Emerging Mechanisms of Therapy Resistance in Metastatic ER+ Breast Cancer

Abstract

Breast cancer is the most frequently diagnosed cancer in women, with more than 316 000 new cases expected to be diagnosed in 2025. Nearly 80% of new breast cancer cases will be estrogen receptor–positive (ER+). While ER+ breast cancer has a high 5-year survival rate, patients are at risk of developing late recurrence and metastasis for 10 to 20 years after initial diagnosis. Late recurrence and metastasis are associated with therapy resistance and disease progression. Understanding the molecular mechanisms that drive therapy resistance and disease progression is essential for the development of therapies that will prevent and treat advanced ER+ breast cancer. This review will focus on mechanisms of therapy resistance associated with standard treatments for advanced ER+ breast cancer, including CDK4/6 inhibitors and PI3K/AKT/mTOR pathway inhibitors. Additionally, we will highlight how therapy resistance enriches for breast cancer stem-like populations and how targeting this population of cells may be advantageous for preventing breast cancer progression.

Estrogen (E2)-induced estrogen receptor (ER) contributes to normal mammary gland development and breast tumorigenesis through genomic (eg, target gene expression) and nongenomic (eg, signaling) actions. Given the reliance of ER+ breast cancer on E2 and ER for tumor growth, antiestrogens and ER blocking strategies remain a clinical mainstay. Endocrine therapies such as selective ER modulators (SERMs; tamoxifen), selective ER downregulators (SERDs; fulvestrant), and aromatase inhibitors (AIs; letrozole, anastrazole) are used for adjuvant treatment of ER+ breast cancers. Premenopausal patients are typically treated with tamoxifen, whereas AIs are standard of care for postmenopausal women. Alternatively, fulvestrant is an approved first-line treatment and can be used when tamoxifen and AIs are no longer effective (1). More recently, the CDK4/6 inhibitor abemaciclib, alone or in combination with endocrine therapies (tamoxifen or AI), was FDA approved for adjuvant treatment of high-risk early ER+ breast cancer (2).

ER+ breast cancers initially respond well to endocrine therapies; however, primary resistance can occur in 15% to 20% of ER+ tumors, and acquired resistance can occur in 30% to 40% (3). This subtype is distinct from HER2+ and triple-negative breast cancers because ER+ patients tend to experience late recurrence (4), sometimes relapsing 20 or more years after diagnosis (5). Therapy resistance contributes to recurrence and metastasis, with the latter being the ultimate cause of death in most patients. Loss of ER expression is only observed in a fraction of patients (10%-15%) who develop endocrine therapy resistance (6). High ER levels have been reported to confer resistance to E2 deprivation, resulting in increased ligand-independent ER activity (7). Late recurrence and metastasis of therapy-resistant ER+ breast cancer is attributed to factors such as ER gain-of-function mutations (8-10), ER fusions (11, 12), altered coregulator interactions (13), crosstalk with growth factor and oncogenic signaling pathways, and dysregulation of the cell cycle.

First-line treatment of ER+ metastatic breast cancer (MBC) combines CDK4/6 (eg, palbociclib, abemaciclib, ribociclib) or phosphoinositide-3-kinase/protein kinase B/mammalian target of rapamycin (PI3K/AKT/mTOR) pathway inhibitors (eg, alpelisib, capivasertib, everolimus) with endocrine therapies (Fig. 1). These combinations increase progression-free survival (PFS) and overall survival (OS) in patients with metastatic disease (14-16) when compared to endocrine therapy alone (Table 1). Unfortunately, resistance to CDK4/6 and PI3K/AKT/mTOR inhibitors occurs in most metastatic patients, highlighting the need for alternative treatments. An emphasis should be placed on understanding the factors that sustain dormancy of ER+ tumor cells that contribute to late recurrence. For example, the breast cancer stem cell (CSC) population is poorly proliferative. The dormant status of CSCs (G0 phase) is thought to be refractory to conventional therapies whose actions are dependent on proliferation (17). This review will address emerging mechanisms of resistance in ER+ MBC with a focus on (i) CDK4/6 inhibitor resistance; (ii) PI3K/AKT/mTOR inhibitor resistance; and (iii) dormant CSC populations.

Figure 1.

(A) FDA-approved therapies for ER-positive metastatic breast cancer. Inhibitors targeting the PI3K/AKT/mTORC1 pathway and CDK4/6 converge on CDK4/6 to suppress cell-cycle progression. Abbreviations: AI, aromatase inhibitor; E2, estradiol; ER, estrogen receptor; RTK, receptor tyrosine kinase; SERD, selective estrogen receptor degrader; SERM, selective estrogen receptor modulator; T, testosterone. (B) Mechanisms of resistance and ongoing clinical trials investigating kinase inhibitors. The pink star indicates FGFR as a therapeutic target implicated in CDK4/6 inhibitor resistance. CDK2, CDK7, CDK9, and Aurora kinase A (AURKA) are emerging targets to overcome acquired resistance to CDK4/6 inhibitors.

Abbreviations: CTD, C-terminal domain; ERE, estrogen response elements; FGFR, fibroblast growth factor receptor; Rb, retinoblastoma protein. Created in BioRender. Lee, C. (2025) https://BioRender.com/w7u0b6q.

Table 1.

