An Open-label Phase I Study of GDC-0927 in Postmenopausal Women with Locally Advanced or Metastatic Estrogen Receptor–Positive Breast Cancer

Abstract

Purpose:

GDC-0927 is a novel, potent, nonsteroidal, orally bioavailable, selective estrogen receptor (ER) degrader that induces tumor regression in ER⁺ breast cancer xenograft models.

Patients and Methods:

This phase I dose-escalation multicenter study enrolled postmenopausal women with ER⁺/HER2⁻ metastatic breast cancer to determine the safety, pharmacokinetics, and recommended phase II dose of GDC-0927. Pharmacodynamics was assessed with [¹⁸F]-fluoroestradiol (FES) PET scans.

Results:

Forty-two patients received GDC-0927 once daily. The MTD was not reached. The most common adverse events (AE) regardless of causality were nausea, constipation, diarrhea, arthralgia, fatigue, hot flush, back pain, and vomiting. There were no deaths, grade 4/5 AEs, or treatment-related serious AEs. Two patients experienced grade 2 AEs of special interest of deep vein thrombosis and jugular vein thrombosis, both considered unrelated to GDC-0927. Following dosing, approximately 1.6-fold accumulation was observed, consistent with the observed half-life and dosing frequency. There were no complete or partial responses. Pharmacodynamics was supported by >90% reduction in FES uptake and an approximately 40% reduction in ER expression, suggesting ER degradation is not the mechanistic driver of ER antagonism. Twelve patients (29%) achieved clinical benefit; 17 patients (41%) showed a confirmed best overall response of stable disease. Baseline levels of ER and progesterone receptor protein and mutant ESR1 circulating tumor DNA did not correlate with clinical benefit.

Conclusions:

GDC-0927 appeared to be well tolerated with pharmacokinetics supporting once-daily dosing. There was evidence of target engagement and preliminary evidence of antitumor activity in heavily pretreated patients with advanced/metastatic ER⁺/HER2⁻ breast cancer with and without ESR1 mutations.

Translational Relevance.

Breast cancer is commonly driven by activation of estrogen receptor (ER) signaling triggered by its ligand, estrogen. Mutations in the ligand-binding domain of the ESR1 gene confer ligand independence while retaining dependence on the ER pathway. GDC-0927 is an oral selective ER antagonist and degrader (SERD), a therapeutic agent designed to inhibit ligand-dependent as well as ligand-independent ER-mediated signaling. In this phase I study, GDC-0927 has shown predictable pharmacokinetics and a tolerable safety profile with evidence of robust target engagement. The study benchmarks clinical data vital for the continued development of the SERD landscape, including [¹⁸F]-fluoroestradiol (FES)–PET utility in the characterization of ER-expressing tumors and the activity of a selective and potent SERD in patients with ESR1 mutations.

Introduction

Patients with breast cancer predominantly present with tumors that depend on the estrogen receptor (ER) for tumor growth and progression. Such tumors are positive for estrogen receptor alpha (ER⁺) encoded by the ESR1 gene, and negative for HER2⁻ encoded by the ERBB2 gene (1). Endocrine therapy, the preferred first-line treatment for advanced or unresectable ER⁺ breast cancer (2), yields superior efficacy and toxicity profile in comparison with chemotherapy (3, 4), and exploits biochemical mechanisms that include selective ER modulation (5), aromatase inhibition (AI; ref. 6), and selective estrogen receptor degradation (SERD; ref. 7). Patients initially experience therapeutic benefit, but eventually develop resistance to endocrine treatments due to any of several mechanisms (8), including acquisition of ESR1 mutations (9). Studies have demonstrated that patients with tumors expressing ESR1 mutations have improved progression-free survival (PFS) after treatment with fulvestrant, a SERD therapeutic, versus treatment with exemestane, an AI therapeutic, whereas patients with tumors expressing wild-type ESR1 have similar PFS after treatment with either therapeutic (10, 11). The success with fulvestrant in patients who acquire ESR1 mutations (10, 11) as well as the limitations of this therapeutic with respect to route of delivery and suboptimal exposure (12, 13) highlight an unmet need for novel SERD therapeutics with improved oral bioavailability for treatment of tumors harboring ESR1 mutations (14).

GDC-0927 is an orally bioavailable small-molecule SERD that immobilizes ER, fully antagonizes its response to estrogen, and induces its proteasomal degradation (15, 16). GDC-0927 demonstrated dose-dependent antitumor activity in ESR1 mutant and wild-type patient-derived xenograft models of ER⁺ breast cancer; the in vivo efficacious dose range was 10–100 mg/kg/day, which was well tolerated. This first-in-human study investigated GDC-0927 in postmenopausal women with ER⁺/HER2⁻ locally advanced or metastatic breast cancer (MBC).

