Treatment of Luminal Metastatic Breast Cancer beyond CDK4/6 Inhibition: Is There a Standard of Care in Clinical Practice?

Abstract

Background

CDK4/6 inhibitors have become the standard for first-line treatment of metastatic luminal breast cancer based on consistent data from several phase 3 trials demonstrating clinically meaningful improvement of progression-free as well as overall survival. In addition, they are about to become a part of adjuvant treatment for patients with high-risk luminal disease based on positive results from the first randomized phase 3 trial on abemaciclib. Nevertheless, the majority of patients with advanced or metastatic luminal breast cancer and prospectively a relevant proportion of patients treated in the adjuvant setting will eventually develop resistance to this endocrine based combination within 12–36 months, depending on the line of treatment.

Conclusion

Potential subsequent therapies include PI3K inhibitors, mTOR inhibitors, endocrine monotherapy, PARP inhibitors, and chemotherapy. However, these therapies have mainly been developed in the pre-CDK4/6 inhibitor era and little is known about potential cross-resistance. The concept of continuing CDK4/6 inhibition beyond progression is supported by some preclinical data, but to date there is very limited clinical evidence to support this strategy. Therefore, treatment of metastatic luminal breast cancer after progression on CDK4/6 inhibitors remains a challenge.

Key Messages

Here we review current evidence from pro- and retrospective studies and give an outlook on future developments with respect to novel therapeutic agents, including oral SERD and AKT inhibitors, which have the potential to change the therapeutic landscape in the future. Furthermore, clinical treatment algorithms and current research will also be discussed.

Introduction

Metastatic breast cancer (MBC) accounts for more than 600,000 deaths per year worldwide and it is the leading cause of cancer mortality among women [1]. Classification of breast cancer subtypes, initially based on gene expression profiling [2, 3], routinely relies on the expression of estrogen receptor (ER), progesterone receptor, and human epidermal growth factor receptor 2 (HER-2), as well as the proliferation index Ki67, assessed by immunohistochemistry. In combination with a limited number of targetable mutations, this allows for individualized therapeutic strategies [4]. Up to 70% of MBC patients have luminal breast cancer, defined by expression of ER and HER-2 negativity [5], with a median overall survival (OS) as long as 57 months [6]. Estrogen deprivation, including selective ER modulators, such as tamoxifen [7], aromatase inhibitors (AI) like letrozole [8], anastrozole [9], and exemestane [10], and selective ER degraders (SERD) like fulvestrant [11], has long been the hallmark of treatment, albeit with a limited clinical efficacy. In endocrine-naive metastatic luminal breast cancer (MLBC) patients receiving anastrozole or fulvestrant as first-line treatment, the median progression-free survival (PFS) was 13.8 and 16.6 months [12], respectively. The PFS is considerably shorter, i.e., only 6.5 months, in endocrine-pretreated patients receiving fulvestrant as second-line therapy [11]. In BOLERO-2, addition of the mTORC1 inhibitor everolimus to exemestane in patients resistant to a nonsteroidal AI demonstrated a significant PFS benefit of 6.5 months [13] but fell short of demonstrating a statistically significant OS benefit. Subsequently, dual mTORC1/2 inhibitors like vistusertib have been developed to circumvent possible negative feedback loops that might result in resistance to mTORC1 inhibition. However, vistusertib did not lead to improved efficacy when directly compared to everolimus [14].

With the introduction of the CDK4/6 inhibitors (CDK4/6i) palbociclib, ribociclib, and abemaciclib added to the backbone of endocrine therapy, response and survival rates substantially improved in both first- and second-line treatments. All 3 CDK4/6i have consistently increased PFS in several phase 3 clinical trials [15, 16, 17, 18, 19, 20]. For example, abemaciclib added to a nonsteroidal AI as first-line treatment for postmenopausal women led to a PFS improvement of 13.4 months in the MONARCH-3 study [19]. In addition, ribociclib and abemaciclib have provided a statistically significant OS benefit in the first-line setting as well as in the second-line setting within MONALEESA-3, MONALEESA-7, and MONARCH 2 [20, 21, 22, 23].

These excellent results have rendered CDK4/6i combined with endocrine treatment as the new standard of care (SOC) for both endocrine-naive and endocrine-pretreated MLBC patients. In the clinical routine, AI are the preferred endocrine backbone among de novo MLBC patients or among patients with disease recurrence >12 months upon completion of adjuvant endocrine treatment. In a direct comparison of letrozole and fulvestrant as the endocrine backbone for palbociclib as first-line therapy in the PARSIFAL study, fulvestrant did not demonstrate improved efficacy or noninferiority compared to letrozole [24]. Fulvestrant might be preferred as the endocrine combination partner in patients with disease progression on AI or with recurrence <12 months after completion of adjuvant endocrine treatment [4]. However, fulvestrant also functions as the endocrine partner for potential subsequent therapies after CDK4/6i, as it is approved in combination with alpelisib in PIK3CA-mutant patients. Therefore, strategic considerations about further treatment lines will also influence the choice of the best endocrine combination partner for CDK4/6i in an individual patient. In any case, selection of the most appropriate CDK4/6i should consider the toxicity profile of each compound (gastrointestinal toxicity for abemaciclib and hepatotoxicity and QTcF prolongation for ribociclib) with regard to the patient's comorbidities. Under special circumstances, selected patients with a low tumor burden or multiple limiting comorbidities and difficult access to their treating physicians may also be considered for endocrine monotherapy [4]. On the other hand, chemotherapy is still the SOC in patients presenting with a visceral crisis [4]. Today, the vast majority of MLBC patients receive CDK4/6i as first-line treatment and the time to subsequent chemotherapy has been substantially increased, thus reducing systemic toxicities and the improving quality of life [15].

