Serial expression analysis of breast tumors during neoadjuvant chemotherapy reveals changes in cell cycle and immune pathways associated with recurrence and response

Abstract

Introduction: The molecular biology involving neoadjuvant chemotherapy (NAC) response is poorly understood. To elucidate the impact of NAC on the breast cancer transcriptome and its association with clinical outcome, we analyzed gene expression data derived from serial tumor samples of patients with breast cancer who received NAC in the I-SPY 1 TRIAL.

Methods: Expression data were collected before treatment (T1), 24-96 hours after initiation of chemotherapy (T2) and at surgery (TS). Expression levels between T1 and T2 (T1 vs. T2; n = 36) and between T1 and TS (T1 vs. TS; n = 39) were compared. Subtype was assigned using the PAM50 gene signature. Differences in early gene expression changes (T2 - T1) between responders and nonresponders, as defined by residual cancer burden, were evaluated. Cox proportional hazards modeling was used to identify genes in residual tumors associated with recurrence-free survival (RFS). Pathway analysis was performed with Ingenuity software.

Results: When we compared expression profiles at T1 vs. T2 and at T1 vs. TS, we detected significantly altered expression of 150 and 59 transcripts, respectively. We observed notable downregulation of proliferation and immune-related genes at T2. Lower concordance in subtype assignment was observed between T1 and TS (62 %) than between T1 and T2 (75 %). Analysis of early gene expression changes (T2 - T1) revealed that decreased expression of cell cycle inhibitors was associated with poor response. Increased interferon signaling (TS - T1) and high expression of cell proliferation genes in residual tumors (TS) were associated with reduced RFS.

Conclusions: Serial gene expression analysis revealed candidate immune and proliferation pathways associated with response and recurrence. Larger studies incorporating the approach described here are warranted to identify predictive and prognostic biomarkers in the NAC setting for specific targeted therapies.

Clinical trial registration: ClinicalTrials.gov identifier: NCT00033397 . Registered 9 Apr 2002.

Figures

Fig. 1

Serial gene expression analysis in locally advanced breast cancer patients undergoing neoadjuvant chemotherapy (NAC). a Study schema. Gene expression analysis was performed on breast cancer tumors collected before treatment (T1), 24–96 hours after initiation of anthracycline-based NAC (T2) and at the time of surgery (TS). b Heat map showing results of supervised clustering analysis of expression profiles of tumors before treatment (T1) and 24–96 hours after initiation of NAC (T2). Rows indicate expression levels for each gene, and columns indicate individual samples. Blue indicates downregulation of gene expression, and red indicates upregulation of gene expression. The upper color bar indicates response to NAC as defined by residual cancer burden (RCB 0/I or RCB II/III). Bars on the left indicate assignment of genes to an ontology group (i.e., immune system– or proliferation-related genes). Her2 Human epidermal growth factor receptor 2, LumA luminal A, Lum B luminal B. c Subtype assignments of matched tumors at T1 and T2. Dark gray boxes running diagonally downward from top left indicate no change between two time points. d Heat map showing result of supervised clustering analysis of expression profiles of known nonresponding tumors at T1 and TS. The upper color bar indicates hormone receptor (HR) and HER2 status (blue = HR+HER2−; green = HR−HER2+; red = HR−HER2−; white = no data). The 10 most significant differentially expressed genes are indicated at the right of the heat map. Blue indicates downregulation of gene expression, and red indicates upregulation of gene expression. e Subtype assignments of matched tumors at T1 and TS. Dark gray boxes running diagonally downward from top left indicate no change between two time points

Fig. 2

Association between gene expression and response to chemotherapy as defined by residual cancer burden (RCB). a Scatterplot showing the expression of a representative cell proliferation gene, CDKN2B, at baseline (T1) and 24–96 hours after initiation of chemotherapy (T2) with the color of the points indicating response [red = RCB 0/I (responders); blue = RCB II/III (nonresponders)]. The diagonal line represents no change in gene expression levels (T1 = T2). The length of the vertical lines between points and the diagonal line represents the magnitude of expression change from T1 to T2. b–d Box plots show expression of CDKN2B in RCB 0/I and RCB II/III at T1 (b), at T2 (c) and the change in gene expression between two time points (T2 − T1) (d)

Fig. 3

Association of changes in gene expression between pretreatment and residual tumors (TS − T1) and recurrence-free survival (RFS). a Scatterplot showing the expression of a representative interferon signaling gene, IFIH1, at TS and T1, with the color of the points indicating outcome (red = no recurrence; blue = recurrence). The diagonal line represents no change in gene expression levels (T1 = TS). The length of the vertical lines between points and the diagonal line represents the magnitude of expression change from T1 to TS. Kaplan-Meier analysis of RFS among patients with high (top tertile: blue) or low expression (red) levels of IFIH1 at T1 (b), at TS (c) and the change in gene expression between two time points (TS − T1) (d)

Fig. 4

Association between gene expression in residual tumors (TS) and recurrence-free survival (RFS). Kaplan-Meier analysis of RFS among patients with high (top tertile: blue) or low expression (red) levels of a representative cell proliferation gene, CENPF, at T1 (a), at TS (b) and the change in gene expression between two time points (TS − T1) (c)

Fig. 5

Assessment of Ki-67 protein expression. a Scatterplot showing Ki-67 scores at baseline (T1) and 24–96 hours after initiation of chemotherapy (T2), with the colors and shapes of the points indicating increases (red squares) or decreases (blue circles). The diagonal line represents no change in Ki-67 score (T1 = T2). The length of the vertical lines between points and the diagonal line represents the magnitude of expression change from T1 to T2. b Kaplan-Meier analysis of recurrence-free survival (RFS) among patients with low (bottom tertile: red) or high (blue) Ki-67 scores at TS