ESR1 Y537S and D538G mutations drive resistance to CDK4/6 inhibitors in estrogen receptor-positive breast cancer

Abstract

Purpose:

Breast cancers with ESR1 mutations are resistant to antiestrogen therapy. In this study, we aimed to investigate the association of ESR1 mutations with resistance to CDK4/6 inhibitors (CDK4/6i) using real-world data analysis and experimental validation.

Patients and Methods:

A total of 3,958 patients with estrogen receptor-positive (ER+) metastatic breast cancer with DNA sequencing data were analyzed. Breast tumor DNA and circulating tumor (ct) DNA were sequenced using the Tempus xT tumor assay and Tempus xF liquid biopsy, respectively. Patients were stratified into either treated with CDK4/6i (tumor tissue: 1,070; ctDNA: 1,885) or CDK4/6i naïve (tumor tissue: 750; ctDNA: 253). Engineered MCF7 cells carrying ESR1 Y537S or D538G knock-in mutations were used to study antitumor efficacy of the CDK4/6i palbociclib in vitro and in vivo.

Results:

In both xF and xT assays, ESR1 mutations were the only somatic alterations significantly more frequent in patients who received CDK4/6i compared to those who did not. Knock-in of ESR1 Y537S or ESR1 D538G in MCF7 cells resulted in upregulation of cell cycle-related gene signatures upon treatment with CDK4/6i ± antiestrogen compared to cells with non-mutant ESR1. MCF7 xenografts harboring ESR1 Y537S and D538G mutations established in nude mice were resistant to palbociclib.

Conclusions:

We report herein real-world and preclinical evidence that ESR1 mutations, particularly Y537S and D538G, can drive resistance to CDK4/6 inhibitors.

Introduction

Breast cancer is one of the leading causes of cancer related deaths worldwide. Approximately 75% of breast cancers are comprised of the estrogen receptor-positive (ER+) subtype [1], which is heavily dependent on both ERα signaling and the cyclin D-CDK4/6-Rb-E2F axis [2]. Treatment of ER+ breast cancers with the combination of antiestrogens and CDK4/6 inhibitors (CDK4/6i) has provided substantial survival benefit to patients and is the current first-line standard of care in the metastatic setting [3–6]. However, virtually all advanced tumors eventually acquire resistance to this combination therapy.

Mutations in ESR1 (the gene encoding ERα) arise in approximately 20–40% of patients with recurrent ER+ breast cancer after adjuvant endocrine therapy. Most of these mutations occur in the ligand-binding domain (e.g., Y537S and D538G), resulting in estrogen-independent binding of ERα to coactivators and transcriptional activity. Consistent with this notion, ESR1 mutations are enriched in patients treated with long-term estrogen suppression with aromatase inhibitors (AI) and are associated with worse clinical outcomes [7, 8]. Despite studies suggesting that ESR1 mutations can drive resistance to antiestrogens, it is less clear whether the mutations also render ER+ breast cancer refractory to treatment with CDK4/6i.

In this study, we analyzed real-world DNA sequencing data from breast tumor biopsies and circulating tumor DNA (ctDNA). In both datasets, ESR1 mutations were significantly enriched in patients treated with CDK4/6i compared to those that were CDK4/6i naïve. Using genetically engineered cells and tumors, we further validated that ESR1 Y537S and D538G hotspot mutations can promote resistance to CDK4/6i ± antiestrogens both in vitro and in vivo. Taken together, this report sheds light on the role of ESR1 mutations in resistance to the current standard of care used in ER+ advanced breast cancer.

