Most patients with prostate cancer treated with androgen receptor (AR) signaling inhibitors develop therapeutic resistance due to restoration of AR functionality. Thus, there is a critical need for novel treatment approaches. Here we investigate the theranostic potential of hu5A10, a humanized mAb specifically targeting free PSA (KLK3).
LNCaP-AR (LNCaP with overexpression of wildtype AR) xenografts (NSG mice) and KLK3_Hi-Myc transgenic mice were imaged with 89Zr- or treated with 90Y- or 225Ac-labeled hu5A10; biodistribution and subcellular localization were analyzed by gamma counting, PET, autoradiography, and microscopy. Therapeutic efficacy of [225Ac]hu5A10 and [90Y]hu5A10 in LNCaP-AR tumors was assessed by tumor volume measurements, time to nadir (TTN), time to progression (TTP), and survival. Pharmacokinetics of [89Zr]hu5A10 in nonhuman primates (NHP) were determined using PET.
Biodistribution of radiolabeled hu5A10 constructs was comparable in different mouse models. Specific tumor uptake increased over time and correlated with PSA expression. Treatment with [90Y]/[225Ac]hu5A10 effectively reduced tumor burden and prolonged survival (P ≤ 0.0054). Effects of [90Y]hu5A10 were more immediate than [225Ac]hu5A10 (TTN, P < 0.0001) but less sustained (TTP, P < 0.0001). Complete responses were observed in 7 of 18 [225Ac]hu5A10 and 1 of 9 mice [90Y]hu5A10. Pharmacokinetics of [89Zr]hu5A10 were consistent between NHPs and comparable with those in mice. [89Zr]hu5A10-PET visualized the NHP-prostate over the 2-week observation period.
We present a complete preclinical evaluation of radiolabeled hu5A10 in mouse prostate cancer models and NHPs, and establish hu5A10 as a new theranostic agent that allows highly specific and effective downstream targeting of AR in PSA-expressing tissue. Our data support the clinical translation of radiolabeled hu5A10 for treating prostate cancer.
This study presents a novel theranostic approach for the treatment of prostate cancer utilizing the humanized mAb hu5A10 to target androgen receptor–regulated free PSA (fPSA). Using murine models of prostate cancer, we developed and evaluated alpha- and beta-emitting radionuclides for radioimmunotherapy with radiolabeled hu5A10. Radioimmunotherapy targeting fPSA is a highly effective therapeutic approach for improved survival and to facilitate local disease control and complete response. We report important differences in the therapeutic outcome following alpha- versus beta-radioimmunotherapy, with alpha-radioimmunotherapy exhibiting delayed but more potent and more sustained antitumor effects. PET with [89Zr]hu5A10 complements fPSA-radioimmunotherapy by enabling diagnosis and patient stratification for subsequent fPSA-targeted therapy. Notably, [89Zr]hu5A10 PET uptake in the prostate of nonhuman primates was high despite several 1,000-fold lower fPSA levels in monkeys in comparison with humans, further supporting clinical translation of hu5A10.
Prostate cancer is a leading cause of cancer death among men in the United States (1). With metastatic prostate cancer 5-year survival rates under 30%, there is a critical need for more effective systemic treatments (2); over the past decade, combinations of more potent androgen receptor (AR) antagonists, androgen synthesis inhibitors, and taxane-based chemotherapy were shown to improve patient outcomes and changed standards of care. None provide sustained disease control and virtually all tumors regrow, the majority of which through the restoration of AR signaling (3).
Because of their very strong correlation with downstream AR-pathway activity, AR-governed enzymes such as human kallikrein 2 (hK2, KLK2) and PSA (KLK3) have been explored as both diagnostic biomarkers and therapeutic targets of prostate cancer (4, 5). These kallikrein-related peptidases are highly selectively and abundantly expressed in both healthy and malignant prostate tissues. However, common age-related pathologic conditions of the prostate such as inflammation, benign age-associated enlargement, and cancer result in the retrograde occurrence of PSA and hK2 in blood at levels ≈10–6 of those in prostate fluid. Immediately upon reaching the blood circulation, uncomplexed and catalytically active PSA (“free” PSA, fPSA) is permanently inactivated by protease inhibitors or enveloped by a2-macroglobulin and converted into “complexed” PSA (4, 6, 7). Measurements of the different forms of PSA and hK2 in blood can be used to predict risk of clinically significant prostate cancer and outcome, and to monitor prostate cancer. Recently, antibody-based methods for specific in vivo targeting of fPSA and hK2 in tissue have been successfully developed and applied for in vivo radioimmunotheranostics (RIT; refs. 5 and 8). This approach relies on the use of high-specificity and high-affinity antibodies developed to specifically bind to the catalytic clefts of hK2 and fPSA that are uniquely exposed on the free forms of PSA and hK2, abrogates binding of the complexed form of these enzymes in the blood, and enables RIT utility in the setting of high PSA levels in the blood. This antibody technology exploits the inherent mechanism of the neonatal Fc-receptor (FcRn) to route antigen-bound mAbs from recycling to lysosomal pathway compartments, resulting in internalization into target cells and accumulation of diagnostic or therapeutic radionuclides at sites of disease (9, 10). Progress in antibody and small-molecule design for targeted delivery and the increased availability of radionuclides with potent therapeutic properties have fueled interest in the field of targeted radiotherapy. In particular, research has focused on high linear energy transfer (LET) therapies which deliver ablative doses to cancerous cells over a small range, sparing adjacent nontargeted tissues (11, 12). Radium-223 dichloride, a bone-seeking calcium mimetic for treatment of bone metastatic castrate-resistant prostate cancer, is the first approved alpha-particle emitter (13) setting a precedent for other alpha-particle emitters undergoing clinical investigation (14). However, many questions remain, in particular, the requisite properties of a RIT construct that would enable therapeutic doses of radiation to be delivered using administration schedules that are safe and effective.
