Estimating disease risk associated with mutated genes in family-based designs

Many clinical decisions require accurate estimates of disease risk associated with inherited gene mutations. While several family-based designs have been proposed, their relative advantages remain unclear.

We considered four commonly-used family-based designs and evaluated their performance in terms of accuracy and efficiency under several genetic models via simulation studies. We also derived and assessed several ascertainment-corrected likelihood methods for analyzing the simulated data and real data from 12 HNPCC pedigrees from Newfoundland.

We found that the design efficiency depends on the question of interest: the clinic-based family design with random probands yields the most efficient estimate of genetic relative risks, whereas the population-based family design with mutation carrier probands provides the most efficient penetrance estimates. For a particular question, an ascertainment correction seems possible using regular likelihood methods but the presence of genetic heterogeneity due to a strong second gene effect can lead to some bias in the risk estimation.

This work gives a general methodological framework for analyzing family-based designs in gene characterization studies and provides more rationale for the choice of an efficient design and an appropriate likelihood method to estimate the risk associated with an inherited gene mutation.