S4-3: Prospective Comparison of Risk Assessment Tools in Early Breast Cancer (Recurrence Score, uPA/PAI-1, Central Grade, and Luminal Subtypes): Final Correlation Analysis from the Phase III WSG-Plan B Trial.

Abstract

Abstract Background: Both the Recurrence Score® (RS) multi-gene assay and invasion factors uPA/PAI-1 are recommended by guidelines (ASCO, AGO) for decision support regarding adjuvant chemotherapy (cht) in patients with early breast cancer (BC). In the EC-DOC trial, we previously suggested central grade 3 and high Ki-67 levels as independent predictive markers for enhanced benefit of taxane-based cht in intermediate risk BC. Here, we present the final WSG-Plan B trial correlation analysis of these risk assessment tools. Methods: Plan B trial (evaluating anthracyline-free adjuvant cht, 6xTC, vs. 4xEC-4xDOC in HER2−negative BC; n=2,448 to be randomized for chemotherapy). RS has been used as the selection criterion for cht vs. endocrine therapy alone in HR+ BC (if RS≤11 in pN0 or pN1) since an amendment in August 2009. uPA/PAI-1 (by ELISA) is obtained as an optional risk factor. Evaluation of central grade and luminal B subtype (using 14% or 20% Ki-67 cut-offs) are performed by the independent trial pathologist in all HR+ tumors. Results: From April 2009 to June 2011, 3037 patients have been recruited and 2290 randomized. The study will be completed by August 2011. Data on RS are available in 2361 patients with HR+ tumors: RS is distributed as follows: 0–11 (18%), 12–25 (61%), >25 (21%). In 257 patients with 0–3 involved LN, cht was omitted based on RS results (12.3% of patients after amendment). By the last interim analysis in February 2011, data on central grade are available in 1509 patients and Ki-67 in 592 patients. An only moderate positive correlation was observed between Ki-67 and RS (Spearman's coefficient rs=0.336; p<0.001) as continuous variables and between RS and central grade (rs=0.498, p<0.001). High RS is predictive of high grade G3 (66% of RS25 high risk and 74% of RS30 high risk are G3), but 30–33% of tumors were assessed as G3 within the low-risk RS11-18 group and 14%-40% of G3 tumors were classified as low-risk RS11-18. When considered as risk categories, there was only a weak concordance between RS and uPA/PAI-1, using either standard RS (18; 30) or PlanB cut-offs (low risk ≤11 RS), with 67% of patients having high uPA/PAI-1 within the low/intermediate-RS subgroups. High-risk RS is predictive of high-risk by uPA/PAI-1, poor grade and luminal B subtype. However, the converse is not true, since clinically relevant proportions (between 33–66%) of patients identified by uPA/PAI-1 or Ki-67 as being at high-risk had low or intermediate RS. Final correlation analyses in all Plan B patients with HR+ tumors (n>2.500) will be presented at the meeting. Discussion: For the first time, risk groups according to RS, Ki-67 and uPA/PAI-1 have been prospectively compared within a phase III trial setting both in N0 and N+ early BC. These data show high-risk status according to RS is predictive of high risk by uPA/PAI-1, high grade by central pathology and luminal B subtype, but not converse. In the low and intermediate risk RS groups, there are still patients considered high-risk according to uPA/PAI-1, Ki-67 or central grade. Further follow-up of the WSG-Plan B trial will clarify the clinical significance of these findings regarding patient outcomes. Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr S4-3.