Predicting Survival Time in Women Starting First Line Chemotherapy for Advanced Breast Cancer.

Abstract

Abstract Purpose: Women starting first line chemotherapy for advanced breast cancer have widely differing baseline characteristics and ultimate survival times. We sought to develop and validate a pragmatic prognostic index for clinical practice and research based on simple clinical and laboratory data.Methods: Prognostic models were developed in 378 women treated with CMF in the control groups of 3 randomized trials, and validated in 564 women treated with AC or mitoxantrone or capecitabine in the experimental groups of these trials. Multivariable censored linear regression of log-transformed survival times was used to select prognostic variables, which were then dichotomised to define negative prognostic features (NPFs). The effects of NPFs on survival time are expressed with hazard ratios (HR) and p-values from multivariable Cox models.Results: There were 7 independently significant NPFs (see table). In the development dataset, an elevated neutrophil count or history of brain metastasis approximately halved survival time, while the other NPFs reduced survival time by about one third. Women with 0 NPFs (15%) had a median survival of 18 months, about 1.4 times as long as those with 1 or 2 NPFs (63%, 13 months), and about 2.6 times as long as those with 3 or more NPFs (22%, 7 months). In the validation dataset, the effects of brain metastasis and elevated neutrophils were more modest, and most NPFs reduced survival by about one third. The median survivals of women with 0 NPFs (17%), 1 or 2 NPFs (60%) and 3 or more NPFs (23%) were 21 months, 13 months and 8 months respectively.Conclusions: A prognostic index using widely available baseline characteristics could improve prognostication in clinical practice. Stratification of patients using the prognostic index could improve the efficiency of randomised trials. Development (n=378) Validation (n=564) Negative prognostic feature% of womenHRp-value% of womenHRp-valueBrain metastasis32.410.00641.730.002Neutrophil count >ULN121.910.001151.350.03ALP >2 x ULN221.660.001201.380.008Age >60451.490.001451.210.05Hb <120g/L351.480.03291.230.05ER negative401.360.02281.52<0.001ECOG 2 or 3271.320.03251.500.001ULN = upper limit of normal Development (n=378) Validation (n=564) Number of NPFs% of womenHRp-value% of womenHRp-value 150.850.30170.670.0021 or 2631.00-601.00-3 or more222.11<0.001231.97<0.001NPF = negative prognostic feature Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 4066.