Overview of the clinical trial for FDA-approved CDK4/6 inhibitors in the metastatic cancers

Clinical trial name	Phase	Treatment	# of patients	Progression-free survival	Clinical trial identifier
Palbociclib
PALOMA-1	II	Palbociclib/letrozole vs letrozole	165	20.2 months vs 10.2 months	NCT00721409
PALOMA-2	III	Palbociclib/letrozole vs letrozole	666	24.8 months vs 14.5 months	NCT01740427
PALOMA-3	III	Palbociclib/fulvestrant vs fulvestrant	521	9.5 months vs 4.6 months	NCT01942135
Abemaciclib
MONARCH-1	II	Abemaciclib	132	6 months	NCT02102490
MONARCH-2	III	Abemaciclib/fulvestrant vs fulvestrant	669	16.4 months vs 9.3 months	NCT02107703
MONARCH-3	III	Abemaciclib/AI (anastrazole or letrozole) vs abemaciclib	493	28.18 months vs 14.76 months	NCT02246621
Ribociclib
MONALEESA-2	III	Ribociclib/letrozole vs letrozole	668	25.3 months vs 16 months	NCT01958021
MONALEESA-3	III	Ribociclib/fulvestrant vs fulvestrant	725	20.5 months vs 12.8 months	NCT02246621
MONALEESA-7	III	Ribociclib/OFS/AI or tamoxifen vs OFS/AI or tamoxifen	672	23.8 months vs 13 months	NCT02278120

Abbreviations: AI, aromatase inhibitor; CDK, cyclin-dependent kinase.

CDK4/6 Inhibitor Resistance

Dysregulation of the cell cycle through activation of oncogenes and inactivation of tumor suppressors is a hallmark of cancer. The retinoblastoma protein (Rb) is a tumor suppressor that prevents uncontrolled cell growth. Rb is phosphorylated by CDK4 and CDK6, members of the cyclin-dependent kinase (CDK) family. This checkpoint is also regulated by negative regulators of CDK4/6-Cyclin activity, which includes the INK4 family (p16, p15, p18, p19) of tumor suppressors. In breast cancer, CDK4/6 inhibitors prevent Rb phosphorylation and, therefore, block cell cycle progression. Initial CDK inhibitors were pan-CDK inhibitors (Table 2) that performed well in preclinical studies but failed in clinical trials due to poor selectivity and toxicity (18-20). This prompted the design of inhibitors to target specific CDK family members. To date, clinical trials for 3 distinct CDK4/6 inhibitors (palbociclib, ribociclib, abemaciclib) were shown to increase PFS and OS for patients with advanced or MBC and are Food and Drug Administration (FDA) approved for use in combination with endocrine therapy (Table 1).

Table 2.

Pan-CDK inhibitors: initial generations of CDK inhibitors

Name	Alternative name	Targeted CDK	Highest clinical phase	Clinical trial identifier
First-generation
Flavopiridol	Alvocidib, L868275, HMR-1275	CDK1, CDK2, CDK4, CDK6, CDK7, CDK9	II	Poor selectivity and high toxicity
R-roscovitine	CYC202, seliciclib, roscovitine	CDK1, CDK2, CDK5, CDK7, CDK9	II
Second-generation
Dinaciclib	SCH 727965, SCH-727965	CDK1, CDK2, CDK5, CDK9	III
P276-00	riviciclib hydrochloride, P276	CDK1, CDK4, CDK9	II	Safety needs to be further evaluated
AT7519	AT7519, AT-7519	CDK1, CDK2, CDK4, CDK5, CDK6, CDK9	II
TG02	TG-02, SB1317, SB-1317	CDK1, CDK2, CDK5, CDK7, CDK9	II	Other main target: JAK2, FLT3
Roniciclib	BAY1000394	CDK1, CDK2, CDK3, CDK4, CDK7, CDK9	II	Combination treatment
RGB-286638	RGC286638	CDK1, CDK2, CDK3, CDK4, CDK5, CDK9	I	The clinical application still needs further investigation

Abbreviation: CDK, cyclin-dependent kinase.

While CDK4/6 inhibitors are a mainstay for the treatment of advanced ER + breast cancer, clinical evidence suggests that both primary and acquired resistance to CDK4/6 inhibitors are common (21). Primary resistance occurs in ∼20% of breast cancer patients (22). These patients have tumors with preexisting mutations that allow cancer cells to bypass the effects of CDK4/6 inhibitors. In contrast, acquired resistance occurs over time in response to treatment. For example, one study performed whole exome sequencing of 59 ER+/HER2− breast tumors treated with CDK4/6 inhibitors and found an increase in the frequency of RB1 and ESR1 loss and activating mutations in AKT1, RAS, AURKA, CCNE2, ERBB2, and FGFR2 (23). This is an active area of research, and this section, “CDK4/6 Inhibitor Resistance,” will focus on emerging mechanisms of CDK4/6 inhibitor resistance that have been published in the past few years. We will discuss 2 primary areas in this section: (i) modulation of cell cycle proteins; and (ii) alternative pathways to CDK4/6 inhibitor resistance that involve oncogenic signaling, transcription factors, or changes in tumor metabolism.

Modulation of Cell Cycle Proteins

Impaired Rb function through loss of Rb or Rb mutations is a major mechanism associated with resistance (24-26). Other factors include overexpression or amplification of CDK4 and CDK6, INK4 family proteins, and upregulation of alternate CDK complexes.

CDK4 and CDK6

CDK4 and CDK6 levels have independently been shown to impact response to CDK4/6 inhibitors. In a retrospective analysis of ER+ MBC patients treated with first- or second-line CDK4/6 inhibitors, CDK4 amplification was associated with worse PFS (27). In a prospective trial, CDK4 amplification was found more common in ER+/HER2− MBC treated with CDK4/6 inhibitors (9.8% pretreated vs 1.5% untreated) and was linked to poorer outcomes (28). Patients with high baseline CDK4 mRNA expression had significantly longer PFS when treated with fulvestrant and palbociclib compared to those with low CDK4 expression (29). In vitro studies in abemaciclib-resistant cells were the first to show CDK6 amplification as a mechanism of CDK4/6 inhibitor resistance, while CDK6 knockdown restored sensitivity (30). Separately, palbociclib-resistant T47D cells increased resistance in treatment sensitive populations by elevating CDK6 expression via exosome-derived miR-432-5p expression (31).

