Clinical implications of AR alterations in advanced prostate cancer: A multi-institutional collaboration

Background: AR gene alterations can develop in response to pressure of testosterone suppression and androgen receptor targeting agents (ARTA). Despite this, the relevance of these gene alterations in the context of ARTA treatment and clinical outcomes remains unclear. Methods: Patients with castration-resistant prostate cancer (CRPC) who had undergone genomic testing and received ARTA treatment were identified in the Prostate Cancer Precision Medicine Multi-Institutional Collaborative Effort (PROMISE) database. Patients were stratified according to the timing of genomic testing relative to the first ARTA treatment (pre-/post-ARTA). Clinical outcomes such as time to progression, PSA response, and overall survival were compared based on alteration types. Results: In total, 540 CRPC patients who received ARTA and had tissue-based (n=321) and/or blood-based (n=244) genomic sequencing were identified. Median age was 62 years (range 39–90) at the time of the diagnosis. Majority were White (72.2%) and had metastatic disease (92.6%) at the time of the first ARTA treatment. Pre-ARTA genomic testing was available in 24.8% of the patients, and AR mutations and amplifications were observed in 8.2% and 13.1% of the patients, respectively. Further, time to progression was longer in patients with AR amplifications (25.7 months) compared to those without an AR alteration (9.6 months; p=0.03). In the post-ARTA group (n=406), AR mutations and AR amplifications were observed in 18.5% and 35.7% of the patients, respectively. The most common mutation in post-ARTA group was L702H (9.9%). Conclusion: To our knowledge, this is the largest real-world clinicogenomics database-driven study exploring the development of ARalterations and their association with ARTA treatment outcomes. Our study showed that AR amplifications are associated with longer time to progression on first ARTA treatment. Further prospective studies are needed to optimize therapeutic strategies for patients with AR alterations.


Introduction
Prostate cancer is the most commonly diagnosed non-cutaneous cancer among men in the United States, with an estimated incidence of 268,000 cases annually.(1) Suppression of the androgen signaling pathway has been the mainstay of treatment of prostate cancer since the landmark discovery by Huggins and Hodges of the androgen dependence of prostate cancer in 1941. (2) In the setting of advanced or metastatic disease, treatment is focused on reducing androgen levels with surgical orchiectomy or medical castration and blocking androgen receptor activation. (3,4) While multiple treatment agents have been developed, outcomes differ widely, and validated biomarkers for treatment response or resistance are needed to help guide treatment selection.
The growing ease and affordability of genomic pro ling have enabled more prostate cancer patients to undergo blood-based and/or tissue-based genomic sequencing. Approval of agents such as olaparib, the rst biomarker selected therapy for prostate cancer, has catapulted us into the modern era of broad-scale sequencing for all men with advanced disease.(5) These approaches have, in turn, provided a foundation for a body of literature seeking to identify actionable alterations and predictive genomic signatures. (6,7) The most commonly altered genes in primary prostate cancer are ETS fusions, PTEN, SPOP, TP53, and FOXA1.(8) Furthermore, in the metastatic castration resistant disease setting, alterations in AR, TP53, MYC, ZMYM3, APC, and RB1 are reported to be signi cantly enriched compared to primary prostate cancer. (9) Although the predictive role of many of these genomic alterations is still being explored, there is heterogeneity in interpretating the clinical signi cance of AR gene alterations in patients with advanced disease.
The AR gene encodes a nuclear receptor that is activated following the binding of androgenic hormones and regulates the transcription of several growth and differentiation-related genes required for the survival of prostatic cells.(10) Indeed, the majority of prostate cancers are driven by the alterations in ARsignaling. (4,11) Affected AR gene domain types, including mutations in the ligand binding domain, gene body and/or enhancer ampli cations, or intronic structural rearrangements resulting in alternative splicing, can yield distinct biological characteristics and could potentially underlie the differential treatment response observed with AR targeted agents (ARTA).
AR gene ampli cations primarily occur in response to androgen deprivation therapy and are observed in up to 50% of patients with castration-resistant prostate cancer.(12) AR ampli cations have been associated with resistance to enzalutamide and abiraterone acetate treatment.(13-16) Among nonmetastatic prostate cancer patients, AR gene anomalies, while less common, have been associated with poorer clinical outcomes (17).
Several AR mutations have been implicated in therapeutic resistance to ARTA, often through AR activation by glucocorticoids and other sex hormones (e.g. L702H, H875Y) or gain of function mutations that confer AR agonist activity to AR antagonists (e.g. T878A, H875, F877L, W742C). (12) , (15) Discrepancies in the current literature, as well as the con icting results from various genomic analyses, further complicates effective, evidence-based treatment selection among patients with prostate cancer. Given the existing knowledge gaps in the clinical relevance of AR alterations in relation to ARTA use, we embarked on our current study. In this study by utilizing a multi-institutional clinical and genomic database, we examined the association between clinical outcomes and AR alteration types, as well as the time of occurrence of alterations relative to the systemic treatment.

