Comparison of outcome prediction in node-negative breast cancer based on biomarkers uPA/PAI-1 or Adjuvant Online using the 10-year follow-up of the randomized multicenter Chemo N0 trial
534 Background: Chemo N0 was the first multicenter randomized clinical trial in breast cancer using biomarkers uPA/PAI-1 for risk stratification and chemotherapy selection. Chemo N0 validated uPA/PAI-1 as clinically relevant parameters for risk assessment in N0 breast cancer. About 50% of N0 patients have low uPA/PAI-1 in their primary tumor with excellent 5-year OS (>95%), even without adjuvant therapy. Patients with high uPA/PAI-1 benefit from adjuvant chemotherapy. We evaluated whether biomarkers can predict outcome more accurately than a computer-program. Methods: In Chemo N0, patients (n=689, 1993–98) were stratified according to uPA/PAI-1: High-risk patients were randomized (6x CMF vs. observation), low-risk patients observed. Retrospectively, we divided recruited patients into two groups (chemotherapy vs. no adjuvant therapy) and stratified using uPA/PAI-1 (low vs. high). Individual 10-year OS was calculated by www.adjuvantonline.org (Version 7.0); these estimated values were compared to the observed 10-year OS in Chemo N0. Results: 81/151 patients with already available complete follow-up data 10 years after trial start have 10-year OS results. Median 10-year OS estimated by Adjuvant Online was 77.1% (median 78.6%) taking into account administered adjuvant therapy. 10-year Chemo N0 follow-up revealed an observed 10-year OS of 66.7% (27/81 deceased). 64/81 did not receive chemotherapy: In high-risk untreated patients (n=22), 10-year OS was 54.6% vs. estimated 75.0%. In CMF-treated patients (all high-risk, n=17), observed OS was 58.8% vs. estimated 74.6%. Conclusions: For the first time, risk assessment by novel biomarkers is compared to that by Adjuvant Online in data from a randomized clinical trial. In patients with high uPA/PAI-1, the individual 10-year OS calculated by Adjuvant Online seems to be underestimated compared to observed patient outcome. Further follow-up data will show whether uPA/PAI-1 can predict the individual course of disease more precisely than a computer-program based on large clinical data sets. Final Chemo N0 10-year follow-up is currently being completed. No significant financial relationships to disclose.