Prostate cancer (PCa) is the second most commonly diagnosed cancer in men and the fifth leading cause of cancer death worldwide. While localized disease has a near-100% five-year survival rate, survival drops to approximately 30% once the cancer spreads to distant sites, making metastatic disease a major clinical challenge.
Since the 1940s, prostate cancer has been known to depend on androgens (male sex hormones, primarily testosterone and its potent form dihydrotestosterone, or DHT) for growth. This made androgen deprivation therapy (ADT), which reduces androgen levels or blocks their receptor, the cornerstone treatment for advanced disease. ADT includes surgical or chemical castration and newer drugs like enzalutamide and abiraterone.
Although ADT is initially effective in nearly all patients, resistance typically develops within 2-3 years, leading to the lethal form called castration-resistant prostate cancer (CRPC). CRPC is defined as disease progression despite testosterone levels below 50 ng/dL, the threshold considered castrate. Once this stage is reached, the cancer has found ways to keep growing without normal androgen stimulation.
Understanding how prostate cancer cells escape the effects of hormone therapy is critical for developing the next generation of treatments. This review comprehensively maps the molecular mechanisms through which the androgen receptor (AR) pathway persists, mutates, and adapts to drive CRPC.
The androgen receptor (AR) is a protein found inside cells that, when activated by testosterone or DHT, moves to the cell nucleus and acts as a master switch, turning on genes that drive prostate cell growth and survival. It belongs to the nuclear receptor family of transcription factors.
The AR protein has three major functional domains: the N-terminal domain (NTD), which influences how strongly the receptor activates target genes; the DNA-binding domain (DBD), which recognizes specific DNA sequences and ensures only the right genes are turned on; and the ligand-binding domain (LBD), located at the C-terminus, where testosterone and DHT physically bind to the receptor.
When androgen binds the LBD, the AR undergoes a conformational change: it releases its chaperone proteins, dimerizes with another AR molecule, and translocates to the nucleus. There it binds to androgen response elements in the DNA and activates target genes, including PSA (KLK3), which is used clinically to monitor prostate cancer.
AR expression patterns shift as prostate cancer progresses. In normal prostate development, AR is highly expressed in stromal cells but absent in epithelial cells. As cancer develops and advances, AR expression declines in stromal cells but persists and increases in epithelial tumor cells. In CRPC, AR is detectable in nearly all tumors regardless of disease stage, confirming that AR signaling remains active even after castration.
One of the most studied resistance mechanisms is AR gene mutation. Mutations are found in up to 60% of metastatic prostate cancers and become more frequent when patients receive AR antagonist drugs, suggesting the cancer is evolving under treatment pressure.
The most clinically important mutations occur in the ligand-binding domain and fall into two categories. Some mutations make the AR promiscuous, allowing it to be activated by non-androgen steroids like estrogens, glucocorticoids, or progesterone. Others convert AR antagonists (drugs designed to block the receptor) into agonists (molecules that activate it), a phenomenon called antagonist-to-agonist switching.
For example, the T878A mutation allows the AR to be activated by progesterone and estrogens, and converts the first-generation antiandrogen flutamide into an AR activator. The F877L mutation transforms enzalutamide and apalutamide into AR agonists. F877L is found in approximately 30% of metastatic CRPC cases. These mutations explain why some patients initially respond to a drug but then paradoxically improve when it is discontinued.
Newer drugs like darolutamide were specifically designed to maintain antagonist activity against a broad spectrum of AR mutations. In preclinical studies, darolutamide retained efficacy against most known resistance mutations, including F877L, W742L, and T878A, though clinical validation and co-occurring resistance mechanisms limit its advantage in practice.
Even without mutations, AR overexpression is one of the most common alterations in CRPC, found in 17-57% of pre-treated tumors. When cells produce more AR protein than normal, the receptor can be activated even by the trace amounts of androgen that remain after castration or drug treatment, bypassing the therapeutic effect of ADT.
AR overexpression most commonly results from AR gene amplification, where the cancer cell duplicates the AR gene region on chromosome X, producing far more AR protein. Structural rearrangements in the upstream enhancer region of the AR gene are also frequently observed in metastatic CRPC. CDK12 mutations, which impair DNA repair, can promote genomic instability leading to AR amplification.
Clinically, patients with AR gene amplification have significantly shorter progression-free survival compared to those without it, and they respond more poorly to enzalutamide than to abiraterone. AR overexpression also reduces PSMA expression, which is important because PSMA-targeted therapies like Lutetium-177 PSMA-617 depend on high PSMA levels for uptake. Patients with AR amplification were 2.4 times less likely to achieve PSA responses from PSMA-targeted therapy.
Beyond gene amplification, AR transcription can be upregulated through a feedback loop: when AR-targeted therapies block the receptor's normal self-suppression mechanism, AR transcription increases. Additionally, oncogenic signals like c-MYC overexpression can independently activate AR transcription even in low-androgen environments.
AR splice variants are abnormal forms of the androgen receptor produced when RNA processing errors cause parts of the AR gene to be skipped. The most important variant is AR-V7, which lacks the entire ligand-binding domain but retains the DNA-binding domain. Without an LBD, AR-V7 cannot be blocked by any anti-androgen drug that works by competing for the LBD, and it is constitutively active without requiring any androgen to function.
