Abstract 3537
Background
Chemotherapy (CT) combined with anti-HER2 drugs (H) is the treatment of choice for HER2+ early breast cancer (BC) patients (pts). However HER2+ tumors are clinically and biologically heterogeneous, and treatment response largely varies according to ER status. Predictive biomarkers are urgently needed in this context. We have recently developed a meta-dataset of clinical trials of neoadjuvant CT +/− H (trastuzumab, lapatinib, or both) in HER2+ BC pts annotated for gene expression, hormone receptor status and pathological complete response (pCR) rates. We have shown that a gene-signature (GS) of RB-1 loss-of-function (RBsig) seems to be predictive of response to neoadjuvant CT +/− H in ER+/HER2+ BC pts in this meta-dataset. Here we report the results of additional analyses aimed to evaluate RBsig’s predictive value against 10 previously developed GSs in the subset of ER+/HER2+ BC pts.
Methods
The association of RBsig with pCR was evaluated in comparison with previously described GSs of: low ER signaling, p53 mutation, high PI3K pathway signaling, high expression of HER2 amplicon genes, PAM 50 and 5 immune-related GSs. For each GS, samples were classified as High or Low group using a previously described classifier. Odds Ratio (OR) performance was calculated using the ROCR (v. 1.0) package in R and plotted by forest plot using the survcomp (v. 1.24) package.
Results
RBsig and the HER2 amplicon GS were best at predicting response to neoadjuvant CT +/− H (211 pts; p:0.017). In the subgroup of pts treated with CT alone (n = 94), only the PI3K pathway GS was significantly associated with pCR (p:0.026). In pts treated with CT + H (n = 117) only the HER2 amplicon GS significantly correlated with pCR (p:0.042). RBsig showed a similar trend in both these subgroups (p: 0.078 and 0.104, respectively) The immune GSs and PAM50 were not associated with pCR, independent of treatment received.
Conclusions
RBsig and HER2 amplicon GS are strongly associated with pCR in ER+/HER2+ tumors unselected for treatment. These results support the potential use of such GSs as predictive markers of response to CT +/−H in ER+/HER2+ BC pts. Validation studies are warranted.
Clinical trial identification
Legal entity responsible for the study
Angelo Di Leo
Funding
None
Disclosure
A. Di Leo: Consultant/Advisory Board: Novartis, Pfizer, Lilly. All other authors have declared no conflicts of interest.