Quantitative In Vivo Imaging of the Androgen Receptor Axis Reveals Degree of Prostate Cancer Radiotherapy Response

Abstract

Non-invasive biomarkers for androgen receptor (AR) pathway activation are urgently needed to better monitor patient response to prostate cancer (PCa) therapies. AR is a critical driver and mediator of resistance of PCa but currently available non-invasive PCa biomarkers to monitor AR activity are discordant with downstream AR pathway activity. External beam radiotherapy (EBRT) remains a common treatment for all stages of PCa, and DNA damage induced by EBRT upregulates AR pathway activity to promote therapeutic resistance. [⁸⁹Zr]11B6-PET is a novel modality targeting prostate-specific protein human kallikrein 2 (hK2), which is a surrogate biomarker for AR activity. Here, we studied if [⁸⁹Zr]11B6-PET can accurately assess EBRT-induced AR activity.

Genetic and human PCa mouse models received EBRT (2–50 Gy) and treatment response was monitored by [⁸⁹Zr]11B6-PET/CT. Radiotracer uptake and expression of AR and AR target genes was quantified in resected tissue.

EBRT increased AR pathway activity and [⁸⁹Zr]11B6 uptake in LNCaP-AR and 22RV1 tumors. EBRT increased prostate-specific [⁸⁹Zr]11B6 uptake in PCa-bearing mice (Hi-Myc x Pb_KLK2) with no significant changes in uptake in healthy (Pb_KLK2) mice, and this correlated with hK2 protein levels.

Introduction

External beam radiotherapy (EBRT), a mainstay in PCa therapy, activates AR; this increases the expression of DNA repair genes, which may promote radio-resistance^1–4 and explain the synergy between ionizing radiation and AR signaling inhibitors (ARSI)^5,6. Non-invasive biomarkers for monitoring DNA damage-induced AR activity may allow monitoring response to EBRT and early detection of treatment resistance, and thus, providing PCa patients with individualized treatment options. In the clinical setting, AR activity is currently monitored through the assessment of serum prostate specific antigen (PSA, KLK3) levels over time⁷. However, measurements of serum kallikreins provide limited information as they reflect a global average of multiple heterogenic lesions in the metastatic setting with limited correlation to protein production⁸. The development of imaging-based response criteria (RECIST, PERCIST) reflects this challenge; the increasing application of RECIST/PERCIST for assessing response (to radionuclide therapies) - for example in the VISION trial, in which blood PSA levels were a secondary endpoint only – supports the relevance of these criteria. The capacity to decipher which lesions at which anatomical sites respond and which do not respond would significantly augment future patient management.

Similar to PSA, human kallikrein 2 (hK2; KLK2) is a prostate gland- and cancer cell-specific trypsin-like serine protease that is tightly governed by the functional status of the androgen receptor (AR) hormone response circuit, and is elevated in the serum of >20% of patients following EBRT⁹. We previously developed 11B6, an IgG1 antibody with high selectivity and specificity for the active cleavage site of hK2. 11B6 uniquely binds to hK2 directly at the cell surface – i.e., at the site of hK2 production and thus, AR activity - and avoids interaction with serum kallikreins. When derivatized with medically relevant radionuclides, this platform can be used for radio-immunotheranostics for detection, delineation, and treatment of diverse models of AR-expressing adenocarcinoma^10–12. Positron emission tomography (PET) with [⁸⁹Zr]11B6 enables monitoring of prostate cancer growth (PCa) and quantification of lesion-specific AR-activity^10–12.

We hypothesize that [⁸⁹Zr]11B6-PET can be used to noninvasively monitor EBRT-induced changes in AR-activity in individual PCa lesions. Using quantitative imaging and genomic analyses of human xenograft and genetically engineered mouse models of PCa, EBRT-induced AR activity was visualized and correlated to transcriptomic alterations following therapy with near-term implications for PCa treatment paradigms.

Materials and Methods

Radiochemistry

Radiosynthesis of [⁸⁹Zr]-DFO-11B6 ([⁸⁹Zr]11B6) has previously been described¹³. 11B6 antibody was provided by Dr. Kim Pettersson, University of Turku, Finland. All labeling reactions achieved >99% radiochemical purity. Average specific activity of the final radiolabeled conjugate was 51.8 MBq/mg (1.4 mCi/mg).

