Impact of treatment characteristics on response of different breast cancer subtypes: Pooled multilayer analysis of the German neoadjuvant chemotherapy trials.

501 Background: Eight German neoadjuvant chemotherapy trials including 6,625 patients with uniform protocol and outcome definitions were conducted between 1998 and 2006 with pathological complete response (pCR) as primary endpoint. Methods: We used a more sophisticated multilevel model with random effects at the level of trials and regimens to pooled individual patient data. The model fully accounted for heterogeneity between trials in pCR. The analysis was restricted to a data set of 3,332 patients from 7 trials and 12 treatment arms with complete baseline parameters and no confounding by treatment indication based on interim response. Results: pCR was associated with an increase in cycle numbers (odds ratio [OR ] 1.2 with every two additional cycles; p = 0.009), with regimens using higher cumulative anthracycline doses (≥ 300 mg/m2 doxorubicin or ≥ 450 mg/m2 epirubicin; OR 1.6; p = 0.002), regimens with higher cumulative taxane doses (≥ 400 mg/m2 docetaxel or ≥ 700 mg/m2 paclitaxel; OR 1.6; p = 0.009), ...