Endocrine monotherapies

Inhibition of ER signaling remains the mainstay of palliative therapy in HR+/HER2– breast cancer and can be achieved through estrogen suppression or direct ER targeting, which represents the mechanism of action of several novel agents, including SERDs, proteolysis-targeting chimeras (PROTACs), selective ER modulators (SERMs) and complete ER antagonists (CERANs) [33].

Mutations in the ESR1 gene encoding the ER are an acquired resistance mechanism to ET. ESR1 mutations induce conformational changes, constitutive activation, and increased proteolytic stability of the ER, thereby inducing resistance to estrogen-depleting approaches with AI, and to a variable extent, different agents directly targeting the ER [34]. ESR1 mutations are rare in treatment-naïve HR+/HER2− breast cancer, but occur at a prevalence up to 20–40% in advanced disease treated with AI [34]. They have been shown to arise during AI therapy, and to precede and correlate with clinical progression [35, 36]. In contrast to AI, which reduces ER signaling via estrogen deprivation, SERDs inhibit ER-induced downstream signaling by directly binding to the ER and inducing its proteasomal degradation [37].

Fulvestrant had long been the only SERD available for treatment of HR+/HER2– metastatic breast cancer. Following progression on a CDK4/6i, a review of recent randomized phase II–III trials using fulvestrant as control therapy has shown a median PFS of around 3.2 months (range 1.9–5.3 months), consistent with real-world outcomes [38–40]. Given the limited clinical activity of fulvestrant after CDK4/6i and its intramuscular route of application due to poor oral bioavailability, oral SERDs were developed to improve potency and enable more convenient oral administration [37].

Elacestrant was the first oral SERD to demonstrate efficacy as monotherapy after mandatory prior CDK4/6i exposure and a maximum of one prior line of chemotherapy in a phase III trial. In the EMERALD study, elacestrant was compared with investigator’s choice (IC) of ET (IC ET) (AI or fulvestrant) in patients with disease progression on prior ET plus a CDK4/6i. The trial met its two primary endpoints, demonstrating improved PFS in both the overall population and with greater benefit in the ESR1-mutated subgroup (median PFS 3.8 months with elacestrant vs. 1.9 months for IC ET; hazard ratio (HR) 0.55 [95% confidence interval (CI) 0.39–0.77]; P = 0.0005). Landmark analyses demonstrated a 6-month PFS rate of 40.8% with elacestrant vs. 19.1% with IC ET in the ESR1-mutated subgroup. A secondary analysis excluded the 31% of patients receiving AIs in the control arm, accounting for the recognized resistance associated with ESR1 mutations. Elacestrant continued to demonstrate benefit over fulvestrant, again with a more pronounced effect in the ESR1-mutated cohort [6-month PFS rate 40.8% vs. 20.8%; HR 0.50 (95% CI 0.34–0.74); P = 0.005] [41]. The most common adverse events (AEs) (any grade) with elacestrant were nausea (35%), fatigue (19%), vomiting (19%), decreased appetite (14.8%) and arthralgia (14.3%). Among grade ≥ 3 events, nausea (2.5%) and back pain (2.5%) occurred most frequently, followed by increase in alanine aminotransferase (ALT) and aspartate aminotransferase (AST) (2.1% and 1.7%, respectively) and headache (1.7%). Treatment-related AEs led to treatment discontinuation in 3.4% of patients (Table 1) [41]. No OS benefit was shown at the final OS analysis in either the intention-to-treat (ITT) or the ESR1-mutated population (both key secondary endpoints; 51.3% of the ITT population experienced an event) [42]. A separately published, post-hoc exploratory analysis by pretreatment duration in exclusively ESR1-mutated patients has been considerably discussed in the field. Longer prior CDK4/6i exposure (≥ 12 months) was associated with improved PFS on elacestrant vs. IC ET [8.6 months vs. 1.9 months, HR 0.41 (95% CI 0.26–0.63)] in ESR1-mutated patients, compared to a shorter pretreatment period [43].

The oral SERD camizestrant has likewise shown clinical activity in the randomized phase II SERENA-2 trial, in which two dose levels were compared with fulvestrant following disease progression on ET, with optional prior CDK4/6i and chemotherapy exposure for advanced disease [44]. ESR1 mutations were detected in 48% of patients in the fulvestrant group and in 30% of those receiving the camizestrant dose [44] that was subsequently selected for further evaluation in the SERENA-4 and 6 trials [28, 45]. At this dose level, camizestrant improved PFS compared with fulvestrant [median PFS 7.2 vs. 3.7 months for camizestrant and fulvestrant, respectively; HR 0.59 (90% CI 0.42–0.82); P = 0.017]. In the prespecified exploratory analyses by ESR1 mutation status and prior CDK4/6i exposure (51% of trial population), the relative benefit of camizestrant seemed maintained in patients with ESR1 mutation [HR 0.33 (90% CI 0.18–0.58)], but not in patients without ESR1 mutation [HR 0.80 (90% CI 0.51–1.27)]. Treatment effect appeared sustained in patients with prior CDK4/6i exposure [HR 0.49 (90% CI 0.31–0.75)], but not without prior CDK4/6i treatment [HR 0.81 (90% CI 0.49–1.33)]. Most common treatment-emergent AEs (any grade) with camizestrant 75 mg were photopsia (12%), anemia (10.8%), cough (9.5%), asthenia (8.1%), and nausea (8.1%). Treatment-related AEs leading to discontinuation in the camizestrant 75 mg group occurred in 3% of patients. No specific treatment-emergent or treatment-related grade ≥ 3 event was observed in > 1 patient (Table 1). OS data were immature [44]. Camizestrant has been evaluated in combination with a CDK4/6i upon molecular progression in the SERENA-6 trial, as discussed below, and is currently being assessed with palbociclib in the first-line setting in the SERENA-4 trial [28, 45].