In this article, we present the results of preclinical studies comparing PSA-targeted RIT compounds carrying radionuclides with high or low LET. Specifically, hu5A10—a humanized fPSA-targeting IgG1-mAb designed to route antigen-bound hu5A10 to the FcRn to enable its internalization into target cells—was labeled with the alpha-particle emitter Actinium-225 ([225Ac]hu5A10) and beta-particle emitting Yttrium-90 ([90Y]hu5A10), representing high-LET (∼100 keV/μm) and low-LET (∼0.2 keV/μm) PSA-RIT, respectively. We further evaluate the possibility of using Zirconium-89–labeled hu5A10 ([89Zr]hu5A10) as a companion diagnostic PET reporter to guide therapeutic dose planning for fPSA-RIT.
Materials and Methods
MDAPCa2b were purchased from ATCC. LNCaP-AR (LNCaP with overexpression of wildtype AR) was a kind gift from Charles Sawyers (15). The cell lines were cultured according to the developer's instructions and frequently tested for Mycoplasma.
All mouse studies were approved by the IACUC, MSKCC (New York, NY; #04-01-002). For xenograft studies, male athymic BALB/c nude mice [NU(NCr)-Foxn1nu; 6–8 weeks old, 20–25 g; Charles River) were inoculated with LNCaP-AR or MDAPCa2b in the flank by subcutaneous injections of 1–5 × 106 cells (100 μL medium/100 μL Matrigel). Tumors developed 3–7 weeks postinoculation.
Transgenic KLK3 mouse models
Site-directed mutagenesis of APLILSR to APLRTKR at positions 4, −3, and −2 of the zymogen sequence of KLK3 (Quick Change Lightning Mutagenesis Kit; Stratagene) enabled furin, an ubiquitously expressed protease in rodent prostate tissue, to cleave the short activation peptide at the cleavage site (-1 Arg/+1 Ile) resulting in constitutive conversion from noncatalytic zymogen to functional PSA enzyme. A transgenic mouse model was established by cloning the described construct into a SV40 T-antigen cassette downstream of the short rat probasin (pb) promoter. This construct was microinjected into fertilized mouse embryos (C57BL/6) and implanted into pseudopregnant female mice to yield the pb_KLK3 genetically modified mouse model (GEMM). Pb_KLK3 mice were crossed with the Hi-MYC model (ARR2PB-Flag-MYC-PAI transgene) to create KLK3_Hi-MYC mice, a cancer-susceptible GEMM with PSA expression. Integration of genes into the genome of the offspring was confirmed by Southern blot analysis and PCR.
Preparation of radiolabeled and fluorescently labeled antibody constructs
Humanized 5A10 IgG1 mAb was developed by DiaProst AB and produced by Innovagen AB (7). The hu5A10 radioimmunoconstructs were prepared according to previously published protocols unique to the specific radionuclide (see Supplementary Materials and Methods). A competitive binding assay was utilized to determine the affinity of radiolabeled huA10 for fPSA; no significant loss of affinity for recombinant fPSA was noted after labeling (8).
Biodistribution and dosimetry studies
Biodistribution studies were conducted to evaluate uptake and pharmacologic distribution of fPSA targeted [90Y]hu5A10, [225Ac]hu5A10, and [89Zr]hu5A10. Each radioconjugate was administered intravenously and contained 30–60 μg of total protein with an injected activity of 5.55 MBq (150 μCi; [90Y]hu5A10), 11 kBq (300 nCi; [225Ac]hu5A10) or 4.07 MBq (110 μCi; [89Zr]hu5A10), respectively. Biodistribution studies in LNCaP-AR tumor models were conducted at 48, 120, and 350 hours for [90Y]hu5A10; 4, 120, and 360 hours for [225Ac]hu5A10; and at 72, 120, and 350 hours for [89Zr]hu5A10 after injection (n = 3–5 per time point). To evaluate the targeting specificity of [89Zr]hu5A10 (350 hours after injection) a blocking dose (1 mg) of unlabeled hu5A10 was coadministered with the radiotracer in an additional group of LNCaP-AR mice. Biodistribution of [89Zr]hu5A10 was further evaluated in the MDAPCa2b model at 120 hours after injection Biodistribution of all three hu5A10 radioimmunoconstructs was studied in the PSA-expressing KLK3_Hi-MYC GEMM at 120 hours after injection Blood was drawn by cardiac puncture immediately after mice were euthanized by CO2 asphyxiation. Organs/tissues were collected, rinsed briefly in water, dried on paper, weighed, and counted in a gamma counter (Packard Instrument) for accumulation of respective radionuclide. Count data were corrected for background activity and decay and the tissue uptake for each sample was calculated by normalization to the total amount of activity injected [measured in units of % injected activity per gram (%IA/g)]. Organ and tumor dosimetry calculations were performed using the Rodent Dose Evaluation Software (16, 17). Briefly, pharmacokinetic data were input into a murine three-dimensional imaging-based dosimetry platform built on accurate MRI volumes to compute the intraorgan and interorgan and tumor-absorbed doses from alpha-particle, beta-particle, and gamma-emission transport. The adult male mouse model and Medical Internal Radiation Dose Committee methodology (18, 19) as implemented in OLINDA/EXM (20) were used to calculate the absorbed doses of [225Ac]hu5A10 and [90Y]hu5A10.
