Family history screening for breast/ovarian cancer.

A176

As of now, no formal trial has been carried out to determine the feasibility of family history screening in primary care for BRCA-related predisposition to breast/ovarian cancer. Here, we show further data relating to one critical concern associated with such screening - the proportion of women in the primary care population that would be called positive and recommended for genetic counseling and, possibly, mutation testing. The target screen-positive rate recommended by the United States Preventive Services Task Force (USPSTF) is 2%. Several published family history algorithms are available for use in practice, but, until our earlier publication, none of these was applied to a primary care cohort to determine the screen-positive rate. We now present an expanded experience that might prove useful in guiding this aspect of implementation. We applied six family history algorithms to each of four separate population-based cohorts of women, 21 to 55 years of age, for whom personal as well as first- and second-degree family histories of breast/ovarian cancer were available. The four cohorts included 3,073 women (age and race/ethnicity were collected on 2,752). In three of the cohorts, personal history of breast/ovarian cancer ranged from 2.6% to 4.1%; in the fourth, the figure was 17.6%. This fourth cohort was collected in the lobby of an outpatient facility where breast cancer patients are treated and where high-risk breast cancer surveillance clinics are held. Individual algorithms yielded widely varying screen-positive rates among the cohorts (rates for one algorithm ranged from 4.1%, to 8.4%, another from 7.8% to 20.8%). The same wide variation was seen among different algorithms applied to the same cohort (4.1% to 9.2% in one cohort; 6.9% to 20.8% in another). Agreement between protocols was assessed using the kappa statistic: 9 comparisons were poor (<0.4), 48 were fair (between 0.4 and 0.75), and 2 were good (>0.75). If a screen positive result was restricted to women whose family history was positive by all six algorithms, rates for the four cohorts ranged from 1.9% to 4.0%. If family history is to be systematically introduced as a screening test for hereditary breast/ovarian cancer, a prime consideration ought to be establishing a highly selective policy for classifying women as candidates for BRCA1/2 mutation testing. Modeling suggests that between a 1% and 2% positive rate would result in identifying a high risk group in which as many as one in 10 women would be found to carry a BRCA1/2 mutation. One possible way to achieve that goal might be to require that several family history screening protocols agree (viewing each protocol as an "expert" and all protocols as an "expert panel"). The present study, for example, indicates that a 2 to 4% screen positive rate might be achieved when six family history screening protocols all agree. This is still slightly higher than the 2% that was targeted by the USPSTF.