¹⁸F-Fluoroestradiol PET/CT Measurement of Estrogen Receptor Suppression during a Phase I Trial of the Novel Estrogen Receptor-Targeted Therapeutic GDC-0810: Using an Imaging Biomarker to Guide Drug Dosage in Subsequent Trials

Phase I clinical trials are first-in-human trials to assess safety, tolerability, pharmacokinetics, and pharmacodynamics of novel therapeutics (1, 2). In the traditional phase I model, the dose of the novel therapeutic may initially be given at subtherapeutic doses and is then increased in subsequent participants, with the goal of determining the highest dose that can be tolerated without incurring dose-limiting toxicities. If the phase I trial is successful, the maximum tolerated dose of the novel therapeutic is chosen for subsequent efficacy trials.

However, a therapy may reach an optimal on-target molecular effect at a dosage lower than the maximum tolerable dose, which is a potential flaw of this study design. If this occurs, then further escalation of the drug dosage may result in additional side effects without additional benefit. For example, therapeutics that function through suppression of the estrogen receptor (ER) should achieve maximal effect when the ER is no longer available for estrogen binding and/or downregulated, depending on the type of antagonist used. Any additional increases of the drug dose beyond this point should not produce further benefit, but may result in greater toxicity. If the drug dose at which maximal ER suppression occurs could be determined, then a dosage could be selected that maximizes benefit and minimizes harm, as the optimal biological dose.

16α-[18F]-fluoro-17β-estradiol (FES) PET/CT has been validated as an accurate method for localizing ER-expressing tumors (3) and as a predictive assay for breast cancer endocrine therapy (4–7). In breast cancer, the uptake of ¹⁸F-FES, as measured by SUV on PET, has been shown to correlate with ER expression in biopsy material assayed by in vitro radioligand binding and by immunohistochemistry (8–11), providing evidence of the value of FES SUV to measure specific binding to ER. FES uptake in breast cancer metastases declines after therapy with ER blocking agents such as tamoxifen and fulvestrant as well as estrogen-depleting agents such as aromatase inhibitors (12). Recently, serial FES PET/CT imaging of the availability of ER in patients with breast cancer demonstrated residual ER availability during fulvestrant therapy in nearly 40% of patients, suggesting that the current dose of fulvestrant therapy was inadequate for complete block of ER in many patients (13). These studies suggest a role for FES PET as a pharmacodynamic biomarker for breast cancer (14, 15), which can help determine the dosage of ER targeted therapies needed for maximal ER occupancy and/or downregulation.

GDC-0810 is a potent ER antagonist and degrader being developed for the treatment of postmenopausal women with ER-positive advanced breast cancer (16, 17). GDC-0810 binds to the ER to limit hormone action and induces conformational changes that lead to the degradation of the receptor, thus limiting both modes of ligand-dependent and -independent ER signaling. This agent has been shown to induce tumor regression in both tamoxifen-sensitive and tamoxifen-resistant in vivo tumor models (16).

In this article, we used FES PET/CT to evaluate ER occupancy and guide dose selection for future trials of GDC-0810, a novel ER-targeted therapeutic.

This prospective, multi-institutional study (Clinicaltrials.gov NCT1823835) was performed with institutional review board approval at all three participating institutions and with written informed consent of enrolled patients. Medical oncologists from each institution identified study participants based on the following inclusion criteria: (i) postmenopausal females with pathologically proven ER-positive, HER2-negative adenocarcinoma of the breast; (ii) evidence of either locally recurrent disease not amenable to resection or radiotherapy with curative intent or metastatic disease, progressing after at least 6 months of hormonal therapy for ER-positive breast cancer; (iii) at least a 2-month interval since the last use of tamoxifen; (iv) at least a 6-month interval since the last use of fulvestrant; (v) at least a 2-week interval since the last use of any other anticancer hormonal therapy; (vi) at least a 3-week interval since the last use of chemotherapy; (vii) Eastern Cooperative Oncology Group (ECOG) performance status of 0–2; and (viii) adequate organ function. Exclusion criteria included: (i) untreated or symptomatic CNS metastases; (ii) endometrial disorders; (iii) any significant cardiac dysfunction within 12 months prior to enrollment; (iv) active inflammatory bowel disease or chronic diarrhea, short bowel syndrome, or upper gastrointestinal surgery including gastric resection; (v) known human immunodeficiency virus (HIV) infection; (vi) known clinically significant history of liver disease; (vii) major surgery within four weeks prior to enrollment; or (viii) radiotherapy within 2 weeks prior to enrollment.