Materials and Methods

Study design

All patient demographic, clinical, and sequencing data were provided by Tempus AI, Inc. (www.tempus.com). This study was conducted on de-identified health information subject to an institutional review board (IRB) exempt determination (Advarra Pro00072742) and did not involve human subjects research. We retrospectively analyzed de-identified next generation sequencing (NGS) data of tumor tissue DNA (n=1,820) and ctDNA (n=2,138) from patients with ER+ metastatic breast cancer generated with the Tempus xT tumor assay (595–648 genes) or Tempus xF liquid biopsy assay (105–523 genes) between December 2017 and July 2023. For patients with multiple samples sequenced, the most recently collected sample was chosen for analysis and stratified by assay type (xT or xF). Patients were grouped into treated with CDK4/6i (xT: 1,070; xF: 1,885) or untreated (xT: 750; xF: 253), where treated is defined as earliest initiation of abemaciclib, palbociclib, or ribociclib prior to biopsy. It is possible that patients in the treated group were exposed to more than one CDK4/6i prior to sample collection. Patients in the untreated group were restricted to those who were biopsied within 30 days of metastatic diagnosis.

Cell lines and reagents

MCF7 parental, ESR1 Y537S, and ESR1 D538G cells were kindly provided by Dr. Ben Ho Park [9]. The cells were genotyped monthly using ddPCR in Dr. Park’s laboratory. We performed Sanger-seq to confirm the presence of the ESR1 Y537S and D538G knock-in mutations upon receiving the cell lines. We routinely tested the cells for mycoplasma using MycoAlert mycoplasma detection kit (Lonza, Cat. No. LT07–710) every 3 months. For all the described experiments, we used the cells within 10 passages after thawing. The cells were maintained in DMEM containing 10% FBS and 1x antibiotic-antimycotic (Invitrogen). Charcoal-stripped serum was purchased from Invitrogen. Fulvestrant and palbociclib were purchased from Tocris and MedChemExpress, respectively.

Dose response assays

Cells were seeded at a density of 1,000 cells/well in 96-well plates. Twenty-four hours after seeding, cells were treated with a dose range of palbociclib ± estrogen-free media or ± 1 nM fulvestrant in full media. Media and drugs were replenished on day 3 and cell viability was determined on day 6 using the CyQuant cell proliferation assay (Invitrogen).

RNA-sequencing

RNA extraction, library preparation, sequencing, and data analysis were conducted as described in our previous study [10]. The basal-like gene signatures used for the gene set variation analysis were described in previous studies [11, 12].

Xenograft studies

Animal studies were conducted according to a protocol approved by the UTSW Institutional Animal Care and Use Committee (IACUC protocol 2018–102359). Briefly, estrogen pellets (0.25 mg, 21-day release; Innovative Research of America) were implanted s.c. in the flank of ovariectomized female nude mice (Envigo, RRID: IMSR_ENV:HSD-069) one day before tumor cell inoculation. One million cells mixed in PBS:matrigel (1:1) were injected s.c. Mice with tumors measuring ≥200 mm³ were randomized to treatment with vehicle (0.5% methylcellulose) or palbociclib (50 mg/kg/day, via orogastric gavage). Tumors were serially measured and with calipers and volume in mm³ calculated using the formula: volume = width² × length/2.

Immunohistochemistry (IHC)

Tumors were fixed in formalin and then embedded in paraffin. Five-μm tumor sections were used for IHC. Nuclear positivity of phosphorylated Rb (Cell Signaling, s8516, RRID: AB_11178658) and Ki67 (Agilent, IR62661-2, RRID: AB_2890068) was quantified by an expert breast pathologist blinded to treatment arm following the standard College of American Pathologists (CAP) breast tumor biomarkers scoring guidelines.

Data availability

Raw RNA-seq data generated in this study have been deposited in the Gene Expression Omnibus and are available under accession code GSE281302. Other raw data generated in this study are available upon reasonable request. The de-identified patient data utilized in this research were collected in a real-world healthcare setting and are subject to controlled access due to privacy and proprietary reasons. Public availability is restricted by ethics committee guidelines and/or informed consent. Where possible, derived data supporting the findings of this study have been included within the article and its Supplementary Figures/Tables.

