Oral Estrogen Receptor PROTAC Vepdegestrant (ARV-471) Is Highly Efficacious as Monotherapy and in Combination with CDK4/6 or PI3K/mTOR Pathway Inhibitors in Preclinical ER+ Breast Cancer Models

Abstract

Purpose: Estrogen receptor (ER) alpha signaling is a known driver of ER-positive (ER+)/human epidermal growth factor receptor 2 negative (HER2-) breast cancer. Combining endocrine therapy (ET) such as fulvestrant with CDK4/6, mTOR, or PI3K inhibitors has become a central strategy in the treatment of ER+ advanced breast cancer. However, suboptimal ER inhibition and resistance resulting from the ESR1 mutation dictates that new therapies are needed.

Experimental design: A medicinal chemistry campaign identified vepdegestrant (ARV-471), a selective, orally bioavailable, and potent small molecule PROteolysis-TArgeting Chimera (PROTAC) degrader of ER. We used biochemical and intracellular target engagement assays to demonstrate the mechanism of action of vepdegestrant, and ESR1 wild-type (WT) and mutant ER+ preclinical breast cancer models to demonstrate ER degradation-mediated tumor growth inhibition (TGI).

Results: Vepdegestrant induced ≥90% degradation of wild-type and mutant ER, inhibited ER-dependent breast cancer cell line proliferation in vitro, and achieved substantial TGI (87%-123%) in MCF7 orthotopic xenograft models, better than those of the ET agent fulvestrant (31%-80% TGI). In the hormone independent (HI) mutant ER Y537S patient-derived xenograft (PDX) breast cancer model ST941/HI, vepdegestrant achieved tumor regression and was similarly efficacious in the ST941/HI/PBR palbociclib-resistant model (102% TGI). Vepdegestrant-induced robust tumor regressions in combination with each of the CDK4/6 inhibitors palbociclib, abemaciclib, and ribociclib; the mTOR inhibitor everolimus; and the PI3K inhibitors alpelisib and inavolisib.

Conclusions: Vepdegestrant achieved greater ER degradation in vivo compared with fulvestrant, which correlated with improved TGI, suggesting vepdegestrant could be a more effective backbone ET for patients with ER+/HER2- breast cancer.

Figure 1.

Vepdegestrant engages the ER transcription factor and degrades wild-type (WT) ER and ET-resistant ER mutants. A, Chemical structure of vepdegestrant and schematic representation of the ER:CRBN interaction induced by vepdegestrant. B, T47D-KBluc cellular luciferase reporter assay demonstrating a vepdegestrant dose-dependent decrease of estradiol-bound ER/ERE-driven luciferase expression after 24-hour treatment (IC₅₀ 1.1 nmol/L). C, Vepdegestrant dose-dependent in vitro degradation of ER^WT after 72-hour treatment of MCF7, BT474, CAMA-1, ZR-75-1, and T47D breast cancer cell lines. D, Quantified ER levels in MCF7 cells following 72-hour treatment with vepdegestrant by in-cell western immunofluorescence. The average of eight independent experiments showed a vepdegestrant DC₅₀ of 0.9 nmol/L and D_max of 95%. ER levels were normalized to α-tubulin and expressed as a percentage of DMSO controls. E, T47D breast cancer cells with CRISPR knock-in ESR1 mutants stably expressing ER^Y537S or ER^D538G variants, incubated with increasing concentrations of vepdegestrant for 72 hours. F, Western blot levels of ER^WT and ER mutants expressed in stably transfected embryonic kidney T-REx-293 cells treated for 4 hours with vepdegestrant. β-Actin is shown as a loading control. D_max, maximum degradation.

Figure 2.

Vepdegestrant demonstrates PROTAC mechanism of action and induces ER:vepdegestrant:CRBN trimer formation. A, AlphaLISA CRBN-DDB1:probe displacement assay demonstrates a 17-fold left shift in the IC₅₀ of vepdegestrant in the presence of ER-LBD. B, Vepdegestrant and its E3-inactive analog, A5927, demonstrate dose-dependent antagonism of estradiol-bound ER/ERE-driven luciferase expression in T47D-KBluc cells with IC₅₀ values of 3 and 33 nmol/L, respectively, at 4 hours. Results are representative of three independent experiments. C, Vepdegestrant’s mode of action is dependent on the recruitment of ER and CRBN and active proteasomes. The E3-inactive analog of vepdegestrant A5927 fails to degrade ER (lane 5 vs. lane 2) in MCF7 cells. Vepdegestrant in the presence of the proteasome inhibitor carfilzomib (lane 3), or 10-fold excess of competing ER or CRBN ligands, lasofoxifene (lane 7) or pomalidomide (lane 9), respectively, fails to degrade ER. D, Top, size-exclusion chromatography demonstrates ER:vepdegestrant:CRBN ternary complex formation; bottom, the ER-LBD co-migrates in the left-hand fractions with CRBN-DDB1 as a higher-order complex in the presence of vepdegestrant, as shown by SDS-PAGE. E, Cryo-EM structure of the CRBN-DDB1:vepdegestrant:ER ternary complex in transparent surface representation. DDB1 and CRBN shown in ribbon representation. Image prepared by ChimeraX.

Figure 3.

