Evaluation of a cancer genetics referral screening tool

There is increasing evidence of the effectiveness of enhanced management strategies in reducing the morbidity and mortality associated with hereditary risk for breast and ovarian cancer. However, the impracticality of utilizing existing hereditary risk models in primary care settings, imposes a significant barrier to the identification of at-risk individuals who might benefit from these advances. This study evaluated the reliability and accuracy of a Referral Screening Tool (RST) designed to allow for easy identification of individuals who may be at hereditary risk for breast/ovarian cancer, in order to facilitate referral for cancer genetics services. The RST was administered to 2,467 unselected women undergoing screening mammography. Telephone interviews were conducted on a random subset of 296 women, and included collection of a detailed four-generation cancer pedigree, demographic characteristics, and cancer screening history. Four established risk models (BRCAPRO, Myriad II, BOADICEA and FHAT) were used to analyze the pedigrees for validation purposes. Reliability as measured by concordance of randomly repeated RSTs was 95.5%, kappa = 0.75 (95% CI 0.61--0.80). Using a dichotomous risk classification scheme, with a 10% BRCA1 or BRCA2 mutation probability or FHAT score of ≥ 10 as the measure of "high-risk", the RST demonstrated an overall (high-risk by any model) sensitivity of 81.2% (individual model range 81.3--91.3%) and specificity of 91.9% (individual model range 76.2--90.3%). Examination of the numerical values obtained by the model assessments, revealed no overlap of the inter-quartile ranges between RST screen-positive and screen-negative subjects. Preliminary data suggests the RST may also provide a means to identify individuals at moderate levels of risk for breast/ovarian cancer. Given the RST is relatively easy to administer and interpret, these data appear to validate the potential effectiveness of the RST as a screening tool to assist primary care providers in identifying individuals who would benefit from cancer genetics services.