The three-arm phase III EMBER-3 trial compared the oral SERD imlunestrant to IC ET (exemestane or fulvestrant) and to imlunestrant plus abemaciclib. Patients were eligible with endocrine-resistant disease after adjuvant or first-line AI therapy, with or without prior CDK4/6i therapy, and no chemotherapy for advanced disease was permitted. Ninety percent of patients in the control arm received fulvestrant, and 10% received exemestane [46]. The trial was negative for its primary endpoint PFS with imlunestrant vs. IC ET in the overall population, but a statistically significant PFS benefit [median PFS 5.5 months with imlunestrant and 3.8 months with IC ET; HR 0.62 (95% CI 0.47–0.82); P < 0.001] in the ESR1-mutated group—another primary endpoint—was demonstrated with initial drop and later separation of the Kaplan–Meier curves [46, 47]. Exploratory subgroup analysis of the ESR1-mutated cohort by prior CDK4/6i therapy showed an HR of 0.42 (95% CI 0.25–0.72) in CDK4/6i-naïve and 0.72 (95% CI 0.52–1.01) in CDK4/6i-pretreated patients (70% of the ESR1-mutated population, predominantly palbociclib) [46]. A numerical improvement in OS (key secondary endpoint) was observed in favor of imlunestrant without reaching statistical significance in the ESR1-mutated cohort [median OS 34.5 vs. 23.1 months, respectively; HR 0.60 (95% CI 0.43–0.86); P = 0.0043, not significant] [47]. Single-agent ET with imlunestrant was inferior to the combination therapy with abemaciclib, as will be outlined below [46, 47]. The most common AEs of any grade were fatigue/asthenia (22.6%), diarrhea (21.4%), nausea (17.1%), arthralgia (14.1%), and AST increase (12.5%). AEs of grade ≥ 3 were anemia and neutropenia (2.1% each), AST increase and thrombopenia (0.9% each) [46]. Treatment-related AEs led to treatment discontinuation in 2% of patients (Table 1) [47].

The randomized trials evaluating the oral SERDs amcenestrant and giredestrant as monotherapy yielded negative results against IC ET in endocrine-resistant populations with varying degrees of prior CDK4/6i exposure [48, 49]. Giredestrant was positive for its primary endpoint, PFS, in the phase III evERA trial in combination with everolimus (full publication results pending), as described below [50], and is currently investigated with a CDK4/6i in phase III trials in the first-line setting in endocrine-sensitive and endocrine-resistant disease [29, 51].

PROTACs represent an additional class of novel endocrine agents with proven clinical efficacy in a phase III trial [52]. PROTACs act by recruiting a ubiquitin ligase to a target protein, promoting its polyubiquitination and subsequent proteasomal degradation [33]. Vepdegestrant, a first-in-class PROTAC targeting the ER [33], demonstrated superiority over fulvestrant in the phase III VERITAC-2 trial in an endocrine-resistant, post-CDK4/6i population [52]. Patients had to be on their most recent ET for ≥ 6 months before progression, and no prior chemotherapy or fulvestrant for advanced disease was allowed. Vepdegestrant did not meet the primary endpoint of PFS in the ITT population but demonstrated superiority in the primary endpoint of PFS in the ESR1-mutated subgroup [median PFS 5.0 vs. 2.1 months for vepdegestrant and fulvestrant, respectively; HR 0.58 (95% CI 0.43–0.78); P < 0.001]. Data on OS were immature. Most common AEs were fatigue/asthenia (26.6%), ALT or AST increased (14.4% each), nausea (13.5%), and anemia (12.2%). The most common AE grade ≥ 3 were neutropenia, hypokalemia (1.9% each), anemia, QT prolongation, and hypertension (1.6% each, Table 1) [52].

Lasofoxifene is a next-generation SERM and has been tested against fulvestrant in the phase II ELAINE-1 trial in an exclusively ESR1-mutated cohort progressing on AI and CDK4/6i therapy. The trial was negative for its primary endpoint, PFS [median 5.6 months for lasofoxifene and 3.4 months for fulvestrant; HR 0.70 (95% CI 0.43–1.15); P = 0.14] [53].

CERANs are another novel class of drugs targeting the ER and function as complete estrogen antagonists by blocking the ligand-binding pocket of the ER. Palazestrant is currently being evaluated in the phase III OPERA-01 trial in endocrine-resistant, CDK4/6i pretreated patients against endocrine monotherapy [33, 54].

