The effects of requisite assumptions on linkage analyses of manic-depressive illness with HLA

Two large pedigrees with a bipolar proband were selected on the basis of compatibility with the hypothesis of autosomal dominant inheritance of a depressive taxon. We compared the results of linkage analyses of the depressive taxon with HLA and GLO in these pedigrees when penetrance function, diagnostic classification, and HLA haplotype frequencies were varied. The analyses show that all three types of assumptions affect the lod scores and likelihoods obtained. Using the most favorable assumptions, we obtain a lod score of 3.901 at theta = 0.10 for HLA with the depressive taxon. This lod score would conventionally be considered strong evidence of linkage of a subtype of manic-depressive illness with HLA but the dependence of these results on several untested assumptions suggests more cautious interpretation. Nonetheless, we feel it does represent evidence in favor of linkage because this high a lod score cannot readily be dismissed as statistical artifact. The results point out the importance of careful delineation of the assumptions made in linkage analyses of complex disorders, including the diagnostic criteria, marker allele frequencies, and age-of-onset function used.