Results

The genetic landscape of ER+ metastatic breast cancer

Patient demographics of the xF and xT cohorts are summarized in Table 1 and Table 2, respectively. The representativeness of study participants is summarized in Supplementary Table 1. Overall, 3,958 patients diagnosed with ER+/HER2-negative metastatic breast cancer were included in the analysis. The proportion of patients in the xF and xT cohorts was similar, with xF (n=2,138, 54%) being slightly higher than xT (n=1,820, 46%). In each cohort, patients were further stratified as treated or not with CDK4/6i. The majority of patients with prior CDK4/6i therapy in the xF (n=1,391, 74%) and xT (n=881, 82%) groups had received palbociclib. Mutation enrichment analysis identified TP53, PIK3CA, and ESR1 as the most frequent mutations in both xF and xT groups. There was no statistical difference in the frequency of PIK3CA mutations between CDK4/6i treated and CDK4/6i naïve patients in either the xF or xT groups, and no statistical difference in frequency of TP53 mutations between the treated and untreated groups within the xF cohort. In contrast, ESR1 was more frequently mutated in the treated vs. the untreated group in both xF (29.2% vs. 9.5%, q<0.001) and xT (32.1% vs. 4.8%, q<0.001) (Fig. 1A,B). Further analysis revealed a consistent enrichment of D538G, Y537S, and Y537N as the most frequent ESR1 mutations in the treated group vs. the untreated in both the xT and xF cohorts (Fig. 1C,D). In addition to ESR1, gene alterations in RB1 (5.6% vs. 1.6%, xF), FGFR1 (14.9% vs. 11.6%, xT), FGF3 (18.8% vs. 14.4%, xT), FGF4 (18.9% vs. 13.9%, xT), and CCND1 (19.7% vs. 15.6%, xT) were more frequent in the treated vs. the untreated group (Fig. 1A,B). Of note, FGF3, FGF4, and CCND1 are in the same 11q13 amplicon. Consistent with the TCGA database (Fig. 1E), the great majority of alterations in FGFR1, FGF3, FGF4, and CCND1 were copy number amplifications (Fig. 1F).

Table 1.

xF patient demographics

	Overall, N = 2,1381	not treated with CDK4/6i, N = 2531	treated with CDK4/6i, N = 1,8851	p-value2
Age at Diagnosis	57 (47, 65)	59 (48, 66)	57 (47, 65)	0.2
Unknown	162	17	145
Gender				>0.9
Female	2,106 (99%)	250 (99%)	1,856 (98%)
Male	32 (1.5%)	3 (1.2%)	29 (1.5%)
Race				>0.9
White	1,121 (76%)	118 (78%)	1,003 (76%)
Black or African American	159 (11%)	14 (9.2%)	145 (11%)
Other Race	126 (8.5%)	14 (9.2%)	112 (8.4%)
Asian	61 (4.1%)	5 (3.3%)	56 (4.2%)
American Indian or Alaska Native	10 (0.7%)	1 (0.7%)	9 (0.7%)
Native Hawaiian or Other Pacific Islander	1 (<0.1%)	0 (0%)	1 (<0.1%)
Unknown	660	101	559
Ethnicity				0.7
Not Hispanic or Latino	749 (84%)	89 (86%)	660 (84%)
Hispanic or Latino	140 (16%)	15 (14%)	125 (16%)
Unknown	1,249	149	1,100
Hormone Status				<0.001
ER+, PR+, HER2-	1,784 (83%)	181 (72%)	1,603 (85%)
ER+, PR-, HER2-	293 (14%)	64 (25%)	229 (12%)
ER+, HER2-	61 (2.9%)	8 (3.2%)	53 (2.8%)
Months from Metastatic Disease to Sample Collection	25 (8, 48)	1 (0, 1)	29 (14, 52)	<0.001
Treated with abemaciclib	494 (23%)	0 (0%)	494 (26%)	<0.001
Treated with ribociclib	291 (14%)	0 (0%)	291 (15%)	<0.001
Treated with palbociclib	1,391 (65%)	0 (0%)	1,391 (74%)	<0.001

¹Median (IQR); n (%)

²Wilcoxon rank sum test; Fisher’s exact test; Pearson’s Chi-squared test

Table 2.

