controls, that the predicted change in the encoded protein is significant, and that mutations in the same gene are found in more than 1 affected family. Such observations have been sufficient to propose a causal role for missense mutations in cardiac actin in DCM.2 However, without evidence of cosegregation (linkage) or an observed biological effect, some uncertainty remains, and actin mutations have not yet been found in other DCM series.1,3 Demonstration of de novo mutation confirms the role of a disease gene,4 but this has not been achieved for DCM. The second candidate gene to reveal a mutation in DCM is that encoding desmin.1 Li et al describe a missense mutation, Ile451Met, in the desmin gene in a single proband with DCM. The variant was not present in normal controls, but this alone does not indicate a disease-causing role. Three relatives carried the mutation, but only 1 of these was clinically affected. The presence of the mutation in equal numbers of affected and unaffected individuals raises concern about the statement that the mutation segregates with DCM in this family. One risks falsepositive findings in studies of plausible candidate genes (such as desmin in DCM5 ) by assuming incomplete penetrance rather than postulating nonsegregation of the mutation with the disease, ie, the null hypothesis. However, the family was large enough (4 generations) to have used linkage analysis to test the role of the desmin variant. With only the data presented, 2 interpretations remain: the variant may be disease causing, albeit partially penetrant, or it may be incidental and therefore not likely to be transmitted from the deceased ancestor to the 2 living affected and 3 deceased, presumed-affected cousins. Three simple tests can and should be applied to resolve these alternatives. First, if tissue blocks are available on any deceased, presumed-affected relative, then demonstration of the mutation would support linkage. Second, analysis of relatives of the mother of the proband (II.4) would check that the variant was not inherited from her (indicating an incidental role). Third, haplotype analysis of the desmin gene region in the 5 surviving siblings of the deceased (III.4–8), who do not themselves have the mutation, would test whether 4 copies of the desmin gene are present, revealing that the mutation was not present on that side of the family and could not have caused the disease. If there is no linkage to desmin, there is a high chance that haplotype analysis will reveal it, because there are sufficient individuals to reconstruct genotypes of deceased parents.4 It would be very helpful to know what can be learned by a more comprehensive genetic analysis in this pedigree.