Principal Mechanisms of Resistance to CDK4/6i

Despite this significant clinical progress, the majority of patients will show tumor progression within up to 36 months of CDK4/6i treatment [15, 16, 18, 20, 23]. Preclinical models suggest multiple resistance mechanisms against CDK4/6i in MLBC involving either hyperactivation of the CDK4/6 G1 checkpoint kinase, thus mandating the development of more potent CDK4/6i, or bypassing of the CDK4/6 G1 checkpoint kinase through CCNE1/CDK2 leading to retinoblastoma protein (Rb) phosphorylation or through loss of target by acquired loss-of-function RB1 mutations [25]. In addition, ineffective ER inhibition by de novo ERα (ESR1) driver mutations [26], or due to constitutive activation of the PI3K/AKT/mTOR signal cascade through extensive cross-talk between the ER and HER-2, IGF-1R, or FGFR1 pathways [27, 28], has been recognized as being of clinical importance. Under treatment with CDK4/6i, acquired ERS1 mutations, detected in up to 30% of patients, primarily reflect a resistance to endocrine therapies [26, 29], whereas de novo RB1 mutations, found in up to 10% of CDK4/6i-pretreated patients, indicate CDK4/6i resistance [25, 26, 30]. Increased CCNE1 mRNA levels, documented among half of the PALOMA-3 patients, were shown to correlate with a short median PFS of 7.6 months and, conversely, low CCNE1 mRNA levels correlated with an increased PFS of 14.1 months under palbociclib [31], suggesting CCNE1 expression as a potential predictive biomarker for CDK4/6 inhibition. This supports the need for development of dual CDK2- and CDK4/6 inhibitors, such as PF-06873600, currently being tested in a phase 1/2 trial (NCT03519178; Table 1). Interestingly, neither CCND1 nor CDK4 or CDK6 amplification could be associated with PFS in the PALOMA-3 trial [31], although resistance mechanisms through CDK6 overexpression via the FAT1/Hippo pathway have been described [32, 33]. In addition, acquired resistance mechanisms specific to certain CDK4/6i have also been postulated; abemaciclib has been demonstrated to maintain in vitro anti-tumor activity in Rb-deficient, palbociclib/ribociclib-resistant cell lines [34] since it can effectively inhibit further CDK complexes, such as CDK7/Cyclin H1 and CDK9/Cyclin T1 [35, 36]. This preclinical finding supports the rationale for continuing CDK4/6 inhibition beyond progression, e.g., with an alternative CDK4/6i.

Table 1.