Cell lines

22Rv1 cells were purchased from ATCC. LNCaP-AR (LNCaP with overexpression of wildtype AR) was a kind gift from Charles Sawyers¹⁴. Cells were cultured according to the providers’ instructions and frequently tested for mycoplasma contamination. Cell lines were authenticated using GenePrint10 Short Tandem Repeat analysis (Laragen Inc, Culver City, CA)

Mouse models

All animal experiments were conducted in compliance with MSKCC guidelines, IACUC-established guidelines, and RARC animal protocol (# 04–01-002). Xenografts were established in male athymic BALB/c (nu/nu) mice (6–8 weeks old, 20–25 g; Charles River) by subcutaneous injection of LNCaP-AR or 22Rv1 cells (1–5×10⁶ cells, 1:1 = media : Matrigel). Tumors developed after 3–7 weeks. The transgenic PCa mouse models used, Hi-Myc × Pb_KLK2 with prostate-specific AR-driven hK2 expression, as well as Pb_KLK2 mice with abundant AR-driven hK2 expression specific to murine prostate tissue, have been previously reported¹⁰. Irradiated animals ranged from 35–42 weeks at study outset.

EBRT

Irradiation of disease sites was performed as previously described¹⁵. Briefly, a whole-body CT was acquired (XRad225Cx, Precision X-Ray, Inc.; dual focal spot x-ray tube at 45 kVp with a flat-panel amorphous silicon imager mounted on a C-arm gantry), tumor fields were identified and a treatment plan with >3 angles and a dose rate of $3 Gy/min (tube voltage, 225 kVp) was devised. Radiation dosimetry was performed using Gafchromic EBT film (ISP Inc.); a clear film that polymerizes with increasing optical density to a degree linearly with dose. The Gafchromic film verified the targeting accuracy, the magnitude of dose delivered and the geometry of the planned dose plan.

Magnetic resonance imaging

Prostate tumor volumes were defined using T2-weighted MR scans (Bruker BioSpin 4.7 T). An interleaved T2-weighted turbo spin echo sequence (3,200/57.1) with 8 averages was used, with slice dimensions of 8.5 × 3.99 × 0.8 cm. A total scan duration of 10  minutes 14 seconds generated 220 µm and 800 μm in and out of plane slices, respectively. A trained reader calculated prostate volumes by segmenting the prostate (OsiriX, v8.1)¹⁶.

Gene expression analysis

RNA was purified using the RNeasy Mini Kit (Qiagen), and quantitative PCR to determine expression of KLK2, KLK3, and FOLH1 was performed as previously described.

For RNA-sequencing, raw read count RNA-sequencing data were generated from untreated (NT; n = 3) LNCaP-AR tumor samples and 5 × 10 Gy (n=3) treated samples. A total of 58,828 genes were acquired and analyzed as previously reported¹⁷. Both hierarchical clustering analysis (based on Euclidean distance) and multi-dimensional scaling (MDS) plots demonstrated a clear division between the samples from the two cohorts (Suppl. Fig. 1, 2). Differentially expressed genes (DEGs) were defined at an adjusted p<0.001 and an absolute value of log2 fold-change >1. A positive fold-change represented up- and a negative fold change represented downregulation in EBRT-treated tumors. Pathway analysis was performed using enrichR¹⁸ and the KEGG 2021 database.

Bioluminescence imaging

Activity of the AR-dependent reporter construct expressed in LNCaP-AR tumors was quantified by bioluminescence imaging (Living Image^® 4.5.2) following retro-orbital injection of D-Luciferin (30 mg/mL, 10 µL; exposure times 1, 5, 10, 20, and 40 seconds). Data were expressed as radiance (photons/s) divided by tumor volume measured by caliper (V = length x width²).