xT patient demographics

	Overall, N = 1,8201	not treated with CDK4/6i, N = 7501	treated with CDK4/6i, N = 1,0701	p-value2
Age at Diagnosis	56 (46, 64)	58 (48, 67)	55 (45, 63)	<0.001
Unknown	149	55	94
Gender				0.7
Female	1,799 (99%)	741 (99%)	1,058 (99%)
Male	20 (1.1%)	9 (1.2%)	11 (1.0%)
Unknown	1	0	1
Race				0.063
White	891 (79%)	325 (75%)	566 (80%)
Black or African American	121 (11%)	57 (13%)	64 (9.1%)
Other Race	70 (6.2%)	27 (6.3%)	43 (6.1%)
Asian	45 (4.0%)	20 (4.6%)	25 (3.6%)
Native Hawaiian or Other Pacific Islander	5 (0.4%)	0 (0%)	5 (0.7%)
American Indian or Alaska Native	3 (0.3%)	2 (0.5%)	1 (0.1%)
Unknown	685	319	366
Ethnicity				0.4
Not Hispanic or Latino	641 (90%)	241 (92%)	400 (89%)
Hispanic or Latino	69 (9.7%)	22 (8.4%)	47 (11%)
Unknown	1,110	487	623
Hormone Status				<0.001
ER+, PR+, HER2-	1,422 (78%)	527 (70%)	895 (84%)
ER+, PR-, HER2-	354 (19%)	207 (28%)	147 (14%)
ER+, HER2-	44 (2.4%)	16 (2.1%)	28 (2.6%)
Months from Metastatic Disease to Sample Collection	9 (0, 35)	0 (0, 0)	30 (15, 51)	<0.001
Treated with abemaciclib	213 (12%)	0 (0%)	213 (20%)	<0.001
Treated with ribociclib	125 (6.9%)	0 (0%)	125 (12%)	<0.001
Treated with palbociclib	881 (48%)	0 (0%)	881 (82%)	<0.001

¹Median (IQR); n (%)

²Wilcoxon rank sum test; Pearson’s Chi-squared test; Fisher’s exact test

Figure 1. ESR1 mutations are enriched post treatment with CDK4/6i.

A-B) The ten most frequent genes with pathogenic/likely pathogenic (P/LP) somatic short variant or copy number alterations in xF (A) and xT (B). C-D) Percentage of the ten most frequent ESR1 mutations in xF (C) and xT (D). E) Frequencies of alterations in FGFR1, FGF3, FGF4, and CCND1 in the TCGA database. The tile plot was generated using cBioPortal (https://www.cbioportal.org/). F) Alteration types of FGFR1, FGF3, FGF4, and CCND1 in xF and xT.

To elucidate whether different genetic alterations are enriched as a function of which CDK4/6i received, we analyzed both xF and xT cohorts by stratifying patients based on the CDK4/6i they had received. For this analysis, we excluded patients who received none or more than one kind of CDK4/6i. In the xF cohort, of the genes with significantly different frequencies, ESR1 was most frequently altered in the palbociclib group (q=0.005), whereas BRCA2 was more frequently altered in the ribociclib group (q=0.005). Further analysis of individual ESR1 mutations showed that Y537S and D538G were more frequently altered in the palbociclib group (q=0.002 and q=0.047, respectively) compared to ribociclib and abemaciclib. In the xT cohort, alterations in ESR1 trended higher in the palbociclib group but this difference was not statistically significant. There were no statistically significant differences in frequency of Y537S or D538G alterations between different CDK4/6i-treated groups. In contrast to the xF cohort, the lack of statistical significance in the xT cohort is likely due to a smaller number of treated patients in this cohort (Supplementary Tables 2-5).

ESR1 hotspot mutations promote resistance to CDK4/6i

Recent studies suggest that ESR1 mutations are associated with resistance to antiestrogens in ER+ breast cancer [7], yet it remains unclear whether ESR1 mutations can also promote resistance to CDK4/6i. In a cell proliferation assay, ESR1 Y537S or D538G knock-in mutations in MCF7 cells resulted in a 2.7-fold increase in the IC₅₀ of palbociclib as opposed to cells with wild-type (WT) ESR1. The increase of palbociclib IC₅₀ was magnified up to 46.9- and 29.1-fold when cells were in estrogen-deprived media (mimicking patients treated with an AI) or treated with the ER degrader fulvestrant, respectively (Fig. 2A). Potential clonal effects were ruled out by testing two individual clones for each Y537S and D538G cells in the same experimental settings (Supplementary Fig. 1). Consistent with the results with palbociclib, both ESR1 Y537S and D538G also reduced sensitivity to ribociclib alone or in estrogen-deprived media, compared to WT (Supplementary Fig. 2).