Li and colleagues examined protein interactions involving CDK4 and CDK6 in both CDK4/6 inhibitor sensitive and resistant cell lines using immunoprecipitation mass spectrometry (32). Analysis from CDK4/6 inhibitor–resistant, CDK6-high-expressing cells identified interactions between CDK6 and INK4 proteins, specifically p15INK4B and p18INK4C; these were not observed with CDK4. Guided by structural studies, the authors developed Proteolysis Targeting Chimeras (PROTACs) linked to palbociclib that led to the degradation of CDK4 and CDK6. Degradation of CDK6 reduced INK4 binding and inhibited tumor growth in vivo (32). Of note, p16 (p16INK4A) overexpression has been associated with intrinsic resistance to CDK4/6 inhibitors in patient-derived xenografts (PDX) and ER+ breast cancer cell lines (25).

Cyclin E/CDK2 complex

CCNE1 encodes Cyclin E, which promotes cell cycle progression in complex with CDK2. Studies suggest that CCNE1 amplification promotes CDK4/6 inhibitor resistance (33). Guarducci et al showed that the CCNE1/RB1 ratio was an adverse prognostic indicator in the NeoPalAna trial (34). In the PALOMA-3 and POP trials, baseline tumor tissue expression of CCNE1 was associated with reduced PFS and poor anti-proliferative activity, respectively (35). Analysis of 307 patient samples from the PALLET trial found that high baseline levels of CCNE1 were associated with resistance to letrozole combined with palbociclib (36). Further, the PEARL study showed that high levels of CCNE1 in non-luminal ER+ patients have poor PFS with palbociclib plus endocrine therapy compared to capecitabine (chemotherapeutic) (37). The identification of Cyclin E/CDK2 as a biomarker of response to CDK4/6 inhibitors has led to CDK2 inhibitors being tested in clinical trials (38). Currently, there are at least 5 additional CDK2 inhibitors in phase 1 clinical trials for breast cancer patients (Fig. 1 and Table 3).

Table 3.

Overview of the clinical trial for the CDK2/7/9 inhibitors in the metastatic cancers

Name of drug	Phase	Treatment	# of patients	Type of advanced breast cancer	Status	Clinical trial identifier
CDK2
TQB3616	II	TQB3616/fulvestrant	33	HR+/HER2−	Active	NCT06702618
AZD8421	I/II	AZD8421/camizestrant/ribociclib/palbociclib/ vs AZD8421	204	ER+/HER2−	Active	NCT06188520
AVZO-021	I/II	AVZO-021	430	HR+/HER2− and CCNE-1 altered	Active	NCT05867251
BG-68501	I/II	BG-68501/fulvestrant vs fulvestrant	138	HR+/HER2−	Active	NCT06257264
Atirmociclib & PF-07104091	I/II	Atirmociclib/PF-0710409	240	HR+/HER2−	Active	NCT05262400
CDK7
SY-1365	I/II	SY-1365/fulvestrant	107	HR+/HER2−	Terminated	NCT03134638
Samuraciclib (CT7001)	I/II	Samuraciclib/fulvestrant vs samuraciclib	124	locally advanced or HR+/HER−	Completed	NCT03363893
	II	Samuraciclib/fulvestrant vs samuraciclib	60	Metastatic or locally advanced HR+/HER2−	Active	NCT05963984
	I/II	Samuraciclib/elacestrant	49	locally advanced or metastatic, ER+, HER2−	Active	NCT05963997
	I/II	Giredestrant/samuraciclib vs giredestrant	316	locally advanced or metastatic, ER+, HER2−	Active	NCT04802759
SY-5609	I	SY-5609/fulvestrant vs SY-5609	105	HR+/HER2−	Completed	NCT04247126
Q901	I/II	Q901/pembrolizumab vs Q901	130	HR+/HER2−	Active	NCT05394103
XL102	I/II	XL102/fulverstrant vs XL102/abiraterone/prednisone vs XL102	52	HR+, TNBC	Terminated	NCT04726332
CDK9
PRT2527	I	Ribociclib/letrozole vs letrozole	30	HR+/HER2−	Completed	NCT05159518

Abbreviations: CDK, cyclin-dependent kinase; ER, estrogen receptor; HR, hormone receptor.

CDK7

CDK7 is another potential target for acquired CDK4/6 inhibitor resistance. CDK7 plays key roles in cell cycle control, RNA transcription, and DNA repair mechanisms. CDK7 was identified as a lead driver of palbociclib resistance from a small-interfering RNA (siRNA) kinome knockdown screen performed in ER+ breast cancer cells (39). Sensitivity to CDK7 inhibition was associated with ER and Rb1 loss in palbociclib-resistant models. Guarducci et al investigated CDK7 in the context of ER mutations and reported that CDK7 inhibition leads to a compensatory increase in ER activity (40). Combining CDK7 inhibition (SY-1365) with fulvestrant reduces this feedback and enhances antitumor effects in vivo (40), suggesting that combining CDK7 inhibitors with selective ER downregulators could have clinical benefit.

To date, several CDK7 inhibitors have been tested in clinical trials (Fig. 1 and Table 3). SY-1365 is the first selective CDK7 inhibitor (41) to enter phase 1 clinical investigation (NCT03134638) for patients with ovarian or breast cancer. While 37% of patients showed clinical benefit, the trial did not progress to phase 2 due to lack of an optimal safety profile. More recently, phase 1 dose escalation of samuraciclib, a CDK7 inhibitor with low activity against other CDKs was reported (42). This study included an expansion cohort of samuraciclib in combination with fulvestrant in 25 ER+/HER2− patients who previously received a CDK4/6 inhibitor. Three patients achieved partial response, and another 12 patients had stable disease. The clinical benefit rate for this subgroup was 36%, with a median PFS of 3.7 months. These results have led to additional trials of samuraciclib in combination with endocrine therapy in advanced ER+ breast cancer with results pending (NCT05963984, NCT05963997, NCT04802759). Several other CDK7 inhibitors have either been terminated, completed, or are currently recruiting patients (Table 3). Together these findings suggest CDK7 inhibitors could be a treatment option for CDK4/6 inhibitor–resistant cancers.