Patients and Methods

Study design

This was an open-label, multicenter phase I clinical trial. The primary objective was to investigate the MTD and/or recommended phase II dose (RP2D) in patients with advanced or metastatic breast cancer. Secondary objectives included evaluation of the safety, antitumor activity at the RP2D, and pharmacokinetics of single and multiple doses of GDC-0927. The protocol was approved by Institutional Review Boards at participating institutions and written informed consent was obtained from patients prior to performing any procedures. The study was conducted in conformance with the ICH E6 guideline for Good Clinical Practice and the principles of the Declaration of Helsinki or the laws and regulations of the country in which the research was conducted, whichever afforded greater protection to the individual.

Patients

Postmenopausal women with histologic or cytologic diagnosis of locally assessed ER⁺/HER2⁻ adenocarcinoma of the breast, with evidence of metastatic disease or locally recurrent disease not amenable to resection or radiotherapy, with progressive disease after greater than or equal to 6 months of treatment with hormonal therapy, were eligible to enroll. An ER⁺ tumor was defined as staining in ≥1% cells by IHC. HER2⁻ breast cancer was defined as an IHC result of 0 or 1+ for cellular membrane protein expression, or a FISH result showing HER2/CEP17 ratio of <1.8, or an average of <4 copies of HER2 gene per nucleus for systems without an internal central probe. Inclusion criteria were ≥18 years of age with an Eastern Cooperative Oncology Group (ECOG) performance status ≤2 and adequate organ function [absolute neutrophil count ≥1,500/μL; platelets ≥100,000/μL; aspartate transaminase (AST) and alanine aminotransferase (ALT) ≤3 × upper limit of normal (ULN) or ≤5 × ULN if liver function abnormalities were due to underlying malignancy; total serum bilirubin ≤1.5 × ULN regardless of liver involvement secondary to tumor; serum creatinine ≤1.5 × ULN]. Other inclusion criteria were ≤2 prior chemotherapies in the advanced or metastatic setting, ≥2 months since last use of tamoxifen, ≥6 months since last use of fulvestrant, ≥2 weeks since last use of any other endocrine therapy, and ≥3 weeks since last use of any chemotherapy. The inclusion of patients with increased serum indirect bilirubin (≤3 × ULN) due to Gilbert syndrome was permitted. In the dose-expansion cohorts, eligible patients had measurable disease or evaluable disease.

Prior treatment with another oral SERD was not allowed. Other exclusion criteria were: untreated brain metastasis; current treatment with any systemic anticancer therapies or experimental treatments; current use of warfarin, phenytoin, or proton pump inhibitors; diagnosis of any secondary malignancy within prior 2 years except for adequately treated basal cell or squamous cell skin cancer, or carcinoma in situ; active inflammatory bowel disease or chronic diarrhea, short bowel syndrome, or upper gastrointestinal (GI) surgery; human immunodeficiency virus infection; clinically significant history of liver disease; major surgery within 4 weeks prior to enrollment; radiotherapy within 2 weeks prior to enrollment; any of the following occurrences within 12 months of enrollment—myocardial infarction, severe/unstable angina, ongoing cardiac dysrhythmias of grade ≥2, atrial fibrillation of any grade, coronary/peripheral artery bypass graft, symptomatic congestive heart failure, or cerebrovascular accident including transient ischemic attack; and any other condition deemed inappropriate for study.

Study treatments

During dose escalation, GDC-0927 was administered orally as a single dose on day −7 for pharmacokinetic evaluation during the lead-in period. Starting on day 1, patients received their treatment once daily. Dose-limiting toxicities (DLT) were evaluated from day −7 through the first cycle (28 days) of treatment, for a total of 35 days. Patients were assigned to escalating doses of GDC-0927 using a standard 3+3 design. The starting dose was 600 mg/day followed by dose escalation in 400 mg increments. An expansion cohort was enrolled at the maximum administered dose (MAD) to further characterize the safety, pharmacokinetics, and antitumor activity of GDC-0927.

Dose reductions were not allowed during the DLT assessment window, but were allowed once the dose cohort was closed to further enrollment. Dose delays were allowed during the DLT assessment window and for <28 days in consecutive cycles.

Study assessments

Toxicity was graded according to the NCI Common Toxicity Criteria for Adverse Events (Version 4.0). A DLT was defined as any of the following adverse events (AE) deemed by the investigator to be related to the study drug: Any grade ≥3 non-hematologic toxicity (excluding alopecia of any grade and grade 3 or 4 nausea, vomiting, and diarrhea that reversed to grade ≤2 within 24 hours with maximal medical therapy); any grade ≥3 hematologic toxicity lasting >7 days; any grade toxicity leading to study drug interruption lasting >7 days; any case of potential drug-induced liver injury (elevated ALT or AST in combination with either elevated bilirubin or clinical jaundice, as defined by Hy's law); treatment-emergent ALT or AST (>3 × baseline value) in combination with total bilirubin (>2 × ULN of which ≥35% is direct bilirubin); treatment-emergent ALT or AST >3 × baseline value in combination with clinical jaundice.

