Progressive Diseases: Interpretation of Genetic Data
Simple modeling is proposed to represent the screening of gene polymorphisms for association with a progressive disease of insidious onset such as Alzheimer's disease. The modeling demonstrates that when a polymorphism affects the rate of progression as well as the risk of disease, the correct interpretation of DNA data requires an accurate sampling of the living, diseased population. Furthermore, in this population, the effect of the polymorphism on disease risk cannot be distinguished from a corresponding effect on the rate of progression of the disease, and a null result does not preclude a significant effect of the gene on the disease. By contrast, when the population is sampled either at time of diagnosis or at autopsy, the effect of the polymorphism on disease frequency can be directly related to the frequency of the polymorphism in the sample, but evaluating the rate of disease progression requires additional data. When the only available data are obtained from a live patient population, substantial differences in interpretation can result from subtle differences in the patient selection protocol. When existing DNA databases are used in which this protocol is not well characterized, there is a corresponding uncertainty introduced into the deduced effect of the polymorphism on disease risk and rate of progression.