Retrospective analysis of molecular scores for the prediction of distant recurrence according to baseline risk factors

Abstract

Clinical variables and several gene signature profiles have been investigated for the prediction of (distant) recurrence in several trials. These molecular markers are significantly correlated with overall and late distant recurrences. Here, we retrospectively explore whether age and body mass index (BMI) affect the prediction of these molecular scores for distant recurrence in postmenopausal women with hormone receptor-positive breast cancer in the transATAC trial. 940 postmenopausal women for whom the Clinical Treatment Score (CTS), immunohistochemical markers (IHC4), Oncotype Recurrence Score (RS), and the Prosigna Risk of Recurrence Score (ROR) were available were included in this retrospective analysis. Conventional BMI groups were used (N = 865), and age was split into equal tertiles (N = 940). Cox proportional hazard models were used to determine the effect of a molecular score for the prediction of distant recurrence according to BMI and age groups. In both the univariate and bivariate analyses, the effect size of the IHC4 and RS was strongest in women aged 59.8 years or younger. Trends tests for age were significant for the IHC4 and RS, but not for the CTS and ROR, for which most prognostic information was added in women aged 60 years or older. The CTS and ROR scores added significant prognostic information in all three BMI groups. In both the univariate and bivariate analyses, the IHC4 provided the most prognostic information in women with a BMI lower than 25 kg/m2, whereas the RS did not add prognostic information for distant recurrence in women with a BMI of 30 kg/m2 or above. Molecular scores are increasingly used in women with breast cancer to assess recurrence risk. We have shown that the effect size of the molecular scores is significantly different across age groups, but not across BMI groups. The results from this retrospective analysis may be incorporated in the identification of women who may benefit most from the use of these molecular scores, but our findings need further evaluation before these scores can be used in clinical decision making.