Average pretherapeutic tumor sizes were 350 ± 107 mm3 (range, 268–471 mm3) and animals were randomized into groups receiving fPSA-targeted RIT or no treatment. A single 11.1 kBq (300 nCi) or 18.5 MBq (500 μCi) activity of [225Ac]hu5A10 or [90Y]hu5A10, respectively, was injected into the tail vein in LNCaP-AR tumor-bearing mice (some with bilateral tumor grafts). In the low-LET study, tumor volume, time to nadir (TTN; i.e., time to smallest post-RIT tumor volume) and time to progression (TTP; i.e., time to two consecutive increases in tumor volume post-nadir), and survival following treatment with [90Y]hu5A10 (n = 6 mice with a total of nine tumors) or no treatment (n = 6 mice, six tumors) were quantified. The same parameters were recorded in high-LET [225Ac]hu5A10-treated LNCaP-AR xenografts (n = 14 mice, 18 tumor) compared with untreated controls (n = 9 mice, 16 tumors). Because of logistic issues, high-LET and low-LET treatment studies were not done in parallel. Tumor volume (V) was calculated by measuring the length (L) and width (W) of tumors by caliper and using the formula for a rotated ellipsoid [V = (W × 2L)/2]. Humane endpoint was defined as weight loss of 20% or a tumor diameter exceeding 15 mm. Tumor volumes were measured twice per week and survival monitored until all mice either had succumbed to tumor burden (either spontaneously or meeting criteria for euthanasia) or remained free of a visible/palpable tumor mass at the time the last mouse without complete response succumbed.
Tissue histology and autoradiography of GEMM prostate
Bulk tissue comprised of nonseparated prostate lobes, seminal vesicle, and prostatic urethra was harvested following euthanasia. Tissue was embedded in optimal cutting temperature compound (Sakura) and incubated on ice for 2 hours before snap-freezing on dry ice in a cryomold. Sets of contiguous 5 μm (autoradiography) or 100 μm (fluorescence microscopy) thick tissue sections were cut with a CM1950 cryostat microtome (Leica) and arrayed onto SuperfrostPlus glass microscope slides. Autoradiographs were acquired approximately 168 hours after injection of KLK3_Hi-MYC GEMM with [89Zr]hu5A10 or [90Y]hu5A10 by immediately placing sectioned tissue in a film cassette against a Fuji film BAS-MS2325 imaging plate covered by 20-μm-thick polyvinylchloride film. The slides were exposed for 48 hours. Exposed phosphor plates were read by a Fujifilm BAS-1800II bio-imaging analyzer generating digital images with isotropic 50 μm/pixel dimensions. For microscopy, mice were injected with 25 μg Cy5.5-hu5A10 or Cy5.5-IgG1 120 hours before sacrifice (9). Sections stained for actin and DNA were incubated with 200 μL of 10 U/mL rhodamine-phalloidin (Life Sciences) in PBS for 2–3 hours at room temperature in a covered container and washed with PBS twice. DNA/nuclei staining was performed by incubating slides for 10 minutes in 5 μg/mL DAPI in PBS, followed by a wash with PBS. Slides were air dried and a drop of Mowiol A-48 (Calbiochem) was placed on the slide before adding a mounting cover glass. Micrographs were acquired using an Eclipse Ti-E fluorescence microscope (Nikon) equipped with a motorized stage (Prior Scientific Instruments), X-Cite light source (EXFO), and filter sets (Chroma). Images were acquired and processed using NIS-Elements AR, version 4.0 (Nikon), FIJI (NIH), or a TCS SP8 (Leica) confocal laser scanning microscopy (Molecular Cytology Core Facility, MSKCC, New York, NY; ref. 9), and MosaicJ (Phillipe Thévenaz, Biomedical Imaging Group, Swiss Federal Institute of Technology Lausanne, Lausanne, Switzerland). All fluorescent images were captured with a fixed fluorophore-dependent exposure time.
PET imaging experiments were conducted on a microPET Focus 120 scanner (Concorde Microsystems). Mice were anesthetized by inhalation of 1%–2% isoflurane/oxygen gas mixture to record PET images (21). PET images were acquired at 120 hours after injection. The PET-bed was then moved for CT imaging using a NanoSPECT/CT (Bioscan). Data were exported in raw format, and the rigid body (3 degrees of freedom) co-registration between PET and CT data was performed in Amira 5.3.3 (FEI). Details on PET and CT acquisition and analysis are provided in the Supplementary Materials and Methods.
Measurement of total and fPSA
NHP studies—pharmacokinetics of [89Zr]hu5A10
The study using cynomolgus macaques was approved by the IACUC of MSKCC (#14-03-006). Animals were maintained in accordance with the USDA Animal Welfare Act and Regulations and the Guide for the Care and Use of Laboratory Animals (24, 25). The animal care and use program at MSK is USDA registered, maintains Public Health Services Assurances, and is fully accredited by AAALAC International. Binding of hu5A10 to cynomolgus PSA was confirmed by time-resolved fluorescence measurements of fPSA and total PSA (23) in ejaculate collected from two monkeys and a purchased reference sample (Charles River) obtained by electroejaculation. The pharmacokinetics and distribution of [89Zr]hu5A10 was evaluated using a longitudinal PET/CT study over 2 weeks at the Weill Cornell Medicine—Citigroup Biomedical Imaging Center. Each animal was imaged dynamically for 60 minutes after injection, and subsequently at three more time points in the following ranges: 48–72, 120–144, and 216–312 hours. All details pertaining nonhuman primates (NHP) are described in the Supplementary Materials and Methods.