Overview of current clinical phase 1–3 trials in MLBC

Trial	Phase	Indication	Regimen	Status	Conclusions/remarks	Reference
NCT03519178	1/2	Pretreated MLBC	PF-06873600 alone and in combination with endocrine treatment	Recruiting	Evaluates the safety of dual CDK2 and CDK4/6i PF-06873600 alone and in combination with endocrine treatment among CDK4/6i-pretreated patients	−
NCT02632045 (MAINTAIN)	2	Pretreated MLBC	Ribociclib/fulvestrant vs. placebo/fulvestrant	Recruiting	Evaluates the efficacy of ribociclib/fulvestrant after progression on CDK4/6i	−
NCT02738866	2	Pretreated MLBC	Palbociclib/fulvestrant	Recruiting	Evaluates the efficacy of palbociclib/fulvestrant after progression on palbociclib/AI	−
NCT02778685	2	MLBC	Pembrolizumab/palbociclib/letrozole	Recruiting	Evaluates the efficacy of pembrolizumab/palbociclib/letrozole among treatment-naive patients	−
NCT03147287 (PACE)	2	Pretreated MLBC	Avelumab/palbociclib/fulvestrant vs. palbociclib/fulvestrant vs. fulvestrant	Recruiting	Evaluates 3 regimens after progression: − on CDK4/6i in the metastatic setting − <12 months after completion of CDK4/6i in adjuvant setting
NCT03294694	1	Pretreated MLBC (cohort B)	PDR001 (spartalizumab)/ribociclib/fulvestrant	Recruiting	Evaluates the safety of anti-PD-1 spartalizumab in combination with ribociclib/fulvestrant among CDK4/6i pretreated patients (cohort B)	−
NCT03280563 (MORPHEUS HR+BC)	1/2	Pretreated MLBC	Atezolizumab/entinostat vs. atezolizumab/fulvestrant vs. atezolizumab/abemaciclib/fulvestrant vs. atezolizumab/ipatasertib vs. atezolizumab/ipatasertib/fulvestrant vs. atezolizumab/bevacizumab/endocrine treatment	Recruiting	Evaluates the ORR in multiple regimens among CDK4/6i-pretreated patients	−
NCT02732119 (TRINITY-1)	1/2	Pretreated MLBC	Ribociclib/everolimus/exemestane	Completed	Evaluated the efficacy of ribociclib/everolimus/exemestane after progression on CDK4/6i	50
NCT03099174	1	Untreated MLBC (cohort D) Pretreated MLBC (cohorts D and F)	Xentuzumab/abemaciclib/fulvestrant	Recruiting	Evaluates the safety and efficacy of anti-IGF-1/2 xentuzumab combined with abemaciclib/fulvestrant in: − de novo MLBC after progression on AI (cohort D) − untreated patients <12 months of AI completion in the adjuvant setting (cohort D) − AI progression when recurrence occurs >12 months of AI completion in the adjuvant setting (cohort D) − progression on CDK4/6i (cohort F)	51
NCT02684032	1	Untreated MLBC (cohort A) Pretreated MLBC (cohorts B, C, and D)	Gedatolisib/palbociclib/letrozole vs. gedatolisib/palbociclib/fulvestrant vs. fulvestrant	Active, not recruiting	Evaluated the safety and MTD of the dual mTOR/PI3K inhibitor gedatolisib in multiple combinations in treatment-naive patients (cohort A), CDK4/6i-naive patients (cohort B), and CDK4/6i-pretreated patients (cohorts C/D)	52
NCT03238196	1	Pretreated MLBC	Erdafitinib/palbociclib/fulvestrant	Recruiting	Evaluates the safety and tolerability of the FGFR inhibitor erdafitinib in combination with palbociclib/fulvestrant in CDK4/6i-pretreated patients with FGFR-amplified MLBC	−
NCT04504331	1	Pretreated MLBC (cohorts 1 and 2)	Infigratinib/tamoxifen (cohort 1), infigratinib/palbociclib/fulvestrant (cohort 2)	Recruiting (cohort 1), planned (cohort 2)	Evaluates the safety and tolerability of the FGFR inhibitor infigratinib in multiple combinations in FGFR-altered MLBC	−
NCT03584009 (VERONICA)	2	Pretreated MLBC	Venetoclax/fulvestrant vs. fulvestrant	Completed	Evaluated the efficacy of fulvestrant with or without the BCL-2 inhibitor venetoclax after progression on CDK4/6i	−
NCT03900884 (PALVEN)	1	Pretreated BCL-2 positive MLBC	Venetoclax/palbociclib/letrozole	Recruiting	Evaluates the safety and tolerability of the BCL-2 inhibitor venetoclax in combination with palbociclib/letrozole in pretreated, CDK4/6i-naive patients	−
NCT02860000	2	Pretreated MLBC	Alisertib/fulvestrant vs. alisertib	Completed	Evaluated the efficacy of the AURKA inhibitor alisertib with or without fulvestrant in endocrine-refractory MLBC	−
NCT01872260	1/2	Untreated MLBC (arms 1, 2, 3, and 4)	Ribociclib/letrozole vs. alpelisib/letrozole vs. ribociclib/alpelisib/letrozole	Completed	Evaluated the safety, tolerability, and preliminary clinical antitumor activity of the 3 combinations in treatment-naive patients	−
NCT03006172	1	Untreated PIK3CA-mutant MLBC (arms B and E) Pretreated PIK3CA-mutant MLBC (arms C and D)	arm B: GDC-0077 (inavolisib)/palbociclib/letrozole, arm E: inavolisib/palbociclib/fulvestrant, arm C: inavolisib/letrozole, arm D: GDC-0077/fulvestrant	Recruiting	Evaluated the safety, tolerability, and preliminary clinical anti-tumor activity of the PI3K inhibitor inavolisib in combination with palbociclib/letrozole or fulvestrant in CDK4/6i-naive patients (arms B and E) or in combination with letrozole or fulvestrant in CDK4/6i-pretreated patients (arms C and D)	−
NCT04191499 (IN AVO 120; WO41554)	2/3	Untreated PIK3CA-mutant MLBC	Inavolisib/palbociclib/fulvestrant vs. palbociclib/fulvestrant	Recruiting	Evaluates the efficacy and safety of inavolisib in combination with palbociclib/fulvestrant vs. palbociclib/fulvestrant in PIK3CA-mutant treatment-naive MLBC with progression after <12 months since completion of adjuvant endocrine therapy	−
NCT03939897	1/2	Pretreated MLBC	Phase 1: copanlisib/abemaciclib/fulvestrant, phase 2: copanlisib/abemaciclib/fulvestrant vs. abemaciclib/fulvestrant	Suspended	Evaluates the efficacy and safety of the PI3Kδ inhibitor copanlisib in combination with abemaciclib/fulvestrant in CDK4/6i- and PI3K inhibitor-pretreated patients (phase 1) and in CDK4/6i- and PI3K inhibitor-nonpretreated patients (phase 2)	−
NCT03778931 (EMERALD)	3	Pretreated MLBC	RAD1901 (elacestrant) vs. SOC (AI or fulvestrant)	Active, not recruiting	Compared PFS between oral SERD elacestrant and SOC in CDK4/6i-pretreated patients	−
NCT02569801 (HydranGea)	2	Pretreated MLBC	GDC-0810 (brilanestrant) vs. fulvestrant	Withheld	Evaluated the efficacy, safety, and tolerability of the oral SERD brilanestrant vs. fulvestrant in ΑI-refractory MLBC	−
NCT02734615	1	Pretreated MLBC	LSZ102 vs. LSZ102/ribociclib vs. LSZ102/alpelisib	Completed	Evaluated the safety and tolerability of the oral SERD LSZ102 alone or in combination with alpelisib or ribociclib in fulvestrant- and/or CDK4/6i-pretreated MLBC	119
NCT03284957 (AMEERA-1)	1/2	Pretreated MLBC	SAR 439859 (amcenestrant) alone vs. in combination with palbociclib, midazolam or alpelisib	Recruiting	Evaluates the safety, tolerability, and preliminary clinical antitumor activity of the oral SERD amcenestrant in multiple combinations in CDK4/6i-pretreated MLBC	116
NCT04059484 (AMEERA-3)	2	Pretreated MLBC	Amcenestrant vs. SOC endocrine monotherapy	Recruiting	Compares PFS between amcenestrant and SOC endocrine monotherapy in CDK4/6i-pretreated MLBC	−
NCT04478266 (AMEERA-5)	3	Untreated MLBC	Amcenestrant/palbociclib vs. letrozole/palbociclib	Recruiting	Compares PFS between amcenestrant/palbociclib and letrozole/palbociclib in treatment-naive MLBC	−
NCT04576455	2	Pretreated MLBC	GDC-9545 (giredestrant) vs. SOC endocrine monotherapy	Recruiting	Compares PFS between the oral SERD giredestrant and the SOC endocrine monotherapy in CDK4/6i-pretreated MLBC
NCT04546009 (BO41843)	3	Untreated MLBC	Giredestrant/palbociclib vs. letrozole/palbociclib	Recruiting	Compares PFS between giredestrant/palbociclib and letrozole/palbociclib in treatment-naive MLBC	−
NCT04188548 (EMBER)	1	Pretreated MLBC	LY3484356/abemaciclib/AIs (part A) vs. LY3484356 vs. LY3484356/alpelisib vs. LY3484356/everolimus (part B)	Recruiting	Evaluates the safety and tolerability of the oral SERD LY3484356 combination with abemaciclib/AI in CDK4/6i-naive patients (part A), or alone, with alpelisib or everolimus in CDK4/6i-pretreated patients (part B)	−
NCT04305496 (CAPItello-291)	3	Pretreated MLBC	Capivasertib/fulvestrant vs. placebo/fulvestrant	Recruiting	Compares PFS between AKT inhibitor capivasertib in combination with fulvestrant vs. fulvestrant in CDK4/6i-pretreated MLBC	−
NCT03959891 (TAKTIC)	1	Pretreated MLBC	Ipatasertib/fulvestrant vs. ipatasertib/letrozole vs. ipatasertib/palbociclib/fulvestrant	Recruiting	Evaluates the safety and tolerability of the AKT inhibitor ipatasertib in CDK4/6i-pretreated, PI3K inhibitor-naive MLBC	−
NCT04060862 (IPA Tunity 150)	3	Untreated MLBC	Ipatasertib/palbociclib/fulvestrant	Recruiting	Compares PFS in ipatasertib in combination with palbociclib/fulvestrant vs. palbociclib/fulvestrant in untreated patients	−
NCT 04256941 (INTERACT)	2	Untreated MLBC	Palbociclib or ribociclib or abemaciclib with AI or fulvestrant	Recruiting	Assesses PFS under multiple regimens in untreated MLBC patients with ESR1 activating mutations	−
NCT04534283	2	Tumors harboring pathogenic alterations in BRAF, RAF1, MEK1/2, ERK1/2, and NF1	LY3214996/abemaciclib	Recruiting	Assesses the clinical efficacy of the ERK1/2 inhibitor LY3214996 in combination with abemaciclib	−

