CCNE1 and PLK1 Mediate Resistance to Palbociclib in HR+/HER2- Metastatic Breast Cancer

Abstract

Purpose: In hormone receptor-positive (HR+)/HER2- metastatic breast cancer (MBC), it is imperative to identify patients who respond poorly to cyclin-dependent kinase 4/6 inhibitors (CDK4/6i) and to discover therapeutic targets to reverse this resistance. Non-luminal breast cancer subtype and high levels of CCNE1 are candidate biomarkers in this setting, but further validation is needed.

Experimental design: We performed mRNA gene expression profiling and correlation with progression-free survival (PFS) on 455 tumor samples included in the phase III PEARL study, which assigned patients with HR+/HER2- MBC to receive palbociclib+endocrine therapy (ET) versus capecitabine. Estrogen receptor-positive (ER+)/HER2- breast cancer cell lines were used to generate and characterize resistance to palbociclib+ET.

Results: Non-luminal subtype was more prevalent in metastatic (14%) than in primary tumor samples (4%). Patients with non-luminal tumors had median PFS of 2.4 months with palbociclib+ET and 9.3 months with capecitabine; HR 4.16, adjusted P value < 0.0001. Tumors with high CCNE1 expression (above median) also had worse median PFS with palbociclib+ET (6.2 months) than with capecitabine (9.3 months); HR 1.55, adjusted P value = 0.0036. In patients refractory to palbociclib+ET (PFS in the lower quartile), we found higher levels of Polo-like kinase 1 (PLK1). In an independent data set (PALOMA3), tumors with high PLK1 show worse median PFS than those with low PLK1 expression under palbociclib+ET treatment. In ER+/HER2- cell line models, we show that PLK1 inhibition reverses resistance to palbociclib+ET.

Conclusions: We confirm the association of non-luminal subtype and CCNE1 with resistance to CDK4/6i+ET in HR+ MBC. High levels of PLK1 mRNA identify patients with poor response to palbociclib, suggesting PLK1 could also play a role in the setting of resistance to CDK4/6i.

Figure 1.

Non-luminal AIMS-HTG intrinsic subtype is associated with resistance to palbociclib but not to capecitabine. A, Intrinsic subtype distribution in patients with AIMS-HTG performed in primary tumor samples or (B) in metastatic tumor samples. C, PFS according to AIMS-HTG intrinsic subtypes in patients treated with capecitabine or (D) with palbociclib plus ET. In PFS curves, Kaplan–Meier plots are used to represent the survival curves, while Cox models, adjusted by prognosis variables, are used to represent the values.

Figure 2.

CCNE1 mRNA expression is increased in metastases and associated with resistance to palbociclib plus ET. CCNE1 mRNA expression levels according to (A) source of tumor sample or (B) AIMS-HTG subtypes. C, Levels of CCNE1 mRNA expression in a set of 14 paired tumors (primary-metastatic). D, Paired individual mRNA CCNE1 expression levels in 14 paired tumors (primary-metastatic). PFS analysis according to expression levels of CCNE1 in patients treated with (E) palbociclib plus ET or (F) with capecitabine. In PFS curves, Kaplan–Meier plots are used to represent the survival curves, while Cox models, adjusted by prognosis variables, are used to represent the values. LowExpr CCNE1 (below median value); HighExpr CCNE1 (above median value); median value = 6.980949.

Figure 3.

