Correlative Biomarker Analysis of PSMA Expression on CTCs and PSMA Imaging in a Phase I Study of PSMA Targeted Tubulysin Conjugate EC1169

Abstract

Purpose:

PSMA directed therapies provide meaningful clinical benefit in metastatic castration–resistant prostate cancer (mCRPC), yet responses remain limited, underscoring the need for additional biomarkers of PSMA expression heterogeneity. Here, we explored the relationship of PSMA imaging and PSMA expression on circulating tumor cells (CTCs) in our early phase trial using a PSMA–targeted small molecule.

Patients and Methods:

This Phase 1 study investigated EC1169, a small molecule conjugated to tubulysin analog warhead, and ^99mTc–EC0652, a PSMA imaging agent. Part A (dose–escalation) identified the recommended phase 2 dose. Part B (dose–expansion) assessed radiographic progression–free survival (rPFS) as its primary endpoint. We enrolled and treated 103 mCRPC patients. Most Part B patients underwent ^99mTc–EC0652 PSMA imaging. A CTC assay assessed for PSMA–positive CTCs and their association with response and PSMA imaging.

Results:

99mTc–EC0652 SPECT imaging demonstrated increased sensitivity for detecting bone lesions compared with standard scans (CT/bone scans). Using an optimized CTC assay, we observed that patients with a decrease in PSMA+ CTCs at baseline versus C3D1 displayed a longer rPFS (8.0 vs. 2.9 months; P=0.04). Importantly, patients with predominantly PSMA–positive disease on 99mTc–EC0652 imaging also harbored PSMA–negative CTCs, with a subset displaying NEPC–like morphology.

Conclusions:

While EC1169 showed limited activity, CTC and imaging analyses showed significant heterogeneity in PSMA expression on CTCs in patients with predominantly PSMA–positive lesions on SPECT. Our study highlights the importance of assessing both PSMA–based CTC and imaging assays in future validation trials.

INTRODUCTION

While novel androgen receptor signaling inhibitors (ARSIs) have shown a dramatic improvement in overall survival (OS) in men with metastatic castrate–resistant prostate cancer (mCRPC), patients inevitability develop resistance to therapy (1). In this setting, targeting prostate–specific membrane antigen (PSMA), which is a membrane–bound glycoprotein highly expressed in prostate carcinoma, has gained traction both with its sensitivity for imaging disease and in prolonging life in mCRPC patients (2–16). PSMA–directed therapies focus on using conjugates between a cytotoxic compound or radioisotope from the resultant endosome and an antibody or small molecule that binds to cell surface PSMA—this binding results in PSMA internalization and the intracellular release of the cytotoxic agent or radioisotope. PSMA–directed radioisotope conjugates have a unique theranostic advantage in providing the ability to detect and treat PSMA avid disease (3, 4). One such FDA–approved conjugate is ¹⁷⁷Lutetium–PSMA–617 that has a companion imaging biomarker for PSMA (2–4). However, ways to prognosticate response and resistance to PSMA–directed agents remain unclear (17). This area is nonetheless of considerable importance given the frequency of lineage plasticity, often associated with low or absent PSMA, has increased with the dominant use of ARSIs (18).

We conducted a phase I study of EC1169, a novel small molecule conjugated to a microtubule inhibitor tubulysin B hydrazide directed to PSMA, in mCRPC patients who have progressed on ARSIs, namely abiraterone and/or enzalutamide (19). Although EC1169 will not advance in clinical development due to limited ant–itumor activity, we present key biomarkers analyses from this study. Specifically, we assessed PSMA expression using an imaging agent, 99mTc–EC0652, and using contemporary assays, explored PSMA positivity (and neuroendocrine prostate cancer: NEPC–like features) at a single cell level in circulating tumor cell (CTC) pre– and post–treatment with EC1169. Our findings validate a role for PSMA CTCs enumeration pre– and post–treatment with a reduction being associated with a prolonged radiographic progression free survival (rPFS). Furthermore, and importantly, we highlight the significant heterogeneity in PSMA–positive CTCs with a subset displaying NEPC–like morphology, in patients who otherwise display predominately PSMA–avid single–photon emission computed tomography (SPECT) lesions.

PATIENTS AND METHODS

Study Design

This study was a Phase 1 A/B trial of a PSMA–targeted tubulysin conjugate, EC1169, in patients with mCRPC. Subjects were enrolled at 11 study sites in the U.S. The protocol was approved at each site’s Institutional Review Board (IRB) of all participating sites. Written informed consent was obtained from all patients prior to enrollment. Between August 4^th, 2014 and October 18^th, 2017, 148 subjects were screened for eligibility with 45 patients failing screening. In total, 103 patients were treated with at least one dose of EC1169—this included 40 patients in the Part A (dose–escalation) and 63 patients in Part B (dose–expansion). The study was performed in accordance with the Declaration of Helsinki and Good Clinical Practice. Part A was a dose–escalation phase to determine the maximum tolerated dose (MTD) and to recommend the phase 2 dose for EC1169. Part B was a two–cohort (B1 and B2), two–stage (stage I and II) expansion using the recommended phase 2 dose (RP2D) determined in the dose–escalation phase (Part A). Patients enrolled into Part B of cohort B1 were taxane–naïve, while those enrolled in cohort B2 were taxane–exposed. Furthermore, stage I of Part B (including cohorts B1 and B2) further studied the safety of the RP2D, while also evaluating efficacy of EC1169. If there was evidence for anti–tumor response, the trial would proceed to stage II with patient expansion (Figure S1, schematic shown for Part B).