CDK9

CDK9 was recently identified from a high-throughput kinase inhibitor library screen as a key target in AI- and palbociclib-resistant breast cancer (43). CDK9, in complex with Cyclin T, promotes transcription elongation by phosphorylating RNA polymerase II. AZD4573 is a CDK9 inhibitor that shows synergistic effects with palbociclib in ER + breast cancer cells and patient-derived organoid models (Fig. 1) (43). Treatment with AZD4573 in combination with palbociclib and fulvestrant drives tumor regression and downregulates known CDK9 target genes (MYC, MYB, MCL1) in endocrine therapy–resistant and palbociclib-resistant ER+ PDX. CDK9 inhibitors are a growing area of research (44), with several clinical trials completed (Table 3) or underway for breast cancer patients (NCT0559518, NCT04983810). Newer CDK9 inhibitors, like KB-0742 (45), have potent antitumor activity in prostate cancer (46). PROTAC-based approaches have been developed to target CDK9 (47). For example, PROTAC L055 inhibited proliferation, induced cell cycle arrest, and reduced survival of ER+ breast cancer cells in vitro and suppressed tumor growth in mice in vivo (47).

Aurora kinase A

Aurora kinase A (AURKA) is a serine/threonine kinase involved in cell cycle regulation. AURKA has been shown to be amplified or overexpressed in breast cancer, which promotes stemness and therapy resistance, and AURKA mutations have been observed following CDK4/6 inhibitor treatment (23). Alisertib is an AURKA inhibitor that can restore therapy resistance in breast cancer cells (48) and is currently being evaluated in clinical trials in CDK4/6 inhibitor–resistant MBC (49). Two phase 2 trials have demonstrated antitumor activity when combined with fulvestrant or paclitaxel (Fig. 1, Table 4). There is 1 active phase 2 clinical trial (Table 4) that aims to assess efficacy of alisertib in combination with endocrine therapy in patients with ER+/HER2− MBC previously treated with endocrine therapy plus CDK4/6 inhibitor, while also aiming to identify biomarker-defined subgroups most likely to benefit from the combination therapy.

Table 4.

Overview of the clinical trial for the AURKA/FGFR/OXPHOS inhibitors in the metastatic cancers

Name of drug	Phase	Treatment	# of patients	Type of advanced breast cancer	Status	Clinical trial identifier
Aurora kinase A
Alisertib	II	Alisertib/fulvestrant vs alisertib	96	ER+/HER2-	Active	NCT02860000
	II	Alisertib/paclitaxel vs paclitaxel	169	ER+/HER2-	Completed	NCT02187991
	II	Alisertib/endocrine therapy vs alsertib	150	HR+/HER2-	Active	NCT06369285
FGFR
Erdafinib	I	Erdafinib/palbociclib/fulvestrant	35	ER+/HER2-/FGFR-amplified metastatic	Completed	NCT03238196
Infigratinib	I	Infigrantinib/tamoxifen	4	HR+/HER2- negative/FGFR altered	Terminated	NCT04504331
Lucitanib	II	Lucitanib	178	FGFR aberrant vs FGFR non-aberrant	Terminated	NCT02202746
Futibatinib (TAS-120)	II	Futibatinib/fulvestrant vs futibatinib	168	HR+ HER2-/TNBC/ FGFR1/2 amplification	Completed	NCT04024436
OXPHOS
IACS-010759	I	IACS-010759	29	Advanced solid tumors and lymphoma	Completed	NCT03291938

Abbreviations: AURKA, aurora kinase A; ER, estrogen receptor; FGFR, fibroblast growth factor receptor; HR, hormone receptor; OXPHOS, oxidative phosphorylation.

Alternative Pathways to CDK4/6 Inhibitor Resistance

While many CDK4/6 inhibitor–resistant mechanisms are associated with dysregulation of cell cycle components, modulation of growth factor signaling pathways, transcription factors, and cellular metabolism has been observed. This is exemplified by activation of the PI3K/AKT/PTEN signaling pathway where AKT amplification after CDK4/6 inhibitor treatment occurs (23) and patients with PTEN mutations or loss prior to CDK4/6 inhibitor treatment have worse clinical outcomes (50). The PI3K/AKT/PTEN pathway is dysregulated in cancer and has been clinically targeted for patients with ER+ MBC with FDA approval of apelisib (PI3K inhibitor) and capivasertib (AKT inhibitor).

Growth factor signaling pathways

While targeting the PI3K/AKT pathway has been advantageous (discussed separately below), other growth factor signaling pathways are also potential targets. For example, the fibroblast growth factor receptor (FGFR) pathway activation has been described as a mechanism of CDK4/6 inhibitor resistance (51). The MONALEESA-2 trial showed that patients with FGFR1 amplification had shorter PFS (52). Additionally, genomic profiling of paired pre- and post-treatment tumor samples from 60 patients in the PALOMA-3 trial who developed CDK4/6 inhibitor resistance, found FGFR1, FGFR2, or FGF3 amplifications or FGFR2 mutations in 24 (40%) of post-resistance biopsies. Acquired FGFR/FGF alterations were detected in 12 of these 24 tumors, suggesting these arise under selective treatment pressure and may contribute to acquired resistance (53). In a separate retrospective study of ER+ MBC patients receiving CDK4/6 inhibitor therapy, FGFR1 amplification was associated with shorter PFS (27). There are several FDA-approved drugs targeting the FGFR pathway (erdafitinib, pemigatinib, and futibatinib) for urothelial carcinoma and cholangiocarcinoma. While these have also been tested in ER+ MBC (Fig. 1, Table 4), they are not yet FDA approved; several trials are still underway.