Safety was assessed through summaries of AEs, changes in laboratory test results, changes in vital signs, and GDC-0927 exposure; all patients who received study treatment on day 1 of cycle 1 were included in the analyses. A full plasma pharmacokinetic profile was obtained for GDC-0927 and analyzed using noncompartmental methods for all patients in the study. Determination of steady-state trough concentrations (before dosing) of GDC-0927 was performed for all patients in dose-escalation and dose-expansion cohorts up to cycle 3 day 1. Objective response and clinical benefit rates were derived according to RECIST v1.1. [¹⁸F]-fluoroestradiol (FES)–PET imaging was performed at baseline and at cycle 2 day 3 (≥18 hours after dosing) in dose escalation to quantify ER expression in tumors and to assess for pharmacodynamic response. The mean percent change from baseline in SUV_Max (corrected for background activity) for up to five index lesions was calculated for each patient as described elsewhere (17). Disease in the liver and some areas of the GI tract was considered unevaluable in the FES analysis because of confounding normal physiologic uptake of the tracer.

Protein levels for ER, progesterone receptor (PR), and Ki67 were measured by IHC in baseline and available on-treatment (cycle 2 or 3, day 8) biopsies. All IHC cases were stained by a single laboratory and scored by a single pathologist to avoid interlaboratory and interobserver variability. Ki67 was scored as percentage of positively stained tumor cells. ER and PR protein levels are represented by Histoscores (H-scores) calculated with a range of 0–300 using the formula: (3 × percentage of positively stained tumor cell nuclei at strong intensity) + (2 × percentage of positively stained tumor cell nuclei at intermediate intensity) + (1 × percentage of positively stained tumor cell nuclei at weak intensity). Gene expression analysis in available matched baseline and on-treatment biopsies was performed using the RNA Access Library Preparation Kit (Illumina) followed by paired-end (2×50b, 50M reads) sequencing on the HiSeq (Illumina).

Circulating tumor DNA (ctDNA) was analyzed to gain insights into potential causal relationships between the clinical activity of GDC-0927 and resistance mechanisms to prior endocrine therapy. Hotspot mutations in ESR1, PIK3CA, and AKT1 were analyzed in baseline, on-treatment, and end-of-treatment plasma-derived ctDNA samples using the BEAMing digital PCR assay (Sysmex Inostics, Inc.). ESR1 mutations were defined as nucleotide substitutions that resulted in the following amino acid changes: E380Q, S463P, V534E, P535H, L536H/P/Q/R, Y537C/N/S, and D538G. Similarly defined, AKT1 mutations were limited to E17K, and PIK3CA mutations included C420R, E542K, E545G/K, Q546K, M1043I, and H1047L/R/Y. A subset of 1,400 mg dose plasma samples were analyzed with Foundation Assay for Circulating Tumor DNA (FACT), a next-generation sequencing ctDNA assay (Foundation Medicine) which covers genomic alterations in full exonic coding regions of 62 commonly altered genes.

Samples for pharmacokinetic analyses following a single dose of GDC-0927 were collected on lead-in days −7, −6, −5, −4, −3, and on day 1 of cycle 1. Steady-state samples were collected on day 1 of cycle 2. Single-dose and steady-state pharmacokinetic parameters for GDC-0927 were estimated on the basis of concentration data and nominal time. Plasma concentrations were used to perform a noncompartmental pharmacokinetics analysis with the use of Phoenix WinNonlin (version 8.2, Certera).

Antitumor activity was defined as objective response and clinical benefit rates derived according to RECIST v1.1 and summarized by dose level and cohort. Separately, the maximal response in patients with measurable disease at any time on study was reported using waterfall plots. Clinical benefit was defined as either (i) patients with confirmed responses [complete response (CR) or partial response (PR)], or (ii) patients with 24-week or longer outcomes of unconfirmed responses (uCR, uPR), stable disease (SD), or non-CR/non-progressive disease (non-CR/PD).

Statistical analysis

Because of the exploratory nature of this study, no confirmatory inferential analyses were planned, and no imputation for missing data was done. Descriptive statistics (such as means, medians, SDs, and ranges for continuous data and percentages for categorical data) were used to summarize patient characteristics, treatment administration/compliance, safety parameters, pharmacokinetic parameters, antitumor activity endpoints, and exploratory biomarkers. A total of 60 patients were planned for enrollment at the MTD and/or RP2D to provide a more robust safety profile and a sample size for a preliminary assessment of antitumor activity.

Data availability

To request access to data that support the findings of this study, see here (https://vivli.org/members/enquiries-about-studies-not-listed-on-the-vivli-platform/), and for up-to-date details on Roche's Global Policy on the Sharing of Clinical Information, see here (https://www.roche.com/innovation/process/clinical-trials/data-sharing/).