Participants

Eligible patients had histological, pathological, and/or cytological confirmation of prostate cancer, failed therapy with abiraterone acetate or enzalutamide, and/or evidence of progressive, mCRPC with a serum testosterone level < 50 ng/dl. Progression was defined as a 25% increase in prostate–specific antigen (PSA) with an increase in the absolute value of at least 2 ng/ml; ≥ 20% increase of soft tissue disease using the sum of longest diameter of all target lesions or the appearance of ≥ 1 new soft tissue lesion(s) radiographically; and/or new osseous lesions found on bone scan. For Part A (dose–escalation) only, patients must have previously been treated with a taxane, whereas in Part B (dose–expansion), patients enrolled into cohort B1 were taxane–naïve (or received < 2 cycles) and cohort B2 were taxane–exposed (received ≥ 2 cycles). All patients in Part B were required to have ≥ 1 PSMA positive lesion by 99mTc–EC0652 scan. Other eligibility criteria included an absolute neutrophil count ≥ 1500/mm³, platelet count ≥ 100,000/mm³, and hemoglobin ≥ 9 g/dl.

Subjects were excluded if they received previous treatment with samarium–153 or strontium–89. No other anti–cancer therapy or investigational agents were permitted during the study or within 28 days prior to treatment start except for hormonal maintenance therapy. Subjects were required to be ≥ 18 years of age with an Eastern Cooperative Group (ECOG) performance status of 0/1. Full eligibility criteria are in Supplementary Data. All subjects provided written informed consent.

Study Treatment

Part A – Dose Escalation:

EC1169 was administered as an IV bolus injection in 21–day cycles either on day 1 and day 8 (weekly) or on days 1, 3, 5, 8, 10, and 12 (three times per week, TIW). For weekly dosing, there were 14 possible dose levels of EC1169, which ranged from 0.3 to 14.0 mg/m². For TIW dosing, there were 14 possible dose levels, which ranged from 0.2 to 8.5 mg/m². Doses were based on a modified Fibonacci sequence (Tables S1 and S2). Infusion of EC1169 occurred at least 4 days after administration of 99mTc–EC0652. Patients continued treatment until they met criteria for radiographic disease progression and/or no longer clinically benefited from therapy (20).

For earlier protocol versions (v1.0–4.0), dose escalation was determined using the continual reassessment method. For later versions of the protocol (v5.0–6.0), dose escalation was based on the standard 3+3 design, which consisted of 3 to 6 patients per dose level (details in Supplementary Data).

Part B – Dose Expansion:

EC1169 was administered as an IV bolus injection weekly on day 1 and day 8 of a 21–day cycle using the RP2D found in Part A. Infusion of EC1169 occurred at least 4 days after administration of 99mTc–EC0652. As noted earlier, further expansion of Part B to Stage II was determined by evidence of drug efficacy demonstrated in Stage I. Patients continued treatment until they met criteria for radiographic disease progression and/or were no longer benefiting from therapy (20). Of note, for Part B, enrollment into cohort B1 (taxane–naïve) was stopped early due to insufficient efficacy, based on the schema provided earlier (Figure S1), while cohort B2 (taxane–exposed) was enrolled fully.

Outcomes

The primary endpoint of Part A was to identify the RP2D of EC1169 in patients with recurrent mCRPC who have previously been on taxane. The primary endpoint of Part B was rPFS in taxane–naïve and taxane–exposed PSMA–positive patients with recurrent mCRPC per Prostate Cancer Working Group 2 and 3 (PCWG2 and PCWG3) (20, 21). Key secondary endpoints included safety of EC1169/EC0652, pharmacokinetics of EC1169, time to PSA progression, PSA response, and target lesion response. In addition, secondary and exploratory endpoints evaluated the association between both blood and imaging biomarkers with efficacy of EC1169. CTC levels and expression of other blood–based biomarkers were examined throughout EC1169 treatment. Furthermore, EC0652 imaging was compared to standard modalities. Similarly, CTC PSMA–positivity was compared to the number of EC0652–positive lesions on SPECT imaging.

Assessments

During the screening period, subjects received a physical exam, laboratory assessments and blood samples for CTCs, PSA, and serum testosterone. Imaging was performed with computed tomography (CT) or magnetic resonance imaging (MRI) and bone scan [technetium–99m labeled diphosphonate, fluorine–18 sodium fluoride positron emission tomography (PET) or PET/CT] to identify target lesions. Imaging was also performed with 99mTc–EC0652 as described (see Methods). During the treatment period, subjects received a physical exam and laboratory assessments weekly. CT/MRI and bone scans and blood tests as above were carried out every 2 cycles (21–day cycles). During the follow–up period, repeat 99mTc–EC0652 SPECT imaging was offered to all patients.

Pharmacokinetics

Pharmacokinetic studies of EC1169 were performed in Part A patients during cycle 1, day 1 and day 12 for the TIW, and cycle 1, day 1 and day 8 for the weekly dosing schedules. Samples were taken at time points before and after EC1169 (Supplementary Data). Plasma levels of EC1169 and its metabolite EC0347 (unconjugated tubulysin B hydrazide) were quantified by liquid chromatography–tandem mass spectrometry.