Analysis was performed using Prism 8.0 (GraphPad). All data are presented as mean ± SD, unless noted otherwise. Survival, TTN, and TTP were calculated using Kaplan–Meyer analysis. Statistical significance was calculated using one way-ANOVA with Bonferroni correction for multiple testing and a P ≤ 0.05 was considered to indicate statistically significant differences.
Macro- and micro-biodistribution of 90Y-, 225Ac-, and 89Zr-labeled hu5A10
Murine in vivo/ex vivo biodistribution studies
Targeting specificity of the radioconjugated hu5A10 was confirmed by coinjecting [89Zr]hu5A10 and a 20-fold molar excess of unlabeled hu5A10, which reduced accumulation of [89Zr]hu5A10 (mean ± SD: 7.9 ± 0.8 %IA/g blocked, 28.6 ± 11.8 %IA/g unblocked, P = 0.0018) in LNCaP-AR tumors (Supplementary Fig. S1A). Tumor uptake of [90Y]hu5A10, [225Ac]hu5A10, and [89Zr]hu5A10 in LNCaP-AR xenografts, in contrast to uptake in healthy organs, tended to increase over time (for [225Ac]hu5A10 4 hours vs. 120 or 350 hours, P ≤ 0.0115; all other P ≥ 0.8928; Supplementary Fig. S1B–S1D). Administration of the three different fPSA-targeted radioconjugates resulted in almost identical biodistributions; choice of chelate and radionuclide had negligible impact on tumor targeting and organ kinetics of hu5A10 (Fig. 1A). In addition, only minimal differences were noted when comparing [90Y]hu5A10, [225Ac]hu5A10, and [89Zr]hu5A10 in LNCaP-AR xenografts and KLK3_Hi-MYC GEMM (Fig. 1B). In the xenograft model, [90Y]hu5A10 displayed slightly longer blood retention (P = 0.2634), and significantly higher bone uptake (P = 0.0391) was noted for [89Zr]hu5A10.
hu5A10 biodistribution correlated with KLK3 expression levels; high KLK3-expressing MDAPCa2b tumors had a mean [89Zr]hu5A10 uptake of 49.8 ± 29.4 %IA/g compared with 26.2 ± 6.0 %IA/g in the LNCaP-AR xenograft (P = 0.0490; Supplementary Fig. S1E; ref. 26). In addition, uptake of [90Y]hu5A10, [225Ac]hu5A10, and [89Zr]hu5A10 in the in KLK3_Hi-MYC GEMM tended to be higher in ventral prostate lobes compared with the lower KLK3-expressing dorsolateral and anterior lobes (ventral vs. anterior lobes P ≤ 0.0047 for [225Ac]hu5A10, [89Zr]hu5A10; P ≥ 0.4546 for remainder; Fig. 1B).
To further confirm that [89Zr]hu5A10 can be used as an effective theranostic surrogate for [90Y]hu5A10 and [225Ac]hu5A10, mice with LNCaP-AR tumors and KLK3-Hi-Myc GEMM were imaged with [89Zr]hu5A10-PET/CT. The imaging data demonstrated the capacity of [89Zr]hu5A10 to quantitatively visualize hu5A10 biodistribution (Fig. 2; ref. 8).
Calculating absorbed doses for [90Y]hu5A10 and [225Ac]hu5A10 from the ex vivo biodistribution data showed that the tumor received the highest absorbed doses for both [90Y]hu5A10 (13.17Gy) and [225Ac]hu5A10 (9.3Gy), underlining the specific localization of the radiopharmaceuticals (Supplementary Table S1). The liver and the spleen were the nontargeted organs with the highest dose for animals treated with [90Y]hu5A10 (liver 5.36Gy, spleen 3.26 Gy) and [225Ac]hu5A10 (liver 1.98 Gy, spleen 2.77 Gy). The absorbed doses for [90Y]hu5A10 were higher than those calculated for [225Ac]hu5A10 despite the similar biodistribution profiles (Fig. 1A). This is compatible with the higher injected activity of [90Y]hu5A10 (18.5 MBq [150 μCi], vs. 11.1 kBq [300 nCi] [225Ac]hu5A10) tolerated in this model due to the big differences in LET.
Autoradiography and microscopy
We further verified the organ-scale uptake with autoradiographic exposures to visualize RIT agent localization. Autoradiography of whole mounted KLK3_Hi-MYC GEMM prostate sections showed correlations between regions of increased [89Zr]hu5A10 or [90Y]hu5A10 activity and lobes expressing high levels of PSA with little if any difference between the therapeutic and diagnostic radioconjugates (Supplementary Fig. S2). No uptake was noted in adjacent tissues lacking PSA expression, such as seminal vesicles and urethra.
As autoradiographic investigation is limited by phosphor-sheet composition resolution, we conducted high-resolution subcellular localization studies by preparing a fluorescent (Cy5.5) immunoconjugate for comprehensive immunofluorescent investigations of prostate tissue from KLK3_Hi-MYC GEMM following systemic administration of Cy5.5-hu5A10 or isotype control. Cy5.5-hu5A10 localized to the prostate epithelium and lumen of the glandular prostate ducts (Fig. 3A); the concentration of the fluorescently labeled tracer at these sites was specific, as control nonspecific human IgG1 antibody was not detected (Fig. 3B). These data confirm previous reports using antibodies targeting prostate kallikreins and are consistent with prior observations showing that FcRn facilitates transcytosis of unbound huIgG1 (hu5A10) from blood circulation to a glandular lumen, while huIgG1 bound to antigen (hu5A10:fPSA) is subjected to endosomal maturation before fusing with lysosomes (9, 10). Underlining the internalization of the fPSA-specific antibody conjugate (8), diffuse signal was also present in the seminal vesicles themselves. To ascertain luminal cell transcytosis of Cy5.5-hu5A10, we used similarly processed thicker tissue sections and imaged by confocal microscopy (Fig. 3C and D; Supplementary Fig. S3). Internalization of the Cy5.5-hu5A10 was observed as foci in luminal cells throughout the subregion of the dorsolateral prostate.