Patients were assigned sequentially to escalating oral doses of GDC-0810. Each patient was imaged prior to GDC-0810 administration and at cycle 2, day 3 of GDC-0810 therapy (4 weeks of treatment). Dosage for patients undergoing FES PET/CT was 200 mg once per day, which then increased by 200 mg until 800 mg once per day or in divided doses. Patients dosed at 200 mg and 400 mg daily underwent imaging at peak drug concentration, which was performed 2 to 12 hours after dose. Patients on all other doses underwent imaging 18 to 24 hours after dose in order to provide confirmation of receptor occupancy at the time when GDC-0810 was at its trough concentration.

Medical records were reviewed to document patient age and race, tumor histology, the time since last tamoxifen use (by inclusion criteria of at least two months) and the time since last fulvestrant use (by inclusion criteria of at least 6 months).

FES was manufactured using a modified version of the published work by Knott and colleagues, 2011 (18). The drug product is assayed for total radioactivity and visually examined for particulates. Samples are removed for pH measurement, analytical HPLC measurements of radiochemical purity and identity, radionuclidic identity, endotoxin levels, and sterility of the product. Radiochemical purity was required to be greater than 90%. Specific activity was required to be greater than 726 Curies per millimole. Once all of the pre-release Quality Control (QC) testing is successfully completed, the drug product batch is released for human administration.

As the availability of serum SHBG influences FES uptake (19), blood was obtained from patients within one week of each FES PET/CT scan for laboratory analysis of serum SHBG. The blood sample obtained with the on-treatment FES PET/CT was also analyzed for plasma drug concentration.

Patients were administered 185 ± 37 MBq (5 ± 1 mCi) of FES with a mass limit of 5.2 μg. Approximately 60 minutes after tracer administration, a low-dose CT from mid-skull to mid-thigh was performed on a Siemens or GE Healthcare integrated PET/CT scanner for attenuation correction and lesion localization. Immediately after the CT scan, a PET scan was obtained over the same anatomical region. PET data were reconstructed iteratively with segmented correction for attenuation with the CT data and displayed in multiplanar reconstructions. Reconstruction algorithms varied by imaging site, but were consistent within each site and for each individual patient.

One board-certified radiologist at each of the three sites reviewed the FES PET/CT exams performed at their site. Physiologic FES avidity was expected in the liver, bowel, kidneys, ureters, bladder, and often the uterus and adnexa. Focal sites of FES avidity that could not be attributed to physiologic avidity were considered FES-avid malignancy. Up to five malignant lesions were chosen as index lesions on the pretreatment scan. FES avidity was quantified for each index lesion on the pretreatment and posttreatment scans. A volume of interest (VOI) was drawn to encompass each index lesion and a lesional maximum SUV (SUV_max) was recorded. For each lesion, a background VOI was drawn in surrounding normal tissue and background maximum SUV (SUV_background) was recorded. FES uptake for each lesion was calculated as SUV_max − SUV_background = SUV_{corrected for background} or simply SUV.

The mean percent reduction in SUV for all index lesions within a patient was calculated. Complete suppression of ER was defined as greater than 90% or greater reduction in aggregate SUV_{corrected for background}. Less than 90% reduction was defined as incomplete suppression. Waterfall plots were used to illustrate the relationship between reduction and dose.

To assess the relationship between FES SUV and SHBG values, we followed the value transformations provided in Peterson and colleagues (19). SUV was transformed by taking the logarithm of the average corrected value and then calculating the percent change on the transformed values. The square root of each time point's SHBG value was taken and then the percent change calculation was performed. Based on the skewed distributions of the factors, the strength of the relationship was assessed using Spearman's correlations.

The relationship between tamoxifen and fulvestrant use with pretherapy SUV values was also assessed by Spearman's correlations. To assess if any tamoxifen/fulvestrant use affected SUV values, the Wilcoxon rank-sum test was performed.