Safety

Safety was assessed in all patients who received ≥ 1 dose of EC1169. Adverse events (AEs) were monitored from the date of first dose until 30 days after the last dose. AEs were graded according to the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) v4.03. Dose reductions and delays were allowed for toxicities pre–defined in the protocol. The maximum permitted dosing delay for recovery from AEs was two weeks. Details of permitted dosing adjustments and criteria for removal of a patient from treatment due to toxicity are provided in the Supplementary Data.

Dose limiting toxicities (DLTs) included any ≥ grade 4 hematologic toxicity, ≥ grade 3 neutropenic fever or non–hematologic toxicity. DLTs were based on events occurring during the first cycle of therapy and were required to be drug–related (definitely, probably, or possibly–related). DLTs were reported to the sponsor within 24 hours. The trajectory of enrollment in Part A was determined based on the number of DLTs observed. A full list of DLTs is provided in the Supplementary Data.

Statistics and Efficacy Assessment

For Part A, efficacy was evaluated only in subjects enrolled in the weekly dosing schedule who received at least one cycle of EC1169. For Part B, efficacy was evaluated in all subjects who received at least one cycle of EC1169. Measures of efficacy included objective tumor response rate, rPFS (primary endpoint of Part B), time to PSA progression [according to PCWG2 and PCWG3] (20, 21), and PSA response. Response criterion for soft tissue lesions was determined using RECIST 1.1. For osseous disease, progression was met when ≥ 2 bone lesions were observed in the first follow–up bone scan or observed in any subsequent scan when compared to the first scan, confirmed ≥ 6 weeks later. Once patients met these criteria for progression, they were removed from study treatment. PSA progression was defined as a 25% increase in PSA with an absolute increase of ≥ 2 ng/ml from the nadir, confirmed by a second value performed ≥ 3 weeks later. Therapy was not discontinued solely based on PSA. Objective tumor response rates were calculated with corresponding 95% confidence intervals. Time to PSA progression and rPFS were assessed by the Kaplan–Meier method with 95% confidence interval for the median number of months. Waterfall plots were used to analyze the best percent change from baseline in PSA, target lesions, as well as other biomarkers [alkaline phosphatase, lactate dehydrogenase (LDH)].

Circulating Tumor Cells (CTCs)

CTC enumeration was analyzed using Epic Sciences platform. An optimized PSMA CTC assay was further developed based on a prior published assay (22) (Supplementary Data). Briefly, a single tube of blood (Streck Cell–Free DNA BCT) was collected from each patient at baseline and while on treatment with EC1169. Samples were shipped to Epic Sciences as whole blood and processed within 96 hours of the blood draw (23). Nucleated cells deposited onto glass slides were stained with DAPI, pan–cytokeratin (CK) (RRID: AB_258824), CD45 (RRID: RRID:AB_321156), and either PSMA (RRID: AB_3720971 and AB_2798032) or AR–V7 (RRID: AB_2861275) and imaged by a high–throughput scanning microscope. Any cell that was CK+ and CD45− with an intact nucleus was classified as a CTC, and CTC clusters defined as at least two adjacent cells were classified as one event. CTC counts were normalized to blood volume and expressed as the number detected per mL. Initial testing utilized the AR–V7 CTC (CD45, CK, DAPI, AR–V7), and PSMA CTC (CD45, CK, DAPI, PSMA) assays and scoring criteria as previously described (22, 24). Furthermore, a subset of samples were evaluated for NEPC–like morphology (small cell with scant cytoplasm, high nuclear—cytoplasmic ratio, salt and pepper chromatin) as validated in Beltran et al. (22, 25–27).

EC0652 Imaging and Analysis

In Parts A and B prior to treatment with EC1169, subjects received 0.1 mg EC0652 IV bolus injection which was radiolabeled with 740–925 MBq (20–25 mCi) of Technetium–99m. After infusion, conjugate planar images of the whole body were obtained. Following this, hybrid SPECT/CT scans were performed. An optional end–of study SPECT imaging procedure was offered. Data from imaging were analyzed by performing a total lesion count (bone and soft tissue) per patient. Findings were then compared to total lesion counts identified on imaging modalities, such as bone and CT scans, that were able on central review. Reads of 99mTc–EC0652 and conventional imaging were performed by dedicated radiologists as predefined in the investigators listed on FDA form 1572.

RESULTS

EC1169 Trial Pharmacokinetics, Population, Safety and Efficacy

A total of 103 mCRPC patients who failed therapy with abiraterone acetate or enzalutamide were enrolled and treated with at least one dose of EC1169 and underwent SPECT imaging with 99mTc–EC0652 (Figure 1A). Of the 103 patients, 40 patients were treated in the Part A (dose–escalation) and 63 patients in Part B (dose–expansion) (Figure 1B – consort diagram). Of note, all Part B patients were required to have ≥ 1 PSMA positive lesion by 99mTc–EC0652. The baseline characteristics were similar between both treatment groups within Part A and B (Table 1). Pharmacokinetic analysis of EC1169 in Part A was performed in patients before and during cycle 1 and showed increasing mean maximum plasma concentrations with escalating doses. The elimination half–life of EC1169 was ~1.1 hours with low levels of EC0347 (metabolite of EC1169) detectable in the circulation, suggestive of stability. The maximal tolerated dose as determined by Part A was 6.5mg/m² weekly and thus was used in Part B (Table S1, S2). Of the 40 patients enrolled in Part A, 35 and 5 were assigned to weekly or TIW dosing, respectively. However, the TIW dosing schedule A was eventually eliminated (in v6.0) due to difficulty in enrollment. Of the 63 enrolled in Part B, 19 were taxane–naïve (cohort B1) and 44 were taxane–exposed (cohort B2). The median duration of study treatment and cycles are shown in Table S3.