CDK4/6i beyond Progression

In PALOMA-3, four percent of the patients in the palbociclib arm received CDK4/6i beyond progression [37]. Real-world data support this observation. In an analysis of 525 patients who received further systemic therapies after CDK4/6i progression, 12.3% of those having CDK4/6i in combination with an AI as first-line therapy were treated with a CDK4/6i in the subsequent line [38]. To some extent, the concept of “treatment beyond progression” with CDK4/6i has therefore entered routine clinical practice despite the fact that there is only very limited clinical evidence to support this strategy so far.

A multicenter analysis of 58 MLBC patients receiving abemaciclib as monotherapy or in combination with endocrine treatment after progression on palbociclib suggested some clinical benefit based on a median PFS of 5.8 months and a treatment duration of 6 months or longer in one third of the patients [39]. Two further retrospective analyses of small patient cohorts receiving abemaciclib as a monotherapy [40] or in combination with endocrine therapy [41] in the same clinical setting seem to confirm a limited clinical activity of abemaciclib beyond CDK4/6i progression, with a median PFS up to 7.0 months. Continuation of palbociclib beyond progression, while switching the endocrine therapy, was evaluated in a small, retrospective, single-institution study [42]. The estimated median PFS for the entire duration while on CDK4/6i was 23.5 months (95% CI 12.8–27.8), of which 11.8 months (95% CI 5.34–13.13) was the median PFS beyond the first progression. The median OS from first-line CDK4/6i treatment was 45.4 months [42]. A retrospective analysis from MSKCC evaluated 135 patients receiving more than 2 lines of CDK4/6i. Patients who discontinued their first CDK4/6i due to progression had a relatively short duration of the second CDK4/6i therapy, with a median time to subsequent therapy of 19.9–22.2 weeks. The data also provide evidence of radiologically demonstrated responses in some of the patients [43]. In current consensus guidelines, no routine use of CDK4/6i beyond progression in MLBC is recommended [4]. Nevertheless, participation in clinical trials is strongly recommended. This question is currently addressed by phase 2 trials, like MAINTAIN and NCT02738866, investigating the benefit of ribociclib and palbociclib added to fulvestrant as a second-line treatment after progression under upfront CDK4/6i (Table 1).

Investigational CDK4/6i Combinations after CDK4/6i Progression

Preclinical models suggest an interaction of checkpoint inhibitors with CDK4/6i through inhibition of proteasome-mediated PD-L1 degradation via SPOP, a cullin 3 E3 ubiquitin ligase adaptor protein [44], as well as through direct stimulation of PD-1-expressing T cells by CDK4/6i, resulting in an enhanced in vitro intratumoral T-cell infiltration [45]. The first promising results have come from a phase 1/2 open-label single-arm study investigating palbociclib and letrozole in combination with pembrolizumab in CDK4/6i-pretreated patients [46]. Further checkpoint inhibitors like spartalizumab (PDR001), avelumab, or atezolizumab in combination with CDK4/6i and endocrine therapy are currently being investigated in several clinical trials (NCT03294694, PACE, and MORPHEUS HR + BC; Table 1).