Results

Study population

Between March 2015 and January 2020, a total of 42 patients received GDC-0927 in dose-escalation (600 mg, n = 3; 1,000 mg, n = 3; 1,400 mg, n = 6) and dose-expansion cohorts (1,400 mg, n = 30) at six sites in the United States and eight sites in Spain (Supplementary Fig. S1). Patients in dose expansion had advanced or MBC that had progressed following previous treatments with no more than two endocrine therapies. AIs were part of prior therapeutics for all patients (n = 42, 100%). Patients had a median of 4 (range, 1–8) prior cancer therapies across all settings (neoadjuvant, adjuvant, and metastatic), including chemotherapy (n = 35, 83%), tamoxifen (n = 28, 67%), CDK4/6 inhibitors (n = 19, 45%), and fulvestrant (n = 18, 43%). Twenty-nine patients (69%) had measurable disease. Bone-only disease was present in 6 patients (14%), and 32 patients (76%) had visceral disease. The majority of patients were identified as White (n = 33, 79%) with a median age of 56 years (range, 38–78). Other demographics and baseline characteristics, including ESR1, PIK3CA, and AKT1 mutation status, are provided in Table 1. Representativeness of study patients is shown in Supplementary Table S1.

Table 1.

Baseline demographics and patient characteristics.

	All patients (N = 42)
Age, years
Median	56
Range	38–78
Race
Asian	1 (2%)
Black or African American	4 (10%)
White	33 (79%)
Unknown	4 (10%)
Baseline ECOG performance status
0	35 (83%)
1	7 (17%)
Number of ESR1 mutations at baseline
0	20 (48%)
1	13 (31%)
2	1 (2%)
3	3 (7%)
4	2 (5%)
5	2 (5%)
6	1 (2%)
Number of PIK3CA mutations at baseline
0	28 (67%)
1	12 (29%)
2	2 (5%)
AKT1 mutation status at baseline
Mutation detected	4 (10%)
Mutation not detected	38 (90%)

MTD and study drug exposure

All dose-expansion patients were treated with GDC-0927 at the MAD of 1,400 mg/day. The MTD was not reached at doses up to the MAD. The median duration of treatment was 98.5 days (range, 32–1,127 days; Fig. 1). Four patients discontinued from the study following study termination by the sponsor, and transitioned to another Genentech-sponsored oral SERD study.

Figure 1.

Swimmer plot of time on study.

Safety

Forty-two patients who received at least one dose of GDC-0927 were included in the safety analysis. There were no reports of DLTs, treatment-related serious adverse events (SAE), deaths, or AEs leading to withdrawal. Overall, 37 patients (88%) experienced at least one AE regardless of causality, the most frequent of which were nausea (n = 14, 33%), constipation, diarrhea (n = 9, 21% each), arthralgia, fatigue, hot flush (n = 8, 19% each), back pain, vomiting (n = 7, 17% each), anemia, and headache (n = 6, 14% each; Table 2). A total of 30 patients (71%) experienced at least one AE considered related to GDC-0927 by the investigators, the most frequent of which were nausea, hot flushes (n = 8, 19% each), constipation (n = 6, 14%), diarrhea, and fatigue (n = 5, 12% each; Table 2). Nine patients (21%) experienced at least one grade 3 AE considered unrelated to GDC-0927, including 7 patients each with distinct AEs and 2 patients with the same AE of grade 3 back pain; there were no grade 4 or grade 5 AEs reported in this study. Seven patients (17%) experienced at least one SAE considered to be unrelated to GDC-0927.

Table 2.

AEs in ≥10% of patients overall (N = 42).

	Any grade	Grade 3a
Adverse events regardless of causality
Patients with any adverse events, n (%)	37 (88)	9 (21)
Nausea	14 (33)	0
Constipation	9 (21)	0
Diarrhea	9 (21)	0
Arthralgia	8 (19)	0
Fatigue	8 (19)	0
Hot flush	8 (19)	0
Back pain	7 (17)	2 (5)
Vomiting	7 (17)	0
Anemia	6 (14)	0
Headache	6 (14)	0
Aspartate aminotransferase (AST) increased	5 (12)	0
Asthenia	5 (12)	1 (2)
Cough	5 (12)	0
Pain in extremity	5 (12)	0
Adverse events considered related to study drug
Patients with at least one adverse event, n (%)	30 (71)	0
Total number of events	109	0
Hot flush	8 (19)	0
Nausea	8 (19)	0
Constipation	6 (14)	0
Diarrhea	5 (12)	0
Fatigue	5 (12)	0

^aThere were no grade 4 or grade 5 events in this study.

Per protocol, AEs of special interest (AESI) were events reported to the sponsor within 24 hours after learning of the event, regardless of relationship to study drug. Causality was determined by the investigator, and causal attribution guidance was provided in the protocol. Two patients (5%) experienced AESI of grade 2 deep vein thrombosis and grade 2 jugular vein thrombosis, both considered unrelated to GDC-0927 by the investigators. There were laboratory abnormalities noted in both hematologic and clinical chemistry parameters, none of which were considered clinically significant. None of the patients experienced any clinically significant vital sign changes.

None of the AEs led to withdrawal from the study or treatment. There were no dose modifications due to AEs. Seven patients (17%) experienced AEs that led to dose interruptions.

