This retrospective chart review with linked tissue analysis was approved by the Institutional Review Boards at the five participating institutions in the USA and Japan: The University of Texas MD Anderson Cancer Center (protocol number: PA15-0905), University of Hawai’i Cancer Center (protocol number: RA-2017-021), Medical College of Wisconsin (protocol number: PRO00029366), Anne Arundel Medical Center (protocol number: 1056834), and Showa University (protocol number: 2383). A waiver of informed consent was granted based on the study’s retrospective nature. We reviewed the institutional databases to identify patients with ER-positive, HER2-negative, node-negative invasive breast cancer who had 21-gene signature assay data available and who had undergone definitive surgery between April, 2005 and May, 2018 and adjuvant endocrine therapy without any cytotoxic agents. We included only patients for whom archival FFPE tissue from definitive surgery was available. We excluded patients with pathological T4, pathological node-positive, distant metastatic disease, and male patients. Nodal micrometastasis was allowed. We included patients with RS of 11–25 according to TAILORx intermediate group.
From the database, we extracted the 21-gene signature assay RS, age at diagnosis, menopausal status at diagnosis, histological type (ductal, lobular, or others), pathological T stage, progesterone receptor (PR) status, nuclear grade, and lymphovascular invasion status. ER and PR positivity were defined as tumors with > 1% of the cells positive using immunohistochemistry (IHC) staining. HER2 positivity was defined as a HER2/CEP17 fluorescence in situ hybridization ratio of ≥ 2.0 and/or an IHC staining score of 3+ . At the MD Anderson Cancer Center, Breast Medical Oncology database (Protocol number: 2004 − 0541) was used to identify the eligible patients.
Sample handling and processing
Archival FFPE tissue from either presurgical biopsy or resection specimens was prepared locally and sent to Sysmex Corporation (Kobe, Japan) under a Research Agreement. RNA was extracted from archival FFPE tissues. RNA was used for generation of second-strand cDNA, and the generated cDNA was amplified, then biotinylated and fragmented with an Ovation® FFPE WTA System (Tecan Genomics, Inc., Redwood City, CA), followed by hybridization to a GeneChip™ Human Genome U133 Plus 2.0 array (Thermo Fisher Scientific Inc., Waltham, MA) overnight (for 18 hours) according to the manufacturer’s protocol. Finally, the hybridized DNA microarray was fluorescence stained with GeneChip™ Fluidics Station 450Dx v.2 and scanned with the GeneChip™ Scanner 3000Dx v.2 (both from Thermo Fisher Scientific).
The microarray mRNA expression data for 95 genes were used for 95-gene signature analysis. The list of probes set IDs is shown in Table S1. First, total RNA, including mRNA, was extracted from FFPE tissue by using the RNeasy FFPE Kit (QIAGEN, Hilden, Germany). Next, cDNA was amplified from the mRNA using the Ovation® FFPE WTA System (Tecan Genomics) and then biotin labeled using the Encore® Biotin Module (Tecan Genomics). For microarray analysis, the GeneChip Hybridization, Wash, and Stain Kit; GeneChip™ Hybridization Control Kit; GeneChip™ Human Genome U133 Plus 2.0 Array; and GeneChip™ System 3000Dx v.2 (Thermo Fisher Scientific) were used.
After obtaining the microarray data, quality control of data was performed using Expression Console™ software (Thermo Fisher Scientific). Only data satisfying the internal standards were subjected to 95-gene signature analysis. Using the 95-gene signature algorithm, a 95-gene signature score (about − 0.1000 to 0.1000) was calculated from the expression levels of 95 genes on the microarray normalized by the ref-RMA method, and cases were classified into 95GC-H (> 0.0000) or 95GC-L (≤ 0.0000) groups. For details on the method of calculating the 95-gene signature score, refer to U.S. patent no. US20110263444A1.
Standard descriptive statistics and frequency tabulation were used to summarize data. The chi-square test and Fisher’s exact test were used to evaluate the association between two categorical variables. The Kruskal-Wallis test was used to compare the distributions of continuous variables among different groups. Univariate and cox regression models were used to investigate the association between each variable and time to recurrence (TTR). TTR was defined as the time from surgery to invasive disease recurrence or breast cancer–specific death. Patients who had not experienced recurrence or death were censored at the time of last follow-up. A Kaplan-Meier estimate with a log-rank test was used to compare TTR by group. Scatter plots were used to show the distribution of 95-gene signature and 21-gene signature assay scores. Pearson's correlation coefficient was used to assess the association between 95-gene signature score and 21-gene signature assay score. All analyses were carried out using STATA 14 software (StataCorp, College Station, TX).