Abstract P6-01-09: Assessment of the prognostic and predictive ability of a gene signature compared to histological grade in estrogen receptor positive, HER2 negative breast cancer

Abstract

Abstract Background: Genomic biomarkers have been widely adopted to assist in clinical decision making regarding chemotherapy use in estrogen receptor (ER)-positive, HER2-negative breast cancer. First generation genomic signatures (FGGS) serve predominately as prognostic biomarkers and secondarily have a role in prediction of chemotherapy response. The majority of the FGGSs provide similar prognostic information which mainly capture in different ways tumor proliferation. While several studies have compared the prognostic value of FGGSs to clinico-pathological variables, few studies have performed a similar comparison for their predictive value. For this reason, we aimed to compare both the prognostic and predictive value of histological grade and the genomic marker. Methods: We retrieved publicly available cDNA microarray data from 1,373 primary ER+/HER2- breast cancers (n=721 treated with various or unknown, n=350 untreated node negative, n=302 treated with neoadjuvant chemotherapy). We developed a genomic signature simulated from recurrence online (http://www.recurrenceonline.com/) to calculate recurrence score and recurrence risk using pre-defined sets of genes by cDNA microarray (B Gyorffy Breast Cancer Res Treat 2012). Breast cancers were categorized as low, intermediate or high risk for distant recurrence using grade and genomic signature. We compared the prognostic and predictive information provided by histological grade to the genomic signature. The outcome of interest in untreated patient was distant event free survival. The outcome of interest in the anthracycline-taxane treated patients was pathological complete response (pCR) in breast and axilla. Results: Fifty five, 28 and 17% breast cancers were classified as low, intermediate and high risk by genomic signature and 22, 59 and 19% as grade I, II and III respectively. The genomic signature classified 11% of grade I/II cancers (126/1108) and only 42% of grade III cancers (112/265) as high risk, and 29% of GradeIII (77/265) as low risk. Univariate analysis in the untreated cohort, showed both histological grade (overall p=0.007) and the genomic signature (p<0.001) could predict prognosis. In multivariate analyses for tumor size, age, grade and genomic signature, only the genomic signature remained statistically significant for prognosis. As expected a significantly higher rate of pCR was observed in histological grade III cancers (15.9%) compared to grade I (3.4%) and II (3.8%) after neoadjuvant chemotherapy (NAC). Results were similar using the genomic signature with pCR rates of 4.6%, 5.7% and 16.5% for low, intermediate and high risk, respectively. Grade I and II cancers (n=189) classified as high risk by the genomic signature had a pCR rate of only 2.8%. Instead, the grade III tumors which were also defined at high risk by the genomic signature had a pCR rate of 26.5%. Multivariate predictive models showed neither biomarker retained statistical significance in predictive response to NAC. Interpretation: The genomic signature was better at identifying low risk cases compared to histological grade alone. There was no difference in prediction of NAC response between either biomarker. Better predictive biomarkers for NAC response are needed. Citation Format: Takayuki Iwamoto, Catherine Kelly, Giampalo Bianchini, Takeo Mizoo, Tomohiro Nogami, Takayuki Motoki, Tadahiko Shien, Naruto Taira, Naoki Hayashi, Naoki Niikura, Toshiyoshi Fujiwara, Hiroyoshi Doihara, Junji Matsuoka. Assessment of the prognostic and predictive ability of a gene signature compared to histological grade in estrogen receptor positive, HER2 negative breast cancer [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P6-01-09.