Impact of EBRT on [⁸⁹Zr]11B6 tumor uptake

Mice bearing LNCaP-AR and 22Rv1 xenografts, and Hi-Myc × Pb_KLK2 and Pb_KLK2 mice, received [⁸⁹Zr]11B6 (3.7–5.55 MBq [100–150 μCi], 25 µg protein, i.v.; t=0 h), after EBRT (n=4–5/group). To confirm specificity, a control group of mice with 22Rv1 tumors treated with 4 × 5 Gy was co-injected with 1 mg of unlabeled 11B6. [⁸⁹Zr] radioactivity in tumors and organs harvested 120 h post-injection (p.i.) was quantified using a gamma-counter. Data were background and decay corrected, and the percentage injected activity per gram tissue (%IA/g) was calculated.

Monitoring AR-activity using PET/ CT

PET/CT imaging (Inveon MM, IRW Acquisition software) was performed as previously described¹⁹, at 120 h p.i. with Hi-Myc × Pb_KLK2 following administration of [⁸⁹Zr]11B6 (3.7–5.55 MBq [100–150 μCi], 25 µg of protein, i.v.). Duration of PET scans were $1 h or until 20 × 10⁶ coincident events were recorded. A 3D maximum a priori reconstruction was used to generate tomographic datasets. Assessment of hK2 expression for correlation with [⁸⁹Zr]11B6 uptake was reported previously¹⁷.

Histology

Prostate tissues of Hi-Myc × Pb_KLK2 and Pb_KLK2 mice harvested after EBRT (5 × 10 Gy) were fixed in 4% paraformaldehyde and cut into 15 µm sections before staining with hematoxylin and eosin (H&E). Immunohistochemistry (IHC) for detection of AR and c-MYC was performed at the Molecular Cytology Core Facility (MSKCC) using a Discovery XT processor (Ventana Medical Systems). Sections were blocked in 10% normal goat serum in PBS for 30 minutes before staining with an anti-AR (N-20) antibody (1 µg/mL, 3 h; Santa Cruz, #SC-816; secondary: biotinylated goat anti-rabbit IgG, 1:200, 16 minutes; Vector labs, #PK6101), or an anti-c-MYC antibody (1:100, 5h; Epitomics, #P01106; secondary: biotinylated goat anti-rabbit IgG, 1:200, 1 h; Vector labs, #PK6101). Blocker D, Streptavidin-HRP and DAB detection kit (Ventana Medical Systems) were used according to the manufacturer’s instructions.

Statistics

Statistical significance was determined by unpaired two-tailed t-test (2 groups) or, for >2 groups, by one-way ANOVA followed by Dunnett’s test to correct for multiple comparisons and set to p<0.05. Data are presented as mean ± standard deviation (SD). Analysis was performed with GraphPad Prism Version 9.2.0. For RNA-sequencing, differentially expressed genes were considered significant with an adjusted p<0.001 and log2 fold-change >1 as described previously¹⁷.

Data availability statement

The RNA-sequencing data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) database, https://www.ncbi.nlm.nih.gov/geo (accession no.GSE206847). Other data generated in this study are available upon request from the corresponding author.

Results

Changes in AR and AR-driven PCa biomarkers in response to EBRT

PCR analysis of LNCaP-AR tumors treated with 1, 3 or 5 fractions of 2, 5 or 10 Gy EBRT revealed dose-dependent increases in AR, KLK2, KLK3 compared to nontreated (NT) controls (Fig. 1, Table 1). FOLH1 expression after EBRT varied and remained unchanged under EBRT (Fig. 1B, Table 1). After 3 cycles of EBRT in 22Rv1 xenografts, AR gene expression was significantly increased along with KLK2 and KLK3, while there were no significant changes in FOLH1 expression (Fig. 1C, Table 1). The fold change of AR transcription was higher in 22Rv1 than LNCaP-AR tumors, which is likely an effect of lower baseline AR expression in the 22Rv1 model. This outcome corresponds with previously reported findings and provides additional support for the correlation between KLK2 and AR expression when monitoring changes rendered by EBRT⁹.

Figure 1. AR activity and gene expression after EBRT in LNCaP-AR and 22Rv1 xenografts.