Figure 1. PSMA–Targeting Agents and Trial Profile.

(A) Cartoon shown of EC1169, a PSMA–targeting ligand conjugated with tubulysin B hydrazide, a potent microtubule inhibitor. Also shown is a cartoon of EC1169, a ^99mTc–EC0652 PSMA–targeted imaging agent. (B) Patient enrollment logistics and group assignment for Part A and B are shown. Reasons for treatment discontinuation are denoted for each subgroup. NLCB: no longer clinically benefiting as per Prostate Cancer Working Group 2/3. AE: adverse event. ¹Four patients discontinued treatment due to other reasons, including three patients who were clinically progressing and one patient who developed melanoma of the scalp.

Table 1:

Demographic and Clinical Characteristics of Study Patients

		Part A			Part B
		TIW n = 5	Weekly n = 35	Total n = 40	Cohort 1: Taxane Naïve n = 19	Cohort 2: Taxane Exposed n = 44	Total n = 63
Median age, years (Minimum–Maximum)		70.0 (53.0–74.0)	70.0 (57.0–87.0)	70.0 (53.0–87.0)	73.0 (59.0–84.0)	68.0 (49.0–82.0)	70.0 (49.0–84.0)
Race, n (%)	White	5 (100.0%)	33 (94.3%)	38 (95.0%)	15 (78.9%)	38 (86.4%)	53 (84.1%)
	Black or African American	0 (0.0%)	0 (0.0%)	0 (0.0%)	2 (10.5%)	3 (6.8%)	5 (7.9%)
	Native Hawaiian or Other Pacific Islander	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (2.3%)	1 (1.6%)
	Asian	0 (0.0%)	1 (2.9%)	1 (2.5%)	1 (5.3%)	0 (0.0%)	1 (1.6%)
	Not Provided	0 (0.0%)	1 (2.9%)	1 (2.5%)	1 (5.3%)	2 (4.5%)	3 (4.8%)
ECOG, n (%)	1	1 (20.0%)	10 (28.6%)	11 (27.5%)	15 (78.9%)	16 (36.4%)	31 (49.2%)
	2	4 (80.0%)	25 (71.4%)	29 (72.5%)	4 (21.1%)	28 (63.6%)	32 (50.8%)
Median Gleason Score at Diagnosis (Minimum–Maximum)1		7.5 (7.0–9.0)	8.0 (4.0–10.0)	8.0 (4.0–10.0)	7.0 (4.0–10.0)	8.0 (5.0–10.0)	8.0 (4.0–10.0)
Median time, initial cancer diagnosis, years (Minimum–Maximum)		8.0 (4.9–15.0)	7.9 (2.2–25.5)	8.0 (2.2–25.5)	7.0 (1.6–26.4)	8.8 (1.2–21.4)	8.5 (1.2–26.4)
Median prior cancer-related RT, regimens (Minimum–Maximum)2		3.0 (1.0–4.0)	2.0 (1.0–7.0)	2.0 (1.0–7.0)	2.0 (1.0–4.0)	2.0 (1.0–3.0)	2.0 (1.0–4.0)
Median prior cancer-related treatments, regimens (Minimum–Maximum)3,4		5.0 (3.0–7.0)	6.0 (4.0–11.0)	6.0 (3.0–11.0)	4.0 (2.0–12.0)	5.0 (2.0–13.0)	5.0 (2.0–13.0)
Median baseline PSA, ng/ml (Minimum–Maximum)		414.3 (6.3–3449.5)	169.9 (10.5–3180.7)	174.4 (6.3–3449.5)	66.3 (2.6–4715.4)	211.5 (0.2–4507.7)	164.5 (0.2–4715.4)
Median baseline LDH, U/L (Minimum–Maximum)		214.0 (159.0–495.0)	228.0 (118.0–451.0)	221.0 (118.0–495.0)	212.0 (152.0–328.0)	244.0 (112.0–2600.0)	217.5 (112.0–2600.0)

TIW = three times per week; SD = standard deviation; ECOG = Eastern Cooperative Performance Group; RT = radiation therapy; PSA = prostate specific antigen; LDH = lactate dehydrogenase

¹Part A = 38 (4 in TIW, 34 in Weekly dosing); Part B = 58 (18 in Cohort 1, 40 in Cohort 2)

²Part A = 33 (5 in TIW, 29 in Weekly dosing); Part B = 44 (12 in Cohort 1, 32 in Cohort 2)

³Part A = 40 (5 in TIW, 35 in Weekly dosing); Part B = 63 (19 in Cohort 1, 44 in Cohort 2)

⁴Additional details can be found in Table S12.