The synergistic activity between CDK4/6i and PI3K or mTOR inhibitors observed in vitro appears to be mediated by p21-mediated cell quiescence [47], cell senescence [48], or even augmentation of tumor-infiltrating cytotoxic NK cells, CD4+ and CD8+ T cells, and suppression of Tregs [49]. First clinical data from a prospective trial (NCT02871791) evaluating palbociclib in combination with exemestane and everolimus in CKD4/6i-pretreated MLBC patients revealed substantial systemic toxicities including high-grade mucositis and neutropenia among 15.6 and 71.9% of the patients, respectively. Poor overall response rates (ORR) and a median PFS of 3.8 months precluded further investigation of this combination [50]. In contrast, the phase 1/2 TRINITY-1 trial, evaluating ribociclib in combination with exemestane/everolimus in the same clinical setting, revealed a manageable toxicity profile and demonstrated a clinical benefit with a 1-year PFS of 33% [51]. Alternative strategies of concurrent CDK4/6 and PI3K/AKT/mTOR pathway blockade beyond CDK4/6i progression include the monoclonal antibody xentuzumab, which binds IGF-1 and IGF-2 [52] and thus suppresses IGF-1R and IR-A signaling upstream of the mTOR/PI3K pathway, or dual mTOR/PI3K inhibitors like gedatolisib [53], and are currently being investigated in several phase 1 trials (Table 1).

Preclinical data also demonstrate an association of FGFR1 amplification, detectable among 15% of MLBC patients with endocrine resistance [54]; treatment of FGFR-amplified breast cancer cells with FGF-2 strongly induces CCND1 expression and may thus lead to an escape from CDK4/6 inhibition [55]. Among patients in the MONALEESA-2 study, FGFR1 amplification detected in ctDNA was associated with a shorter PFS under ribociclib [56]. Initial studies failed to demonstrate a significant activity of the FGFR1–3 inhibitors lucitanib [57] and dovitinib [58] in combination with fulvestrant in endocrine-resistant patients. Clinical development is now focusing on novel FGFR inhibitors, such as erdafitinib, which is currently being investigated in combination with palbociclib and fulvestrant in CDK4/6i-pretreated patients (Table 1). Preclinical evidence also suggests interactions between the CDK4/6/RB1 axis and AURKA [59], PLK1 [60], and BCL2 [61] inhibitors. Currently, the BLC-2 inhibitor venetoclax and the AURKA inhibitor alisertib are being investigated in clinical trials (VERONICA, PALVEN, and NCT02860000), promising further advances in the treatment of MLBC beyond CDK4/6i progression.

PI3K Inhibitors

The latest advance in the treatment of MLBC is the FDA and EMA approval of the first α-selective PI3K inhibitor alpelisib in 2019/2020 for postmenopausal MLBC patients harboring PIK3CA mutations. PIK3CA encodes the isoform p110α, i.e., the catalytic subunit of class IA PI3K [27]. Somatic, heterozygous gain-of-function PIK3CA mutations lead to a constitutive activation of PI3K and are detectable in up to 47% of patients with MLBC [62, 63, 64]. More than 80% of PIK3CA mutations occur within the helical domain (exon 9, mostly PIK3CA^E542K and PIK3CA^E545K) and the kinase domain (exon 20, mostly PIK3CA^H1047R) [62, 64]. PIK3CA mutations are early events in breast cancer pathogenesis [65, 66]. PIK3CA mutations were detectable among a third of the patients in the PALOMA-3 study and in most cases they persisted throughout the study treatment, with up to 6% of the subjects developing de novo PIK3CA mutations under treatment in both study arms [26]. PIK3CA mutations did not impact the efficacy of CDK4/6i [15]. Similarly, in the BOLERO-2 study, i.e., the registrational trial for everolimus, no association between PIK3CA mutations and response to everolimus/exemestane among MLBC patients could be demonstrated [67]. In the SAFIR02 trial, PIK3CA mutations conferred resistance to chemotherapy and were associated with a poorer median OS [68]. Clinical development of first-generation pan-PI3K inhibitors like pictilisib [69], buparlisib [70, 71], and the α-selective inhibitor taselisib [72] was halted due to significant systemic toxicities, which outweighed the agents' therapeutic benefit. In contrast, in the phase 3 SOLAR-1 trial, the α-selective alpelisib demonstrated significant activity with a manageable toxicity profile, mainly consisting of metabolic, dermatologic, and gastrointestinal toxicities [73]. This international, randomized, double-blind study investigated fulvestrant in combination with alpelisib in postmenopausal patients. Enrollment preferentially focused on patients with a secondary endocrine resistance with progression >6 months upon first-line palliative AI treatment in de novo MLBC or recurrence after ≥24 months of adjuvant endocrine therapy or relapse <12 months after the completion of adjuvant AI therapy. Only 13–14% of the PIK3CA-mutant patients and 22–27% of the PIK3CA wild-type patients had a primary endocrine resistance and only 6% of the patients were pretreated with CDK4/6i, whereas patients with prior palliative chemotherapy, fulvestrant, or mTOR inhibitors were not eligible. Subjects with a visceral crisis, inflammatory breast cancer, a fasting plasma glucose level ≥140 mg/dL and/or glycosylated hemoglobin (HbA1c) ≥6.4% were excluded. Mutational PIK3CA analysis was predominantly performed on archived tumor tissue (>90% of patients) and focused on hotspot exon 7, 9, and 20 mutations. Median PFS was the primary endpoint in the PIK3CA-mutant cohort. Herein, the median PFS was 11.0 months for alpelisib versus 5.7 months for the control arm. In the PIK3CA-mutant cohort, the 12-months PFS was 46.3% for alpelisib versus 28.4% in the PIK3CA wild-type cohort. In the PIK3CA-mutant cohort, alpelisib improved the ORR to 26.6%, with the majority of patients showing a partial remission. The clinical benefit rate at 24 weeks was also increased to 61.5% with the addition of alpelisib. OS, the key secondary endpoint, was prolonged by 7.9 months in the PIK3CA-mutant cohort; however, this did not reach statistical significance (HR = 0.86; 95% CI 0.64–1.15; p = 0.15) [74]. Moreover, 74 and 62% of the patients had dose interruptions and dose adjustments of alpelisib, respectively, whereas a fourth of the patients permanently discontinued alpelisib due to intolerable side effects [73]. These primarily consisted of high-grade hyperglycemia, commonly developing within the first weeks of treatment, as well as diarrhea and rash, all manageable with well-defined therapeutic algorithms [75, 76, 77]. Antidiabetic management of alpelisib-induced hyperglycemia is of special importance; whereas metformin and gliflozins are the preferable agents [77], administration of insulin should be avoided since it has been demonstrated to suppress PI3K inhibition by reactivating the PI3K/mTOR cascade [77].