A, Schematic of EBRT fractionation regimen. B,C, Gene expression analysis of LNCaP-AR (B) and 22Rv1 (C) xenografts after 1, 3, and 5 fractions of 2, 5 or 10 Gy EBRT revealed upregulation of AR and KLK2/KLK3 in a dose-dependent manner. Data were normalized to NT. See Table 1 for mean and p-values. D, Bioluminescence imaging readout of AR activity in LNCaP-AR xenografts after 1 or 4 fractions of EBRT revealed dose-dependent increase in AR activity independent of fractionation (all p=not significant vs. NT). Mean ± SD and individual values are given; statistical significance was calculated using one-way ANOVA and Dunnett’s test for multiple comparisons.

Table 1. Fold-change of AR and AR pathway genes in LNCaP-AR and 22Rv1 tumors after EBRT (vs. controls).

Mean ± SD are given; p-values (treatment vs. NT) are shown in parentheses and were calculated using one-way ANOVA and Dunnett’s test for multiple comparisons.

	1 Fraction	2 Gy	5 Gy	10 Gy
LNCaP-AR	Ar	1.8 ± 0.9 (0.1044)	2.2 ± 0.8 *(0.0128)	2.4 ± 1.0 **(0.0057)
	Klk2	1.5 ± 0.4 (0.4988)	2.1 ± 0.9 *(0.0248)	2.3 ± 1.3 *(0.0102)
	Klk3	1.5 ± 0.8 (0.6229)	2.1 ± 1.5 (0.0855)	2.6 ± 1.1 *(0.0183)
	Folh1	1.3 ± 0.4 (0.5323)	1.2 ± 0.6 (0.8127)	1.1 ± 0.8 (0.9315)
	3 Fractions	2 Gy	5 Gy	10 Gy
	Ar	1.4 ± 0.8 (0.6799)	2.7 ± 1.4 **(0.0022))	2.0 ± 0.4 (0.1640)
	Klk2	1.4 ± 1.0 (0.6169)	2.1 ± 0.9 *(0.0471)	2.1 ± 1.0 (0.0688)
	Klk3	1.9 ± 1.1 (0.1771)	2.1 ± 0.7 (0.0737)	2.8 ± 1.5 **(0.0052)
	Folh1	0.7 ± 0.7 (0.4646)	1.2 ± 0.7 (0.8600)	0.3 ± 0.1 (0.0615)
	5 Fractions	2 Gy	5 Gy	10 Gy
	Ar	2.4 ± 1.1 (0.0637)	3.4 ± 1.5 ***(0.0008))	4.4 ± 2.8 ****(<0.0001)
	Klk2	1.6 ± 0.6 (0.1540)	2.0 ± 1.1 **(0.0079)	2.2 ± 1.3 **(0.0025)
	Klk3	1.7 ± 0.8 (0.3805)	2.5 ± 1.6 *(0.0104)	2.7 ± 2.0 **(0.0028)
	Folh1	1.3 ± 0.6 (0.2652)	0.8 ± 0.4 (0.5133)	0.8 ± 0.7 (0.4302)
22Rv1	3 Fractions	2 Gy	5 Gy	10 Gy
	Ar	3.4 ± 2.2 (0.1296)	3.6 ± 2.2 (0.1180)	7.3 ± 5.0 ***(0.0001)
	Klk2	4.3 ± 2.5 ***(0.0007)	2.9 ± 1.3 (0.0894)	5.2 ± 2.3 ***(0.0002)
	Klk3	3.0 ± 1.8 (0.1732)	5.8 ± 3.2 ***(0.0005)	6.7 ± 3.5 ****(<0.0001)
	Folh1	1.9 ± 1.7 (0.4425)	1.6 ± 1.2 (0.7579)	2.6 ± 2.4 (0.1146)