For Part B, enrollment into cohort B1 (taxane–naïve) was stopped early due to insufficient efficacy (Figure S1), while cohort B2 (taxane–exposed) was enrolled fully. At the time of the data cut–off on November 29^th, 2018, all 103 patients discontinued treatment (Part A, 40; Part B, 63) (refer to Figure 1B for discontinuation reasons). Of the total 103 subjects included in this study, 57.5% (23 out of 40, Part A) and 54% (34 out of 63, Part B) experienced grade ≥ 3 adverse effects (AEs) (Table 2). However, only a minority experienced serious AEs (25% and 27%, Part A and B, respectively). Specifically, in Part B, a greater percentage of subjects in the taxane–exposed cohort (B2) compared to the taxane–naïve cohort (B1) experienced AEs ≥ grade 3 (65.9% versus 26.3%); serious AEs (31.8% versus 15.8%); AEs that led to interruption (54.5% versus 31.6%); and AEs that led to permanent discontinuation (15.9% versus 10.5%). Of note, one fatal treatment AE occurred in the taxane–exposed cohort of Part B, which was due to sepsis unrelated to treatment. No DLTs were reported for patients treated with EC1169 (Table 2). Furthermore, the most common drug–related AEs were fatigue, decreased appetite, nausea, abdominal discomfort, dysgeusia, and alopecia (Tables S4, S5, and S6 for complete AEs). The most common drug–related AEs ≥ grade 3 were anemia, neutropenia, hypophosphatemia, and fatigue (Table S7).

Table 2:

Summary of Adverse Events

	Part A			Part B
	TIW n = 5	Weekly n = 35	Total n = 40	Cohort 1: Taxane Naïve n = 19	Cohort 2: Taxane Exposed n = 44	Total n = 63
Patients with ≥ 1 AE, n (%)	5 (100.0%)	34 (97.1%)	39 (97.5%)	19 (100.0%)	44 (100.0%)	63 (100%)
Grade ≥ 3 AE, n (%)	1 (20.0%)	22 (62.9%)	23 (57.5%)	5 (26.3%)	29 (65.9%)	34 (54.0%)
Serious AE, n (%)	0 (0.0%)	10 (28.6%)	10 (25.0%)	3 (15.8%)	14 (31.8%)	17 (27.0%)
Drug-related AE, n (%)	2 (40.0%)	27 (77.1%)	29 (72.5%)	17 (89.5%)	39 (88.6%)	56 (88.9%)
AE leading to reduction of the study dose, n (%)	0 (0.0%)	2 (5.7%)	2 (5.0%)	0 (0.0%)	2 (4.5%)	2 (3.2%)
AE leading to interruption of the study treatment, n (%)	1 (20.0%)	10 (28.6%)	11 (27.5%)	6 (31.6%)	24 (54.5%)	30 (47.6%)
AE leading to permanent discontinuation of the study treatment, n (%)	0 (0.0%)	4 (11.4%)	4 (10.0%)	1 (5.2%)	8 (18.1%)	9 (14.2%)
Fatal AE, n (%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	1 (2.3%)	1 (1.6%)
DLT, n (%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)	0 (0.0%)

AE = adverse event; DLT = dose limiting toxicity

¹Drug-related AEs are deemed possibly, probably or definitely related to treatment

^*Grade ≥ 3 are defined as severe; medically significant, not immediately life-threatening; hospitalization or prolongation indicated; disabling; limiting self-care activities of daily living

For Part A, efficacy results are presented for all patients treated on the weekly schedule (N=35), including subset analysis for patients treated at the RP2D dose of 6.5 mg/m². While no patients on the weekly schedule experienced a complete response (CR), 1 patient experienced a partial response (PR) and 17 patients demonstrated stable disease (SD) yielding a disease control rate (DCR) of 51.4% (95% CI 34.0–68.6). For subjects treated at the RP2D of 6.5 mg/m², 5 patients showed SD with no patients experiencing either CR or PR (DCR, 71.4%, 95% CI 29.0–96.3) (Table S8).

For Part B, efficacy results are presented for all patients treated in cohort B1 (taxane–naïve) and cohort B2 (taxane–exposed). In cohort B1, 6 patients demonstrated stable SD (DCR, 31.6%, 95% CI 12.6–56.6). In cohort B2, no patients experienced a CR; 2 and 10 patients experienced a PR and SD, respectively (DCR, 27.3%, 95% CI 15.0–42.8) (Table S8). Time to PSA progression and rPFS were evaluated in Part B (dose–expansion) patients. The median rPFS time was 6.3 months (95% CI 2.2–9.0) in cohort B1 and 4.1 months (95% CI 2.2–7.7) in cohort B2 (Table S9 and Figure 2A). The median time to PSA progression, as per PWCG2/3, was 0.9 months (95% CI 0.7–2.1) in cohort B1 and 2.1 months (95% CI 1.6–3.4) in cohort B2 (Table S9 and Figure S2) (20, 21).

Figure 2. Radiographic Progression Free Survival, Change In Target Lesion size, and Percent Change In PSA in Part B mCRPC Patients.

(A) Radiographic progression free survival (rPFS) for Part B participants (shown in months) since receiving the first dose of EC1169. Taxane–naïve (B1) and taxane–exposed (B2) patients are shown as a dashed or solid line, respectively. The median rPFS time was 6.3 months (95% CI 2.2–9.0) and 4.1 months (95% CI 2.2–7.7) in cohort B1 (taxane–naïve) and cohort B2 (taxane–exposed), respectively. (B) Bar graphs demonstrate change in tumor size as measured by the sum of diameters of all tumor target lesions compared to baseline for each participant assigned to Part B. Of the 63 patients in Part B, 30 had radiographically evaluable target lesions. Treatment status is represented in dark blue for taxane–exposed and light blue for taxane–naive. (C) Bar graph represents of the best percent change in serum PSA from baseline for each participant assigned to Part B. Treatment status is represented in dark blue for taxane–exposed and light blue for taxane–naive. Of note, not all patients are represented in the plot because a subset of patients discontinued treatment before confirmatory or follow up PSA draws could be obtained.

