Estimating the sensitivity of a genetic test using gene-carrier probability estimates and its application in genetic counselling.

Abstract

Background: The estimation of the sensitivity of a genetic test is of practical importance. If a 'gold standard' (an exact test) is not available, an estimate of the probability of being a gene carrier may be useful for an individual. This information could be based on individual pedigree and family-history data, and a known genetic model for the disease of interest.

Methods: We develop a maximum-likelihood estimate of the sensitivity of a genetic test that may be applied in a situation without a 'gold standard', in which an estimate of the probability of being a gene carrier for an individual is available.

Results: A maximum-likelihood estimate for the sensitivity can be obtained through an iterative algorithm. We demonstrate the method using data from a project on familial breast cancer. We further present disease-risk estimates incorporating results from a genetic test with different values of sensitivity, and compare these with disease-risk estimates that are solely based on family-history data.

Discussion: We provide a systematic methodology to obtain an estimate of sensitivity of a genetic test when only gene-carrier probability estimates from a genetic model are available. Given a negative result from a genetic test, predictions for lifetime and age-specific disease-risk, accounting for test sensitivity, can then be provided in genetic counselling.