Background and objective:In clinical settings with fixed resources allocated to predictive genetic testing for high-risk cancer predisposition genes, optimal strategies for mutation screening programmes are critically important. These depend on the mutation spectrum found in the population under consideration and the frequency of mutations detected as a function of the personal and family history of cancer, which are both affected by the presence of founder mutations and demographic characteristics of the underlying population. The results of multistep genetic testing for mutations in BRCA1 or BRCA2 in a large series of families with breast cancer in the French-Canadian population of Quebec, Canada are reported. Methods: A total of 256 high-risk families were ascertained from regional familial cancer clinics throughout the province of Quebec. Initially, families were tested for a panel of specific mutations known to occur in this population. Families in which no mutation was identified were then comprehensively tested. Three algorithms to predict the presence of mutations were evaluated, including the prevalence tables provided by Myriad Genetics Laboratories, the Manchester Scoring System and a logistic regression approach based on the data from this study. Results: 8 of the 15 distinct mutations found in 62 BRCA1/BRCA2-positive families had never been previously reported in this population, whereas 82% carried 1 of the 4 mutations currently observed in >2 families. In the subset of 191 families in which at least 1 affected individual was tested, 29% carried a mutation. Of these 27 BRCA1-positive and 29 BRCA2-positive families, 48 (86%) were found to harbour a mutation detected by the initial test. Among the remaining 143 inconclusive families, all 8 families found to have a mutation after complete sequencing had Manchester Scores >18. The logistic regression and Manchester Scores provided equal predictive power, and both were significantly better than the Myriad Genetics Laboratories prevalence tables (p,0.001). A threshold of Manchester Score >18 provided an overall sensitivity of 86% and a specificity of 82%, with a positive predictive value of 66% in this population. Conclusion: In this population, a testing strategy with an initial test using a panel of reported recurrent mutations, followed by full sequencing in families with Manchester Scores >18, represents an efficient test in terms of overall cost and sensitivity.

The discovery of deleterious mutations in the breast and ovarian cancer susceptibility genes, BRCA1 and BRCA2, has facilitated the identification of individuals at particularly high risk of these diseases. There is a wide variation between populations in the prevalence and related risks of various types of BRCA1/2 mutations, so estimates cannot be extrapolated to Canadians, especially not founder populations such as French- Canadians. Polymerase chain reaction (PCR)-based methods were used to detect the majority of these mutations. These approaches usually failed to detect large DNA rearrangements, which have been claimed to be involved in other populations in 5% to up to 36% of BRCA1-positive families. There is very little information about the contribution of this type of mutation in BRCA2-positive families. To investigate if our available mutation spectrum of BRCA1 and BRCA2 in high-risk French-Canadian breast/ovarian cancer families has been biased by PCR-based direct sequencing methods, we first used Southern blot analysis to test DNA samples from 61 affected/obligate carrier individuals from 58 families in which no BRCA1/2 deleterious mutation was found. Finally, 154 individuals from 135 BRCA1/2 nonconclusive families, including all those tested previously by Southern blot analysis, were tested with the new multiplex ligation probe amplification (MLPA) technique. These approaches failed to detect any rearrangement. Moreover, if the frequency of MLPA-detectable rearrangements in our cohort of 135 BRCA1/2 nonconclusive families was 2.2% or higher, we would have had a 95% or greater chance of observing at least one such rearrangement. As no rearrangements were identified, such large rearrangements must be quite rare in our population.