Since only 6% of SOLAR-1 patients were pretreated with CDK4/6i, clarification of the role of PI3Kα inhibition after CDK4/6i therapy is of importance. First data supporting the efficacy of PIK3CA inhibition after CKD4/6 inhibition has been provided by the phase II BYLIEVE trial, testing alpelisib within multiple cohorts and in combination with letrozole or fulvestrant for PIK3CA-mutant MLBC patients after CDK4/6i progression (cohort A, 121 patients) [78]. Preliminary results for this cohort have demonstrated a 6-month PFS of 50.4% for the combination of alpelisib and fulvestrant [79] and a median PFS of 7.3 months, superior to an extrapolated PFS of 3.6 months based on real-world data of patients after progression on CDK4/6i [80]. This supports alpelisib as second-line treatment after CDK4/6 inhibition in PIK3CA-mutant MLBC patients [81] (Fig. 1). Although alpelisib has received FDA approval for PIK3CA-mutant MLBC patients upon progression on or after an endocrine-based regimen, the EMA approval is limited for progression upon an endocrine monotherapy. This complicates the use of alpelisib in patients with de novo PIK3CA-mutant MLBC progressing on first-line CDK4/6i and encourages their recruitment into running clinical trials. Combined inhibition of PI3Kα and the CDK4/6/RB1 axis [82], with the aim of preventing or delaying the development of resistance, has been investigated for the combination of alpelisib, ribociclib, and letrozole in a phase 1 trial (NCT01872260) and is currently being evaluated in the phase 2/3 INAVO120 study for inavolisib (GDC-0077), in combination with palbociclib and fulvestrant (Table 1).

Fig. 1.

Proposed treatment algorithms of MLBC beyond CDK4/6i progression. ET, endocrine therapy.

mTOR Inhibitors

The combination of everolimus and exemestane has long been approved by the FDA and the EMA for patients progressing on or after a nonsteroidal AI. In addition, everolimus in combination with tamoxifen or fulvestrant has been demonstrated to be superior to endocrine therapy alone in randomized trials (TAMRAD and PrE0102), based on significant improvement of the PFS [83, 84]. However, none of these studies has shown a significant OS benefit. In the registrational BOLERO-2 trial everolimus significantly prolonged the investigator-assessed PFS, i.e., the primary endpoint, from 3.2 to 7.8 months (HR = 0.45; 95% CI 0.38–0.54; p < 0.0001) [85] and numerically prolonged OS by 4.4 months, but without reaching statistical significance (HR = 0.89; 95% CI 0.73–1.10; p = 0.14) [86]. In the PALOMA-3 trial, 16% of the patients in the palbociclib arm received everolimus/exemestane as a first subsequent treatment, with a median PFS of 4.3 months [37]. A small (n = 33) single-institution retrospective analysis in patients treated with everolimus and exemestane demonstrated comparable PFS and OS, regardless of prior CDK4/6i exposure [87]. Two further retrospective studies indicated a median PFS between 4.2 and 5.7 months for everolimus-based therapies in CDK4/6i-pretreated patients [88, 89]. Even though data from the pivotal BOLERO-2 trial also stem from the pre-CDK4/6i era, treatment with everolimus/exemestane remains a relevant second-line option [4], especially for patients without activating PIK3CA mutations (Fig. 1). Ongoing research, like in the EMBER study (Table 1), is currently evaluating oral SERD in combination with everolimus upon CDK4/6i progression. This will further elucidate the role of mTOR inhibitors after CDK4/6i.

Endocrine Monotherapy or Cytostatic Treatment

In the PALOMA-3 study, a third of the palbociclib-pretreated patients were subsequently treated with chemotherapy and a further quarter of the patients were treated with endocrine monotherapy upon CDK4/6i progression [37]. Palbociclib-pretreated patients who received chemotherapy as their immediate subsequent line showed a median duration of therapy of 5.6 (4.3–6.1) months. The duration of the subsequent chemotherapy was identical among patients in the placebo arm (5.6 months; 95% CI 3.7–6.9). Patients who were selected for chemotherapy as their first subsequent therapy had a shorter median PFS on study treatment compared to the overall study population, demonstrating a selection bias towards a more rapidly progressing and presumably more endocrine- and CDK4/6i-resistant disease. Patients receiving an endocrine therapy as their immediate subsequent line of therapy, in contrast, had a longer median PFS on study treatment compared to the overall study population, supporting this observation. The duration of endocrine therapy as the first subsequent line was 4.0 (3.2–5.7) vs. 6.2 (4.8–8.3) months in the palbociclib and placebo arm, respectively. In an efficacy analysis of subsequent therapies of patients treated in the multicenter phase 2 TREnd trial, the time to treatment failure (TTF) of the subsequent therapy was similar for both chemotherapeutic and endocrine treatments upon CDK4/6i progression and ranged from 3.7 to 4.2 months [90]. Within 2 retrospective single-center analyses of patients pretreated with palbociclib in a second or later line, the TTF ranged from 4.1 to 4.7 months for subsequent single-agent chemotherapy [91, 92], in line with real-world data [93], whereas TTF data regarding endocrine treatment upon CDK4/6i progression were less consistent [91, 92]. In a population-based observational study with a total of 525 patients receiving treatment after CDK4/6i progression, subsequent chemotherapy was more common among younger patients, with a rapid progression and non-AI backbone under CDK4/6i, whereas elderly patients, with bone-only disease or no prior cytostatic treatment, were more likely to receive subsequent CDK4/6i beyond progression [38]. Thus, age, metastatic site, clinical course of the disease, and prior endocrine treatment influence the selection of subsequent therapy after progression on CDK4/6i [38, 94].

