Evaluation of two different models to predict BRCA1 and BRCA2 mutations in a cohort of Danish hereditary breast and/or ovarian cancer families.

To meet the increasing demand for BRCA1 and BRCA2 mutation analysis, a robust system for selecting families who have a higher chance of a mutation has become important. Several models have been developed to help predict which samples are more likely to be mutation positive than others. We have undertaken a complete BRCA1 and BRCA2 mutation analysis in 267 Danish families with high-risk family history. We found deleterious mutations in 28% (76) of the families, 68% (52) of those in BRCA1 and 32% (24) in BRCA2. We compared our results with two popular manual models developed to estimate the chance of a positive result. One is the recently published Manchester model and the other is the Frank 2 model updated by Myriad Genetic Laboratories, Inc. Neither of the models would have suggested screening all mutation-positive samples. The Manchester model would have suggested screening 124 of the families in the cohort, thereby detecting 54 of 76 mutations (sensitivity 71%; specificity 63%), whereas the Frank 2/Myriad model would have found 60 of 76 mutations by screening 169 samples if a 10% likelihood was adapted (sensitivity 79%; specificity 43%). The updated Manchester model suggested screening 172 families whereby 64 mutations would have been detected (sensitivity 84%; specificity 44%). We conclude that although both models would have reduced the number of samples screened significantly, up to 28% of the mutations would not have been found by applying these models to this Danish cohort of families. This raises the question whether models designed for specific populations can be used in a wider setting.