Background
Targeting CDK4/6 signaling in combination with endocrine therapies significantly improves progression-free survival (PFS) overall survival [
1] in patients with advanced estrogen receptor-positive /HER2-negative (ER+/HER2−) breast cancer [
2‐
4] and is now the standard of care for this disease. Three CDK4/6 inhibitors are now approved for the treatment of ER+/HER2− metastatic breast cancer: palbociclib (Ibrance®), ribociclib (Kisqali®), and abemaciclib (Verzenio®). However, despite the clinical advances associated with the addition of CDK4/6 inhibitors to endocrine therapies, acquired resistance to approved treatments for ER+/HER2− breast cancer remains a significant unmet clinical need, particularly in the metastatic setting.
Preclinical work in our laboratory first demonstrated that hormone receptor-positive breast cancer cell lines are differentially sensitive to the CDK4/6 inhibitor palbociclib when compared to other breast cancer subgroups [
5]. Clinical translation of these data ultimately led to the subsequent approval of palbociclib for the treatment of advanced breast cancer in combination with hormonal targeted therapy [
5,
6]. Multiple lines of preclinical evidence also support the interplay between ER and CDK4/6 signaling. Mitogenic action of estrogen in ER-dependent breast cancers is mediated via the induction of Cyclin D1 that can then bind to CDK4 and CDK6 resulting in the hyperphosphorylation of the retinoblastoma (Rb) tumor suppressor protein [
7]. This in turn leads to cell cycle progression from G1 to S phase and subsequent cell proliferation [
8].
Although ER positivity is a predictive biomarker of response to CDK4/6-based therapies, factors associated with the acquisition of resistance are still poorly understood. Preclinical and clinical studies have implicated a number of molecular alterations that either directly activate CDK4/6 signaling or activate escape signaling pathways in acquired CDK4/6 inhibitor resistance. These include but are not limited to Cyclin E amplification and/or Rb loss [
9‐
11], CDK6 amplification [
12], deregulated Hippo signaling [
13], amplification and aberrant activity of FGFR [
14], MAPK pathway activation [
15], compensatory PI3K-dependent activation of non-canonical Cyclin D1-CDK2 [
9], and alterations in PDK1 and PI3K/AKT signaling [
16]. The phosphatidylinositol 3-kinase/AKT/mammalian target of the rapamycin (PI3K/AKT/mTOR) pathway is a key signaling driver of cellular proliferation and survival of cancer cells. Dysregulation and activation of this pathway can drive tumorigenesis of ER+/HER2− breast cancers and is associated with resistance to anti-estrogen-targeted therapies [
17,
18]. Previous studies in our laboratory have demonstrated that inhibitors of the PI3K/AKT/mTOR pathway have selective activity in ER+ breast cancer cell lines carrying activating mutations in
PIK3CA [
19,
20]. In the SOLAR-1 Phase III clinical trial, patients with
PIK3CA-mutated ER+/HER2− breast cancer had almost a doubling in PFS in response to the p110α-selective PI3K inhibitor alpelisib (BYL719) (Piqray®) in combination with fulvestrant compared to patients treated with fulvestrant with placebo [
21]. Collectively, these data suggest a complex interplay between PI3K, ER, and CDK4/6 signaling that may drive ER+/HER2− breast cancers to respond and then progress through currently approved therapies.
In this study, we used in vitro and in vivo breast cancer models of acquired resistance to CDK4/6-based therapies to characterize molecular mechanisms associated with therapeutic resistance. We assessed the potential of pharmacologically targeting PI3K, ER, or CDK4/6 signaling to prevent or reverse acquired resistance. These data provide insight into the design of optimal therapeutic strategies to overcome therapeutic resistance in ER+/HER2− breast cancer.
