Abstract 5294: Biomarkers in residual disease after neoadjuvant chemotherapy of breast cancer predict long-term outcome

Abstract

Abstract Aim: to assess the prognostic value of 24 candidate genes in residual breast cancers following neoadjuvant chemotherapy. Background Patients who do not achieve pathological complete response after neoadjuvant chemotherapy have increased risk of recurrence and represent a challenge for clinical management. Expression profiling of residual disease may identify genes related to chemoresitance that allow the development of a profile related to long-term outcome. We sought to assess this by evaluating the expression of ESR1, ERBB2 and genes playing a role in stemcellness (ALDHA1, CD44, STAT3), proliferation (TOP2A, MKI67, AURKA), apoptosis (BCL2, BCL2L1, PAWR), immunological response (CD3D, CXCL13, STAT1), epithelial to mesenchymal transition (SNAI1, SNAI2, SOX9, TWIST), stromal activation (DECORIN, SPARC, PLAU), energy metabolism (ACACB, LDHB), and ERK/Ras activation (DUSP4). Patients and Methods A total of 126 primary breast cancer patients treated with neoadjuvant chemotherapy were included, with a median follow up of 4.1 years. Overall, 66% were ER+, 52% PgR+, 12% Her2+; 3% grade 1, 49% grade 2, 48% grade 3; 52% node -ve and 48% node +ve. Gene expression profiles were done using the NanoString nCounter system from 126 post-treatment and 56 matched pre-treatment tumour samples. Differential gene expression changes were tested by paired t-test. The association of the biomarkers in residual disease with time to relapse was assessed by Cox regression analysis. Penalised Elastic Net Cox regression was used to identify independent prognostic gene signatures. Benjamini-Hochberg methodology was used for multiple comparison correction for survival analysis; threshold of q-value was set to have < 1 false positive. Results Fifteen genes were found to be up- and 9 downregulated following chemotherapy. Overall 20 genes changed significantly and 17 of these changed in the ER+/Her2- subgroup. Among those, ALDH1A1, STAT1 and TWIST1 were most significant. 14 genes were prognostic in unvariable analysis of all 126 residual samples. In 118 patients with complete clinico-pathological data available, ACACB, CD3D, MKI67 and TOP2A were informative independent of ER, PgR and Her2 status. PAWR was positively correlated with good outcome in ER+ but negatively in ER- patients (interaction test p<0.001). From the Elastic Net regression analysis in all 126 patients, 6 genes were identified (ACACB, CD3D, DECORIN, ESR1, MKI67, PLAU) to provide significant prognostic information independent of ER, PgR and Her2 status. In 78 ER+/Her2- samples 4 genes (ACACB, ERBB2, MKI67, PAWR) were similarly informative. Conclusions Our findings demonstrate that gene expression of diverse biological functions is affected by chemotherapy and for some is prognostic in residual disease. If validated, our multivariable gene models may potentially identify patients who may need additional post-neoadjuvant treatment to improve prognosis. Citation Format: Richard Buus, Marie Klintman, Amna Sheri, Maggie Cheang, Mitch Dowsett. Biomarkers in residual disease after neoadjuvant chemotherapy of breast cancer predict long-term outcome. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 5294. doi:10.1158/1538-7445.AM2015-5294