Investigating ERBT-induced transcriptomic changes in an unbiased approach, 4,851 DEGs (8.2% of transcriptome gene set) were identified in LNCaP-AR tumors after EBRT (5 × 10 Gy; vs. NT); 2,552 genes were up- and 2,299 were downregulated (Fig. 2). Upregulation of AR-regulated genes such as AR signaling co-activator ETV1 ²⁰, KLK2, and KLK3 (log2 fold-change= 10.01, 1.033, 1.882) indicated that AR signaling was increased after EBRT. Interestingly, other AR target genes, including TMPRSS2 and FKBP5, were downregulated following treatment. Of the 144 previously established AR-associated DNA repair genes²¹, 18 were DEGs with 8/18 upregulated (CHEK1, FANCL, MAD2L1, MBM7, PARP1, RAD18, RAD21, RFC3)^21,22. FOLH1 was also upregulated despite its inverse correlation to AR pathway activity, contrasting qPCR findings. Upregulated MYC expression in EBRT-treated tumors supports a role for MYC in AR-driven EBRT responses, and pathway analysis showed that the top DEGs converged on cell cycle and regulation of DNA replication, both of which are closely intertwined with AR through cyclins and changes in protein expression during replication^23,24, further supporting a role for AR signaling in PCa response to EBRT.

Figure 2. EBRT-induced transcriptomic changes in LNCaP-AR xenografts.

A, Volcano plot showing 4,851 (8.24%) DEGs (FDR=0.01) following EBRT. B, Top 10 up- and downregulated genes (FDR=0.01).

EBRT increases AR activity in PCa in vivo

To confirm EBRT-induced AR signaling in vivo, activation of an AR-reporter gene in LNCaP-AR tumors was assessed using bioluminescence imaging. EBRT increased mean AR-activity without significant differences between 1 and 4 fractions (Fig. 1D).

[⁸⁹Zr]11B6-uptake is an indicator of EBRT-induced AR activity

[⁸⁹Zr]11B6 tissue uptake was assessed in 22Rv1 and LNCaP-AR tumors treated with 2, 5 or 10 Gy (1 or 4 fractions) EBRT or left untreated (Fig. 3). A total EBRT dose >10 Gy significantly increased uptake of [⁸⁹Zr]11B6 by LNCaP-AR tumors (38.61–47.24 %IA/g vs. 17.9%−28.3 %IA/g in NT) and 22Rv1 xenografts (13.2–62.6 %IA/g, vs. 7.9–11.2 %IA/g NT). Co-injection of cold 11B6 significantly decreased [⁸⁹Zr]11B6-uptake by 22Rv1 tumors after 20Gy EBRT (13.2–21.9 %IA/g vs 2.1–13.2 %IA/g blocked), confirming hK2 specificity (Fig. 3B).

Figure 3. [⁸⁹Zr]11B6 localizes to PCa after irradiation.

Ex vivo biodistribution of [⁸⁹Zr]11B6 in LNCaP-AR (A) and 22Rv1 (B) at 120h post-EBRT revealed higher uptake in irradiated tumors that received more than 8 Gy total dose of EBRT. Cold, unlabeled 11B6 confirmed specificity in 22Rv1. Mean ± SD and individual values are given; statistical significance was calculated for tumor uptake (NT vs. EBRT) using one-way ANOVA and Dunnett’s test for multiple comparisons.

EBRT-induced AR activity in PCa can be monitored by [⁸⁹Zr]11B6 positron emission tomography (PET) / computed tomography (CT) imaging

To confirm [⁸⁹Zr]11B6 uptake as a surrogate marker for EBRT-induced AR activity, [⁸⁹Zr]11B6 uptake was quantified in vivo and ex vivo in Pb_KLK2 (non-malignant) and Hi-Myc x Pb_KLK2 (PCa) mice after treatment with 5 fractions of 10 Gy. No significant volumetric changes were observed by MRI (Fig. 4A,B) after EBRT treatment of PCa tissue. EBRT increased AR expression in PCa (Hi-Myc x Pb-KLK2) (Fig. 4C); this was paralleled by significantly higher [⁸⁹Zr]11B6 uptake after EBRT in vivo (before EBRT, 11.04 ± 4.42%; after EBRT (same cohort of mice), 20.23 ± 4.28%).

Figure 4. EBRT treatment of Hi-Myc x Pb_KLK2 and Pb_KLK2 mice.

A, EBRT and imaging schedule for PCa (Hi-Myc x Pb_KLK2) and healthy (Pb_KLK2) mice. B, MR imaging revealed comparable PCa volumes ±50 Gy treatment. Mean ± SD and individual values are given; statistical significance was calculated using unpaired two-tailed t-test (p=0.5872). C, IHC of Hi-Myc x Pb-KLK2 tumors revealed increased intratumor AR and c-MYC expression after EBRT (magnification: overview 10x, insert 40x).

