Problems in detecting etiological heterogeneity in genetic disease illustrated with retinitis pigmentosa.

Simulated data were generated under four etiologic mechanisms for each of 20 different pedigree structures drawn from a study of families ascertained through a proband with retinitis pigmentosa. These simulated data were then used to identify subgroups of pedigrees which best supported each of three genetic mechanisms (autosomal dominant, autosomal recessive, X-linked recessive with 10% penetrance of disease in heterozygous females) and a non-genetic, sporadic mechanism. Results of these studies show that pedigrees identified as supporting one genetic model in a 'model choice' approach tend to be etiologically homogeneous, but are not truly representative of all the phenotypic combinations possible under that mechanism. The problem of etiologic heterogeneity is most acute when dealing with pedigrees less than size 10. Pedigrees lumped under a non-genetic, sporadic mechanism are extremely heterogeneous and studies of the natural history of diseases where both genetic and non-genetic mechanisms may be operating (such as with retinitis pigmentosa) should avoid using this group of largely simplex pedigrees.