In addition, changes in tumor size were also compared to baseline values in both Part A and B. For Part A, 20 out of the total 35 patients had target lesions evaluable by RECIST based on central radiology review (weekly, 18; TIW, 2) (Figure S3). For Part B, 30 out of the total 63 patients had evaluable target lesions (8 in cohort B1, and 22 in cohort B2) (Figure 2B). Changes in serum PSA were also examined. In Part A, two patients on the QW regimen achieved a PSA50 response. In Part B, one patient (cohort B2) achieved a PSA50 response. Of note, a subset of patients discontinued treatment before confirmatory or follow up PSA draws could be obtained (Figure S4, Figure 2C). Furthermore, serum LDH and alkaline phosphatase in Part B patients showed a trend towards decreased levels compared with baseline (Figure S5).

EC0652 PSMA Imaging

A total of 106 patients received baseline 99mTc–EC0652 imaging, of which 103 patients proceeded to treatment with EC1169 (Part A, 40, Part B, 63). Of the 63 patients in Part B, 51 patients had 99mTc–EC0652 imaging that was directly compared to baseline conventional imaging modalities. Of the 51, 61% (N=31) patients had bone scans and 53% (N=27) patients had CT scans, which were available for central reader assessment. More bone lesions were found using 99mTc–EC0652 compared to traditional bone scan (770 vs. 587, respectively; 76.2% overlap) suggesting a greater ability of this type of PSMA–directed imaging to detect bone disease. Furthermore, all bone lesions detected on traditional bone scans were noted by 99mTC–EC0652. However, less total soft tissue lesions (>1.5 cm) were found using 99mTc–EC0652 compared with standard CT (48 vs. 54, respectively) (Figure 3A, Table 3). 99mTc–EC0652 was also performed before and after treatment with EC1169 with a representative image of response shown in Supplementary Figure S6. The safety of 99mTC–EC0652 was monitored for at least 4 days from date of the first dose of EC0652, which demonstrated that a total of 7 (out of 106, 6.6%) patients reported at least one AE: 5 not related to treatment and 2 (1.9%) drug–related AEs (dry mouth and nausea). In patients receiving ≥ 1 dose of 99mTc–EC0652 there were no serious, fatal, or drug–related AEs ≥ grade 3.

Figure 3. PSMA CTC and Imaging Heterogeneity in mCRPC Patients.

(A) Left. The total number of lesions detected by 99mTc-EC0652 (turquoise) or bone or CT scans (purple) is shown, separated by bone lesions and nodes ≥ 1.5cm. Right. Baseline 99mTc–EC0652 SPECT scans demonstrating that PSMA-targeted imaging detects more osseous lesions than conventional bone scintigraphy in a representative patient example. (B) The number of CK+ CD45− and PSMA–positive CK+ CD45− CTCs are shown normalized to blood volume and expressed as the number detected per mL. (C) Change in PSMA+ CTCs after completion of EC1169 treatment at cycle 3 day 1 compared to baseline (positive changes are capped at 50% increase). (D) Radiographic progression free survival (rPFS) for patients post–EC1169 treatment separated by those with an increased (blue) or decreased (orange) PSMA+ CTCs. (E) Bar plot is shown of 12 patients that harbor at least 1 CTC(s) and have undergone SPECT imaging. Y–axis denotes percent of CTCs or imaging lesions positive for PSMA highlighting substantial PSMA CTC heterogeneity even in the context of predominantly positive PSMA imaging. Of note, patients with zero values did not have evaluable lesions on bone or soft tissue SPECT imaging. Furthermore, patient with NEPC–like CTCs are marked with asterisk.

Table 3:

Comparative lesion detection by imaging modality using 99mTc–EC0652, bone scan, or standard CT scan

	99mTc–EC0652 positive	99mTc–EC0652 negative	Bone scan positive	Bone scan negative	CT scan positive	CT scan negative
Bone lesions, total number of lesions (%)	770	0	587	183	N/A	N/A
Nodes > 1.5cm, total number of lesions (%)	48	6	N/A	N/A	54	0

N/A = Not applicable.

CTC Heterogeneity and Comparison to Imaging

CTC analysis was performed retrospectively on bio–banked samples during and after the completion of the trial (22) using an initial and optimized assay. The initial assay was conducted on 63 Part B patients, of which 55 (B1, 17; B2, 38) had samples that passed QC. CTCs were defined as any DAPI+, CK+, CD45− cells and were assessed for AR–V7 or PSMA expression in two separate 4–channel immunofluorescence assays (Supplementary Data). Using the initial assay, 48 patients had detectable CK+ CD45− CTCs (median 11 CTCs per mL, range 1–187). AR–V7 positive CTCs were identified in 11 (20%) patients using the prior clinically validated cutoff of ≥ 1 CTC (28) (Figure S6). Per prior studies, AR–V7 and high CTCs (≥3 CTC/mL vs. < 3 CTC/mL, validated prognostic cutoff) trended towards shorter rPFS (23, 24) (Figure S7).

