Abstract Introduction: CDK4/6 inhibitors, such as palbociclib, target the cyclin-dependent kinases 4 and 6, essential for cell cycle regulation. Preclinical studies have suggested potential determinants of palbociclib sensitivity and clinical data have recently indicated loss-of function mutations in RB1 and increased expression of CCNE1 as potential biomarkers of resistance in a fraction of breast cancer (BC) patients. In line with these studies, using a panel of resistant luminal BC cell lines made resistant to palbociclib we recently demonstrated a common transcriptional signature of resistance to this agent, including the over-expression of CCNE1 and under-expression of RB1. Based on these data, we developed a minimal signature of de-novo resistance to palbociclib (ratio between CCNE1 and RB1, CCNE1/RB1) and demonstrated its prognostic role in both preclinical and clinical datasets. Gene expression signatures, such as CCNE1/RB1, are limited to the analysis of tissue samples because they are not easily detectable in liquid biopsies. Based on the key role DNA-methylation (DNAm) plays in regulating gene expression in human cells, we hypothesized that DNAm could be used as surrogate of our gene expression signature, with the further advantage of being easily adapted to non-invasive screenings. Exploiting this concept, here we report on a DNAm-based signature mirroring CCNE1/RB1 and its potential role in predicting resistance to palbociclib in BC patients. Materials and methods: Seven estrogen receptor positive breast cancer cell lines with acquired in vitro resistance to palbociclib were used. DNAm profiles of untreated parental cells and of their palbociclib-resistant counterparts were generated through Illumina Infinium MethylationEPIC BeadChip. For each cell line, candidate DNAm sites associated with resistance to palbociclib were selected by differential DNAm analysis in resistant versus sensitive counterparts. To identify DNAm sites/regions distinguishing CCNE1/RB1 high vs low patients, RNA-seq data from ER+/HER2- patients in the TCGA-BRCA dataset were considered and patients were classified as CCNE1/RB1 high (≥90% quantile) or low (≤10% quantile). Matched DNAm data were exploited to perform differential DNAm analysis between CCNE1/RB1 high and low samples by Rocker-meth, a tool recently developed by our group to perform differential DNAm analysis. The pharmacogenomic cancerrxgene dataset was used to test the association between the CCNE1/RB1 DNAm signature and the measure of effectiveness of palbociclib across multiple BC cell lines. Results: From the differential methylation analysis of CCNE1/RB1 high versus low ER+/HER2- BC patients from the TCGA-BRCA dataset we identified the set of sites and regions that compose our DNAm-based signature. In-silico functional analysis of differentially methylated sites and regions showed significant enrichment of regulatory-related terms. Both hyper- and hypo- DNAm sites in resistant versus untreated sensitive cell lines were significantly enriched in the set of differentially methylated sites in CCNE1/RB1 high versus low TCGA samples. In the cancerrxgene dataset, the CCNE1/RB1 DNAm signature correlated with higher IC50 in breast cancer cell lines. Conclusions: This study shows that DNAm is a potential predictor of resistance to palbociclib in ER+/HER2- patients. The DNAm-based signature mirroring CCNE1/RB1 developed using TCGA-BRCA dataset was validated in our preclinical BC cell lines and associated with effectiveness of palbociclib in an independent BC cell line dataset. Ongoing studies on the analysis of DNAm in liquid biopsies of metastatic luminal BC patients will evaluate the reliability of this DNAm signature in predicting resistance to palbociclib. Citation Format: Matteo Benelli, Martina Bonechi, Dario Romagnoli, Chiara Biagioni, Ilenia Migliaccio, Francesca De Luca, Francesca Galardi, Angelo Di Leo, Luca Malorni. A DNA-methylation signature to predict resistance to the CDK4/6 inhibitor palbociclib [abstract]. In: Proceedings of the 2019 San Antonio Breast Cancer Symposium; 2019 Dec 10-14; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2020;80(4 Suppl):Abstract nr P4-04-07.
Read full abstract