In contrast, EBRT did not impact uptake in Pb_KLK2 mice (Fig. 5A-C). Correlation of hK2 protein levels in tumors and [⁸⁹Zr]11B6-uptake further confirmed AR activity (Fig. 5D). Taken together, these results indicate that hK2-targeted [⁸⁹Zr]11B6 can noninvasively monitor increased AR signaling after radiotherapy in a Myc-driven model of PCa.

Figure 5. AR activity increase following EBRT visualized by [⁸⁹Zr]11B6-PET/CT.

Representative MR, IHC (40x magnification) and volume rendered PET/CT images before treatment (A) and after irradiation with 50 Gy (B) of a Hi-Myc x Pb_KLK2 mouse. White arrow indicates prostate location in MR images (scale: 0.5 cm). (C) Activity concentration of [⁸⁹Zr]11B6 increased following irradiation (p<0.05). Mean ± SD are given; statistical significance was calculated using unpaired two-tailed t-test. (d) PET signal from [⁸⁹Zr]11B6 corresponds with ex vivo hK2 expression.

Discussion

The current study demonstrates that EBRT-induced AR-activity, which increases in a dose-dependent manner, can be monitored noninvasively using PET. Activation of AR-signaling by EBRT may serve as prognostic biomarker and improve development of EBRT combination regimens. In a phase 3 clinical trial, the combination of EBRT with bicalutamide increased disease-free survival²⁵, and PSA decay rate during salvage radiotherapy has been identified as a predictor of progression-free survival²⁶. EBRT-induced AR-activity might thus negatively impact patient outcomes, and vice versa, inhibition of this response may improve patient care. Attempts to monitor AR noninvasively have been made with [¹⁸F]FDHT, a radio-analog of testosterone²⁷; however, [¹⁸F]FDHT reports AR levels rather than its functional signaling activity. To measure AR pathway activity, several AR target genes are utilized as biomarkers and therapeutic targets in PCa, including prostate-specific membrane antigen (PSMA) and PSA. Recently, FDA-approved PSMA-PET has increased the ability to detect metastatic PCa lesions and is considered as a strategy to monitor AR blockade by ADT. Unfortunately, preclinical and clinical studies demonstrated that PSMA-PET is not an optimal tool for assessment of ADT efficacy^28–31. We observed similar findings in our evaluation of PSMA levels after EBRT; FOLH1 expression increased 2.5-fold in 22Rv1 but not in LNCaP-AR xenografts. Taken together, these results underline the complex links between AR-activity, EBRT resistance, and AR pathway biomarkers. It should however be noted that resistance to EBRT may also be caused by non-AR driven mechanisms.

KLK2 expression and corresponding hK2 protein levels are well-established as biomarkers of AR pathway activity^9,10. In line with a previous study⁹, we showed that EBRT increases KLK2 expression in a dose-dependent manner. To noninvasively target KLK2 expressing cells, we developed 11B6, an antibody that specifically internalizes into PCa cells in response to AR-activity by binding uncomplexed hK2¹⁰. 11B6 can be exploited for PET, single photon emission tomography, intra-operative imaging^10,13, and radioimmunotherapy^{18,26,40–41}. Studies in multiple rodent models and non-human primates showed that [⁸⁹Zr]11B6 rapidly accumulates in PCa¹¹, and changes in PCa [⁸⁹Zr]11B6-uptake correspond to both AR-activity and hK2 protein levels¹⁰. We thus hypothesized that [⁸⁹Zr]11B6 could be used to monitor changes in AR-activity during and after EBRT. We confirmed relevance of [⁸⁹Zr]11B6-uptake as biomarker by correlating its tumor-uptake with EBRT-induced expression of the canonical AR biomarker KLK2. Furthermore, EBRT did not increase [⁸⁹Zr]11B6 prostate uptake in healthy Pb_KLK2 mice while uptake was significantly elevated in PCa of Hi-Myc x Pb_KLK2 mice; this suggests that EBRT-induced AR activation is a radiobiological response unique to malignant prostate tissues.