Discussion
The three approved inhibitors of cyclin-dependent kinases 4 and 6, palbociclib, ribociclib, and abemaciclib, have shown significant activity in combination with endocrine therapy in ER+/HER2− breast cancer with the largest positive impact on the outcome data in the past 45 years [
2‐
4]. As such, they represent a new standard-of-care for this subtype of the disease. Although ER-positive status strongly predicts for response to these drugs, identifying pharmacological targets that may block the eventual onset of resistance to CDK4/6-based therapies is a critical next step. In the current study, we utilized multiple models of acquired resistance to CDK4/6-targeted therapies and confirm that cross-resistance exists among this class of compounds; however, the PI3K/mTOR pathway signaling remains activated in resistance. We show that targeting PI3K alone or in combination with inhibitors of CDK4/6 and endocrine therapies can overcome acquired resistance and potentially prevent the emergence of resistance to CDK4/6-ER-based therapies for ER+/HER2− breast cancer.
Despite the proven clinical benefit of adding CDK4/6 inhibitors to endocrine therapy for ER+/HER2− breast cancer, ultimate progression on therapy remains a significant factor impacting long-term benefits of CDK4/6-based therapies. Results from the pivotal PALOMA-2 trial, show that although the addition of palbociclib to letrozole significantly improved PFS from 14.5 to 24.8 months, progressive disease still occurs in the majority of palbociclib-treated patients [
25]. The molecular mechanisms by which therapeutic resistance to CDK4/6 inhibition acquired are not well understood at present but are likely mediated by several different processes including loss of Rb protein or function; alterations in CDK4/6, Cyclin D, Cyclin E, and/or CDK repressor protein expression; activation of CDK2; or activation of PI3K signaling [
11,
26,
27].
Many of the proposed resistance mechanisms such as Rb loss or CDK2 activation are not easily amenable to targeted therapeutic interventions; however, the PI3K pathway activation in breast cancer has been successfully targeted with small molecule inhibitors. Targeting downstream mTOR signaling via an mTORC1 inhibitor, everolimus, in combination with the aromatase inhibitor exemestane compared to exemestane plus placebo, improved the median progression-free survival (PFS) in ER+/HER2− advanced breast cancer from 3.2 to 7.8 months (hazard ratio 0.38 [95% CI 0.31–0.48],
P < 0.0001) [
28]. Targeting the PI3K pathway upstream at AKT has had limited efficacy in breast cancers as shown with the selective AKT1/2 inhibitor, ipatasertib [
29]. The majority of the clinical investigation associated with targeting the PI3K/AKT/mTOR pathway, upstream of mTORC1, has focused on directly targeting the catalytic subunit of the PI3K enzyme that is frequently activated through somatic mutation of the
PIK3CA gene [
30]. Initial attempts to target the PI3K enzyme complex in ER+/HER2− breast cancer using “pan-PI3K inhibitors” that inhibit all four isoforms (α, β, γ, and δ) of the class IA PI3Ks were limited by toxicity [
31,
32]. Conversely, a significantly greater promise has been observed in ER+/HER2− breast cancers using an α-selective molecule. Combination of the p110α-selective PI3K inhibitor BYL719 (alpelisib; Novartis) with fulvestrant in patients with ER+/HER2− metastatic breast cancer carrying activating
PIK3CA mutations that had progressed on a previous endocrine therapy induced an improvement in PFS of 11.0 months compared to 5.7 months in the placebo plus fulvestrant group (hazard ratio 0.65, 95% CI 0.50 to 1.25,
P = 0.00065) [
21]. These data support the hypothesis that
PIK3CA mutations can drive the progression of a subtype of ER+/HER2− breast cancers, including patients whose disease progressed on hormone therapy. Given that the standard-of-care for many ER+/HER2− breast cancers is now hormone therapy in combination with a CDK4/6 inhibitor, there is a strong rationale to suggest inhibitors of PI3K/mTOR signaling may also be active in this arena.