P values less than 0.05 were considered statistically significant. All analyses were performed using SAS 9.4 (The SAS Institute).

Forty-one postmenopausal female patients with ER-positive, HER2-negative adenocarcinoma of the breast were enrolled in a phase I dose escalation trial. Five patients did not undergo FES PET/CT imaging, and six patients only underwent a baseline FES PET/CT without an on-therapy scan. Thirty patients underwent baseline and on-therapy scans and were thus evaluable for measurement of therapy-induced changes in FES PET/CT. Their mean age was 60 years (range, 33–78 years). The histology of the tumors was invasive ductal carcinoma in 25 patients, invasive lobular carcinoma in 3 patients, mixed ductal lobular carcinoma in 1 patient, and cribriform carcinoma in 1 patient. All 30 patients received orally administered GDC-0810 in doses ranging from 200 to 800 mg. The administration schedule could be once or twice per day under fasting or non-fasting conditions. Dose escalation cohorts are summarized in Table 1.

All patients had at least one FES-avid abnormality at baseline imaging. Disease distribution varied from oligometastatic to widespread FES-avid metastatic disease. Soft tissue, pulmonary, and skeletal FES-avid metastases were observed and included in the analysis. Heterogeneity of FES avidity was observed across the subject population as well as across lesion population within individual subjects. The average per-patient SUV_{corrected for background} of lesions ranged from 0.2 to 9.3 with a median of 3.1.

On cycle 2 imaging, 24 of 30 patients (80%) demonstrated > 90% reduction in aggregate FES SUV (Fig. 1). This included 1 of 3 patients receiving 200 mg once per day, 2 of 4 patients receiving 400 mg once per day, 5 of 5 patients receiving 300 mg twice per day, 9 of 11 patients receiving 600 mg once per day, 3 of 3 patients receiving 400 mg twice per day, and 4 of 4 patients receiving 800 mg once per day. For the 30 patients, the range of SUV reduction was 63.6% to 100% with a median of 98.5% reduction. An example of a patient demonstrating >90% FES SUV suppression is shown in Fig. 2, while an example of incomplete suppression can be seen in Fig. 3.

Pretherapy, background FES SUVmax values demonstrated a mean of 1.36 and standard deviation of 0.51. On-therapy, FES SUVmax values demonstrated a mean of 1.22 and standard deviation of 0.44.

As the availability of serum SHBG influences FES uptake (19), SHBG was recorded within 1 week of both pretherapy and on-therapy FES PET/CT scans. Both pretherapy and on-therapy SHBG levels were available for 22 of 30 patients. Median pretherapy and on-therapy SHBG values were 53.6 nmol/L (range, 23.3–144.0) and 146 (range, 78–242), respectively, resulting in a median percent change of 172.1% (range, 59.7%–334.0%). However, no clear relationship was found between the change in SHBG and the change in FES SUV for either the raw (rho = 0.17, P = 0.44) or transformed values (rho = 0.16, P = 0.48).

No difference was seen in FES reductions based on plasma drug concentrations (data not shown).

Eighteen of 30 (60%) of patients used tamoxifen prior to study enrollment with a median of 61 months (range, 2–193 months) between last tamoxifen use and pretherapy FES PET/CT. Per inclusion criteria, there was an interval of at least 2 months since the last use of tamoxifen. No significant relationship was found between time since last tamoxifen use and FES SUV on the pretherapy PET/CT (rho = 0.42, P = 0.08). Furthermore, no relationship was found between SUV values for patients who used tamoxifen at any point (median SUV, 4; range, 1.1–9.3) compared with those who did not (median SUV, 2.8; range, 0.2–7.3; P = 0.39).

Ten of 30 (33%) of patients used fulvestrant prior to study enrollment with a median of 30 months (range, 13–48 months) between last fulvestrant use and pretherapy FES PET/CT. Per inclusion criteria, there was an interval of at least 6 months since the last use of fulvestrant. Similar to tamoxifen, the scatter plots showed no relationship between time since last fulvestrant use and FES SUV, and no significant correlation was found (rho = −0.06, P = 0.88). Additionally, SUV values were not significantly different between those that previously used fulvestrant (median, 4.7; range, 1.3–9.3) and those who did not (median, 2.1; range, 0.2–7.3; P = 0.09).