Clinicopathologic and genomic characteristics of extreme responders to palbociclib plus ET. A, PFS analysis in extreme responder patients (sensitive vs. refractory), under palbociclib plus ET treatment. B, Unsupervised hierarchical clustering complete linkage and euclidean distance of n = 229 patients treated with palbociclib plus ET according to 51 differentially expressed significant genes by SAM analysis (FDR < 0.05), in refractory versus sensitive tumors (each row represents a patient, and each column a gene). Also includes the measure of proliferation by Ki67, as well as the clinical subtype classification of patients in LumA, LumB, and non-luminal; the presence of ESR1 mutations; the presence of one or multiple metastases; and the type of response to treatment. C, Boxplot showing mRNA expression levels of PLK1 and CCNE1 according to tumor subtype and source of tissue in the whole biomarker cohort (D). Correlation plot between mRNA expression levels of PLK1 and CCNE1 in the whole biomarker cohort. E, Correlation matrix and unsupervised hierarchical clustering (complete linkage and euclidean distance of the 51 differentially expressed genes in refractory versus sensitive tumors to palbociclib plus ET. PFS analysis according to levels of mRNA PLK1 in patients treated with (F) palbociclib or (G) capecitabine. In PFS curves, Kaplan–Meier plots are used to represent the survival curves, while Cox models, adjusted by prognosis variables, are used to represent the values. LowExpr PLK1 (below median value); HighExpr PLK1 (above median value); median value = 8.380255. Sensitive: PFS in the upper quartile; Refractory: PFS in the lower quartile.

Figure 4.

Breast cancer cells resistant to palbociclib and ET depend on PLK1 expression for tumor growth. A, Representative images of breast tumor spheroids from MCF7, MCF7-PalboR (palbociclib resistant), or MCF7-FPR (fulvestrant-palbociclib–resistant) treated with the indicated drugs at 250 nmol/L final concentration, every 72 hours for 10 days. B, Western blot analysis for p-Rb (Ser780), p-Rb (Ser807/811), Rb, ER-α, Plk1 of whole cell lysates from MCF7 parental or resistant cells treated with vehicle, palbociclib alone, or in combination with fulvestrant at 1 μmol/L final concentration for 72 hours. β-Actin was used as a loading control. Images are representatives from three independent experiments. C, Representative images of breast tumor spheroids from MCF7, MCF7-PalboR, or MCF7-FPR transfected with 40 nmol/L siRNA scrambled (siCTRL) or siRNA versus PLK1 mRNA (siPLK1) for 48 hours and treated with vehicle (DMSO), 1 μmol/L of fulvestrant ± 1 μmol/L of palbociclib for another 6 days, every 72 hours. All images were captured at 20× magnification (bars = 200 μm). D, Representative images of breast tumor spheroids from MCF7-CTRL or PLK1-overexpressing cells treated for 6 days with 1 μmol/L fulvestrant + palbociclib. Images are representatives from three independent experiments. Spheroid area was calculated by ImageJ software and reported in the bar plot. Values were expressed as percentage relative to vehicle-treated controls. Data are expressed as mean ± SD of three separate experiments, indicated by error bars, performed in quadruplicate (***, P < 0.001, Student t test).

Figure 5.

Volasertib overcomes resistance to fulvestrant and palbociclib in MCF7, MCF7-PalboR, MCF7-FPR, and PLK1-overexpressing cells. A, Viability assay to assess pharmacologic synergy between fulvestrant-palbociclib and volasertib. MCF7-FPR were seeded in 96-well plates, treated with increasing concentrations of each drug alone or in combination (up to 1 mmol/L fulvestrant, 1 mmol/L palbociclib, 100 nmol/L volasertib) every 72 hours for 1 week. Cells were stained with crystal violet; staining intensities were quantified, and combination indexes were determined using the Chou-Talalay test by CompuSyn software. Numbers inside each box indicate the ratio of viable treated cells to untreated cells, from three independent experiments. B, Representative images of breast tumor spheroids from MCF7, MCF7-PalboR (left), or MCF7-FPR (right) treated with vehicle (DMSO), 500 nmol/L fulvestrant, 500 nmol/L palbociclib, 10 nmol/L volasertib, alone or combined, 10 nmol/L paclitaxel for 6 days, every 72 hours. All images were captured at 20× magnification (bars = 200 μm). C, Spheroid area was calculated by ImageJ software. Values were expressed as percentage relative to vehicle-treated controls. For all panels, data are expressed as mean ± SD of three separate experiments, indicated by error bars, performed in quadruplicate. Differences between data sets were determined by two-way ANOVA Bonferroni multiple comparisons (*, P < 0.05; ****, P < 0.0001).