Pharmacokinetics

Overall, a total of 39 patients were included in the pharmacokinetic analysis and pooled according to treatment dose level; 3 patients with missing pharmacokinetic data were not included in the analysis. The mean GDC-0927 plasma concentration–nominal time profiles following single- and multiple-dose oral administration of GDC-0927 as tablets (600, 1,000, 1,400 mg) in patients with breast cancer are presented in Fig. 2A and B. The average C_max and AUC_0–24 of GDC-0927 increased with increasing dose in the dose range investigated in this study. Single-dose and steady-state mean pharmacokinetic parameters are presented in Supplementary Table S2. Following once-daily dosing, mean steady-state exposure (AUC_0–24) across all dose levels was 1.6-fold higher compared with single-dose exposure, and was consistent with the observed half-life and frequency of dosing.

Figure 2.

GDC-0927 plasma concentration (mean ± SD) following single dose (A) and at steady state (B).

Pharmacodynamics

Twelve patients underwent FES-PET imaging; 9 patients with FES-avid disease at baseline had >90% reduction in FES uptake at cycle 2, day 3 while receiving GDC-0927 doses of 600 mg (n = 1), 1,000 mg (n = 3), and 1,400 mg (n = 5), including 6 patients with baseline ESR1 mutations (Fig. 3A–C). Three patients had no FES-avid disease at baseline despite having ER H-scores of 160, 180, and 300; this discrepancy could be due to all 3 patients having liver (n = 2) or ovarian (n = 1) metastatic lesions that may have been unevaluable due to high tracer uptake in adjacent normal tissue. Two of the 3 patients on study for 182 days (liver) and 204 days (ovarian) had SD and achieved clinical benefit; 1 patient had PD and was on study for 62 days (liver).

Figure 3.

FES-PET scans demonstrate tumor avidity at baseline and target engagement during GDC-0927 exposure. A, Target engagement across all doses; reduction of FES uptake is independent of ESR1 mutation status. B and C, Representative FES-PET scan images at baseline and on treatment (cycle 2, day 3, 18–24 hours after cycle 2, day 2 dose) for 2 patients with tumors harboring ESR1 mutations dosed at 1,000 mg.

Nine patients provided paired baseline and on-treatment (cycle 2 or 3, day 8) tissue samples and ER, PR, and Ki67 protein levels were evaluated by IHC (Supplementary Table S3). Matched gene expression by RNA-sequencing and IHC results were evaluated in baseline and on-treatment tumor biopsy pairs from 6 patients, 4 of whom also had baseline FES-PET. A predefined, experimentally derived 38-gene ER signature z-score was utilized to determine ER signaling pathway activity collated from a core set of 21 estrogen-induced and 17 estrogen-repressed genes (16). In general, patients showed evidence of on-target pathway modulation and reduced proliferative activity by GDC-0927 as demonstrated by decreases in these biomarkers (ER signature z-score, ER H-score, PR H-score, and Ki67 IHC) upon treatment (Fig. 4A–D). PGR, the gene-encoding PR, is a downstream transcriptional target of ER and included in the ER signature z-score; analysis of this gene by itself (Supplementary Fig. S2) demonstrated a trend similar to PR IHC in the 6 patients (Fig. 4C). Furthermore, in an analysis of patients whose on-treatment clinical benefit status (yes/no) was available, baseline protein levels of ER and PR (n = 21 and n = 18, respectively) were not significantly different between patients with or without clinical benefit postdose (Supplementary Fig. S3A).

Figure 4.

Biomarker analyses of tumor tissue and ctDNA patient samples. Changes in ER pathway activity (A) by gene expression and in IHC protein levels for ER (B), PR (C), and Ki67 (D), in matched baseline and on-treatment patient biopsies (n = 6). A–D, Each individual patient is represented by the same line color throughout; 5 patients were dosed at 1,400 mg and 1 patient (black line) at 1,000 mg; 2 patients with baseline PR H-scores of 0 were excluded from the PR panel. MAF in paired baseline and postbaseline patient ctDNA samples assayed for ESR1 MAF (n = 19; E) and PIK3CA/AKT1 MAF (n = 12; F). E and F, Postbaseline samples were the first ctDNA sample available after treatment was initiated (either cycle 3 day 1 or end-of-treatment), available for 19 of 20 patients with baseline ctDNA samples with ESR1 mutations. MAF is the sum MAF of all eligible mutations if >1 mutation detected in a sample. A–F, Dashed lines indicate patients with clinical benefit defined as ≥24-week outcomes of unconfirmed responses (uCR, uPR), SD, or non-CR/PD.

Clinical activity

Twelve patients (29%) achieved clinical benefit, defined as either (i) patients with confirmed responses (CR or PR), or (ii) patients with 24-week or longer outcomes of unconfirmed responses (uCR, uPR), SD, or non-CR/PD (Fig. 1). Four patients (9.5%) with clinical benefit received treatment on study beyond 2 years (34.2, 35.8, 36.2, and 37 months). There were no confirmed CRs or PRs on this study. Seventeen patients showed a confirmed best overall response of SD (41%) that included patients dosed at 600 mg (n = 1), 1,000 mg (n = 2), and 1,400 mg (n = 14). Five patients showed a confirmed best overall response of non-CR/PD (14%); all 5 patients were from the 1,400 mg dose level.