Figure 2. ESR1 alterations promote resistance to palbociclib ± antiestrogens in vitro and in vivo.

A) Dose response curves of palbociclib. MCF7 cells carrying ESR1 wild-type (WT, black), Y537S (red) or D538G (green) were treated with a dose range (0–10 µM) of palbociclib (Palbo) ± estrogen deprivation (E2-) or 1 nM fulvestrant (Fulv). After 6 days of treatment, cell viability was measured by the CyQUANT assay (n=3). B) Tumor volume of MCF7 ESR1 WT (n=12), Y537S (n=8), and D538G (n=8) xenografts in ovariectomized female nude mice. The mice were treated with vehicle or 50 mg/kg palbociclib p.o. for 4 weeks. C) Comparison of the fold change in tumor volume at the end of palbociclib treatment of tumors described in (B). D-E) IHC of Ki67 (D) and pRB^S807 (E). Representative images were shown on the left, and quantitative results of IHC were shown on the right. Data represent mean ± SD; Statistical analysis: One-way ANOVA with a Dunnett’s post-hoc analysis.

Next, we examined whether the ESR1 hotspot mutations promote resistance to CDK4/6i in vivo. For this purpose, we treated ovariectomized female nude mice bearing established MCF7 ESR1 WT, Y537S, or D538G xenografts with vehicle or palbociclib. Treatment of ESR1 WT xenografts with palbociclib arrested tumor growth throughout 30 days of the regimen. On the contrary, ESR1 D538G xenografts were completely resistant to palbociclib (Fig. 2B). Although palbociclib delayed tumor growth of ESR1 Y537S xenografts, the statistically significant increase in tumor volume upon treatment with palbociclib compared to ESR1 WT suggested that this mutation desensitizes MCF7 xenografts to therapeutic inhibition of CDK4/6 (Fig. 2B,C). Consistent with these results, IHC of the MCF7 xenografts revealed that ESR1 Y537S and D538G maintained higher levels of pRB and Ki67 compared to MCF7 ESR1 WT tumors upon treatment with palbociclib (Fig. 2D,E). Collectively, these data indicate that ESR1 Y537S and D538G can blunt the inhibitory effects of CDK4/6i on phosphorylation of RB and tumor growth.

To investigate the mechanisms underlying ESR1 mutations-mediated resistance to CDK4/6i, we performed RNA-seq analysis and found that, upon treatment with CDK4/6i ± estrogen-deprived media or fulvestrant, cells harboring ESR1 Y537S and D538G exhibited upregulation of E2F, G2M, and estrogen response Hallmark gene signatures compared to ESR1 WT cells (Fig. 3A). These results suggest that the ESR1 Y537S and D538G mutations enable continued expression of cell cycle-related genes even when CDK4/6 and ERα signaling are blocked. Of note, the RNA-seq analysis also revealed that ESR1 Y537S and D538G mutations upregulated basal-like gene signatures compared to WT cells (Fig. 3B). Consistent with previous studies, these results suggest that ESR1 Y537S and D538G mutations promote estrogen-independent cell growth and promote a luminal-to-basal subtype switch in ER+ breast cancer cells [11, 13, 14]. In line with this, the basal-like breast cancer subtype has been correlated with worse outcome to CDK4/6i clinically [15, 16], thus suggesting an additional epigenetic mechanism to explain how ESR1 mutations drive resistance to CDK4/6i.

Figure 3. ESR1 Y537 and D538G mutations are associated with upregulation of cell cycle-related and basal gene signatures.