EBRT-induced AR activation exclusively in PCa-bearing mice as well as elevated MYC levels in xenografts and c-MYC expression in the genetic PCa model after EBRT support the known relationship between MYC and AR. MYC upregulation has been shown to antagonize AR signaling and AR target gene expression in patient samples³² but has been positively correlated to AR variant expression in another study³³. Upregulation of MYC may provide rationale for the use of co-treatment concepts using direct or indirect MYC inhibitors to block additional pro-tumorigenic transcription factors that drive PCa³⁴.

The difference in [⁸⁹Zr]11B6 uptake in the LNCaP-AR xenograft tumor model and the well-documented role of AR as a transcription factor led us to hypothesize that there would be a significant transcriptomic impact in the post EBRT-treatment setting. However, analysis of RNA-sequencing of irradiated mice revealed a downregulation of AR, highlighting the variability in tissue response to EBRT. This result exemplifies the need for diagnostic agents that focus on assessing functional AR pathway activity rather than the number of available receptors or AR expression itself. Upregulation of AR pathway target genes KLK2 and KLK3 in our data clearly demonstrate that the AR pathway is being differentially activated in tumor-bearing mice after radiotherapy.

The transcriptional EBRT-signature observed in the current study is in line with that reported for 11B6 alpha-radioimmunotherapy in Hi-Myc x Pb_KLK2 mice¹⁷. Comparison of the top ten up- and downregulated DEGs revealed five common up- (MMP7, ETV1, NTS, PLA2G2A, PEG3) and down-regulated DEGs (PASD1, DENN2D, PTGFR, SLC25A43, FAM213A); this similarity underscores the ability of [⁸⁹Zr]11B6-PET to reflect AR-driven therapeutic responses.

Overall, we demonstrated a highly specific and sensitive approach for noninvasive monitoring of functional AR-activity under EBRT. We propose that a baseline hK2-PET would be utilized at the time of dose planning, with repeated imaging following treatment start to monitor AR-signal. Exclusively in cancerous tissue, [⁸⁹Zr]11B6 tumor-uptake correlated with AR pathway activation after irradiation. Changes in [⁸⁹Zr]11B6 PCa-uptake paralleled increases in KLK2 and AR expression seen in qPCR analysis, as well as ex vivo hK2 protein concentrations and IHC staining. The significantly shorter circulation time of the 11B6 mAb construct in humans (and non-human primates) compared to mice supports feasibility of serial imaging³⁵ in patients.

Most patients receive co-treatment with pharmacological compounds inhibiting the AR-pathway. However, in some lesions the AR-pathway is still active to some degree due to insufficient dosing, and unknown resistance mechanisms and pathobiology. In this scenario, the molecularly specific and spatially defined signal of 11B6-PET may indicate areas that should receive increased radiation, or conversely, areas without (remaining) hK2-PET signal could be de-escalated to reduce risk of radiation-induced toxicity to surrounding tissues. The notion that rodents have about 50-fold lower testosterone levels than humans, rather on par with castrated patients, underlines the utility of this molecular, lesion-specific, cancer imaging strategy.

There is no doubt that the cost of imaging is a contentious issue in modern patient management. However, the cost of a PET scan (or multiple scans) relative to the treatment planning scans and daily radiation dosing is incremental – and the ability to capture personalized precision information of patient response (in particular AR-active lesions that may be outside of the prescribed dose field) is capable of driving significant long term cost savings. Therefore, hK2-PET/CT would facilitate a deeper understanding of response and resistance patterns of individual tumor lesions and patients and ultimately, of how to adjust treatment.

Thus, although clinical trials are needed, monitoring the AR-target gene hK2 in the treatment setting could allow patient stratification based on AR-pathway response, refinement of treatment and dosing strategies, e.g., by selection of AR-targeted treatment combinations and allowing physicians to prescribe the lowest dose needed, and may provide mechanistic insights into enhancement of EBRT in some patients with concurrent or adjuvant ARSI.

Implications.

hK2 expression in PCa tissue is a proxy of EBRT-induced AR activity that can noninvasively be detected using [⁸⁹Zr]11B6-PET; further clinical evaluation of hK2-PET for monitoring response and development of resistance to EBRT in real time is warranted.