All 42 patients were included in the PFS analysis. The earliest contributing event was disease progression in 34 patients (81%); the remaining 8 patients (19%) were without any event. The median PFS was 3.3 months [95% confidence interval (CI): 1.9–5.3 months].

ctDNA and clinical activity

Baseline patient ctDNA samples were available for all patients in this study and samples from 22 (52%) patients had at least one detectable ESR1 mutation (Table 1). The predominant ESR1 mutations were D538G, Y537S, and Y537N, and eight (19%) of the samples classified as ESR1 mutant had more than one ESR1 mutation detected demonstrating tumor heterogeneity (Supplementary Fig. S4A and S4B). PIK3CA and AKT1 mutations were also observed in 14 (33%) and four (10%) baseline patient ctDNA samples, respectively. Six of the 22 (27%) patients with baseline ctDNA ESR1 mutation(s) achieved clinical benefit, as well as 6 of the 20 (30%) patients without detectable baseline ctDNA ESR1 mutations. Similar to the lack of a correlation between baseline ER and PR protein levels and clinical benefit, baseline ESR1 mutant allele frequencies (MAF) were not significantly different between patients with or without clinical benefit (n = 23; Supplementary Fig. S3A). At the 1,400 mg dose, the median PFS was 3.52 months (95% CI: 1.8–6.4) in patients with tumors without detectable ESR1 mutations (n = 17), and 3.55 months (95% CI: 1.8–6.0) in patients with tumors expressing baseline ESR1 mutations (n = 19). ESR1 data were available for 36 of 36 patients dosed at 1,400 mg (Supplementary Fig. S5).

MAF generally declined in the first postbaseline (either cycle 3 day 1 or end-of-treatment) ctDNA samples collected for both ESR1 (16 of 21 patients; postbaseline sample unavailable for 1 patient with baseline ESR1 mutation detected) and PIK3CA/AKT1 (8 of 13; postbaseline sample unavailable for 2 patients with baseline PIK3CA/AKT1 mutations detected; Fig. 4E and F). The majority (15/16) of the ESR1 MAF reductions were greater than 95% relative to baseline and while 5 of 16 (31%) of the patients with a decline in ESR1 MAF also achieved clinical benefit, in general, a decline in ESR1 MAF did not correlate with clinical benefit (Supplementary Fig. S3B). Six patients displayed increased ESR1 MAF in the first postbaseline sample, among whom 1 (17%) achieved clinical benefit and 4 had an increase in more than one variant. Clinical response was not contingent on ESR1 MAF dynamics (χ²P = 0.35, OR = 3.9; Supplementary Fig. S3B).

Of the 23 patients with baseline samples available (all at the 1,400 mg dose) for analysis using the FACT ctDNA assay, 19 harbored between one and four total short variant gene alterations at baseline, the majority of which were substitutions (Supplementary Fig. S6A and S6B). The top mutated genes following ESR1/PIK3CA/AKT1 were TP53 (26.1%), CDH1 (8.7%), NF1 (4.3%), and PTEN (4.3%; Supplementary Fig. S6C). Each patient's mutation landscape and the associated clinical response for the patient are indicated in Supplementary Fig. S6C. ESR1 and PIK3CA mutations were mostly mutually exclusive (log₁₀ OR −0.363). Consistent with BEAMing results, ESR1 mutations were less prevalent in on-treatment samples (35% on-treatment vs. 52% pretreatment). In a comparison of 53 matched ctDNA samples with both BEAMing and FACT results collected at baseline, cycle 3 day 1 or end-of-treatment, an overall concordance of 89% for ESR1 mutation status (mutant vs. no mutation detected) was observed between the two assays. Six samples with ESR1 mutation(s) detected via BEAMing had no alterations detected via FACT, which may be a reflection of the differing limits of detection between digital PCR and next-generation sequencing assays (Supplementary Fig. S4A and S4B).

Discussion

Resistance to endocrine therapeutics (18) is a major clinical challenge in the MBC setting. While there is heterogeneity in the mechanisms of resistance (19), those associated with retention or reactivation of ER signaling, including acquisition of ESR1 mutations, are potentially sensitive to SERD therapeutics. Several SERDs have recently entered clinical development for ER⁺ advanced and MBC, highlighting an immense interest in next-generation SERDs resulting from the fulvestrant experience (20). The limitations of fulvestrant include poor pharmaceutical properties dictating intramuscular injections, thereby limiting the administered dose and potentially achieving an incomplete receptor blockade. Therapeutics that are mechanistically similar to fulvestrant but have an improved bioavailability may have best-in-disease potential (21). In this study, the pharmacokinetic parameters were predictable, with increases in GDC-0927 exposure with dose.