AR-V7 expression is significantly elevated in CRPC compared to hormone-sensitive prostate cancer and increases further after treatment with abiraterone or enzalutamide. It is associated with aggressive disease features including increased risk of biochemical recurrence after surgery, shorter overall survival, and resistance to both antiandrogens and taxane chemotherapy. AR-V7 upregulates 59 genes that drive CRPC progression.
Another splice variant, ARv567es, also lacks the LBD. These two variants have distinct transcriptional profiles. AR-V7-positive patients show greater resistance to taxanes, while AR-V7-negative but ARv567es-positive patients retain better taxane sensitivity, suggesting that testing for specific variants could guide chemotherapy decisions.
The enzyme AKR1C3 plays a dual role, both producing androgens locally within the tumor and physically stabilizing both full-length AR and AR-V7 protein, increasing c-MYC levels. Inhibiting AKR1C3 reduces AR-V7 levels and restores sensitivity to enzalutamide and abiraterone in preclinical models, making it a promising therapeutic target for overcoming splice variant-mediated resistance.
Beyond the AR itself, resistance can arise from changes in coregulator proteins, which modulate AR's ability to turn on target genes. More than 150 coregulators have been identified, including coactivators that enhance AR signaling and corepressors that dampen it. In CRPC, coactivators are commonly overexpressed while corepressors are lost or mutated.
The steroid receptor coactivators (SRC-1, SRC-2, SRC-3) are elevated even in early prostate cancer and become more prominent in CRPC. Pioneer factors like FOXA1, HOXB13, and GATA2 also enhance AR's ability to access and activate DNA in low-androgen environments. Conversely, loss of the corepressors NCOR1, NCOR2, and SPOP removes normal brakes on AR activity. SPOP normally degrades AR protein, so SPOP mutations lead to AR accumulation and enhanced signaling.
The glucocorticoid receptor (GR) represents a distinct AR-bypass mechanism. GR shares many transcriptional target genes with AR. After androgen deprivation, GR expression increases as a compensatory mechanism, activating the same growth genes that AR normally controls even without AR involvement. This GR upregulation under ADT helps explain why prostate cancer can continue growing despite AR-targeted therapy.
Lineage switching is the most dramatic resistance mechanism: under prolonged AR inhibition, approximately 25% of metastatic prostate cancers transform into a completely different cancer subtype called neuroendocrine prostate cancer (NEPC). NEPC cells do not express AR at all and do not respond to any hormone therapy. This transition, driven by loss of the tumor suppressors p53 and Rb, produces an extremely aggressive and therapy-resistant cancer.
Even with systematic castration that removes testicular testosterone production, prostate cancer cells can maintain AR activation by manufacturing their own androgens within the tumor. This is called intratumoral steroidogenesis. Tumor cells overexpress key steroidogenic enzymes to convert weak adrenal precursor steroids into active androgens locally.
The key enzyme CYP17A1 is central to this process. Abiraterone, one of the most important CRPC drugs, works by inhibiting CYP17A1 to block androgen synthesis in the adrenal glands, testes, and tumor tissue. However, abiraterone treatment causes accumulation of upstream steroidogenic precursors that can directly activate the AR through alternative pathways not blocked by abiraterone, eventually enabling resistance.
An alternative pathway discovered recently bypasses testosterone entirely: androstenedione is converted directly to 5alpha-androstanedione by 5alpha-reductase, which is then further processed to DHT without ever passing through testosterone. This pathway operates not just at the primary tumor but also at metastatic sites, meaning even distant spread can be fueled by locally produced DHT.
The potent AR antagonist RD162 and the enzyme inhibitor AKR1C3 inhibitors have shown preclinical efficacy in blocking this intratumoral production, and combining CYP17A1 inhibitors with next-generation antiandrogens is an active therapeutic strategy aimed at preventing AR activation from both endocrine and intratumoral androgen sources simultaneously.
CRPC is driven by multiple overlapping resistance mechanisms. A single patient's tumor may simultaneously harbor AR mutations, AR overexpression, splice variants, altered coregulators, and intratumoral steroidogenesis. This biological complexity explains why targeting any single mechanism typically produces only transient benefits before further resistance develops.
The key clinical insight from this review is that AR signaling persists across virtually all stages and sites of prostate cancer progression, making it an enduringly relevant target. However, the forms that AR takes in resistant disease differ dramatically from hormone-sensitive disease, requiring tailored strategies rather than simply intensifying existing therapies.
Several emerging therapeutic approaches are being developed based on these mechanisms: AR protein degraders, AKR1C3 inhibitors to block intratumoral steroid synthesis, combination strategies pairing AR inhibitors with GR inhibitors or taxane chemotherapy, and novel agents targeting FOXA1 or GATA2 coregulator activity.
Translating these molecular insights into clinical benefit requires better biomarkers that identify which resistance mechanisms are active in an individual patient's tumor. Prospective clinical trials testing resistance-specific therapeutic combinations, guided by liquid biopsy and tumor molecular profiling, represent the most promising path forward for improving outcomes in CRPC.