Study Design and Patient Selection
A retrospective analysis was conducted using the Prostate Cancer Precision Medicine Multi-Institutional Collaborative Effort (PROMISE) database, which includes deidenti ed clinical and genomic data from patients with advanced prostate cancer (metastatic hormone sensitive or castration resistant) treated at 18 academic centers.(18) Patients had germline and somatic genomic testing (tissue, blood and/or germline) through CLIA certi ed commercially available platforms during routine clinical care. Detailed information of the assays can be found in supplementary table 1. Data was collected from registered institutions between 4/1/2020 and 7/7/2021 using a standardized RedCap database. This study was approved by local institutional review boards at participating sites per institutional policy and the Declaration of Helsinki.
For the purpose of this study, we included castration resistant prostate cancer (CRPC) patients at the time of the rst ARTA treatment and had available somatic genomic testing performed on blood or tissue ( Figure 1). Only samples that were collected after androgen deprivation therapy exposure were included in this study. Subjects not meeting these criteria or for whom these details could not be veri ed were excluded. Patients were then categorized as pre-ARTA or post-ARTA depending on whether the genomic sequencing sample was acquired before or after exposure to ARTA. Eligible patients received any ARTA, including abiraterone, apalutamide, darolutamide or enzalutamide. Gonadotropin-releasing hormone agonist/antagonist therapy and rst-generation antiandrogens were not classi ed as ARTAs in this study.
Cases were individually reviewed by a genomics expert in a multi-step quality control process. The RedCap database was queried for all patients marked as having an AR alteration. In a second step, to ensure all cases had been identi ed, the free text eld containing the raw next generation sequencing (NGS) report (excluding protected health information) was separately queried to identify any unmarked AR alterations. At each institution at least 10% of the entries were quality controlled by a physician with a genomic expertise. Patients who had insu cient tumor for analysis or incomplete/missing NGS reports were excluded. Entries needing clari cation were agged and returned to the site for query resolution. After this two-stage process, a second genomics expert at the central site was consulted regarding any unclear entries.

Outcome Measures
Data regarding patient demographic characteristics, PSA values, treatment types, genomic pro le and assay type were extracted from the database. In patients with more than one genomic testing, only the rst somatic genomic testing was included for the analysis.
For patients in the pre-ARTA NGS group, the outcome measures of interest were PSA decline of ≥50%, time to progression (TTP), and overall survival (OS). Further, these outcome measures were compared based on AR alteration status (no alteration vs mutation or ampli cation). In the post-ARTA NGS group, the frequency of speci c AR alterations (L702H, T878A, H875Y, W742Y) was evaluated for their association with prior treatment exposures by the time of genomic testing and OS. TTP was de ned as the time from the initiation of ARTA to treatment discontinuation due to clinical progression or censored at the time of the last documented follow-up. OS was calculated from the time of initiation of ARTA to death or censored at the time of the last follow-up.

Statistical Methods
Descriptive statistics of patient characteristics and outcomes were described using proportions for categorical variables and median and interquartile range for continuous variables. Median TTP during the rst ARTA treatment and median OS were estimated using the Kaplan-Meier method. The log-rank test was used to assess the differences in TTP and OS across subgroups, as de ned by AR alteration status.

Patient Characteristics
In total, 540 patients with CRPC who received an ARTA and had tissue (n=321) and/or blood (n=244) sample collection for genomic testing following androgen deprivation therapy exposure were identi ed using the PROMISE database.  1-10). AR mutations and ampli cations were seen in 18.5% (n=75) and 35.7% (n=145) of the patients, respectively and 14 patients (3.4%) had both types of alterations. The most common AR mutations in the post-ARTA group were L702H (9.9%, n=40), followed by T878A (5.9%, n=24), regardless of prior treatment type. The most commonly received treatments among patients with T787A, W743C, L702H, and H875Y were abiraterone and enzalutamide (Table 3). W743C and L702H mutations were more commonly seen after one line of ARTA, whereas the other mutations showed greater prevalence after exposure to a second ARTA. AR mutations in the post-ARTA group were observed at similar rates in patients with prior docetaxel exposure (0.14%) compared to those without prior exposure (19.9%; p=0.29).
Overall survival for patients with AR alterations detected post-ARTA is shown in Figure 2