This study enrolled a heavily pretreated patient population with ER⁺/HER⁻ advanced or MBC; prior therapeutics included AIs for all patients while many patients received radiotherapy, tamoxifen, fulvestrant, CDK4/6 inhibitors, PI3K pathway inhibitors, and a few patients received investigational therapeutics. Encouraging antitumor activity was seen in this study, which was comparable in subgroups with or without ESR1 mutations. GDC-0927 had a safe AE profile in patients who were treated orally, once daily, at doses up to 1,400 mg/day, which was the MAD. The MTD was not reached. AEs were predominately grade 1 or 2 in severity. There were no grade 4 or 5 AEs. No DLTs, treatment related SAEs, AESIs, death due to AEs, or events leading to withdrawal from the study treatment were reported. Most of the AEs were managed by dose interruption, medication, and/or supportive care.

This study demonstrated that GDC-0927 greatly reduced ER availability, as measured by FES-PET, which indicated a reduction of >90% in uptake by the tumors postdose on cycle 2, day 3, in comparison to baseline. This reduction in ER availability was consistent, regardless of the ESR1 mutation status. FES-PET response was unaffected by the presence of PIK3CA mutations. Other recent studies have also used FES-PET to assess the pharmacodynamic effects of novel SERDs (22, 23), including elacestrant (24).

The presence of mutant ER was supported by ctDNA analyses that demonstrated the predominant ESR1 mutations in this study to be D538G, Y537S, and Y537N at baseline. Seven of 16 patients (44%) who exhibited decreased ESR1 MAF at their first on-treatment visit showed SD. However, similar to another study (25), ESR1 MAF dynamics in this small study were not predictive of clinical response. Reduced ER and PR levels, Ki67 positivity, and expression of ER target genes were observed in on-treatment biopsies, further demonstrating the on-target activity of GDC-0927. While ER tumor activity and ctDNA dynamics were effective in assessing GDC-0927 dosing, baseline parameters including ER IHC, PR IHC, ESR1 MAF levels, and dynamics were not predictive of clinical benefit, strengthening the need to develop better prognostic biomarkers in late line HR⁺/HER2⁻ patients. Furthermore, the conclusions from the biomarker analyses in this study are limited because of small sample sizes. Larger studies, such as those ongoing with the next generation of oral SERDs, will be better powered to interrogate both baseline levels and dynamics of these biomarkers and draw conclusions with clinical outcomes. The current development of ctDNA assays that infer tumor ER target gene expression (e.g., cell-free fragmentomics and epigenomics) may serve as more informative biomarkers in the near future.

One key parameter in drug discovery surrounding SERD compounds has been the optimization of ER degradation, along with attention to pharmacokinetic properties. Fulvestrant is a full ER antagonist; it antagonizes estrogen while lacking any ER agonist potential under estrogen-deprived conditions. Full ER suppression by fulvestrant is achieved by immobilizing ER (16) and promoting ER protein degradation (26). Tamoxifen is a partial ER antagonist that does suppress estrogen, but promotes weak ER agonism (27). In preclinical studies, GDC-0927 demonstrated ER degradation in all ER⁺ cell lines tested and retained a full antagonist profile, even in lines where GDC-0810 displayed partial agonism; notably, GDC-0927 lacked ER agonism in the uterus. While FES-PET imaging showed >90% reduction in FES uptake, ER protein expression only declined by an average of approximately 40%. Importantly, ER IHC represents just a single tumor snapshot, whereas FES-PET accounts for global tumor heterogeneity. Regardless, these results may suggest that some ER protein can remain in the tumor following SERD treatment and exhibit minimal binding of estradiol, and that FES-PET studies may represent a combination of effects on ligand binding and ER protein levels. This is consistent with evidence from previous studies showing that ER degradation by fulvestrant is not the mechanistic driver of ER antagonism (16, 28).

Several oral SERDS are in development, and elacestrant (GS2-02) was recently approved by the U.S. FDA for patients with ER⁺, HER2⁻, ESR1-mutated advanced or MBC as evaluated in the EMERALD trial (29), with elacestrant becoming the first oral SERD with improved efficacy over fulvestrant in the second- and third-line MBC setting. The AMEERA-5 trial (30) evaluating amcenestrant (SAR439859) in a combination study was discontinued in mid-2022 when results did not meet the continuation criteria and development of the amcenestrant program was subsequently stopped. The SERENA-6 (31) and SERENA-4 (32) trials are evaluating camizestrant (AZD9833) in combination treatments, with results expected in the near future; results for the SERENA-2 study showed camizestrant 75 and 150 mg had a median PFS of 7.2 and 7.7 months, respectively, compared with 3.7 months for patients treated with fulvestrant (33). The EMBER-3 trial is evaluating imlunestrant (Ly3484356). In a phase Ia/Ib/IIa study, GDC-0810 was reported to have a favorable safety and tolerability profile with preliminary antitumor activity in heavily pretreated patients with ER⁺ advanced/MBC, regardless of ESR1 mutation status, although the program has since been discontinued (34). Primary results from the acelERA trial investigating giredestrant (GDC-9545) did not meet its primary endpoint, but giredestrant showed a numerical improvement over the comparator arm, which was more pronounced in the population with ESR1-mutated tumors (35).