Treatment efficacy of low- and high-LET fPSA-targeted RIT
The efficacy of fPSA-targeted RIT with [90Y]hu5A10 (low-LET) and [225Ac]hu5A10 (high-LET) was examined in LNCaP-AR xenografts. Compared with the nontreated animals, both [90Y]hu5A10 and [225Ac]hu5A10 significantly increased disease control. The effects of [90Y]hu5A10 on tumor volume seemed to be more immediate than [225Ac]hu5A10 (Fig. 4A) with a median TTN of 7 days for [90Y]hu5A10-treated and of 38 days for [225Ac]hu5A10-treated mice (P < 0.0001; Fig. 4B). However, the treatment effects of [90Y]hu5A10 were less sustained than [225Ac]hu5A10 resulting in median TTP of 24 days ([90Y]hu5A10) versus 72.5 days ([225Ac]hu5A10; P < 0.0001; Fig. 4C). This difference entailed a median survival of 188 days with [225Ac]hu5A10, which was significantly longer than the survival of [90Y]hu5A10-treated animals (64 days, P = 0.0009; Fig. 4D). At the end of the study, 7 of 18 animals (38.9%) receiving the high-LET treatment exhibited complete responses with unpalpable tumor burden, whereas only 1 of 9 animals (11.1%) in the low-LET group presented with a similar outcome. However, both treatments conferred an overall survival benefit when compared with untreated controls with a median survival of 32 days (P ≤ 0.0054).
In vivo biodistribution and imaging studies in NHP
Binding to PSA in seminal plasma confirmed that hu5A10 can detect cynomolgus monkey PSA. Measured PSA levels in a reference sample of seminal fluid were 44.0 ± 8.66 ng/mL, while 4.25 ± 1.84 ng/mL and 28.8 ± 9.62 ng/mL fPSA were detected in samples from monkeys included in this study; the latter samples were of lower quality and suspected to having been subject to degradation in vitro. Despite these very low levels relative to man, we endeavored to investigate whether hu5A10 uptake could be visualized in a NHP model. [89Zr]hu5A10 (0.2 mg/kg; 115 ± 11 MBq) was administered to three adult, male cynomolgus macaques and imaged longitudinally by PET/CT over 2 weeks. No adverse events or physiologic responses were observed. All observed parameters (e.g., blood pressure, heart rate, body temperature) remained in the normal range and unchanged during injection and throughout the imaging procedure. All NHP subjects recovered normally from anesthesia on the day of the initial administration of [89Zr]hu5A10 and on subsequent imaging days. The pharmacokinetic behavior of [89Zr]hu5A10 was similar between the three NHP subjects studied (Fig. 5A–G; Supplementary Fig. S4) and the aggregate data were generally very consistent with the slow bulk tissue clearance observed in mice; in the absence of a tumor sink, the liver had the highest relative uptake in both species. [89Zr]hu5A10 rapidly entered the blood pool and distributed into most tissues within the first hour followed by a slow clearance over several weeks from most tissues. The NHP prostate gland was visualized clearly by [89Zr]hu5A10 PET/CT over the entire observation period with SUVmean reaching 5.9 ± 2.3 at 1 hour after injection and decreasing thereafter to SUVmean 4.5 ± 1.4 (2–3 days), 2.9 ± 1.2 (5–6 days), and 2.0 ± 0.6 (9–13 days). The time–activity curves (Fig. 5E) demonstrated significant uptake in epididymis, less so in testis and minimal uptake in the seminal vesicles. The early 1 hour pattern of uptake in the epididymis paralleled the blood pool and did not follow the slower uptake curve in prostate and testicles. Radioactivity transit into bladder and intestine was insignificant (SUVmean < 1); the majority of the off-target radioactivity residualized in liver and was not excreted. Spleen accumulation peaked at 30 minutes (SUVmean 7.5 ± 0.5) and dropped to SUVmean of 3.9 ± 0.4. Kidney uptake reached an early plateau (SUVmean 4.8 ± 0.5) several minutes after injection; the spike in the time–activity curve 30 seconds after injection was due to a significant spillover effect from the adjacent large vessels as the high initial input activity transits in close proximity (Supplementary Fig. S4). A measure of uptake in bone was determined in four lumbar vertebrae (L1–L4) which reached a SUVmean of 2.3 ± 0.7 on the day of administration, but stabilized at a low plateau of 1.2–1.3 SUVmean afterward. There was negligible accumulation in brain, lung, or salivary glands as anticipated, so these tissues were not included in the volume-of-interest analysis.