PARP Inhibitors

More than half of the patients in the phase III OlympiAD trial, evaluating olaparib versus single-agent chemotherapy (capecitabine, vinorelbine, or eribulin) in patients with germline BRCA1/2 (gBRCA 1/2) mutations had hormone receptor-positive, HER-negative disease. For this cohort, the ORR was 65.4%, compared to 36.4% in the control arm, demonstrating a significant activity of olaparib in MLBC [95]. A second randomized phase 3 PARP inhibitor (PARPi) trial with a very similar study design, i.e., the EMBRACA trial, investigating talazoparib in the same clinical setting yielded comparable results. In the overall population, talazoparib significantly improved PFS from 5.6 to 8.6 months (HR = 0.54; 95% CI 0.41–0.71; p < 0.001). The treatment effects in EMBRACA were identical for hormone receptor-positive and hormone receptor-negative patients, with an HR for PFS of 0.47 (95% CI 0.32–0.71) and an ORR of 63.2% in hormone receptor-positive disease [96]. Even though these data also stem from the pre-CDK4/6i era, due to a sustained PFS benefit of both PARPis over physicians' choice chemotherapy, olaparib and talazoparib remain relevant treatment options for taxane- and anthracycline-pretreated, endocrine-resistant patients with gBRCA 1/2 mutations beyond CDK4/6i progression [94].

Novel Oral SERDS beyond CDK4/6i Progression

Under selective pressure of endocrine AI treatment, up to 30% of patients develop ESR1^D538 or ESR1^Y537S/N/C mutations [97]. In CDK4/6i trials, the prevalence of ESR1 mutations varies from 4% among first-line patients in the MONALEESA-2 study [17] to 25.3% among pretreated patients in the PALOMA-3 study [26]. The prevalence of ESR1 mutations in the BOLERO-2 study was 28.8% [98], and it was even higher among heavily pretreated patients in the FERGI and SOFeA trials (i.e., 37 and 39%, respectively) [99]. The most common mutations, i.e., ESR1^Y537S and ESR1^D538G, cluster in the ligand-binding domain of ER, leading to ligand-independent receptor activity [100]. Since AI only reduce levels of the ligand but cannot inhibit constitutively activated mutant ER, ESR1 mutations are predictive of resistance to AI [101, 102]. ESR1 mutations have no predictive value as biomarkers for CDK4/6i, but rather they function as factors of endocrine resistance [29]. A retrospective analysis of the SoFEA study suggests that fulvestrant might be superior to exemestane in patients with ESR1 mutations [103]. However, the numbers in this analysis are too small to come to a conclusion. The recently published plasmaMATCH study investigated an extended 500-mg dose of fulvestrant in patients with activating ESR1 mutations [104]. That study cohort did not meet its predefined response rate, but it demonstrated an objective response in 8% of the patients, suggesting at least some activity for SERD in this setting. More promising results are expected from novel oral SERD, able to overcome endocrine resistance owed to ESR1 mutations.

RAD1901 (elacestrant), demonstrating antitumor activity both against ESR1^Y537S and ESR1^D538G mutations, is the first oral SERD being tested against fulvestrant, anastrozole, letrozole, or exemestane in a phase 3 trial (EMERALD) recruiting CDK4/6i-pretreated patients with or without ESR1 mutations [105] (Table 1). In contrast, the clinical development of brilanestrant has recently been halted. SAR 439859 (amcenestrant) has also shown clinical activity irrespectively of the ESR1 mutational status among heavily pretreated MLBC patients in the AMEERA-1 trial [106] and it is currently being evaluated in the phase 2 AMEERA-3 trial versus physicians' choice among CDK4/6i-pretreated MLBC patients, as well as in the phase 3 AMEERA-5 trial in combination with palbociclib versus palbociclib/letrozole among CDK4/6i-naive MLBC patients. Another oral SERD, i.e., GDC-9545 (giredestrant), with a 10-fold higher potency than fulvestrant, showing promising results in a phase Ib trial [107] is similarly being tested in a phase 2 trial (NCT04576455) versus SOC endocrine therapy in CDK4/6i-pretreated MLBC patients and in the phase 3 BO41843 study as a first-line treatment in combination with palbociclib versus palbociclib/letrozole alone for CDK4/6i-naive MLBC patients (Table 1). LSZ 102 was the first oral SERD evaluated in escalating doses in combination with alpelisib among fulvestrant- and/or CDK4/6i-pretreated MLBC patients in a phase Ib trial [108, 109] with good tolerability and sustained antitumor activity. The phase I EMBER trial is evaluating yet another SERD, i.e., LY3484356, in combination with alpelisib or everolimus in CDK4/6i-pretreated patients. Data from these trials will hopefully provide additional information regarding the incorporation of oral SERD into the current endocrine treatment landscape and for patients who progress on a CDK4/6i.