As part of a phase I clinical trial of a novel ER-modulating agent, GDC-0810, we used FES PET/CT to monitor ER suppression and help select the optimal dose engaging the target for GDC-0810. This is a novel application of targeted imaging. We use an imaging agent as a pharmacodynamic biomarker to determine the dose of a novel therapeutic needed for optimal target suppression. In theory, determination of a dose that achieves optimal target suppression that is lower than the maximum tolerated dose determined by phase I dose escalation may allow for a biologically optimal drug effect while reducing dose-related toxicities. In this trial, only 1 of 3 patients in the 200 mg and 2 of 4 patients in the 400 mg per day cohorts reached the goal of >90% FES SUV suppression, while the 600 mg and 800 mg per day cohorts achieved that goal in the majority of patients (14 of 16 and 7 of 7). The 600 mg daily dose has been selected as the recommended phase II dose of GDC-0810 in part due to decreases in FES SUV achieved in phase I subjects, in addition to other factors, including safety and pharmacokinetics.

In the pretherapy FES PET/CT scans, there was heterogeneity of intensity in FES-avid lesions between and within patients. Heterogeneity of ER-positive metastatic breast cancer is well described in the literature. Mortimer and colleagues found that 4 of 17 (24%) patients with metastatic breast cancer had discordance in FES uptake between sites in individuals (4). Linden and colleagues found an absence of FES uptake in one or more metastatic sites in 10% of patients with primary ER-positive tumors; the same group showed in a follow-up study that 13% percent of patients (6 of 47) with ER-positive primaries had one or more sites of FES-negative disease (20).

Posttherapy scans were performed 2 to 12 hours after dose during peak concentrations. Starting at the 600-mg dose level, the posttreatment FES PET/CT was performed 18 to 24 hours after dose in order to provide confirmation of receptor occupancy at the time when GDC-0810 was at its trough concentration. The difference in time of imaging could affect measured FES avidity. We might expect the reduction in uptake seen at 200 and 400 mg per day would be even less if FES PET/CT was performed at trough drug levels. In other words, the dose effect may be even more pronounced than we report.

Reduction in FES avidity on-therapy may be due to receptor occupancy and/or loss. Joseph and colleagues demonstrate in preclinical models that ER is still detectable by immunohistochemistry following treatment with GDC-0810 (17), thus both processes may be occurring. It is difficult to distinguish these processes by PET.

Within our cohort were 22 patients with both pretherapy and on-therapy serum SHBG levels. We did not find a significant relationship between changes in FES SUV and changes in SHBG. A prior study of 312 FES PET scans of patients with ER-positive breast cancer found a relationship between FES SUV and SHBG, with SHBG inversely associated with FES SUV (19). However, very large increases in SHBG were required to result in relatively small changes in FES SUV. For example, a study by Peterson and colleagues (19) reported a more than doubling of SHBG resulted in an expected FES SUV decrease of only about 15%. Furthermore, in the range of SHBG levels between 64 nmol/L and 225 nmol/L, which were the majority of patients in our cohort, there was little if any change in FES SUV. The dramatic decreases we observed in FES SUV on GDC-0810 therapy could not be accounted for by the increases in SHBG alone, even when conservatively using the Peterson equation developed by Peterson and colleagues. Thus, we propose the decreases in FES SUV can be principally attributed to the pharmacological effects of GDC-0810.

One important consideration in the calculation of percent SUV reduction on-therapy is the use of mean versus median percent reduction. Our study used mean percent reduction, while another recent study used median (13). In our study, up to five target lesions were chosen for quantitative analysis. In the hypothetical situation where the percent SUV reduction of five targets is 100, 100, 100, 30, and 20, using median percent reduction would result in a calculated 100% reduction in FES SUV, which would be an overestimation of the drug effect. Using mean percent reduction would result in only a 70% reduction in FES SUV, which would be an incomplete suppression. It is well known that the mean is sensitive to outliers, while the median is insensitive. In this study, we wanted to incorporate outliers, as it made our estimate of effect more conservative. This is in contrast to a recent study that used median percent reduction (13). It is important to consider the method used to calculate SUV changes as it alters the perceived drug effectiveness. Additionally, our calculations of lesional FES SUV were corrected for background FES avidity. The calculated percent reductions in FES avidity would differ if background was not considered. Thus, correction for background avidity is also important to consider when comparing the statistical methods assessing effectiveness between studies.