Other growth factor signaling proteins found amplified or mutated in response to CDK4/6 inhibitors include the known oncogenes ERBB2 and Ras (23), which are enriched in endocrine therapy–resistant breast cancers (54). Conversely, the presence of ERBB2 alterations in patients from the MONALESSA phase 3 trial was associated with greater PFS with ribociclib vs placebo (55). A separate study of 106 ER+/HER2− MBC patients receiving palbociclib in combination with fulvestrant found KRAS mutations as potential predictors of palbociclib resistance within 6 months (56). Together, these studies suggest growth factor signaling pathways are a mechanism of acquired resistance.

Transcription factors

Oncogenic signaling pathways or direct amplification modulate transcription factors involved in proliferation, survival, migration/invasion, and therapy resistance. For example, ZNF703 amplification was associated with reduced clinical benefit from ribociclib in the MONALEESA-2 trial (52, 55). Similarly, ZNF703 amplification was associated with CDK4/6 inhibitor resistance in a separate smaller study (27). Watt et al reported that CDK4/6 inhibition with abemaciclib triggers widespread chromatin remodeling in luminal breast cancer cell lines and enhancer remodeling via AP-1 transcription factor proteins (57). These results suggest that ZNF703 and AP-1 transcription factors are involved in mediating crucial biological effects of CDK4/6 inhibition.

Another factor that contributes to CDK4/6 inhibitor resistance is the tumor suppressor, FAT1. Analysis of 348 ER+ patients found that loss-of-function mutations in FAT1 and Rb1 were associated with resistance to CDK4/6 inhibitors (58). FAT1 loss elevated CDK6 levels via YAP/TAZ transcription factors of the Hippo pathway, and CDK6 knockdown restored sensitivity to CDK4/6 inhibitors. An exploratory genomic study of 87 ER+ MBC patients that had received prior CDK4/6 inhibitor treatment identified FAT1 as a potential predictive biomarker of CDK4/6 inhibitor retrial failure (59). Of note, this study also identified TP53 mutations, CDK4 amplification, and RB1 loss as other predictive biomarkers of retrial failure.

Cellular metabolism

Changes in cellular metabolism (eg, glycolysis or oxidative phosphorylation) contribute to acquired CDK4/6 inhibitor resistance. Lorito et al found that hexokinase 2 (HK2), a critical enzyme in glycolysis, showed variable regulation in palbociclib-resistant cells depending on HER2 status (60). ER+/HER2− cells displayed enhanced glycolysis during palbociclib sensitivity, whereas ER+/HER2+ cells increase glycolysis as they develop resistance. Higher expression levels of HK2 are linked to poorer OS and relapse-free survival in ER+/HER2+ breast cancer patients (60). In a separate study, transmembrane protein TMEM45A was identified as a driver of palbociclib resistance. TMEM45A overexpression was associated with enhanced glycolysis through activation of the AKT/mTOR pathway (61). Transcriptomic analysis identified several pathways such as nucleotide metabolism and inositol metabolism enriched in ER+ palbociclib-resistant cells compared to palbociclib-sensitive cells (62).

Oxidative phosphorylation (OXPHOS) also plays a critical role in metabolic reprogramming during endocrine and CDK4/6 inhibitor resistance. Inhibition of OXPHOS with IACS010759 (complex I inhibitor) was shown to block tumor growth in endocrine and palbociclib-resistant PDX models (63). Mutations in PIK3CA/AKT1 genes were found to be significantly associated with better response to IACS-010759, and treatment increased OXPHOS dependency. Further, high expression of OXPHOS-associated genes predicted poor prognosis in ER+ breast cancer patients (63). Both glycolytic and OXPHOS pathways contribute to CDK4/6 inhibitor resistance. Understanding context-specific metabolic vulnerabilities has the potential to lead to better treatment strategies, but more research is needed in this area.

As discussed above, diverse mechanisms of CDK4/6 inhibitor resistance have been identified and led to clinical trials targeting these pathways (Table 3, Table 4). Importantly, there are several ongoing clinical studies aimed at identifying biomarkers to predict response to endocrine therapy in combination with CDK4/6 inhibition (NCT04660435, NCT03439735, NCT05601440). Upon disease progression on CDK4/6 inhibitors, there is no standard of care for further systemic therapy. Treatment options are informed by molecular profiling to identify clinically relevant mutations (eg, PIK3CA mutations). The following section will discuss resistance mechanisms of PI3K/AKT/mTOR pathway inhibitors in ER+ MBC.

PI3K/AKT/mTOR Inhibitor Resistance

The PI3K/AKT/mTOR pathway controls several functions in breast cancer, including proliferation, survival, and metabolism. The significance of this pathway in endocrine therapy resistance has made it a desirable target (64), and there are currently 3 FDA-approved treatments for ER+ MBC targeting the PI3K/AKT/mTOR pathway (Fig. 1). These include the mTOR inhibitor everolimus approved in combination with tamoxifen or exemestane (AI) in patients with advanced ER+ breast cancer refractory to previous endocrine therapy. The PIK3CA inhibitor alpelisib and the AKT inhibitor capivasertib have been separately approved for combined use with fulvestrant for treating advanced ER+ breast cancer after patients have progressed on aromatase inhibitors. However, like other treatments for advanced or MBC, therapy resistance is common in response to PI3K/AKT/mTOR inhibitor treatment and will be the focus of our discussion in this section.