In this study, we present fPSA as a novel target for alpha- and beta-RIT in prostate cancer representing high and low LET therapy, respectively and introduce a viable companion diagnostic to report pharmacokinetic accumulation and clearance. Therapy with fPSA-targeted mAb hu5A10 IgG1 resulted in a significant reduction in tumor burden and improvement in survival with both alpha- and beta-particle emitting radionuclides. [89Zr]hu5A10 is a highly specific and complementary imaging agent for diagnosis and dose planning. Yttrium-90–, Actinium-225–, and Zirconium-89–labeled hu5A10 localized in a KLK3-dependent manner in murine prostate cancer models and primates ([89Zr]hu5A10) with similar pharmacokinetics and biodistribution profiles. Despite the observed similarity in radiotracer uptake, [225Ac]hu5A10 treatment prolonged progression-free and overall survival compared with [90Y]hu5A10 and offered larger overall reduction in tumor volume demonstrating important therapeutic differences between high- and low-LET therapeutic radionuclides. In addition, hu5A10 demonstrated accumulation in target cells as a result of FcRn-mediated internalization of these RIT complexes, increasing therapeutic benefit and preventing off-target tissue damage that may accompany alpha-radionuclide therapy.
Efforts to exploit biomarkers of prostate cancer for targeted radionuclide theranostics have led to many promising advances in management and treatment. However, congruent strategies for administering therapeutics and monitoring disease response are desperately needed in the field given the rates of prostate cancer recurrence and acquired resistance to conventional therapies (27, 28). Given the role of the AR pathway in tumor development and progression, prostate cancer treatments have focused on targets downstream of AR activity such as prostate-specific membrane antigen (PSMA) and six-transmembrane epithelial antigen of the prostate-1 (STEAP1) with varying success. Radionuclide therapies targeting PSMA have shown some successes in disease management. However, resistance to PSMA-targeted therapies might develop due to (therapy-induced) AR (re)activation (29) and subsequent loss of PSMA expression (30–33). Furthermore, alpha-particle therapies targeting PSMA are often hindered by dose-limiting toxicities (see below; ref. 34). Approaches directed to STEAP1, a membrane protein highly upregulated in multiple cancer types, seemingly demonstrated AR-dependent STEAP1 expression in the prostate (35, 36); however, studies investigating STEAP1 regulation have shown both AR-dependent and AR-independent regulation in various preclinical prostate cancer models, suggesting a more nuanced role for AR in regulating STEAP1 expression (37, 38). PSA has been primarily used as a prostate cancer biomarker in the blood, yet serum levels of total PSA do not reliably discern PSA from malignant versus healthy tissue nor provide reliable correlations with AR expression. Instead, targeting fPSA before it has been released and complexed in the blood is a unique approach and of significant clinical interest. The possibility to target fPSA with AR pathway–dependent localization while avoiding circulating PSA in the blood was initially reported by us using murine 5A10 (8). 5A10 selectively binds to PSA at sites of prostate and prostate cancer cells (and potentially the minor fraction of inactive fPSA). Subsequent modification of the Fc region of the mAb (9) enabled the tissue-specific internalization of hu5A10 paving the way for use of hu5A10 in RIT.
Internalization of a radioimmunoconjugate by the target cell is critical to the delivery of radionuclide therapies to ensure that decay of the radionuclide is on target (39). Given the short path length and high LET of alpha-particles, success of [225Ac]hu5A10 particularly depends upon internalization mechanisms; internalization of Actinium-225 also ensures that the radionuclide daughters remain in the cell and contribute to the therapeutic response (12). As shown by confocal microscopy, our data suggest that internalization and uptake of hu5A10 into the prostate epithelium is facilitated by IgG binding to FcRn. The FcRn mediates IgG recycling as well as transport and transcytosis across epithelial cells through IgG:FcRn binding in low pH conditions allowing hu5A10 to reach target tissues and avoid washout (10). FcRn is widely expressed on many tissues, which enables transient internalization and recycling of proteins that interact with the receptor in specialized tissue structures. FcRn-mediated internalization of the hu5A10 complex resulted in significant accumulation in tumor cells while healthy organs had minimal uptake, thus mitigating off-target effects and sparing healthy tissue. Other studies of alpha particle–based radionuclide therapies in prostate cancer have found dose-limiting accumulation of radionuclide in salivary glands or kidneys with unexpectedly higher uptake in healthy organs than beta-emitting RIT with the same biomarker (34, 40–42). While initial uptake in healthy tissue was noted at 4 hours after injection, clearance of [225Ac]hu5A10 from nontarget tissue occurred within 120 hours after injection resulting in high tumor-to-tissue ratios. Clearance from the liver was not as pronounced in the LNCaP-AR xenograft model indicating that hepatic toxicity could be a potential limitation of [225Ac]hu5A10. However, no overt toxicities were observed following radioimmunotherapy administration in any of the animal subjects despite the long circulating half-lives of the antibody constructs. The liver observations could be a result of the prevalent FcRn expression in the liver, a function critical for regulating albumin homeostasis (43).
The results with Actinium-225 and Yttrium-90 yield critical distinctions for prostate cancer RIT. While beta-emitting [90Y]hu5A10 had a more immediate effect on tumor volume, treatment with [225Ac]hu5A10 sustained tumor suppression and provided a significant increase in median survival time (Fig. 4). The faster response time seen in Yttrium-90 treatment could be attributed to the difference between the chosen radionuclides in half-life and path length. Yttrium-90 has a half-life of 2.5 days as opposed to Actinium-225 at 9.9 days and thus, could deliver the absorbed dose with a higher absorbed dose rate than Actinium-225 (42, 44). In addition, the 5 mm path length of Yttrium-90 (42) provided a larger immediate effect on the prostate cancer tumors given their average size of approximately 200 mm3 at treatment initiation. While the 50 μm path length of Actinium-225 may not have an immediate effect on tumor burden, the emission of four alpha-particles and two beta-particles upon decay and ability to cause double-stranded breaks (44) likely induced irreversible cell damage and prevented prostate cancer tumor progression. These findings are in agreement with the greater relative biological effectiveness of Actinium-225 compared with Yttrium-90 (42).