To enhance our ability to detect PSMA+ CTCs, we developed an optimized PSMA CTC assay utilizing a tyramide–based detection system and reanalyzed residual bio–banked samples. The methodology was adjusted to include re–titration of antibody concentrations, and optimization of the secondary detection (Supplementary Data). Using this optimized assay, PSMA CTC detection was compared to the initial assay in both cell lines and 57 mCRPC patient samples across multiple cohorts. Our optimized assay was found to be ~1.6–fold more sensitive (54% vs. 33% PSMA CTCs in new vs. old assay) without compromising specificity (Supplementary Data).

Utilizing this optimized assay, we reanalyzed 49 patients with residual sample at baseline, of whom, 22 (45%) had detectable PSMA+ CTCs (median ~5 [range: ~1-675] per mL excluding zeros; Figures 3B, Figure S6A). Furthermore, 33 paired samples at Cycle 3/Day 1 were also reanalyzed with 18/33 (~55%) having ≥ 1 PSMA+ CTC (median ~8 [range: ~1–165] per mL excluding zeros). Consistent with our initial assay, in those patients with ≥ 1 PSMA+ CTC, the percentage of PSMA+ CTCs compared with total CTCs varied substantially (median 69%, 12–100%) (Figure 3B). We further noted that the total number of CTCs did not significantly correlate with the percent of PSMA–positive CTCs (Figure S8). These findings demonstrate true heterogeneity in PSMA expression on CTCs, rather than reflecting any technical limitations of PSMA detection in our optimized CTC assay.

Additionally in these patients with residual sample at baseline, we analyzed 23 samples using a 4–marker panel of CK/DAPI/CD45/AR for detection of NEPC–like CTCs using the assay described by Beltran et al (26). This phenotype is defined as having high nuclear–to–cytoplasm ratio with small circular size and, importantly, low AR expression. It was previously defined based on CTCs detected in transformed or de novo NEPC patients. In our study, eleven patients harbored CTCs with NEPC–like features (median ~1 [range, ~1–10] per mL) (examples shown in Figure S6B). In prior studies, the presence of at least one cell with NEPC–like features prior to treatment was a strong negative prognostic for survival and response with abiraterone or enzalutamide (26, 27). Using our data, patients with NEPC–like CTCs versus no NEPC–like CTCs showed a trend toward worse rPFS (p=0.57) (Figure S9). Patients with NEPC–like CTCs demonstrated worse outcomes in almost all variables (PFS, rPFS, OS) (Table S10). Given our small sample size, power to detect significance was however limited.

While our drug displayed limited efficacy, several patients showed a reduction in tumor size (refer to Figures 2, 3, and S6C). We therefore restricted our analysis to patients harboring PSMA–positive CTCs at baseline to investigate whether any patients showed an absolute reduction in number post–EC1169 treatment (at Cycle 3/Day 1), which existed in a subset of patients (Figure 3C). We further noted that the mean rPFS was 6.3 and 4.1 months in cohort B1 and B2, respectively. Comparison of patients with an increase versus decrease in the PSMA CTCs revealed statistically longer rPFS (8.0 vs. 2.9 months, P = 0.04) (Figure 3D).

Last, we explored heterogeneity of PSMA expression on CTCs compared with 99mTc–EC0652 SEPCT imaging. Notably, we found that patients with PSMA–imageable disease harbored both PSMA–positive and PSMA–negative CTCs. In the trial, 21 patients with 99mTc–EC0652 SPECT imaging results had matched optimized PSMA–CTC assays. Of these, only 12 patients had detectable CTCs, and 7 of the 12 had at least ≥1 PSMA–positive CTC (Figure 3E). Among these 7 with ≥1 PSMA–positive CTC(s), the percent of PSMA–positive CTCs over total CTCs varied substantially from 13% to 97%.

Although PSMA expression on CTCs was heterogeneous, nearly all patients with ≥1 PSMA–positive CTC(s) showed 100% PSMA positivity in bone and soft tissue (≥1.5cm) lesions, except for Pt 5 and Pt 6. Pt 5 had 67% PSMA+ CTCs, with 6 out of 6 bone and 5 out of 6 soft tissues lesions positive for PSMA on SPECT imaging. Pt 6 had 24% PSMA+ CTCs with only 2 of 5 soft tissue lesions detected on PSMA imaging. The remaining 5 out of 12 patients with detectable CTCs showed no PSMA–positive CTCs, yet all demonstrated 100% PSMA positivity on SPECT imaging (Figure 3E).

Furthermore, within the 12 patients with detectable CTCs and 99mTc–EC0652 SPECT, there was no clear trend between PSA response and PSMA expression on CTCs and SPECT imaging (Table S11). Lastly, all patients with NEPC–like CTCs with available 99mTc–EC0652 SPECT imaging (N = 5) showed 100% PSMA positivity on imaging (Figure 3E). Together, these data suggest that patients with predominately PSMA–positive lesions by imaging may still harbor significant PSMA heterogeneity, including those with NEPC–like morphology, on CTCs analyses.

Discussion

PSMA–directed therapies, namely PSMA theranostics have been shown to be effective in prolonging OS and imparting clinical benefit in mCRPC patients. Notable in this regard is two phase 3, open–label randomized theranostic trials in which post–ARSI/taxane (VISION) or post–ARSI/pre–taxane (PSMAfore) mCRPC patients with PSMA–positive radiographic disease (by ⁶⁸Ga–labeled PSMA–11 PET) were treated with a PSMA–directed radioligand therapy using ¹⁷⁷Lu–PSMA–617 (2, 29). In both trials, a significant prolongation, compared with standard care, was achieved in the rPFS endpoint. However, the response of such agents only occurs in a subset of patients, even when selected by PSMA PET imaging for PSMA–positive lesions, and absence of PSMA–negative lesions (with a size threshold). This underscores the need to develop biomarkers, in addition to PSMA PET imaging, for patients that respond poorly to PSMA directed therapy.

