부산대학교 도서관

Background: Poor inter-laboratory comparability of common clinically used breast cancer biomarkers led to a proposal of statistical standardization (SS) of laboratory results, similar to bone mineral density (BMD) z-scores. This analysis is the first utilization of SS in a trial where all women received TAM. Methods: MA.14 allocated 667 postmenopausal women to TAM +/- Octreotide LAR (OCT) based on locally determined ER/PR, without HER2 status. At 9.8 yrs median follow-up, the secondary endpoint of relapse-free survival (RFS) had a non-significant hazard ratio (HR) for TAM-OCT to TAM of 0.87 (95% CI 0.63-1.21; p = 0.40). 299 patients who were representative of MA.14 patients by treatment and stratification factors (exact Fisher p-values=0.19-0.90) had their tumors centrally assessed for ER, PR, and HER2 by RT-PCR. Continuous values were used for SS of each biomarker. Univariate (uni) assessment used similar categorizations as those for BMD, assigning ER/PR/HER2 values by number of standard deviations (SD) about the mean (Group 1, z-score ≥ 1.0 SD below mean; Group 2, z-score <1.0 SD below mean; Group 3, z-score ≤ 1.0 SD above mean; Group 4, z-score >1.0 SD above mean). A log- rank statistic was used to test for differences between SS biomarker groups with K-M plots for graphical description. Multivariate (multi) effects of SS biomarkers and baseline patient characteristics on RFS were examined with exploratory (un)stratified Cox step-wise forward regression, adding a factor if likelihood ratio criterion was p ≤ 0.05. Sensitivity analyses used a prior external HER2+ cut-point of ≥ 1.32 SD. Results: 292 patient samples passing internal analytical quality control were included in this analysis. Uni analyses indicated SS ER was not associated with RFS (p = 0.31). SS PR had a significant uni effect on RFS [p = 0.03; Group 4 compared to Group 1, HR of 0.33 (95% CI 0.12 -0.90); Group 3 compared to Group 1, HR of 0.42 (95% CI 0.21-0.83); and Group 2 compared to Group 1 HR of 0.70 (95%CI 0.36-1.37)]. SS HER2 also had a significant uni effect on RFS [p = 0.004; Group 4 compared to Group 1, HR of 0.90 (95% CI 0.37-2.16)]; Group 3 compared to Group 1, HR of 0.39 (95% CI 0.18-0.84); and, Group 2 compared to Group 1, HR of 0.34 (95% CI 0.16-0.70)]. Multi stratified/unstratified Cox models indicated T1 tumours (p = 0.02/p = 0.0002) and higher SS PR (p = 0.02/0.01) were associated with significantly longer RFS; other unstratified results showed that N-ve patients had better RFS (p < .0001), while local ER/PR status did not impact RFS (p > 0.05). The HER2+ cut-point of ≥ 1.32 SD indicated directionally worse RFS (uni p-value=0.05; multi p-value=0.06). [ABSTRACT FROM AUTHOR]