Previous work demonstrated a similar strategy to target hK2, an AR pathway-dependent peptidase closely related to PSA. Treatment with anti-hK2 [225Ac]hu11B6 in prostate cancer xenografts and Hi-Myc KLK2-expressing GEMM demonstrated alpha-particle damage-induced upregulation of AR and hK2 expression, leading to further uptake of the hK2-targeting radioimmunoconjugate (5). Double-strand DNA breaks by alpha-emitting [225Ac]hu5A10 may induce the same AR activation and increase in PSA expression, leading to its intratumoral accumulation and contributing to sustained treatment effects following [225Ac]hu5A10 RIT compared with [90Y]hu5A10.
Labeling the hu5A10 construct with the PET-tracer Zirconium-89 provided a strategy for noninvasive imaging of prostate cancer cells and complemented the PSA-targeted RIT described. Earlier work demonstrated transient localization of mouse monoclonal 5A10 in an AR-dependent manner to both prostate lesions and metastatic disease of prostatic origin (8). In this study, a transgenic KLK3-expressing Hi-Myc GEMM showed increased accumulation of [89Zr]hu5A10 in the ventral prostate lobes compared with adjacent dorsolateral and anterior prostate lobes, correlating with higher expression levels of PSA in this region. In addition, MDAPCa2b xenografts showed higher [89Zr]hu5A10 uptake correlating to the higher KLK3 expression in MDAPCa2b versus LNCaP-AR demonstrating the sensitivity of hu5A10 accumulation to reflect changes in PSA expression. Using the described radioimmunoconjugate, [89Zr]hu5A10 PET could be used to monitor changes in AR pathway activity.
Imaging studies in primates show that [89Zr]hu5A10 is well tolerated at the 0.2 mg/kg dose level and good image quality is obtained with 115 MBq (3 mCi) on a clinical PET/CT scanner. This activity was high enough to ensure that imaging up to 2 weeks after injection was feasible. Importantly, prostate visualization becomes clear within the first hour of imaging and the prostate is clearly discernable at all imaging time points over 2 weeks, which is clear from the transaxial imagery (Fig. 5A–D). The highest quality [89Zr]hu5A10 prostate scan (signal-to-background) was obtained 2–3 days after injection, which is within the first half-life of Zirconium-89. This suggests that long-term imaging is likely unnecessary and injected activity might be reduced without degrading diagnostic value. Our in situ assessments of PSA obtained in ejaculates confirmed previously reported findings that PSA levels in cynomolgus monkeys are approximately 5,000-fold lower than in humans (45, 46). In addition, cynomolgus monkeys express two alternative splice variants of PSA and only one of these products constitutes an intact catalytic triad. It is also unclear whether the minor amino acid sequence dissimilarity in the mature cynomolgus monkey PSA results in structural differences that affects enzymatic activity or affinity of hu5A10. Nevertheless, no adverse reactions were observed in the subjects following antibody administration.
Previously translated prostate cancer RIT such as 7E11 and J591 have suffered from low uptake in soft-tissue tumors and off-target binding in kidneys (3, 47). The target-specific accumulation of [89Zr]hu5A10 in NHP, despite the low levels of fPSA, is highly encouraging for future clinical translation to humans. Small-molecule ligands targeting PSMA are under clinical evaluation for prostate cancer detection and targeted therapy; these agents typically clear rapidly and show nonprostatic uptake in kidney and salivary glands (48). In contrast to these small molecules, our results also showed that hu5A10 has an extended blood circulation time, a feature shared by all IgG1 with intact FcRn binding. It is not unlikely that this will result in dose-limiting myelosuppression when utilizing hu5A10 labeled with beta-emitting radionuclides, such as Lutetium-177 and Yttrium-90. However, these side effects are commonly transient and can be successfully limited by fractionated dosing strategies or utilizing radionuclides emitting alpha-particles (22, 49). We also noticed uptake in epididymis, vas deferens and the seminal vesicles; measurable PSA levels in lysates of these organs have been reported in humans, albeit at very low levels in relation to prostate tissue (50). Because of the differences in prostate kallikrein biology and expression levels in humans and NHPs, the translational relevance of these findings is most likely insignificant. It is currently unknown whether the relative PSA expression in these organs compared with prostate are different in NHPs.
In summary, this study presents a complete preclinical evaluation in multiple rodent disease models and NHPs of fPSA-targeting hu5A10 IgG1. Our results show that this novel prostate cancer theranostic is highly applicable and effective for radioimmunotherapy and diagnostic PET imaging. The study further demonstrates the different therapeutic outcomes from beta- and alpha-emitting versions of the hu5A10 mAb; while utilization of the former radionuclide results in more instant therapeutic events, the latter shows slower but more protracted antitumor effects with higher chance of cure.