In the present study, we generated in vitro and in vivo models of ER+/HER2− breast cancer with acquired resistance to both CDK4/6 inhibitor monotherapy and combination therapy with hormone blockade. Acquired resistance to inhibitors of CDK4/6 is characterized by a loss of dependence on ER-CDK4/6-Rb signaling, as demonstrated by cross-resistance to other CDK4/6 inhibitors in this class in both in vitro and in vivo assays. Proteomic profiling of the acquired resistant cell lines indicates that loss of dependence on ER-CDK4/6-Rb signaling in our models can occur either through direct loss of total and phosphorylated Rb protein or a loss of dependence on Rb by a loss of a negative regulator protein such as p27, which can lead to activated Cyclin E1-CDK2 signaling [
33,
34]. Other preclinical studies have shown that both Rb protein and its transcript are lost in acquired resistance to palbociclib [
9,
35]. Collectively, these data indicate that switching patients with disease progression on palbociclib therapy to another CDK4/6 inhibitor is unlikely to provide benefit. These findings are particularly important given that there are currently three clinically approved CDK4/6 inhibitors; loss of dependence on ER-CDK4/6-Rb signaling would confer cross-resistance of each of these compounds. However, significantly upregulated PI3K-AKT-mTOR signaling was found to be common in both models with acquired resistance to palbociclib that were assessed by RPPA analysis, indicating a potential role as an escape signaling pathway for PI3K/AKT/mTOR signaling in acquired resistance to CDK4/6 inhibitors. These data coupled with the recent clinical data emerging around p110α-selective molecules in ER+/HER2− breast cancer indicate that exploring PI3K inhibition as a means to overcome and/or prevent resistance to CDK4/6-based therapies could be a very attractive approach that might be fast-tracked towards clinical translation.
The data presented in this study show that pharmacologically targeting PI3K/mTOR signaling inhibits the growth of cell lines and xenografts conditioned to progress on CDK4/6-based therapies. ER+/HER2− breast cancer xenografts, progressing on palbociclib/fulvestrant, were unresponsive to either an immediate switch to an alternative CDK4/6 inhibitor or a switch following a dosing holiday. The rate of tumor progression in each of these acquired resistance models could only be impacted by the inclusion of a PI3K/mTOR pathway inhibitor in the treatment schedule. The addition of alpelisib to ribociclib/fulvestrant induced significant xenograft regressions in two separate models tested. Similar efficacies were also observed with the alpelisib/fulvestrant doublet combination in the absence of continued CDK4/6 inhibitor treatment, underscoring the importance of the PI3K inhibition. Moreover, monotherapy with either alpelisib or everolimus completely blocked xenograft growth in both PIK3CA mutant and wild-type models. These data provide further evidence that continued treatment with inhibitors of CDK4/6 is unlikely to provide benefit either a single-agent or as a combination partner post-progression on CDK4/6-based therapies.
Targeting PI3K in CDK4/6-resistant breast cancer cells appears to be effective in both
PIK3CA mutant and wild-type models, whereas targeting mTORC1 is effective in
PIK3CA wild-type xenograft and one of two
PIK3CA-mutant models. Additional preclinical and clinical studies are required to investigate this observation. It is likely that targeting PI3K/mTOR signaling at different nodes in the pathway may have different affects on tumor progression depending on the driving alteration. Although our data show that targeting PI3K is likely to benefit patients with
PIK3CA-mutant ER+/HER2− breast cancer that has progressed on CDK4/6-targeted therapy, targeting mTORC1 may provide another therapeutic option for patients with
PIK3CA wild-type breast cancer. Previous studies have shown that activated PI3K/mTOR signaling may be playing a role in resistance to CDK4/6-based therapies in ER+ breast cancer [
9,
36]. Although it has been reported that targeting PI3K may be ineffective once acquired resistance to CDK4/6 inhibitors occurs [
9], we found that PI3K inhibition can reverse resistance in multiple models of acquired CDK4/6 inhibitor resistance. It is possible that differences observed in the present study may be due to the use of the p110α-selective inhibitor alpelisib, over a pan-PI3K inhibitor. Additional evidence for a strong interplay between the CDK4/6 and PI3K pathways is provided by our observation that ER+/HER2− xenografts progressing on an alpelisib/fulvestrant combination could be effectively reversed by the addition of the CDK4/6 inhibitor, ribociclib. These data are consistent with previous reports of in vitro studies that show CDK4/6 inhibitors can reverse acquired resistance to inhibitors of PI3K [
36]. Data from two clinical studies with small numbers of patients indicate that palbociclib-based therapies have limited efficacy after progression on the mTORC1 inhibitor everolimus (RAD001) [
37,
38]. However, the efficacy of CDK4/6 inhibition post-progression on PI3K inhibitor-based therapies is yet to be investigated.