Resistance to mTOR Inhibitors

The mammalian target of rapamycin, or mTOR, is a protein kinase that forms 2 distinct protein complexes called mTOR Complex 1 (mTORC1) and mTOR Complex 2 (mTORC2). The mTORC1 inhibitors, such as everolimus and temsirolimus, have been tested in clinical trials for advanced ER+ breast cancer. Everolimus is an allosteric irreversible inhibitor of mTORC1-dependent phosphorylation of S6K, while temsirolimus is a kinase inhibitor that forms a complex with intracellular protein FKBP-12 to inhibit mTOR. In preclinical models, combination of AIs and everolimus synergistically inhibited proliferation and induced apoptosis (65, 66). Similar results were obtained using temsirolimus and the ERα antagonist pipendoxifene and were mechanistically linked to blocking ERα-mediated transcription and increased G1 arrest (67). While the phase 3 HORIZON trial (NCT00083993) using a combination of letrozole and temsirolimus showed no benefit over letrozole alone in PFS (68), clinical trials with everolimus indicated therapeutic benefit for advanced ER+ breast cancer. Findings from the BOLERO-2 (69, 70) and TAMRAD (71, 72) trials showed that combining everolimus with exemestane or tamoxifen improved PFS compared to exemestane or tamoxifen alone in patients with advanced ER+ breast cancer refractory to previous endocrine therapy (Table 5). These results led to FDA and EMA approval of combined everolimus and endocrine therapy for ER+ MBC after progression on AI. However, low response rates, toxicity, and resistance to everolimus have restricted its use in clinical settings (73, 74). Thus, understanding pathways driving everolimus sensitivity and resistance has the potential to lead to more effective therapies for ER+ MBC.

Table 5.

Overview of the clinical trial for FDA-approved PI3K/mTOR/AKT inhibitors

Clinical trial name	Phase	Treatment	# of patients	Progression-free survival	Clinical trial identifier
BOLERO-2	III	Everolimus/exemestane vs exemestane	724	7.8 months vs 3.2 months	NCT00863655
TAMRAD	II	Everolimus/tamoxifen vs tamoxifen	111	PFS 8.6 vs 4.5 months	NCT01298713
SOLAR 1	III	Alpelisib/fulvestrant vs fulvestrant	572	11 months vs 5.7 months	NCT02437318
FAKTION	II	Capivasertib/fulvestrant vs fulvestrant	183	10.3 months vs 4.8 months	NCT01992952
CAPItello-291	III	Capivasertib/fulvestrant vs fulvestrant	708	7.3 months vs 3.1 months	NCT04305496

Abbreviations: AKT, protein kinase B; FDA, U.S. Food and Drug Administration; mTOR, mammalian target of rapamycin; PI3K, phosphoinositide-3-kinase; PFS, progression-free survival.

MYC

Epigenetic regulation of Myc has been shown to mediate everolimus resistance in preclinical models and patient samples. Bihani et al developed everolimus-resistant (eveR) and long-term estrogen deprived (LTED)-eveR ER + cell lines. Analysis of bulk RNAseq from these models uncovered enrichment of Myc signatures in eveR and LTED-eveR cell lines. They also showed that bromodomain-containing protein 4 (BRD4)-mediated the upregulation of Myc in eveR cell lines. Inhibition of mTOR and BRD4, using everolimus and JQ1 respectively, demonstrated greater efficacy in xenograft models, supporting the potential of adding a BRD4 inhibitor to mTOR inhibitor therapy (75).

Reports also show that Myc is a clinical driver of mTOR inhibitor resistance in breast cancer (76). KEP cell lines, derived from a transgenic mouse model (K14-cre;Cdh1^Flox/Flox;Trp53^Flox/Flox) of metastatic invasive lobular carcinoma (77), were treated with PI3K (alpelisib, buparlisib) and mTOR inhibitors (everolimus, AZD8055). Their findings showed that Myc overexpression had a marginal effect on response to PI3K inhibitors, but AZD8055 and everolimus caused significant suppression of the mTOR pathway in a Myc-independent manner. Further investigations using multi-omic data of KEP tumors subjected to long-term AZD8055 therapy showed recurrent Myc amplifications in tumors that developed resistance to AZD8055 (76). In addition, whole genome sequencing analysis suggested that Myc status was significantly associated with poor response to everolimus in MBC patients. In summary, they concluded that Myc promotes mTOR inhibitor resistance by restoring protein synthesis during mTOR blockade.

Oxidative phosphorylation inhibition and stress response activation

Targeting cellular metabolism is another potential approach to use in combination with mTOR inhibitors. The combination of ONC201/TIC10, which belongs to the impiridone class of inhibitors, and everolimus significantly inhibits the growth of everolimus-resistant cancer cells in 3D growth conditions (78). Farmaki et al demonstrated that ONC201/TIC10 treatment of ER+ MBC cells activates a stress response and suppresses oxidative phosphorylation. This combination efficiently targeted breast cancer cells isolated from patients who progressed after everolimus treatment as determined by transcriptomic analyses and mitochondrial assays. Taken together, these studies offer the potential to improve therapeutic efficacy of mTOR inhibitors (75, 76).

Resistance to PI3K and AKT Inhibitors

Although a significant number of inhibitors have been developed to target PI3K and AKT, only 2 have been approved for ER+ breast cancer patients (alpelisib and capivasertib) (79-82). Alpelisib is a selective PIK3CA inhibitor that inhibits the p110α subunit and became FDA approved following the SOLAR 1 trial among postmenopausal women with ER+/HER2− PIK3CA-mutated MBC who had prior fulvestrant treatment. Data from this phase 3 trial showed that adding alpelisib to fulvestrant extended PFS from 5.7 to 11 months (83). Just like mTOR inhibitors, resistance to PI3K/AKT therapies can limit their clinical applications. Loss-of-function mutations of the tumor suppressor gene PTEN have been reported as a resistance mechanism to alpelisib. Metastatic lesions from a patient treated with alpelisib showed that these lesions had a copy loss of PTEN and had become refractory to the drug (84, 85) when compared to pretreatment tumors. Another mechanism of alpelisib resistance was identified by Zanudo et al using mathematical modeling. They observed that downregulation of FOXO3 reduced response to alpelisib when combined with BH3 mimetics like MCL1 inhibitor (S63845) in ER+ breast cancer cell lines (86).