In conclusion, the three endocrine agents elacestrant, imlunestrant, and vepdegestrant have each demonstrated efficacy as monotherapy in randomized phase III trials; with an ESR1-mutated subgroup included as the primary analysis population. While elacestrant showed a PFS benefit across a biomarker-unselected and the ESR1-mutated population, the benefit with imlunestrant and vepdegestrant was confined to the ESR1-mutated subgroup [41, 46, 52]. These findings should be interpreted in the context of some methodological considerations regarding trial design and cross-trial heterogeneity complicating treatment comparisons (Table 2). All three trials employed ET monotherapy as control treatment, which may not reflect current clinical practice where combination approaches are increasingly preferred. Comparator selection differed across trials, which is a relevant aspect as suboptimal treatment in the control arm may inflate observed treatment effects. While prior fulvestrant was an exclusion criterion in both the vepdegestrant and imlunestrant trials—consistent with fulvestrant being the control treatment—a minority of patients in the imlunestrant trials received AI therapy as control treatment despite mandatory AI preexposure; and in the elacestrant trial, both fulvestrant and AI were used as control treatments despite prior exposure being allowed. In addition, CDK4/6i pretreatment was mandatory only in the vepdegestrant and elacestrant trials, limiting applicability of the imlunestrant results to a post-CDK4/6i setting [41, 46, 52]. Moreover, the included population was heavily pretreated in the elacestrant trial, which may have attenuated the observed treatment effect, as treatment efficacy generally decreases with increasing lines of systemic therapy [41]. Across the three trials, patients with indeterminate ESR1 mutational status were classified and analyzed as ESR1-non-detectable. However, reporting of the prevalence of ESR1-negative and ESR1-indeterminate within the ESR1-non-detectable group was inconsistent between trials, and ESR1 mutational status was not a stratification factor in all three trials (Table 3) [41, 46, 52]. Camizestrant as monotherapy only demonstrated activity in a phase II setting, with the clinically relevant subgroup of patients with CDK4/6i preexposure limited to fewer than 40 patients per arm; the clinical significance of these findings therefore remains to be confirmed in a larger trial [44].

CDK4/6i post CDK4/6i-progression: identical CDK4/6i, switch of ET

Given the modest efficacy of endocrine monotherapy following progression on ET and CDK4/6i, the selection of a combination therapy may be favored as a further line therapy. A biomarker-independent strategy consists of either switching to a different endocrine agent and a different CDK4/6i, or switching to an alternative ET while maintaining the same CDK4/6i beyond progression on both agents.

The phase II PALMIRA trial evaluated the continuation of palbociclib beyond disease progression by randomizing patients previously treated with palbociclib plus an AI or fulvestrant to receive either ET alone or ET with continued palbociclib. The trial did not meet the primary endpoint of PFS [55]. Similarly, the phase II PACE trial randomized patients progressing on an AI plus CDK4/6i therapy (palbociclib in 91% of patients) to fulvestrant alone or fulvestrant plus palbociclib, but did not demonstrate a PFS benefit for the combination therapy [56].

Two extensively debated trials introduced a novel treatment approach, in which therapy was not switched upon clinically evident progression, but upon emergence of a molecular marker of progression [45, 57]. Similar to PALMIRA and PACE, the PADA-1 trial evaluated an ET switch with continued CDK4/6 inhibition. Notably, the treatment was changed at the time of molecular progression—defined by rising ESR1 mutations in circulating tumor DNA (ctDNA)—rather than at clinical progression on first-line AI and palbociclib. Patients with rising ESR1 mutations were randomized to continue AI plus palbociclib or switch to fulvestrant plus palbociclib. This strategy was associated with a significant improvement in the primary endpoint PFS in favor of the molecularly triggered treatment switch [updated analysis: median PFS 12.8 months and 5.9 months for switch to fulvestrant and continuation of AI, respectively; HR 0.54 (95% CI 0.38–0.75); P = 0.0003] [57, 58]. Sixty-eight percent of patients in the control arm crossed over to fulvestrant plus palbociclib after clinically evident progression, with a median PFS of 3.5 months in this subsequent line. The secondary endpoint median time to strategy failure, i.e., time from randomization to time of palbociclib and ET discontinuation or death, was 11.9 months for the switch-fulvestrant-group and 10.6 months for the AI-continuation group [HR 1.02 (95% CI 0.71–1.45); P = 0.9; from primary analysis]. Similarly, the median chemotherapy-free survival was 14.6 months in the fulvestrant-switch group compared with 13.1 months in the AI-continuation group [HR 0.91 (95% CI 0.62–1.33); secondary endpoint, P = 0.6, from primary analysis] [57]. OS data were still immature [58].

Analogous to PADA-1, SERENA-6 investigated a treatment switch triggered by the detection of ESR1 mutations by ctDNA prior to clinically evident progression. Patients on first-line AI plus CDK4/6i therapy (palbociclib in 76% of patients) were randomized upon ESR1 mutation detection to either continue the same regimen or to switch the endocrine backbone to the oral SERD camizestrant, while maintaining the same CDK4/6i. Crossover to an oral SERD with CDK4/6i continuation was not allowed in the control arm. In patients with a detectable ESR1 mutation who were randomized to the respective treatments, PFS was significantly prolonged with camizestrant and CDK4/6i in comparison to continuation of the initial regimen [median PFS 16 months vs. 9.2 months; HR 0.44 (95% CI 0.31–0.60); P < 0.0001]. Results for PFS2, which in principle compares an unequal number of sequentially administered therapeutic substances between groups, were immature at first analysis [HR 0.52 (95% CI 0.33–0.81); P = 0.004, prespecified significance threshold P = 0.0001]. OS data were immature [45].