Although our data suggest that the benefit of continued use of selective inhibitors of CDK4/6 in the acquired resistant setting will be of little or no clinical benefit, combined targeting of CDK4/6 and PI3K/mTOR signaling with hormonal blockade may provide benefit to treatment-naive ER+/HER2− breast cancers. Here, we report that upfront triple combination therapy can prevent/delay the onset of resistance in xenograft models. Sustained tumor regressions were observed for over 9 weeks post-drug withdrawal in xenografts treated with CDK4/6:PI3K:ER combination therapy. Similar responses were observed in a PDX model of ER+/HER2− breast cancer when fulvestrant was replaced by the aromatase inhibitor, letrozole. These data indicate that targeting CDK4/6 and PI3K may be effective in combination with multiple classes of endocrine-based therapies. Moreover, we have shown that triple combination therapy is effective in both “PI3K-activated” (
PIK3CA-mutated or PTEN-null) and “PI3K-normal” (
PIK3CA/PTEN wild-type) ER+/HER2− breast cancer xenografts. The data presented here indicate that the mechanisms by which the triple combination blocks xenograft regrowth are through enhanced inhibition of both PI3K/AKT/mTOR and CDK4/6-Rb/ERα signaling, as opposed to the combination hitting a novel target/signaling pathway. It is possible that this simultaneous inhibition on each pathway leads to the complete arrest of cell cycle progression and ultimate induction of apoptosis. Although long-term treatment with this triple combination was well tolerated in mice, careful dose management of the combination strategy will be required in human studies. Encouraging data have been reported from phase I/II clinical studies investigating triplet CDK4/6:PI3K-mTOR:ER combination strategies. Clinical activity and an acceptable safety profile were observed in response to triple combination treatment with ribociclib (3 weeks on, 1 week off) plus alpelisib (continuous) and letrozole (continuous) in heavily pretreated ER+/HER2− breast cancer patients [
39]. Continuous treatment of triple combination therapy with ribociclib/everolimus/exemestane has also been shown to be well tolerated and demonstrate clear clinical benefit in patients with advanced ER+/HER2− breast cancer [
40].
Competing interests
Neil A. O’Brien: Novartis—research funding.
Martina S.J. McDermott: none to declare
Dylan Conklin: none to declare
Tong Luo: none to declare
Raul Ayala: none to declare
Suruchi Salgar: none to declare
Kevin Chau: none to declare
Emmanuelle DiTomaso: former Novartis employee
Naveen Babbar: full-time Novartis employee
Faye Su: full-time Novartis employee
Alex Gaither: former Novartis employee
Sara A. Hurvitz: clinical research funding paid to UCLA from Eli Lilly, Novartis, and Pfizer
Ronald Linnartz: full-time Novartis employee
Kristine Rose: full-time Novartis employee
Samit Hirawat: former Novartis employee
Dennis J. Slamon: Pfizer—stock ownership, research funding, and travel expenses. Novartis—honoraria, consulting advisory role, research funding, and travel expenses. Eli Lilly—consulting advisory role
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.