Predictors of long-term outcome following high-dose chemotherapy in high-risk primary breast cancer

We report on a predictive model of long-term outcome in 114 high-risk breast cancer patients treated with high-dose chemotherapy between 1989 and 1994. Paraffin-blocks from 90 of the 114 primaries were assessed for the presence of five risk factors: grade, mitotic index, protein expression of p53, HER2/neu, and oestrogen/progesterone receptor status; we could analyse the effect of risk factors in 84 of these 90 tumours. Seven-year relapse-free and overall survival was 58% (95% confidence interval 44–74%) and 82% (95% confidence interval 71–94%) vs 33% (95% confidence interval 21–52%) and 41% (95% confidence interval 28–60%) for patients whose primary tumours displayed ⩾3 risk factors vs patients with ⩽2 risk factors. For the entire group of 168 high-risk breast cancer patients, inflammatory stage IIIB disease and involved post-mastectomy margins were associated with decreased relapse-free survival and overall survival; patients treated with non-doxorubicin containing standard adjuvant therapy experienced worse overall survival (RR, 2.08; 95% confidence interval 1.04 to 4.16; P=0.04), while adjuvant tamoxifen improved overall survival (RR, 0.65; 95% confidence interval 0.41–1.01; P=0.054). Future trial designs and patient selection for studies specific for high-risk breast cancer patients should include appropriate prognostic models. Validation of such models could come from recently completed randomised, prospective trials. British Journal of Cancer (2002) 87, 281–288. doi:10.1038/sj.bjc.6600450 www.bjcancer.com © 2002 Cancer Research UK