K. Lückerath reports personal fees from Sofie Biosciences outside the submitted work. U. Lamminmäki reports a patent for WO2017060247A1 pending. S.-E. Strand reports grants from Swedish Cancer Foundation and Swedish Research Council during the conduct of the study; in addition, S.-E. Strand had a patent for HUMANIZED ANTI PSA (5A10) ANTIBODIES issued to PCT/EP2016/073684. R. Damoiseaux reports a patent for Antibodies pending. H.I. Scher reports personal fees from Asterias Biotherapeutics, Bayer, Pfizer, Inc, Sun Pharmaceuticals Industries, Inc., WCG; nonfinancial support from Amgen, ESSA Pharma Inc, Janssen Research & Development, LLC, Janssen Biotech, Inc, and Menarini Silicon Biosystems; and grants from Epic Sciences, Illumina, Inc., Janssen, Menarini Silicon Biosystems, and Thermo Fisher Scientific outside the submitted work; in addition, H.I. Scher has a patent 10,736,972 issued and licensed to Elucida Oncology, a patent for 16/463,865 pending and licensed to Elucida Oncology, and a patent for 16/769,501 pending and licensed to Elucida Oncology. P. Scardino reports other from OPKO Biotech, Advantagene, and INsightec outside the submitted work. S.M. Larson reports grants from NIH during the conduct of the study, and grants from YMABS Therapeutics Inc and royalties from Elucida, SAMOs, and YMABS Therapeutic Inc outside the submitted work; in addition, S.M. Larson has several patents in the field of Radioimmunotherapy and Drug delivery pending, issued, licensed, and with royalties paid from YMABS Therapeutic; a patent for Nanoparticles issued, licensed, and with royalties paid from Elucida Inc; and a patent for radiotracer drugs from SAMOS; and reports consultation regarding drug products with Progenics, Janssen, and Exini (Lantheus) during the conduct of this work and preparation of article. H. Lilja is named on patents for intact PSA assays and a statistical method to detect prostate cancer (4KScore test) that has been commercialized by OPKO Health; receives royalties from sales of the test and has stock in OPKO Health; was a consultant to Diaprost AB and has stock options and stock in Diaprost AB; and received a speakers honorarium from Janssen R&D LLC. D.L.J. Thorek reports grants from NIH NCI (R0128335, R0128238, R0128539) during the conduct of the study and is scientific advisor for and has equity in Diaprost AB. D. Ulmert reports grants from Prostate Cancer Foundation and Department of Defense during the conduct of the study; in addition, D. Ulmert has a patent for EP2771688A4 issued to Diaprost, a patent for US20190381200A1 pending, and a patent for US20180326102A1 pending, and is a board member for Diaprost AB, the Swedish company that holds and has licensed patents on the humanized version of the antibody (hu5A10) utilized in the study. No disclosures were reported by the other authors.
D.R. Veach: Conceptualization, resources, data curation, formal analysis, methodology, writing–original draft. C.M. Storey: Resources, formal analysis, writing–original draft. K. Lückerath: Resources, formal analysis, writing–original draft. K. Braun: Data curation. C. von Bodman: Data curation. U. Lamminmäki: Methodology. T. Kalidindi: Data curation. S.-E. Strand: Data curation, formal analysis, writing–original draft. J. Strand: Writing–original draft. M. Altai: Writing–original draft. R. Damoiseaux: Writing–original draft. P. Zanzonico: Resources, data curation. N. Benabdallah: Formal analysis. D. Pankov: Data curation. H.I. Scher: Writing–original draft. P. Scardino: Data curation. S.M. Larson: Resources, formal analysis, supervision, funding acquisition, writing–original draft. H. Lilja: Resources, data curation, supervision, funding acquisition, methodology, writing–original draft. M.R. McDevitt: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, methodology, writing–original draft. D.L.J. Thorek: Conceptualization, Resources, Formal analysis, Methodology, Writing–original draft. D. Ulmert: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, methodology, writing–original draft, project administration, writing–review and editing.
We thank Muc Du, Simon Moritz, Ed Fung, Heather Martin (CBIC), and Brad Beattie, Daniel LaFontaine (MSKCC) for their excellent assistance with animal imaging/care, and data acquisition/processing, respectively.
This study was supported in part by the Imaging and Radiation Sciences Program, U.S. NIH grant P30 CA008748 (MSKCC Support Grant). The MSKCC Small-Animal Imaging Core Facility is supported in part by NIH grants P30 CA008748-48, S10 RR020892-01, S10 RR028889-01, and the Geoffrey Beene Cancer Research Center. We also acknowledge William H. Goodwin and Alice Goodwin and the Commonwealth Foundation for Cancer Research, the Experimental Therapeutics Center, and the Radiochemistry & Molecular Imaging Probe Core (P50-CA086438), all of MSKCC. M.R. McDevitt: NIH R01CA166078, R01CA55349, P30CA008748, P01CA33049, F31CA167863, the MSKCC for Molecular Imaging and Nanotechnology. D.L.J. Thorek: NCI R01CA201035, R01CA240711, and R01CA229893. H. Lilja, H.I. Scher: NIH/NCI CCSG to MSKCC (P30 CA008748). H.I. Scher: SPORE in Prostate Cancer (P50 CA092629). H. Lilja: Sidney Kimmel Center for Prostate and Urologic Cancers. David H. Koch Prostate Cancer Foundation Award, Swedish Cancer Society (CAN 2017/559), Swedish Research Council (VR-MH 2016-02974), General Hospital in Malmö Foundation for Combating Cancer. D.L.J. Thorek: NCI R01CA201035, R01CA240711, R01CA229893. D. Ulmert, M.R. McDevitt: DoD W81XWH-18-1-0223. D. Ulmert: UCLA SPORE in Prostate Cancer (P50 CA092131), JCCC Cancer support grant from NIH P30 CA016042 (PI: Teitell), Knut and Alice Wallenberg Foundation, Bertha Kamprad Foundation, David H. Koch Prostate Cancer Foundation Young Investigator Award. S.M. Larson: Ludwig Center for Cancer Immunotherapy (MSKCC), NCI P50-CA86438. S.-E. Strand: Swedish Cancer Society, Swedish National Health Foundation, Swedish Research Council.
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.