The PADA-1 and SERENA-6 trials were notable for the pioneering approach to guide treatment decisions based on detection of molecularly defined progression. Beyond the aspiration to improve survival outcomes, potential advantages of a molecularly guided switching approach may theoretically include the prevention of tumor-progression-related complications, symptoms and clinical deterioration, and possible reduction of clinical and financial toxicity from an increasingly ineffective treatment. Despite both trials being positive for their primary endpoint, PFS, this endpoint design does not allow for a conclusion whether switching upon molecular instead of clinical progression is superior in efficacy, given the inherent imbalance of active substances between treatment arms. In contrast, the lack of differences in median time to strategy failure and chemotherapy-free survival between the two arms in PADA-1 does not appear to indicate a benefit of therapy adaptation upon molecular rather than clinically evident progression. Furthermore, the definition of molecular progression represents a conceptual challenge. Longitudinal ctDNA analyses from PADA-1 suggest that ESR1 mutation detectability is dynamic, with fluctuations and occasional loss of detection occurring independently of treatment changes, including in the AI plus CDK4/6i arm [58]. This might challenge the classification of ESR1 positivity, its direct attribution as a biomarker of relevant molecular progression, and underscores the need for standardized assays and uniform definitions to reduce risk of possible overtreatment. In addition, the feasibility of regular ctDNA monitoring needs to be determined across all healthcare settings, and associated surveillance demands might carry implications for patient experience and health systems.

In summary, switching the endocrine backbone to fulvestrant while maintaining the same CDK4/6i upon clinical progression has only been investigated for palbociclib and did not improve efficacy [55, 56]. Molecularly guided switching strategies of the endocrine agent on first-line AI and CDK4/6i represent a novel approach with potentially beneficial implications, yet convincing evidence of superiority in efficacy is currently lacking, and questions regarding practical implementation and implications for patients and health system resources remain open [45, 57, 58].

CDK4/6i post CDK4/6i-progression: switch of CDK4/6i and switch of ET

In the phase II MAINTAIN trial, patients progressing on ET and CDK4/6i (palbociclib in 87% of patients) in the advanced setting underwent a switch to an alternative ET (fulvestrant or AI), either as monotherapy or in combination with ribociclib. Switching ET plus ribociclib led to improved PFS compared to switching to an alternative ET monotherapy [median PFS 5.3 vs. 2.8 months; HR 0.57 (95% CI 0.39–0.85); P = 0.006]. No OS analysis was reported [59].

Similarly, in the phase III postMONARCH trial, patients progressing on AI plus CDK4/6i therapy were switched to fulvestrant monotherapy or fulvestrant in combination with abemaciclib. Palbociclib was the most common prior CDK4/6i (59% of patients), followed by ribociclib (33%) and abemaciclib (8%). The addition of abemaciclib led to a statistically significant PFS benefit [median PFS 6.0 vs. 5.3 months, and 6-month PFS rates 50% vs. 37% for the combination therapy and fulvestrant, respectively; HR 0.73 (95% CI 0.57–0.95); P = 0.017]. Exploratory subgroup analyses suggested a potential benefit among patients with prior palbociclib exposure [HR 0.62 (95% CI 0.44–0.86)], while no effect was observed in the small subgroup of ribociclib-pretreated patients [HR 1.01 (95% CI 0.67–1.51)]. However, the interaction p-value was not significant. OS data were immature [60].

The approach of simultaneous endocrine and CDK4/6i switch has been modified to the use of oral SERDs instead of fulvestrant for potentially better efficacy and a more convenient oral formulation. In the three-arm phase III EMBER-3 trial, imlunestrant plus abemaciclib was superior to imlunestrant monotherapy after progression on an AI, which had been combined with a CDK4/6i in 65% of patients (palbociclib in 63%, ribociclib in 27%, and abemaciclib in 8% of cases) [47]. The combination of imlunestrant and abemaciclib was superior to imlunestrant alone with respect to the primary endpoint PFS in the ITT population [median PFS 10.9 vs. 5.5 months; HR 0.59 (95% CI 0.47–0.74); P < 0.001] [47]. A benefit for the combination therapy was observed in a prespecified exploratory analysis both among patients with prior CDK4/6i treatment [median PFS 9.1 vs. 3.7 months; HR 0.53 (95% CI 0.40–0.69)] and across ESR1-mutated and non-mutated subgroups [46, 47]. OS data were not mature yet [47].

In the single-arm phase II ELAINE-2 trial, which exclusively enrolled patients with ESR1-mutated disease, lasofoxifene plus abemaciclib demonstrated encouraging activity (median PFS 13 months and overall response rate (ORR) 56%) in a more heavily pretreated population allowing previous ET, chemotherapy, and multiple CDK4/6i exposures [61]. This combination will be compared against abemaciclib and fulvestrant in the phase III ELAINE-3 trial in ESR1-mutated disease following progression on CDK4/6i and AI, addressing the need for head-to-head comparisons of combination therapies [62].

Taken together, after progression on ET plus CDK4/6i, switching ET and adding abemaciclib or ribociclib improved PFS vs. endocrine monotherapy [47, 59, 60]. However, this benefit has mainly been shown after prior palbociclib, and its efficacy following abemaciclib or ribociclib remains less certain [47, 59, 60]. The approach of combining imlunestrant with abemaciclib demonstrated a median PFS of 10.9 months, which appears to be an encouraging result. However, the notable limitation of only 61% CDK4/6i pretreatment complicates comparison with other regimens and limits applicability to the post-CDK4/6i setting [47].