A) RNA-seq based (ESR1-mutant vs. WT cells) gene set enrichment analysis of the Hallmark pathways. Total RNA was extracted after 48 h treatment with vehicle (DMSO) or 200 nM palbociclib ± estrogen deprivation or 1 nM fulvestrant (n=3). B) Gene set variation analysis (GSVA) of published basal gene signatures. RNA-seq data of vehicle-treated MCF7 ESR1 WT and mutant cells described in (B) were used for this analysis. NES: normalized enrichment score. Statistical analysis: One-way ANOVA with a Dunnett’s post-hoc analysis.

Discussion

In this study, we used real-world data to show that ESR1 mutations are enriched in ER+ breast cancer post treatment with CDK4/6i. The Tempus xT dataset used in this study has been partially reported by Rao X et al. [17]. We analyzed an updated xT dataset with more patient samples and with a complementary large liquid biopsy dataset not reported previously. Consistent with the study by Rao X et al., mutations in ESR1, particularly Y537S and D538G, were significantly enriched post treatment with CDK4/6i. In addition, we show cellular and gene expression studies to address the causal role of ESR1 Y537S and D538G in CDK4/6i resistance.

In contrast to our findings, Wardell SE et al. reported that a patient-derived xenografts (PDX) harboring ESR1 Y537S remained sensitive to palbociclib monotherapy [18]. Of note, the dose of palbociclib used by Wardell SE et al. (125 mg/kg/day) was 2.5-fold higher than that used in our study (50 mg/kg/day), which effectively arrested growth and suppressed RB phosphorylation and Ki67 in MCF7 xenografts carrying ESR1 WT but not ESR1 Y537S and D538G. Of note, a mouse equivalent dose of palbociclib given to human is 25.7 mg/kg after adjustment by body surface area [19]. Therefore, it is possible that the potentially supra-pharmacological doses of palbociclib used by Wardell SE et al. do not mirror those used in patients and can result in off-target effects, potentially explaining its antitumor action.

A few clinical studies have reported the correlation between ESR1 mutations and resistance to CDK4/6i. In the TREnd trial, ESR1 mutations were significantly correlated with worse prognosis in patients treated with palbociclib plus fulvestrant or an AI. When treated with palbociclib monotherapy, ESR1 mutations also trended with a worse prognosis (HR, 1.88; 95% CI, 0.47–7.56). Of note, an acquired ESR1 Y537N mutation was observed in a patient who progressed on palbociclib monotherapy after approximately 18 months of the treatment [20]. In another study, Park YH et al. showed that 33% of patients (7/21) who progressed on palbociclib plus an AI acquired ESR1 hotspot mutations [21]. In line with these findings, a recent retrospective study reported that in 5,910 patients with ER+ metastatic breast cancer, the prevalence of ESR1 mutations was 1.71-fold (95% CI, 1.11–2.81, p=0.0066) higher in those treated with first-line CDK4/6i plus an AI (35.8%) than those treated with first-line AI only (21.0%) [22]. In contrast, the MONARCH 3 trial reported a lower frequency of acquired ESR1 mutations in the abemaciclib plus AI arm (19.2%) vs. the placebo plus AI arm (30.4%) [23]. Notably, however, abemaciclib is also more active against other kinases (e.g., CDK1/2/7/9, GSK-3α/β, DYRK, etc.) [24], can be given continuously, and exhibits divergent toxicity profiles [25] compared to palbociclib and ribociclib.

A recent study by Wander SA et al. stratified a cohort into sensitive or resistant to CDK4/6i plus antiestrogen therapy and found that ESR1 mutations were more frequent in resistant (34.1%, 14/41) compared to sensitive tumors (16.7%, 3/18). Further subgroup analysis of the sensitive tumors discovered that ESR1 mutations were exclusive to patients who received palbociclib plus fulvestrant (60%, 3/5) vs. palbociclib plus AI (0%, 0/13) [26]. This result suggests that ER+ breast cancers with ESR1 mutations may still be sensitive to CDK4/6i plus fulvestrant or other selective ER degraders (SERDs). In line with this observation, analysis of PALOMA-3 [5] and MONARCH 2 [27] showed concordant results that fulvestrant plus CDK4/6i improved progression-free survival (PFS) when compared to fulvestrant alone, irrespective of ESR1 status.