Capivasertib is an ATP-competitive AKT inhibitor which inhibits all 3 AKT isoforms and was recently approved by the FDA following the phase 3 CAPItello-291 clinical trial (87). When combined with fulvestrant for postmenopausal women with ER+/HER2− MBC, capivasertib showed benefits in patients with activating PIK3CA mutations, activating AKT1 mutations, or PTEN truncating mutations (82, 88, 89). Although there are ongoing studies that suggest acquired resistance to capivasertib, there is sparse literature available. Dunn et al showed that deleting TSC1/2 genes conferred resistance to both AZD8186 (PI3Kβ inhibitor) and capivasertib in PTEN-null breast cancer through reactivation of the AKT-mTORC1 pathway (90). Additional studies are needed to identify mechanisms of resistance to capivasertib.

Clinical Trials for PI3K/AKT/mTOR Inhibitors

There have been clinical trials to test other mTOR inhibitors, such as the MANTA trial with vistusertib, a dual mTORC1/2 inhibitor (91). While vistusertib showed superior activity to everolimus in preclinical models, it did not outperform everolimus in a head-to-head trial. Another clinical trial (NCT01797120) evaluated everolimus in combination with fulvestrant following AI resistance (92). Here, everolimus plus fulvestrant was superior to fulvestrant alone. These randomized clinical trials have demonstrated the benefit of adding mTOR inhibitors to endocrine therapy. However, there are sparse reports on the mechanisms of mTOR inhibitor resistance among these patients.

Interestingly, following a phase 1/2 clinical trial, loss-of-function PTEN mutations were observed in 25% of patients with resistance to alpelisib when combined with AI. They also reported an increase in ESR1 activating mutations, which were linked to resistance, suggesting that these alterations could promote progression of MBC (85).

Clinical trials for other PI3K inhibitors, such as taselisib, were evaluated through the POSEIDON trial, a phase II trial investigating the efficacy of combining taselisib and tamoxifen in patients with ER+ MBC who have previously received endocrine therapy (NCT02285179, NCT02301988) (93). While the POSEIDON trial was discontinued due to associated toxicity profiles, the SANDPIPER randomized phase 3 trial (NCT02340221) investigated patients with or without a PIK3CA mutation were randomized between taselisib plus fulvestrant and placebo plus fulvestrant. Genomic alterations in ER, PI3K, and p53 pathway genes were linked to resistance to taselisib plus fulvestrant in patients with PIK3CAmut circulating tumor DNA (94).

CSCs in ER+ Breast Cancer

One frequently observed characteristic of therapy resistance in breast cancer is the presence of cancer stem-like cells (CSCs). This sub-population of cells shares properties of normal stem cells such as the ability to self-renew, differentiate, and initiating capacity (95). Other characteristics of breast CSCs include the ability to form tumorspheres (96), expression of stem cell markers such as ALDH (97) or CD44^hi/CD24^lo (98), and expression of epithelial to mesenchymal transition (EMT) and mesenchymal-epithelial transition (MET) markers (99). Elevated ALDH1 expression has been correlated with poor prognosis in breast cancer (97, 100) and predicts resistance in women treated with tamoxifen (101). Other biomarkers for profiling breast CSCs include CD133, CD49f, CD61, and EpCAM (102) but these are reported less frequently. Although there is no clear consensus, CD44^hi/CD24^lo ratios and ALDH remain the most widely used markers to identify breast CSCs. For this section, we will focus primarily on CSC studies in ER+ breast cancer.

CSCs Are Enriched in Therapy-Resistant ER+ Breast Cancer

A few studies have reported breast CSCs to be ER-negative (103, 104), which may explain their poor response to antiestrogens. ER+ breast cancer cells can contribute to CSC biology through E2-dependent paracrine factors (105). Other studies have reported that E2 alone does not mediate CSC properties and requires progesterone (106, 107). Breast CSC activity is increased following short-term endocrine therapy in response to tamoxifen- and fulvestrant-treated breast cancer cells, patient samples, and PDX models (101, 108). Tumor samples from patients treated with chemotherapy or endocrine therapy (letrozole) contained increased CD44^hi/CD24^lo breast CSCs (109). Together, these studies suggest chemotherapy and endocrine therapies initially slow growth in actively dividing tumor cells but concurrently evoke CSC activity. Tamoxifen- (110, 111), letrozole- (112, 113), and fulvestrant-resistant (114) breast cancer models have all been reported to be enriched for CSCs. These studies indicate that antagonizing E2/ER action results in increased CSC activity and may select for cells that contain ligand-independent ER activity.

Evidence suggests an independent role for progesterone receptor (PR) as a context-dependent driver of ER+ breast cancer biology (115, 116). Several studies have reported progesterone-induced expansion of breast CSCs in ER+ breast cancer cell models (106, 117, 118). PR itself has been shown to be a mediator of breast CSCs in ER+ breast cancer models (114, 119-121). Both PR isoforms (PR-A, PR-B) contribute to CSC expansion (119, 121) and evidence indicates that PR-A induces basal-like CSCs whereas PR-B drives a more luminal-enriched CSC population (120). These results suggest that PR isoforms likely regulate distinct gene programs to promote divergent lineages which may be driven in part by PR phosphorylation. In summary, delineating PR and its pathways may be a promising approach to target breast CSCs.

