Relationship of Ox-E/ER signature expression with clinical parameters and outcome in age-stratified cohorts

Christina Yau — 2008-09-22T00:00:00Z

In their recent editorial (Breast Cancer Res 2008, 10:109), Neven et al. highlight our recent finding (Breast Cancer Res 2008, 10:R61) that an experimentally derived oxidative stress gene expression signature, ‘Ox-E/ER’, identifies an aggressive subset of primary estrogen receptor (ER)-positive breast cancers associated with poor clinical outcome, and appears to outperform loss of progesterone receptor (PR) expression as a prognostic variable. Given their earlier studies suggesting an age-related association between PR status and HER2 overexpression in ER-positive breast cancers that could confound our oxidative stress analysis, they suggested that we further explore the prognostic relationship between our Ox-E/ER signature index and breast cancer PR status, stratifying for age-at-diagnosis. Returning to the public expression microarray datasets reported in our study, we were able to stratify most of the ER-positive breast cancers into either younger (<=age 45) or older (> age 45) age groups to correlate age group Ox-E/ER index values with array determined PR mRNA levels, ERBB2 overexpression (determined by mRNA content) and the average expression of the previously reported 71-gene proliferation signature. We also used the optimized Ox-E/ER index cutpoint to dichotomize available ER-positive age cohorts annotated for outcome -- either disease-specific survival (DSS; n = 201) or relapse-free survival (RFS; n = 263)-- in order to compare the prognostic value of the Ox-E/ER index with PR status. Within both age cohorts the Ox-E/ER index values showed significant positive correlations with the proliferation gene signature (younger: Pearson r = 0.32, p = 0.0009; older: r = 0.17, p = 0.0036) and ERBB2 status (younger: r = 0.29, p = 0.0028; older: r = 0.21, p = 0.0003), as well as a significant negative correlation with PR expression (younger: r = -0.25, p = 0.0097; older: r = -0.18, p = 0.0023). With regard to both DSS or RFS, the Ox-E/ER index cut-point (high vs. low status) appears to perform better as a prognostic marker for younger age-at-onset ER-positive breast cancers cases (DSS: younger: log-rank p= 0.005; older: log-rank p= 0.019; RFS: younger: log-rank p= 0.02; older: log rank p= 0.07). In comparison to PR status, while the Ox-E/ER index cut-point appears to perform better than PR status as a prognostic marker for the entire group (as published), this prognostic superiority may be restricted to the younger age cohort (DSS: log-rank p=0.005 for Ox-E/ER vs. 0.25 for PR status; RFS: log-rank p=0.02 for Ox-E/ER vs. 0.60 for PR status) , since we fail to see significant differences in their respective abilities to determine the outcome of older ER-positive cases. An important caveat to note is the small number of PR-negative cases in the younger cohort; additional studies utilizing age cohorts better balanced for PR status are warranted to confirm this finding.

Comment by Christina Yau and Christopher C Benz

Article Comments - 'A gene-expression signature of loss of oestrogen receptor function and oxidative stress links ER positive breast cancers with an absent progesterone receptor and a poor prognosis.'

Relationship of Ox-E/ER signature expression with clinical parameters and outcome in age-stratified cohorts