Therapeutic combinations targeting the PI3K/AKT/mTOR signaling pathway

The PI3K/AKT/mTOR signaling pathway plays an important role in cell proliferation, growth, survival, and metabolism [63]. Intracellular activation of the lipid kinase PI3K leads to recruitment of the serine/threonine kinase AKT, which is further activated through phosphorylation by mTORC2. mTORC1 and 2 represent two complexes with distinct functions that both incorporate the serine/threonine kinase mTOR. Activated AKT is a central signaling node of the pathway and mediates further downstream signaling on multiple target proteins, including activation of mTORC1 among downstream consequences. In contrast, the phosphatase PTEN inhibits PI3K/AKT signaling by antagonizing AKT activation [63, 64].

In HR+/HER2– ABC, the mTOR inhibitor everolimus was the first clinically introduced substance to interfere with the PI3K/AKT/mTOR pathway and inhibits mTORC1 [64, 65]. Everolimus showed therapeutic efficacy in clinical trials in an endocrine-resistant population in combination with ET in the pre-CDK4/6i-era [66–68]. Treatment with everolimus has not been biomarker-guided, and an association with treatment efficacy with PIK3CA mutation has not been established [69, 70].

More recently, approaches targeting the PI3K/AKT/mTOR pathway in combination with fulvestrant in a biomarker-selected population have been implemented in the treatment landscape of HR+/HER2– ABC: In contrast to ESR1 mutations acquired during treatment, PIK3CA mutations—encoding the catalytic subunit of PI3K (p110α)—are mostly considered early, tumor-intrinsic (“truncal”) events and have been observed in up to approximately 35–45% of HR+/HER2– breast cancers, although there is evidence demonstrating spatial and temporal heterogeneity [71–75].

The PI3K inhibitor alpelisib was the first PI3K inhibitor implemented in the treatment of HR+/HER2– ABC [76]. Alpelisib targets the p110α subunit of PI3K [64]. In endocrine-pretreated patients, alpelisib plus fulvestrant demonstrated superiority over fulvestrant for the primary endpoint PFS in a PIK3CA-mutated population [median PFS 11 vs. 5.7 months for the combination therapy vs. fulvestrant, respectively; HR 0.65 (95% CI 0.50–0.85); P < 0.001] in the phase III SOLAR-1 trial. For the secondary endpoint of PFS in the PIK3CA wild-type (WT) patients, no significant effect could be demonstrated [76]. No significant OS benefit was observed [77]. Given that only 6% of patients had received prior CDK4/6i therapy, this treatment was further evaluated in the single-arm BYLieve trial requiring prior CDK4/6i exposure, confirming efficacy in this population (median PFS 8.0 months) [76, 78]. However, alpelisib was associated with relevant toxicity, with a treatment discontinuation rate due to AEs of 26%. The most common grade 3–4 AEs were hyperglycemia (37%), rash (10%), and diarrhea (7%) [77]. Results of the phase III EPIK-B5 trial, available in abstract form only, provided evidence of a significant PFS benefit (primary endpoint) of alpelisib when added to fulvestrant in a post-CDK4/6i and AI setting in an exclusively PIK3CA-mutated population [median PFS 7.4 vs. 2.8 months; HR 0.52 (95% CI 0.37–0.72); P < 0.001]. The authors stated a positive trend for OS, with a median OS of 29.5 vs. 23.8 months in the control arm [HR 0.64 (95% CI 0.41–0.99); P = 0.021]. Statistical significance was not explicitly claimed, and formal interpretation of this P-value awaits full publication [79].

Inavolisib, a selective p110α inhibitor that also degrades mutant p110α, has been incorporated into first-line therapy for HR+/HER2– breast cancer in combination with palbociclib and fulvestrant in patients with a PIK3CA mutation [26]. Inavolisib plus fulvestrant is being evaluated vs. alpelisib plus fulvestrant in patients with PIK3CA-mutated disease after CDK4/6i therapy in the phase III INAVO121 trial, thereby generating important clinical evidence for direct comparisons between endocrine combination therapies [80].

The novel, allosteric PI3Kα inhibitor zovegalisib is designed to target mutant PI3Kα while sparing the WT variant, with the aim of reducing the side effects of PI3Kα-inhibitors [64]. In the phase III ReDiscover-2 trial, zovegalisib plus fulvestrant is being tested against capivasertib and fulvestrant in a CDK4/6i pretreated population with PIK3CA mutation, providing important direct comparisons of combination therapies [81].

Apart from mutations in PIK3CA, the PI3K/AKT/mTOR pathway can be activated by less prevalent mutations in AKT1 (5–7% in ABC) and by loss of PTEN function (5–10% of breast cancers), both resulting in AKT activation [64]. The selective AKT inhibitor capivasertib targets PI3K/AKT/mTOR pathway activation driven by genomic alterations in PIK3CA, AKT1, or PTEN. In combination with fulvestrant, capivasertib demonstrated superiority over fulvestrant in the randomized phase III CAPItello-291 trial. Eligible patients had endocrine-resistant advanced HR+/HER2– breast cancer; 70% had prior exposure to CDK4/6i and 18% had received chemotherapy for advanced disease. Forty-one percent of patients had a confirmed PI3K-AKT pathway alteration, with activating PIK3CA mutations in 31%, activating AKT1 mutations in 5%, and inactivating PTEN alterations in 5% of patients. Of note, PI3K-AKT pathway alteration status was not assessed before inclusion and not used as a stratification factor at randomization; 15% of patients enrolled had an unknown alteration status (Table 3). Capivasertib plus fulvestrant significantly improved PFS for both primary endpoints: in the ITT population [median PFS 7.2 vs. 3.6 months; HR 0.60 (95% CI 0.51–0.71); P < 0.001] and in patients with a PI3K-AKT pathway alteration [median PFS 7.3 vs. 3.1 months; HR 0.50 (95% CI 0.38–0.65); P < 0.001]. Exploratory subgroup analysis suggested a benefit irrespective of CDK4/6i pretreatment in the ITT population. An exploratory analysis in the PI3K-AKT pathway-non-altered population, performed after exclusion of the 25% of patients with unknown molecular status, suggested no clear PFS benefit in this subgroup [HR 0.79 (95% CI 0.61–1.02)]. The most common grade ≥ 3 AEs were rash (12%), followed by diarrhea (9%) and hyperglycemia (2.3%). Treatment discontinuation due to AEs occurred in 13% of patients [82]. Capivasertib in combination with fulvestrant and a CDK4/6i is investigated against fulvestrant and CDK4/6i in patients experiencing endocrine-resistant disease in the adjuvant setting in the phase III CAPItello-292 trial [83].