In a pooled ctDNA analysis of the MONALEESA-2, -3, and -7 trials, the ribociclib arm provided similar PFS benefit in patients regardless of presence of ESR1 mutations. Although this analysis also included patients treated with ribociclib plus AI or tamoxifen (MONALEESA-2 and -7), the majority of evaluable patients with ESR1 mutations (62/92) were treated with ribociclib plus fulvestrant (MONALEESA-3) [28]. In addition, the PADA-1 trial, which enrolled patients with ER+ metastatic breast cancer harboring early identification of ESR1 mutations, revealed a better PFS in patients treated with fulvestrant plus palbociclib (11.9 months; 95% CI, 9.1–13.6) than with AI plus palbociclib (5.7 months; 95% CI, 3.9–7.5) [29]. Overall, these studies suggest that patients with ESR1 mutations may still benefit from an ERα degrader combined with a CDK4/6i.

Although we show that MCF7 cells harboring ESR1 Y537S and D538G are more resistant to palbociclib in the presence of fulvestrant compared to ESR1 WT, this preclinical finding is not concordant with the observation of the aforementioned clinical studies, which considered distinct ESR1 mutations as a whole. Of note, analysis of the PALOMA-3 trial suggests that ESR1 mutations are frequently subclonal with high levels of polyclonality, rendering ESR1 mutations only partially representative of the cancer overall [30]. Therefore, further investigation of individual ESR1 mutations is warranted, as their relative causality of a drug resistance phenotype may vary [7]. For example, analysis of the PALOMA-3 trial by O’Leary B et al., revealed both loss and newly acquired ESR1 mutations at progression compared to baseline. The clear enrichment of ESR1 Y537S mutation in patients treated fulvestrant ± palbociclib highlights that certain ESR1 mutations are advantageous under drug selection [31].

Collectively, strengths of this study include integrating real-world data and laboratory-based validation to elucidate a causal relationship between common ESR1 mutations and resistance to the standard treatment of ER+ breast cancer. These results, however, should be interpreted with caution. Only engineered MCF7 cells were used in our study. Confirmation using other ER+ breast cancer cells and xenografts with ESR1 knock-in mutations [32] should strengthen the evidence in support of our conclusions. MCF7 xenografts were used as a surrogate for metastatic lesions to evaluate activities of CDK4/6i in vivo. We recognize that these data may not necessarily inform clinical use of palbociclib and that there could be differences with abemaciclib and ribociclib. Additional in vivo studies to investigate the effects of ESR1 mutations on sensitivities to ribociclib and abemaciclib are warranted, as they may provide insights into the design and interpretation of future clinical studies.

In summary, this report provides real-world and preclinical evidence that supports ESR1 Y537S and D538G mutations as drivers of resistance to CDK4/6i. Future studies using additional in vitro and in vivo models are warranted to investigate the roles of individual ESR1 mutations, and to determine whether mutant ESR1-induced CDK4/6i resistance can be reversed by combined treatment with SERDs.

Translational relevance.

Combination treatment of CDK4/6i plus antiestrogen therapy is the current standard of care for patients with ER+ metastatic breast cancer. Mutations in ESR1 confer resistance to antiestrogen therapy in ER+ breast cancer. However, whether ESR1 mutations can directly promote CDK4/6i resistance remains unclear. We analyzed real-world data and identified ESR1 as the gene with the most significantly enriched mutations in patients with prior treatment with CDK4/6i. Using genetically engineered breast cancer cells tumors, we validated that ESR1 Y537S and D538G mutations can blunt the antitumor efficacy of CDK4/6i ± antiestrogens both in vitro and in vivo. These findings shed light on the role of ESR1 mutations in promoting resistance to CDK4/6i in ER+ breast cancer.

Abbreviations list:

AI

aromatase inhibitors

CAP

College of American Pathologists

CDK4/6i

CDK4/6 inhibitors

CI

confidence interval

ctDNA

circulating tumor DNA

ER+

estrogen receptor-positive

HR

hazard ratio

IHC

Immunohistochemistry

IRB

institutional review board

NGS

next generation sequencing

PDX

patient-derived xenografts

SERDs

selective ER degraders

WT

wild-type