Discussion
Alterations in AR gene have long been noted as drivers in the progression of prostate cancer to its castration-resistant state. The increased availability of next generation sequencing and the advent of liquid biopsy has led to several publications outlining evolution of AR alterations during the course of prostate cancer treatment (19,20). However, many studies have lacked granular clinical data in order to clearly de ne the association of AR alterations with speci c treatment outcomes. Genomic reports often indicate whether bene t or lack of bene t is expected with various ARTA in the context of AR mutations or ampli cation, yet there are not enough data to clearly determine this, which is re ected in the con icting recommendations seen in reports from different next generation sequencing providers. The PROMISE database, which included 1329 prostate cancer patients who had genomic testing data linked to detailed clinical treatment information, provides a unique opportunity to add new data to this area of uncertainty.
Prior studies evaluating the impact of AR mutations on treatment outcomes have been small and/or included patients with prior ARTA exposure, since this is the group in which these mutations are most frequently seen. In our dataset we were able to identify 11 patients with AR mutations and 27 patients with AR ampli cations that arose prior to any ARTA exposure. Our ndings yielded the novel insight that AR ampli cations are associated with longer time to clinical progression on rst ARTA treatment compared to patients without an AR alteration. Whereas, in patients with AR mutation median TTP was similar to those without an AR alteration. Similar to ours, in one study which only a single genomic testing platform was used, median TTP in pre-ARTA mCRPC patients with ligand binding domain mutations was 6.2 months. (21) Although it should be noted that our sample size is small, and these results might be driven by some outliers in our dataset.
Several studies have demonstrated that patients with AR ampli cations had worse clinical outcomes with ARTA treatment, however the majority of these studies included a mixed cohort of patients who were exposed to prior ARTA. (13,14,(22)(23)(24) Further, in these studies, liquid biopsy was used to determine AR alterations. Since AR copy number gain is correlated with higher cell-free DNA levels which are often associated with higher tumor volume, liquid biopsy can create a confounding effect for the prognostic value of AR ampli cations. (25) Similarly in our study, liquid biopsy was used in 40% of the patients and might be creating a similar confounding effect in this group. In another study, Jayaram and colleagues looked at a cohort from PCR2023 study (ClinicalTrials.gov identi er: NCT01867710) which included 133 baseline liquid biopsy samples of treatment naïve CRPC patients who were randomized to receive abiraterone acetate and one of four different glucocorticoid regimens. (25) Patients with AR copy number above ≥1.92 (n=22) had shorter PFS and OS compared to those with AR copy number of <1.92. Although the increased copy number gain in this study was associated with worse clinical outcomes, authors did not use patients without an AR alteration as the comparator arm, making it hard to interpret the outcomes for these groups.
AR mutations remain infrequent among treatment naïve prostate cancer patients but are detected in almost 20% of those with castration resistant disease, and in up to 40% of patients who have received ARTA.(12) There did not appear to be a unique genomic landscape among patients who develop AR alterations at castration resistance without prior ARTA compared to those who develop AR alterations after exposure to ARTA. It should be also noted that, in the current landscape where ARTA is being used in the metastatic hormone sensitive prostate cancer, we may see earlier emergence of these AR alterations and impact of these on other types of treatments used for mCRPC has not yet been explored. Similar to published literature, in our study cohort with castration resistant prostate cancer patients we found an increased rate of AR mutation frequency among those who received ARTA (19%) compared to those who did not (8%) as well as co-occurrence of AR ampli cation together with mutation which was not seen pre-ARTA exposure. In line with previous studies, the most common mutation in post-ARTA genomic reports in our study was L702H, which occurred with similar rates post abiraterone or enzalutamide. (14) In previous studies, this mutation was shown to be associated with resistance to abiraterone, and further androgen receptor signaling was shown to be activated by prednisone or progesterone in the presence of this mutation (15,26,27). Although the numbers were small in our cohort, L702H was associated with higher 2year OS compared to H875Y and W742C, which may imply better behavior or greater responsiveness to subsequent lines of therapy. Nevertheless, future drug development to target L702H will be critical since it is the most common, and the current AR degraders such as ARV-110 may be less effective in this subset.
Strengths of this analysis include having multiple commercial assays in a wide geographic range of prostate cancer patients being treated with standard therapies as opposed to restricting analysis to patients from one center and using one assay. This may increase the clinical applicability. Furthermore, genomic review by an expert ensured high quality data is the most critical aspect of this precision medicine effort.
The current study represents real-world practice, and as such, sample collection sites, genomic testing platforms and their associated methodologies were heterogenous. The sample size was limited by not all patients having a genomic testing following ARTA treatment. As a result, temporal and spatial heterogeneity, as well as differences in assay methodologies, lack of transcriptomic, and splice variant analysis may have limited the ability to identify all relevant alterations. Additionally, imaging practices were not uniform throughout this retrospective database, thus making the use of validated endpoints, including radiographic progression-free survival, impossible to ascertain. Despite this, we were able to examine overall survival among our study cohort along with clinically relevant endpoints such as PSA decline and time to treatment change, providing meaningful insight into disease progression. Future research efforts in this domain should include serial testing and enhanced inclusion of patients treated with speci c therapies, as well as increasing the number of patients registered in databases such as PROMISE.

Conclusions
The current study utilized the PROMISE database to gain novel insight into the impact of AR ampli cations and mutations on treatment response to ARTA. These ndings highlight the need for prospective data or further rigorous database analyses to truly de ne whether treatment decision making should be impacted by detection of AR alteration on next-generation sequencing. Such steps will advance the potential for personalized medicine, and access to next generation sequencing platforms in advanced prostate cancer will remain central to its success.

Declarations
Compliance with Ethical Standards: This research did not receive any speci c grant from funding agencies in the public, commercial, or not-for-pro t sectors.     cTime to progression, which was de ned as initiation of rst ARTA to treatment discontinuation