Evidence on another AKT inhibitor, ipatasertib, is currently available as an abstract only, and demonstrated a significant PFS benefit in combination with fulvestrant vs. fulvestrant alone in the phase III CCTG/BCT MA.40/FINER trial. Eligible patients had advanced HR+/HER2– breast cancer progressing on AI and CDK4/6i. The trial was positive for PFS in the ITT population [primary analysis, median PFS 5.3 vs. 1.9 months; HR 0.61 (95% CI 0.46–0.81); P = 0.0007], as well as in the PI3K-AKT pathway-altered cohort (secondary analysis), i.e., with genomic alterations in PIK3CA, AKT1, or PTEN [median PFS 5.5 vs. 1.9 months; HR 0.47 (95% CI 0.31–0.72); P = 0.0005]. The most common non-hematological grade ≥ 3 AEs were diarrhea (16%), fatigue (3%), vomiting, and rash (2% each). Treatment discontinuation rate due to AEs was 6.5% [84].

In contrast to the previously discussed trials evaluating PI3K and AKT inhibitors with fulvestrant as ET backbone [79, 82, 84], the mTOR inhibitor everolimus has been investigated in combination with the oral SERD giredestrant in the phase III evERA trial [50]. Patients were included upon progression following ET and CDK4/6i and were randomized to either giredestrant plus everolimus or IC ET (i.e., exemestane, fulvestrant or tamoxifen) plus everolimus. The trial was positive for both primary endpoints PFS in the ITT [median PFS 8.8 vs. 5.5 months; HR 0.56 (95% CI 0.44–0.71); P < 0.0001] and in the ESR1-mutated cohort [median PFS 10 vs. 5.5 months; HR 0.38 (95% CI 0.27–0.54); P < 0.0001, presented as abstract] [50]. Full trial publication, along with its clinical positioning within the treatment landscape, remains pending. A clear strength of the trial was the incorporation of a combination-therapy regimen as comparator treatment [50]. Everolimus in combination with the oral SERD elacestrant vs. elacestrant monotherapy is currently investigated in the randomized phase III ADELA trial in the post-CDK4/6i setting in an exclusively ESR1-mutated population [85].

Similar to the triplet combination strategy of ET, CDK4/6i and PI3K/AKT/mTOR pathway targeting evaluated in the first-line INAVO120 trial [26], the phase III VIKTORIA-1 trial investigated gedatolisib, palbociclib, and fulvestrant in CDK4/6i pretreated patients [86]. Gedatolisib is a multitarget inhibitor targeting both PI3K and mTORC1/2. The two-part, three-arm trial investigated both i) the triplet combination gedatolisib, palbociclib, and fulvestrant and the doublet ii) gedatolisib and fulvestrant vs. iii) fulvestrant monotherapy. No prior exposure to chemotherapy or ADCs in the advanced setting was allowed. Fulvestrant pretreatment was allowed, but the number of fulvestrant-pretreated patients was not reported. Forty percent of patients had prior palbociclib exposure; 49% ribociclib and 17% prior abemaciclib. Parts 1 and 2 of the study enrolled patients with PIK3CA-WT and PIK3CA-mutated tumors, respectively. In part 1, PFS was significantly prolonged with the triplet therapy compared to fulvestrant [median PFS 9.3 vs. 2.0 months; HR 0.24 (95% CI 0.17–0.35); P < 0.001] and with the doublet therapy compared to fulvestrant [median PFS 7.4 vs. 2.0 months; HR 0.33 (95% CI 0.24–0.48); P < 0.001]. The most common grade ≥ 3 treatment-related AEs were neutropenia (62.3% and 0.8% in the triplet and doublet combination group, respectively), stomatitis (19.2% and 12.3%), leukopenia (6.2% and 0%), rash (4.6% and 5.4%), nausea (3.8% and 0.8%), hyperglycemia (2.3% for both groups) and diarrhea (1.5% and 0.8%). Treatment-related AEs led to treatment discontinuation in 2.3% of patients in the triplet combination group and 3.1% in the doublet combination group. Two treatment-related deaths occurred with the triplet combination (pneumonia and hepatic failure) [86].