ER Coactivators Mediate CSC Activity

Altered coregulator interactions are a known contributing factor to therapy resistance. A number of studies have implicated ER coregulators as mediators of breast CSCs. For example, coactivator SRC-3 promotes CSC formation and tumor outgrowth in triple-negative and ER+ breast cancer models (122). PELP1, another ER coactivator, has also emerged as a driver of CSCs in ER+ breast cancer when in complex with SRC-3 within the cytoplasm (123). Follow-up studies identified PFKFB3 and PFKFB4, metabolic bifunctional kinase/phosphatases, as key components of the cytoplasmic PELP1/SRC-3 complex (124). Importantly, blocking PFKFB or SRC-3 activity disrupted PELP1/SRC-3 complexes and PELP1-induced CSCs (123, 124). These studies also demonstrated that PELP1/SRC-3-driven CSC biology was phenocopied in paclitaxel- and tamoxifen-resistant cell models, highlighting the importance of coactivator contributions to CSCs in therapy resistance.

Targeting CSCs in the Face of CDK4/6 and mTOR Inhibitor Resistance

CDK4/6 inhibitors have been shown to effectively reduce CSC activity in ER+ breast cancer cells (125-127). Paired knockdown studies of CDK4 and CDK6 showed that CDK4 was the primary driver of CSC populations (126). The effect of CDK4/6 inhibitors on CSCs can also be enhanced through combination with other compounds. Tamoxifen-resistant cells have higher levels of β-catenin and elevated activity of the Wnt/β-catenin pathway (128). Treatment with palbociclib and ICG-001 (β-catenin inhibitor) reduced proliferation and CSC activity compared to individual drugs in ER+ tamoxifen-resistant cells (129). However, when CDK4/6 inhibitor resistance emerges, the ability of CDK inhibitors to target CSCs is reduced and leaves patients susceptible to disease progression.

Understanding CSC-specific mechanisms that contribute to CDK4/6 inhibitor resistance is crucial to identifying alternative strategies that can effectively target these cells. As mentioned earlier, aberrant FGFR signaling has been linked to CDK4/6 inhibitor resistance. FGFR1 overexpression in ER+ cell line models induced palbociclib resistance and was found to be associated with increased stemness and Wnt/β-catenin signaling (130). Combination of palbociclib and AZD4547 (FGFR inhibitor) exhibited synergistic effects and was effective in reducing CSCs. In a separate study, chemokine receptor CXCR4 was shown to play a vital role in breast cancer stemness and palbociclib resistance through the Wnt/β-catenin pathway (131). Targeting CXCR4 was effective in reducing CSCs and reversing palbociclib resistance. Other mechanisms are reported to contribute to CSCs in CDK4/6 inhibitor resistance. For example, PFKFB4 was identified through transcriptomic and metabolomic studies to play an important role in stemness and palbociclib resistance through reprogramming of glycolysis (132). Paracrine secretion of IL-8 by breast CSCs promotes endocrine therapy resistance and palbociclib resistance, and pharmacological inhibition reversed resistance and reduced metastasis in xenograft models (133). Thus, identifying components that coordinate with pathways (eg, Wnt/β-catenin, cytokines/chemokines) and other signaling factors (metabolic kinases) may be a viable strategy in targeting CSCs in the face of CDK4/6 inhibitor resistance in MBC.

The PI3K/AKT/mTOR pathway also plays an important role in influencing CSC biology, often contributing to resistance by enabling CSC self-renewal and survival in ER+ breast cancer (134-136). The insulin/IGF signaling pathway promotes CSC self-renewal through PI3K-mediated regulation of Myc (137). Studies with alpelisib demonstrate that inhibition of the PI3K/AKT/mTOR pathway reduces breast CSCs, particularly those driven by crosstalk with other pathways such as Notch, Jak/STAT, and MAPK/Erk (138). Moreover, mTOR inhibition using sirolimus abrogated mammary stem cell activity in postmenopausal ER+ patients (139). Clinical trials in other solid cancers have selectively targeted CSCs through suppression of CSC-associated pathway markers (140). However, there have been no trials to date that assess the therapeutic benefit of targeting CSCs in advanced ER+ breast cancer patients.

Conclusions

Therapy resistance remains a significant clinical problem and there remains a crucial need to define the genomic alterations and altered signaling mechanisms that drive the development of resistant tumor cell populations. While ER is still present in the majority of therapy-resistant tumors, estrogen/ER-independent signaling can drive tumor regrowth, particularly within CSC populations. Identifying non-ER molecular targets for pharmacological targeting is one approach to blocking metastatic tumor cell escape. Importantly, like approved therapeutic targets (ie, CDK4/6, PI3K/AKT/mTOR), non-ER targets may also impinge on ER to promote therapy resistance and understanding how these signaling pathways ultimately impact ER function is essential to understanding tumor evolution. Furthermore, clinical trials collecting pre- and post-treatment tissue and liquid biopsies are vital for identifying mechanisms of resistance and developing biomarkers of response to therapy.

Abbreviations

AI

aromatase inhibitor

AKT

protein kinase B

AURKA

aurora kinase A

CDK

cyclin-dependent kinase

CSC

cancer stem cell

E2

estrogen

ER

estrogen receptor

eveR

everolimus-resistant

FDA

Food and Drug Administration

FGFR

fibroblast growth factor receptor

MBC

metastatic breast cancer

mTOR

mammalian target of rapamycin

OS

overall survival

OXPHOS

oxidative phosphorylation

PDX

patient-derived xenograft

PFS

progression-free survival

PI3K

phosphoinositide-3-kinase

PR

progesterone receptor

PROTAC

Proteolysis Targeting Chimera

Rb

retinoblastoma protein

Funding

This work is supported by National Cancer Institute grants R01CA236948 (J.H.O.), K22CA248615 (T.H.T.), T32CA009138 (N.R.O.), and Masonic Cancer Center, University of Minnesota (T.H.T.).

Disclosures

Julie H. Ostrander is a member of the Endocrinology Editorial Board and played no role in the Journal's evaluation of the manuscript.

Data Sharing

Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.