AKT Inhibitors beyond CDK4/6i Progression

The serine/threonine kinases AKT 1, AKT 2, and AKT 3, downstream effectors of the PI3K/AKT/mTOR pathway, mediate cell proliferation and resistance to apoptosis and can be activated by many upstream receptor tyrosine kinases [110]. AKT activation is mediated by phosphorylation at at least 2 sites (i.e., pT308 from PDK1 and pS473 from mTORC2) [111], promoting breast cancer cell survival as well as resistance to endocrine and cytostatic treatments [112, 113, 114]. Multiple negative feedback loops have been described within the PI3K/AKT/mTOR signal pathway, including PTEN, IRS-1, FOXO, and PHLPP1 [111], which may in fact explain the limited efficacy of agents such as everolimus [114]. In addition, different levels of negative feedback loop activity during acute or chronic PI3K/AKT/mTOR pathway inhibition may indeed have an effect on dosing schedules for AKT inhibitors [111]. Whereas immunohistochemical loss of PTEN expression or AKT1–3 amplification is frequent in breast cancer, somatic gain-of-function AKT1^E17K mutation is found only in 1–8% of breast cancer patients [115] and results in a constitutive activation of this pathway. AKT1^E17K activity can be successfully suppressed by both allosteric AKT inhibitors, impeding AKT localization to the plasma membrane, and catalytic, ATP-competitive AKT inhibitors in breast cancer models [116]. The latter encompasses oral AKT inhibitors like GDC-0068 (ipatasertib) and AZD5363 (capivasertib). Capivasertib synergizes with fulvestrant in breast cancer models irrespectively of endocrine sensitivity and has been shown to overcome endocrine resistance by reducing PTEN levels [117]. The randomized phase 2 trial FAKTION evaluated the addition of capivasertib to fulvestrant in CDK4/6i-naive, AI-resistant MLBC patients after a maximum of 3 lines of endocrine treatment and after 1 palliative chemotherapy line [118]. The addition of capivasertib to fulvestrant leads to an improved PFS by 5.5 months, irrespectively of the PIK3CA or PTEN mutational status. The ORR and the clinical benefit rate were both 41% in favor of capivasertib. The toxicity of capivasertib was similar to that reported for the α-selective PI3K inhibitor alpelisib and included rash, diarrhea, hyperglycemia, nausea/vomiting, and stomatitis. In addition, hypertension and a single case of atypical pulmonary infection were reported. Capivasertib is currently being evaluated in the phase 3 trial CAPItello-291 in combination with fulvestrant as a second-line treatment among CDK4/6i-pretreated MLBC patients. In parallel, the phase 3 IPATunity150 trial is investigating ipatasertib in combination with SOC palbociclib/ fulvestrant among CDK4/6i-untreated MLBC patients either with relapse during adjuvant AI therapy or with progression after <12 months of first-line palliative AI therapy (Table 1). An additional phase 1 trial (TAKTIC) is also evaluating ipatasertib in multiple combination regimens among CDK4/6i-pretreated, PI3Ki-naive MLBC patients. Since the addition of ipatasertib to palliative taxane chemotherapy failed to improve PFS in PIK3CA/AKT1/PTEN-altered MLBC as well as TNBC in the phase 3 IPATunity130 trial [119], clinical development of AKT inhibitors is expected to focus on increasing the efficacy of and overcoming the resistance to endocrine therapy.

Conclusion and Perspectives

With CDK4/6i becoming the preferred standard in first-line therapy and even moving into the adjuvant setting, treatment of MLBC beyond CDK4/6i progression remains challenging. CDK4/6i treatment beyond progression cannot be recommended and should be further evaluated within clinical trials [4]. The combination of alpelisib with fulvestrant constitutes a preferred second-line treatment option for PIK3CA-mutant patients after progression on endocrine monotherapy. The requirement for progression on endocrine monotherapy according to the EMA label implies some strategic considerations for its use in de novo PIK3CA-mutant MLBC patients who received CDK4/6i in the first line. As an alternative to first-line CDK4/6i therapy, patients with a rapid relapse on or after adjuvant endocrine therapy should be strongly considered for enrollment into one of the current phase 3 trials investigating the upfront concurrent use of CDK4/6i and PI3K inhibitors (e.g., NCT04191499). For PIK3CA wild-type MLBC patients as well as PIK3CA-mutant patients with no access to alpelisib, or concomitant diseases precluding its use, the combination of everolimus/exemestane as a second-line treatment beyond CDK4/6i progression remains a valid option [4]. For younger patients with aggressive disease and a rapid clinical progression or a visceral crisis, cytostatic monotherapies (potentially in combination with bevacizumab) should probably be prioritized, whereas elderly patients in selected cases, e.g., with slowly progressing disease and a limited disease burden, may be considered as candidates for endocrine monotherapies [38, 94]. In taxane- and anthracycline- pretreated patients harboring gBRCA1/2 mutations, PARPis like olaparib and talazoparib constitute a preferred treatment option after progression on CDK4/6i [94].

Current phase 1–3 clinical trials will hopefully reshape the treatment landscape of MLBC. Oral SERDS and AKT inhibitors are being developed beyond CDK4/6i in phase 2–3 trials. Novel PI3K inhbitors, such as inavolisib, but also BCL-2 inhibitor venetoclax and the AURKA inhibitor alisertib, have already reached clinical phase 2 trials with CDK4/6i-pretreated patients, whereas FGFR inhibitors and PD-1/ PD-L1 antibodies are still in early clinical development in this setting. These new advances are expected to pave the way for the development of further biomarkers in MLBC. Regarding the expected shift of CDK4/6i toward adjuvant treatment among high-risk early breast cancer patients within the next years [120], accelerated incorporation of novel therapeutic agents into the treatment of MLBC is mandated, allowing for the development of highly-efficient, individualized antitumor therapies.

Conflict of Interest Statement

F.M. has received honoraria from Roche, Amgen, Astra Zeneca, Eisai, Celgene, Novartis, Pfizer, Genomic Health, Myriad, Seagen, and Lilly and acted as a consultant for Roche, Astra Zeneca, and Novartis. A.M. has received travel sponsorship from Pfizer, Eisai, and Celgene, as well as honoraria from Roche. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in this paper apart from those disclosed.

Funding Sources

This review was not funded.

Author Contributions

All of the authors contributed significantly to the literature review and the writing of this work and approved the final version.

verified