In conclusion, alpelisib represents a treatment option in the endocrine-resistant management of PIK3CA-mutated HR+/HER2– ABC, supported by a PFS benefit, but in the absence of an OS benefit [76, 77]. Fully published trial evidence in the post-CDK4/6i setting still derives, at present, from single-arm phase II data [78]. Capivasertib has been incorporated in the same therapeutic setting for tumors harboring a PI3K-AKT pathway alteration, after demonstration of a PFS benefit in an only partially CDK4/6i pretreated population [82]. While alpelisib and capivasertib exhibit a partially overlapping, class-related toxicity profile, the safety profile reported for capivasertib is considered more favorable, particularly with regard to severe hyperglycemia [77, 82]. It should be clearly emphasized, however, that this observation is based on cross-trial comparison. With PFS benefits shown for both its dual and triplet combinations, gedatolisib emerges as a potential treatment option in the absence of PIK3CA mutation [86]. In patients with PIK3CA-WT disease, it would have its position as an alternative to previously biomarker-unselected strategies. Though promising, to date, gedatolisib has not yet received Food and Drug Administration (FDA) nor European Medicines Agency (EMA) approval [87, 88]. Across the three trials, fulvestrant served as the control arm, not reflecting the current SOC in most cases. The trials further differed in their pretreatment inclusion criteria—especially relevant with respect to prior fulvestrant and CDK4/6i exposure—and biomarker stratification (Tables 2 and 3); possibly affecting observed treatment effects [76, 82, 86]. The clinical positioning of the ipatasertib-fulvestrant and everolimus-giredestrant combinations within the treatment landscape awaits the full publication of the respective trials [50, 84].

PARP inhibitors

Poly(ADP-ribose)-polymerase inhibitors (PARPi) represent a non-endocrine-based alternative for patients with pathogenic or likely pathogenic germline BRCA1/2 (gBRCA1/2) and somatic BRCA1/2 (sBRCA1/2) mutations, and mutations in germline PALB2 (gPALB2) [89–92].

Two phase III trials tested the PARPi olaparib (OlympiAD) and talazoparib (EMBRACA) against chemotherapy in HER2– ABC in patients with gBRCA1/2-mutations with a maximum of three previous chemotherapy regimens and no limit on previous hormonal therapies (EMBRACA) or a maximum of two previous chemotherapies and at least one previous hormonal therapy (OlympiAD). Importantly, these studies were performed before the introduction of CDK4/6i. Both trials reported a significant PFS benefit in the ITT population [olaparib: median PFS 7.0 vs. 4.2 months, HR 0.58 (95% CI 0.43–0.80); P < 0.001; talazoparib: median PFS 8.6 vs. 5.6 months, HR 0.54 (95% CI 0.41–0.71); P < 0.001], but no improvement in OS was observed [89, 91, 93, 94]. Exploratory subgroup analyses of the respective trials suggested a PFS benefit for olaparib in the triple-negative breast cancer (TNBC) population, but not in HR+/HER2– disease, whereas talazoparib led to improvement in median PFS regardless of subtype [93, 94].

To contextualize the role of PARPi in a treatment landscape after CDK4/6i exposure, a retrospective multi-site analysis suggested a real-world PFS of 8.5 months for talazoparib in HR+/HER2– patients who were mainly pretreated with CDK4/6i [95]. In the phase IIIb LUCY study, no signal of reduced olaparib efficacy was observed in patients with prior CDK4/6i exposure (n = 25, median PFS 9.1 months) compared with CDK4/6i-naïve patients (n = 12, median PFS 8.4 months), although subgroup sizes were unequal and not formally compared [96].

The phase II TBCRC 048 trial provided a signal of anti-tumor efficacy regarding ORR (primary endpoint) of olaparib in patients with mutations in gPALB2 (ORR 68% in HR+/HER2– disease, n = 19) or with tumors that contain a mutation in sBRCA1/2 (ORR of 36% in HR+/HER2– disease, n = 33). The vast majority of HR+/HER2– patients were CDK4/6i pretreated [92]. No relevant activity was observed for ATM or CHEK2 mutations [90]. Based on these data, testing for gBRCA1/2 is recommended to guide treatment strategy in HR+/HER2– breast cancer, as is testing for gPALB2 and sBRCA1/2, although with a lower level of evidence and weaker recommendation [13, 14].

Chemotherapy-based regimens

Chemotherapies, including ADCs, become relevant once endocrine (combination) therapies are exhausted or a rapid tumor response is clinically indicated. Single-agent chemotherapy has been preferred over combination chemotherapy in this setting due to the lack of an OS benefit with combination regimens, except if a rapid response is needed. Generally, taxanes and anthracyclines are the preferred agents, unless there was a recent exposure [13]. Other treatment options include capecitabine, eribulin, vinorelbine, gemcitabine, or platinum agents [13, 14]. Adding bevacizumab to capecitabine or a taxane is an option, having demonstrated a PFS benefit in the absence of an OS benefit [13].

