Figure 2

Kaplan-Meier curves may identify the best fitting genetic model for a polymorphism. For simplicity, survival curves are shown as straight lines. AA = major allele homozygous genotype, AB = heterozygous genotype, BB = minor allele homozygous genotype, assuming allele ‘A’ is the common allele. (a) The effect of the AB genotype on survival is approximately half between the AA and BB genotypes, thus the additive model is appropriate for this polymorphism in the multivariate analysis. (b) The curves of AB and BB genotypes cluster closer to each other when compared to the AA genotype’s curve, thus, the effect of the polymorphism is likely to be dominant. (c) AA and AB genotype survival curves cluster together and clearly separate from the BB genotype curve. Thus, the inheritance pattern is likely to be recessive. (d) In this case, the effect of AB genotype is somewhat in between the effects of AA and BB genotypes, thus, analyzing this polymorphism assuming the codominant model is suitable. (e) This is an interesting polymorphism where the heterozygotes are associated with worse survival compared to either homozygous genotypes (AA and BB). The codominant genetic model is the appropriate model to investigate such polymorphisms in multivariate analyses. Exact biological and genetic reasons for such associations are not clear, but it may be due to heterozygote disadvantage where the heterozygotes display phenotype but not the either homozygotes. (f) The heterozygotes have better survival than AA and BB homozygotes. This case may represent a ‘heterozygote advantage’ situation, where the heterozygotes have favorable survival characteristics. Similar examples are observed in Mendelian diseases, such as sickle cell anemia [56]. In both (e) and (f), presence of another genetic variation in close proximity acting as a prognostic factor (which is not highly correlated with this polymorphism) may be an alternative explanation.