Abstract P2-10-08: Prospective comparison of Recurrence Score and different definitions of luminal subtypes by central pathology assessment of single markers in early breast cancer: results from the phase III WSG-planB Trial

Background: Routine use of multigene RT-PCR based assays as Recurrence Score (RS) vs. single markers (grade, uPA/PAI-1, HR, HER2, Ki-67) is controversially discussed in early BC. Several definitions of luminal A/B subtypes have been proposed by St. Gallen guidelines and individual researchers (grade 1/2 vs. 3, Ki-67 cut-offs 14 or 20%). Recently, integration of PR>20% was proposed as an immunhistochemical luminal A subtype definition (Ki-67 Methods: planB trial (04/09 to 11/11: n=2,449 randomized for 6xTC vs. 4xEC-4xDOC in locally HER2− BC; n=3197 registered; n=3072 available for central tumor bank). RS was used as selection criterion for chemotherapy (CTx) omission in HR+ BC (if RS Results: RS distribution in 2569 HR+ tumors: 0–11 (18%), 12–25 (60%), >25 (22%) (RS1853%/34%/13%). Luminal A subtype based on Ki-67 cut-off of 20% was included in the luminal A/Ki-67 Luminal A/B based on local/central grade: 79%/21%; 67.9%/32.1%. In 354 pN0-1 patients, CTx was omitted based on low RS (88% compliance). In this group, 62% were high-risk by clinical-pathological criteria. Allocation to luminal A/B subtypes based on local/central grade varied substantially: overall concordance was 72%, but 40% of locally luminal B (HR+/G3) were luminal A (HR+/G1/2) by central and 60% of centrally luminal B were luminal A locally. Moderate correlations were found between RS and central grade (Spearman correlation rs=0.317; p < 0.001), local grade (0.321; p < 0.001) and Ki-67 (rs = 0.374; p < 0.001), particularly due to poor correlations in the RS group Inclusion of PR>20% moderately improved the correlation between luminal subtypes and RS to rs=0.34; p < 0.001. Data on allocation to luminal subtypes by local/central Ki-67 will be presented at the meeting. Conclusions: Our results represent the first prospective correlation of different luminal A/B definitions vs. a guideline mentioned gene signature (RS). High RS usually implies high grade and luminal B classification, the converse is not true. There is substantial heterogeneity in risk assessment by luminal A/B classification and grade in low/intermediate RS groups. Concordance for central/local assessment of grade and luminal A/B status was limited. There is strong need for standardization of molecular subtype's immunhistochemical/clinical definition prior to implementation into daily routine. Clinical significance of these findings will be addressed by further follow up of the WSG-planB trial. Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr P2-10-08.