| Summary: | The statistical problems of studies on associations between diseases and genetic polymorphisms such as the HL A system are reviewed. Special emphasis is placed on the bias involved in the assessment of the true significance of an extreme association picked from a number of comparisons. Serological difficulties connected with HL A typing of diseased patients could also complicate detection of HL A and disease associations. An overall assessment of published data on the HL A antigen distribution in Hodgkin's disease patients and controls seems to confirm the association with the '4c' related set of antigens, HL A5, W5, W15, and W18. This association is interpreted in terms of the effects of a disease susceptibility locus, DSA, in the HL A region.
|
|---|