Towards exploring such biomarkers, we performed correlative analyses in a Phase 1A/B study of the PSMA–targeted tubulysin EC1169 in men with mCRPC, by utilizing both CTC and imaging biomarker (99mTc–EC0652) platforms to assess for PSMA disease heterogeneity. While EC1169 did not show any major clinical efficacy, our study demonstrated a discordant relationship between PSMA CTCs and PSMA–imaging. Although this discrepancy could be due to sensitivity issues inherent to each method, our analyses indicate that late–stage disease exhibits a biologically meaningful PSMA heterogeneity in CTCs that is otherwise not evident on imaging.

Perhaps the most striking findings relate to our use of CTCs as a blood–based biomarker for mCRPC. First, we found that patients who demonstrated a reduction in PSMA–positive CTCs in pre– vs. post–EC1169 at start of cycle 3 (21–day cycles) showed a prolonged rPFS. This is consistent with data from the Phase 1B PRINCE trial using ¹⁷⁷Lu–PSMA–617, in which the clearance of PSMA–positive CTCs correlated with longer PFS (30) and other recent studies analyzing CTCs in the context of ARSI therapy (31). This suggests that PSMA CTC detection and its kinetics, albeit limited in our cohort, might aid in guiding treatment monitoring with PSMA–directed compounds (2). At the same time, in cases with low CTC counts, PSMA expression heterogeneity should be interpreted cautiously, as limited detected events may reflect assay sensitivity or sampling variation rather than true biology. The second key finding was that patients with predominantly PSMA–imageable disease harbored significant heterogeneity in PSMA expression on CTCs. These findings are suggestive of notable PSMA loss on tumor cells that cannot be captured at the level of PSMA imaging.

Of note, while our PSMA imaging (99mTc–EC0652) was SPECT based, it demonstrated a strong detection ability akin to PET PSMA imaging, with an increased ability to detect osseous lesions compared with conventional imaging. Nonetheless, validation of discordant PSMA detection on CTC and imaging will require larger trials that utilize FDA–approved PSMA PET and theranostics agents. Lastly, similar to the loss of PSMA on tumor cells, a number of patients with CTCs displayed NEPC–like morphology (26), which often lack AR or PSMA expression. While we cannot conclusively identify the neuroendocrine lineage given the lack of canonical stains (ASCL1 and NEUROD1) and correlation with patient tumor histopathology, these cells demonstrated validated features of poor prognostic outcomes with loss of AR — findings consistent with increasing incidence of lineage plasticity in the era of potent ARSI therapy (18). Identification of such plasticity patients will be critical in the enrollment of such patients for clinical trials. CTC–based approaches to identify neuroendocrine transformation or aggressive variant prostate may be able to be combined with additional liquid biomarkers, such as methylation or whole genome–based assays (32–34). Furthermore, these findings of frequent PSMA loss of CTCs raise the importance of understanding how PSMA and AR signaling relate –not all PSMA loss is suggestive of AR independent signaling, but rather may be targeted by other AR–targets, such as STEAP1/2 (35). In fact, co–detection of PSMA, STEAP1/2, and other cell surface markers (CECAM5, DLL3, etc.) may be critical to understand the repertoire of tumor cells in circulation (36).

Overall, our data align with ongoing efforts to delineate CTCs by high copy number instability, neuroendocrine morphology (INSM1+DLL3+), and staining of other androgen receptor associated and plasticity markers. Furthermore, given the current activity of PSMA–directed drugs, the clearance of CTCs and the assessment of CTC diversity at different stages of disease progression will be crucial to elucidate patient–specific biology and its relation to response and efficacy of ¹⁷⁷Lu–PSMA–617. Although suggestive of the utility of PSMA–positive CTCs and imaging, together, as a prognostic and therapeutic biomarker, our studies along with others will require validation in larger trials with validated PSMA–imaging and active targeted theranostics.

STATEMENT OF TRANSLATIONAL RELEVANCE.

PSMA–targeted imaging and therapeutics are well–established in men with metastatic castration–resistant prostate cancer (mCRPC). However, clinical benefit is limited to a subset of patients, highlighting the urgent need for predictive biomarkers of response and resistance. We conducted an early–phase theranostic trial of a small molecule drug conjugate targeting PSMA (EC1169) combined with PSMA SPECT imaging. While we observed only limited anti–tumor activity, our comprehensive assessment of PSMA expression on both circulating tumor cells (CTCs) and nuclear imaging revealed a substantial degree of PSMA heterogeneity at the level of CTCs, including the appearance of NEPC–like morphology, that would otherwise be missed on PSMA–targeted imaging. Although validation of our findings will require larger trials, we highlight the importance of integrating both imaging and CTC–based assays to fully understand the evolution of disease progression and potentially inform treatment decision–making.

Data Availability Statement

Owing to patient privacy considerations and limitations of informed consent, the clinical data generated in this study are not publicly available. De–identified datasets can be obtained from the corresponding author upon reasonable request and subject to institutional approvals.