Vepdegestrant cellular ER target engagement. A, TMT proteomic profile of MCF7 cells following a 7-hour incubation with 100 nmol/L vepdegestrant. The ER (encoded by ESR1) protein is the most quantitatively and statistically significantly decreased protein. Close to threshold cutoffs, although still significant, is the PGR. B, RNA transcript levels of ER-regulated genes PGR and GREB1 in MCF7 cells following vepdegestrant treatment for 24 hours. C, Western blot of IMiD neosubstrates GSPT1, IKZF3, IKZF1, and CK1α in Ramos cells after 24-hour treatment with vepdegestrant and positive control IMiDs lenalidomide and CC-885. Mito.c served as the loading control. D, Western blot of SALL4 protein levels in neuroblastoma SK-N-DZ cells treated with vepdegestrant or lenalidomide (positive control) for 16 hours. GAPDH served as a loading control. Representative of two independent experiments. TMT, tandem mass tag.

Figure 4.

Vepdegestrant inhibits in vitro proliferation of WT ER and ET-resistant mutant ER^Y537S– and ER^D538G–dependent breast cancer cells. Vepdegestrant DRC of MCF7 (A), T47D (B), T47D ER^Y537S (C), and T47D ER^D538G (D) breast cancer cells in a 5-day proliferation assay. E, Five-day proliferation of MCF7 cells with a 4-fold serial dilution of vepdegestrant or the non-degrading E3-inactive analog of vepdegestrant, A5927. Statistically significant differences are shown by t-test; *, P < 0.05. Results are representative of three independent experiments.

Figure 5.

Vepdegestrant lacks agonist activity, potently degrades ER in vivo, and induces tumor regressions in WT ER CDX, and ER^Y537S ST941/HI and palbociclib-resistant ER^Y537S ST941/HI/PBR PDX models. A, MCF7 tumor-bearing NOD SCID mice were orally administered vehicle or 10 mg/kg vepdegestrant qdx3. Top, Western blot of ER protein levels in MCF7 tumor lysates and β-actin loading control, 18 hours post-last dose. Bottom, plot of ER levels normalized to β-actin. Vepdegestrant reduced ER levels by ≥90% compared with vehicle controls. B, Immature female rats were administered PEG/Tween or EBB/Castor oil vehicles, 10 mg/kg AZD9496 po qdx3 as a positive control of ER agonist activity, 30 mg/kg vepdegestrant po qdx3, or 100 mg/kg fulvestrant sc qdx1. On day 4, tissues were harvested and uterine weights recorded (left). ER protein levels in uterine tissue lysates determined by western blot (right). Vepdegestrant and fulvestrant decreased ER levels by 76% and 65%, respectively, compared with their respective vehicles. C, Orthotopic MCF7 CDX model dosed orally with 3, 10, or 30 mg/kg vepdegestrant daily for 28 days. Right, ER levels at 16 hours post-last dose in tumor lysates, determined by western blot (not shown) and densitometry analysis. D,In vivo efficacy of vepdegestrant in the START ST941/HI ER^Y537S mutant PDX model. Mice were dosed with vepdegestrant at 10 or 30 mg/kg po qdx27, or 200 mg/kg fulvestrant sc biwx2, qwx2. E,In vivo efficacy of vepdegestrant in the palbociclib-resistant ST941/HI/PBR ER^Y537S mutant PDX model. Mice were dosed qdx28 with vepdegestrant 10 mg/kg po, abemaciclib 50 mg/kg po, or palbociclib 60 mg/kg po. Body weights were well maintained (Supplementary Fig. S4). Statistically significant differences compared with vehicle controls are indicated (one-way ANOVA). *, P < 0.05; **, P < 0.01; ****, P < 0.0001. Means are plotted and error bars represent ± SEM. CDX, cell line–derived xenograft; PDX, patient-derived xenograft.

Figure 6.

Vepdegestrant synergizes with targeted small molecule inhibitors of CDK4/6, mTOR, and PI3K in ER-dependent breast cancer models. A, Relative cell growth kinetics over 120 hours by live-cell imaging of MCF7 dosed with vepdegestrant and/or palbociclib at their respective approximate GI₅₀ concentrations. Representative of three independent experiments. B, Differences in relative MCF7 cell growth compared with vehicle control following 120 hours of treatment as indicated. Graphs show the mean of three independent experiments. Means are plotted and error bars represent ± SEM. **, P = 0.005; ***, P < 0.0007; ****, P < 0.0001 (one-way ANOVA test). C, Cell viability and synergistic analysis of MCF7 cells at day 5 dosed with vepdegestrant in combination with palbociclib in an 8 × 8 block matrix. Top, Vepdegestrant dose response shifts with the addition of palbociclib. Bottom, Drug synergy representation from Bliss model analysis run through Combenefit synergy analysis software (representative of three independent experiments). D–F,In vivo drug combination efficacy studies using the MCF7 orthotopic xenograft model. Mean tumor volumes are reported ± SEM. Indicated drugs were dosed as single agents or in combination to 10 mice/arm. Black arrows indicate single-day dosing holidays. D, Mice dosed with 30 mg/kg vepdegestrant or 60 mg/kg palbociclib (IBRANCE; Pfizer) po qdx28, or 200 mg/kg fulvestrant (Faslodex; AstraZeneca) sc biwx2 then qwx2. Body weights were well maintained with two single-day dosing holidays (Supplementary Fig. S11A). E, Mice dosed with 30 mg/kg vepdegestrant or 2.5 mg/kg everolimus (Afinitor; Novartis) po qdx26. Body weights were well maintained with a single-day dosing holiday (Supplementary Fig. S14A). F, Mice dosed with 30 mg/kg vepdegestrant or 25 mg/kg alpelisib (Piqray; Novartis) po qdx5on, 2off (×4), as single agents or in combination. Body weights were well maintained (Supplementary Fig. S15A). Statistically significant differences compared with vehicle controls are indicated (one-way ANOVA), ***, P < 0.001; ****, P < 0.0001.