Recent studies have demonstrated the potential of ER-targeted therapies to cause an effect on FES SUV. A study utilizing FES PET/CT during fulvestrant therapy required discontinuation of tamoxifen for at least five weeks before baseline PET imaging (13). Despite this requirement, patients who withdrew from tamoxifen therapy 5 to 6 weeks before baseline FES PET/CT still demonstrated low FES uptake. Our cohort required at least 2 months of tamoxifen withdrawal prior to pretherapy FES PET/CT. In our study, no difference in FES uptake was observed between patients previously treated with tamoxifen (but with at least two months of tamoxifen withdrawal) and patients who were not previously treated with tamoxifen. Thus, 2 months appears to be a reasonable period of tamoxifen withdrawal prior to FES PET/CT. Similarly, no difference in FES uptake was evident between patients previously treated with fulvestrant (but with at least 6 months of fulvestrant withdrawal) and patients who were not previously treated with fulvestrant. Thus, 6 months appears to be a reasonable period of fulvestrant withdrawal prior to FES PET/CT.

The concept of targeted imaging as a pharmacodynamic biomarker to measure the effects of GDC-0810 on its target, such as FES monitoring of ER occupancy and/or downregulation in this study, could be applied to other novel ER-targeted therapeutics as well as to other drug–target combinations. Examples include ⁸⁹Zr-trastuzumab PET/CT (21–24) to monitor HER2 suppression by novel HER2-targeted therapeutics and ¹⁸F-fluoro-α-dihydrotestosterone to monitor suppression of androgen receptor (25).

The limitations of our study include the small sample size, although it was a standard sample size for a phase I trial. In addition, not all patients had both pretherapy and on-therapy SHBG levels drawn, which further limits the strength of the SHBG analysis.

In conclusion, these results support the prospective use of FES PET/CT as a pharmacodynamic biomarker of ER occupancy and/or downregulation during dose escalation of GDC-0810 in a phase I trial. FES PET/CT was useful in determining the drug dosages needed for specified decreases in FES SUV and contributed to dose selection of GDC-0810 for phase II trials.

M.C. Dickler is a consultant/advisory board member for and reports receiving commercial research support from Genentech/Roche. A. Bardia is a consultant/advisory board member for Genentech. C.L. Arteaga is a member of an advisory board for Novartis and a consultant for Genentech.G.A.Ulaner reports receiving commercial research grants from Genentech. No potential conflicts of interest were disclosed by the other authors.

Conception and design: A. Bardia, I.A. Mayer, E. Winer, J. Baselga, H.C. Manning, U. Mahmood, G.A. Ulaner

Development of methodology: H.C. Manning, U. Mahmood, G.A. Ulaner

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Wang, K.L. Ayres, M.N. Dickler, A. Bardia, I.A. Mayer, C.L. Arteaga, H.C. Manning, G.A. Ulaner

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Wang, D.A. Goldman, A. Bardia, J. Baselga, H.C. Manning, U. Mahmood, G.A. Ulaner

Writing, review, and/or revision of the manuscript: Y. Wang, K.L. Ayres, D.A. Goldman, M.N. Dickler, A. Bardia, I.A. Mayer, E. Winer, J. Fredrickson, C.L. Arteaga, J. Baselga, H.C. Manning, U. Mahmood, G.A. Ulaner

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): D.A. Goldman, I.A. Mayer, J. Fredrickson, H.C. Manning, G.A. Ulaner

Study supervision: G.A. Ulaner

We gratefully acknowledge the work of the radiochemistry cores and teams at the Memorial Sloan Kettering Cancer Center (led by Dr. Jason Lewis and Dr. Serge Lyashchenko), Massachusetts General Hospital (led by Dr. Daniel L. Yokell), and Vanderbilt (led by Dr. Mike Nickels) for supply of FES.

This research was funded in part through the NIH/NCI Cancer Center Support Grant P30 CA008748 and Genentech.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.