The sponsor discontinued the development of GDC-0927 in October 2017. The decision to halt enrollment was not due to any safety concerns, but was based on the totality of clinical data, including data from another SERD under development (36–38). Four patients who were still receiving GDC-0927 and discontinued because of study termination were transitioned to a different Genentech sponsored oral SERD study with giredestrant.

Supplementary Material

Supplementary Data 1

Supplementary Tables S1, S2, and S3.

Supplementary Fig. 1

Supplementary Fig. S1. Study scheme.

Supplementary Fig. 2

Supplementary Fig. S2. The PGR z-score for six patients analyzed in Fig. 4.

Supplementary Fig. 3

Supplementary Fig. S3. Clinical benefit does not correlate with ctDNA ESR1 mutant allele frequency (MAF) baseline levels or dynamics upon treatment, or with baseline ER and PR protein levels by IHC in tumor tissue.

Supplementary Fig. 4

Supplementary Fig. S4. ESR1 mutation analysis in ctDNA.

Supplementary Fig. 5

Supplementary Fig. S5. Kaplan-Meier estimate of progression-free survival for patients with and without ESR1 mutations in baseline ctDNA samples collected from the 1400 mg dose group.ESR1 data was available for 36 of 36 patients dosed at 1400 mg.

Supplementary Fig. 6

Supplementary Fig. S6. Detection of short variant gene alterations using the FACT next generation sequencing assay on ctDNA samples from patients dosed at 1400 mg dose (n=23).

Authors' Disclosures

S. Chandarlapaty reports personal fees and non-financial support from Novartis; grants from Daiichi Sankyo, AmbrX, and paige.ai; personal fees from Prelude Therapeutics, Nuvalent, Inivata, Boxer Capital, and Lilly; grants and personal fees from AstraZeneca; and other support from Odyssey Biosciences and Totus Medicines outside the submitted work. M.N. Dickler reports other support from Genentech/Roche during the conduct of the study and outside the submitted work. J.A. Perez Fidalgo reports grants, personal fees, and non-financial support from AstraZeneca and GSK; personal fees from Roche, MSD, Clovis, and Abilify pharma; non-financial support from Novartis during the conduct of the study; grants and personal fees from Pharmamar outside the submitted work; and has a patent for predictive signature of response to chemotherapy in triple negative breast cancer pending. R. Villanueva-Vázquez reports personal fees from Lilly, Novartis, Eisai, and Pfizer outside the submitted work. J. Giltnane reports other support from Genentech/Roche outside the submitted work; and is an employee and stockholder of Roche/Genentech. S. Cheeti reports other support from Genentech during the conduct of the study. J. Fredrickson reports personal fees from Genentech, Inc., and F. Hoffmann-La Roche Ltd. outside the submitted work. A. Collier reports other support from Roche/Genentech during the conduct of the study and outside the submitted work. H.M. Moore reports other support from Genentech, Inc., F. Hoffmann-La Roche Ltd, and Pfizer, Inc., outside the submitted work. C. Metcalfe reports other support from Genentech during the conduct of the study and outside the submitted work; and has a patent for diagnostic and therapeutic methods for the treatment of breast cancer (11081236) issued. J. Lauchle reports other support from Genentech during the conduct of the study and outside the submitted work; and is an employee of Genentech. E.W. Humke reports personal fees from Genentech during the conduct of the study; and has a Genentech patent on SERD pending to Genentech. A. Bardia reports grants and personal fees from Pfizer, Novartis, Genentech, Merck, Radius Health, Immunomedics/Gilead, Sanofi, Daiichi Pharma/AstraZeneca, Phillips, Eli Lilly, and Foundation Medicine during the conduct of the study. No disclosures were reported by the other authors.

Authors' Contributions

S. Chandarlapaty: Conceptualization, supervision, investigation, project administration, writing–review and editing. M.N. Dickler: Investigation, project administration, writing–review and editing. J.A. Perez Fidalgo: Investigation, project administration, writing–review and editing. R. Villanueva-Vázquez: Investigation, project administration, writing–review and editing. J. Giltnane: Conceptualization, writing–review and editing. M. Gates: Supervision, writing–review and editing. C.-W. Chang: Formal analysis, visualization, methodology, writing–review and editing. S. Cheeti: Formal analysis, supervision, visualization, writing–review and editing. J. Fredrickson: Formal analysis, visualization, writing–review and editing. X. Wang: Supervision, writing–review and editing. A. Collier: Formal analysis, visualization, writing–review and editing. H.M. Moore: Formal analysis, visualization, writing–review and editing. C. Metcalfe: Conceptualization, writing–review and editing. J. Lauchle: Conceptualization, supervision, investigation, methodology, project administration, writing–review and editing. E.W. Humke: Conceptualization, supervision, investigation, methodology, project administration, writing–review and editing. A. Bardia: Investigation, project administration, writing–review and editing.