Two recently identified and tested gene-expression signatures may guide clinicians in selecting which patients may show a response to palbociciib, according to findings presented at the IMPAKT Breast Cancer Conference held 7–9 May 2015 in Brussels, Belgium.
Ilenia Migliaccio, Translational Research Unit, Hospital of Prato, Istituto Toscano Tumori, Prato, Italy, headed a team of investigators from Italy that aimed to tackle the problem of resistance to palbociclib, an inhibitor of proteins kinase 4 and 6 (CDK4/6), which play a role in cell division and proliferation. Palbociclib has shown promise in the treatment of patients with breast cancer and oestrogen receptor (ER)-positive/HER2-negative tumours.
The CyclinD/CDK4-6/Rb/E2F pathway is frequently deregulated in breast cancer, particularly in luminal subtypes in which cyclin D1 amplification and CDK4 gain are common. Pre-clinical data suggest that, beyond ER, other biomarkers of response to palbociclib might be cyclin D1 gene-amplification/overexpression, low levels of p16 and high levels of retinoblastoma (RB). Hints from clinical trials in breast cancer patients suggest that screening for CCND1 amplification or p16 loss does not seem to improve patient selection in the PALOMA1 trial and low clinical benefit rate of palbociclib monotherapy in a ER-positive, RB-positive heavily pretreated population. Additional studies are needed to better identify sensitive/resistant patients.
One known marker of palbociclib resistance, RB genetic loss, led the investigators to study whether a genetic signature of functional RB loss would predict resistance to palbociclib. They further sought to identify a gene expression signature that would be predictive of response to palbociclib.
The team identified two different signatures: One signature showed functional RB loss from breast cancer that was identified in the TCGA dataset (RBsig), the other signature indicated palbociclib sensitivity that was identified by comparing 13 palbociclib-sensitive and 13 palbociclib-resistant cell lines in the Cancer Cell Line Encyclopedia (CCLE) dataset (PDSENSsig).
These signatures were tested for prognostic and predictive value in a broad breast cancer gene expression meta-dataset (3458), and the ability of these signatures to discriminate between palbociclib sensitive or resistant cell lines was also tested using a validation dataset of breast cancer cell lines (GSE48213), according to a previous procedure by Finn et al. (PMID:19874578).
The gene expression signatures showed good predictive value and expression of the signatures in patient tumours correlated with palbociclib response
These tests revealed that the RB signature, which comprised 87 genes, was predictive of RB status in the TCGA dataset across all molecular subtypes. Untreated patients and patients that had received endocrine treatment with ER-positive tumours that also expressed high levels of the RB signature demonstrated significantly worse recurrence-free survival (RFS) compared to similar patients with tumours expressing low levels of the RB signature; hazard ratio (HR) 2.34; p = 7.5 e−09 and HR 2.56; p=5.4 e−11, respectively. In the validation dataset, the RB signature was a strong predictor in evaluating sensitivity or resistance to palbociclib; ROC area under curve (AUC) = 0.93.
The second gene signature tested, PBSENSsig had 20 genes that were found to be upregulated in palbociclib sensitive cell lines. Untreated or endocrine treated patients with ER-positive tumours also expressing high PDSENSsig demonstrated significantly better RFS compared to patients with ER-positive tumours expressing low levels of PDSENSsig; HR 0.63 (p = 0.0017) and HR 0.66 (p = 0.0061), respectively. The PDSENSsig predictor also performed well in the validation dataset; ROC AUC = 0.76.
Dr Lisa Carey of the University of North Carolina, USA, who discussed the study results, said that functional Rb loss signature associated with luminal B tumours is prognostic, the finding is plausible and consistent. CDK4/6 inhibitor resistance may relate to subtype. Signature based on E2F, may reflect proliferation more than Rb. Palbociclib sensitivity and resistance signatures can be created and modestly tracked with expected phenotype and behavior. It is not validated, but it is important as a finding. She questioned if it will be better than just Rb-based assay. With this emerging class of drugs, predictive biomarker efforts are important but not yet successful, and may need to take known variables into account (e.g. subtype).
Conclusion
Treatment decisions regarding adjuvant chemotherapy in breast cancer were historically based on clinico-pathologic factors but genomic analyses are currently being used to characterise molecular breast cancer subtypes and guide treatment choice. These investigators derived two gene expression signatures that may aid in selecting patients more likely to benefit from palbociclib treatment. These findings warrant further validation in breast cancer cell lines and in patients with breast cancer being treated with palbociclib.
Reference
40O Identification of gene expression signatures of palbociclib (PD) response in breast cancer (BC)