The role of chemotherapy as “first non-endocrine based therapy” has been challenged by the results of the DESTINY-Breast06 trial investigating the anti-HER2 ADC trastuzumab deruxtecan (T-DXd) in HR+/HER2– metastatic breast cancer patients with progression on endocrine-based therapy, including CDK4/6i, but no prior use of chemotherapy for advanced disease. Additionally, tumors were required to demonstrate HER2-low staining assessed by immunohistochemistry (IHC) (IHC score 1+ or 2+ and negative in situ hybridization/ISH) or HER2-ultralow expression (IHC 0 with membrane staining; IHC > 0 and < 1+). Eighty-two percent of patients had HER2-low disease, 89% had prior CDK4/6i exposure, and 41% and 47% had prior anthracycline and taxane therapy, respectively, in the curative setting. T-DXd significantly improved PFS vs. physician’s choice of chemotherapy (capecitabine, nab-paclitaxel, or paclitaxel) in the HER2-low population [primary endpoint, median PFS 13.2 vs. 8.1 months; HR 0.62 (95% CI 0.52–0.75); P < 0.001]. ORR was 57% for T-DXd and 32% for chemotherapy in the HER2-low population. The prespecified exploratory analysis of patients with HER2-ultralow disease (18% of patients) showed a numerical trend in PFS in favor of T-DXd [median PFS 13.2 months for T-DXd and 8.3 months for chemotherapy; HR 0.78 (95% CI 0.50–1.21)]. Exploratory analysis of ORR in the HER2-ultralow population was 62% with T-DXd (vs. 26% with chemotherapy) [97]. The three most commonly observed drug-related AEs of any severity with T-DXd were nausea, fatigue, and alopecia; and the most common grade ≥ 3 events were neutropenia (20.7%), anemia (5.8%), and fatigue (3.7%). Drug-related interstitial lung disease (ILD) or pneumonitis occurred in 11.3% of patients (grade ≥ 3 in 1.4%, and grade 5 events in 0.7%); and left ventricular dysfunction/ejection fraction decrease was observed in 8.1% of patients (grade ≥ 3 in 0.7%, no grade 5 events and no reporting of cardiac failure) [97]. Given the pending OS data from this trial, and considering the OS benefit observed in HER2-low disease in DESTINY-Breast04—which evaluated T-DXd after one prior line of chemotherapy—clinicians may currently decide on a case-by-case basis between T-DXd and chemotherapy after CDK4/6i therapy and potentially other endocrine-based therapy lines [98]. In DESTINY-Breast04 and 06, patients with brain metastases were included, but comprised a minority of patients (5–8% of the HR+ population) [97, 98]. An exploratory subgroup analysis of patients with brain metastases (n = 24, 75% HR+/HER2– tumors) from the DESTINY-Breast04 trial (presented as an abstract), showed an intracranial ORR of 25% with T-DXd, but no intracranial response in patients treated with single-agent chemotherapy (n = 11) [99]. A single-arm phase II trial with 12 patients with more heavily pretreated, HER2-low disease (75% HR+) showed an ORR of 50% in asymptomatic untreated and 33% in patients with progressing brain metastases [100]. Overall, these results indicate that T-DXd demonstrates significant intracranial activity in patients with HR+/HER2-low ABC and brain metastases. Another HER2-directed ADC, BNT323/DB-1303, is under investigation against single-agent chemotherapy in HR+/HER2-low ABC after prior ET and possibly CDK4/6i treatment, but no chemotherapy exposure for advanced disease [101].

In contrast, based on data available in abstract form only, the TROP2-directed ADC sacituzumab govitecan (SG) did not demonstrate a benefit vs. single-agent chemotherapy when being evaluated in the phase III ASCENT-07 trial in patients with HR+/HER2– ABC after progression on ET and mostly prior CDK4/6i therapy, but no previous chemotherapy for advanced disease [102]. The TROP2-directed ADC datopotamab deruxtecan (Dato-DXd) has previously shown a significant PFS benefit, but no OS benefit against chemotherapy of physician’s choice in an HR+/HER2– advanced disease population with progression on 1–2 lines of chemotherapy in the advanced setting and optional prior CDK4/6i exposure (83% of patients) [103, 104]. The TROPION-Breast06 trial is a single-arm phase IIIb trial now evaluating Dato-DXd in advanced HR+/HER2– IHC 0 disease (IHC 0 defined as no staining or incomplete/faint membrane staining in ≤ 10% of tumor cells, reflecting HER2-null or HER2-ultralow status) in chemotherapy-naïve patients after progression on ET [105, 106].

The incorporation of immune checkpoint inhibitors (ICIs) in addition to chemotherapy-based regimens represents an additional strategy under clinical evaluation. The efficacy of ICI therapy in HR+/HER2– breast cancer has been rather disappointing, also reflected in the phase II PACE trial in the post-CDK4/6i setting, in which combining avelumab with palbociclib and fulvestrant failed to improve PFS compared with fulvestrant monotherapy [56, 107]. Adding pembrolizumab to single-agent chemotherapy is being evaluated in the phase III KEYNOTE-B49 trial in PD-L1-positive patients with prior endocrine and CDK4/6i therapy, but no prior chemotherapy for advanced disease [108]. Preliminary data from the phase II SACI-IO trial (in abstract form) compared SG with or without pembrolizumab after mandatory ET pretreatment and in mostly CDK4/6i- and partially chemotherapy-pretreated patients, and showed no significant PFS benefit at a preliminary analysis. Final PFS results are awaited [109]. Similarly, the TROP2–directed ADC sacituzumab tirumotecan is currently being evaluated alone and with pembrolizumab vs. chemotherapy in the phase III TroFuse-010 trial in chemotherapy-naïve patients after progression on endocrine and CDK4/6i therapy [110].

Clinical decision making

Determinants of biomarker performance

As outlined above, second-line systemic therapy strategies following progression on CDK4/6i have evolved as a biomarker-oriented field, in which therapeutic algorithms incorporate molecular and/or immunohistochemical biomarker testing and one-to-one matching of therapeutic options, which is a consequence of the design from current clinical trials and supports in its simplicity clinical decision making. However, clinical utility of a predictive biomarker might be determined by several, often interrelated considerations.