Prostate cancer is the second most common malignancy in men. Its growth, particularly in early and intermediate stages, is fundamentally dependent on male sex hormones called androgens, with testosterone and its more potent derivative dihydrotestosterone (DHT) being the primary drivers.
Androgens exert their effects through the Androgen Receptor (AR), a protein that sits in the cytoplasm of prostate cells. When testosterone or DHT binds to AR, the receptor undergoes a structural change, pairs with an identical receptor, and moves into the cell nucleus where it activates genes that promote cell survival and proliferation.
Androgen deprivation therapy (ADT) exploits this dependency by removing androgens from the bloodstream, which starves prostate cancer cells of the signal they need to grow. Early-stage prostate cancer responds well to ADT because the cells are predominantly androgen-dependent.
However, advanced prostate cancer becomes heterogeneous, containing a mix of androgen-dependent and androgen-independent cells. The development of castration resistance, where cancer continues growing despite low androgen levels, is driven by genetic changes in AR and activation of alternative signaling pathways that allow cancer cells to thrive without normal androgen signaling.
Two classes of drugs target androgen signaling at different points. Androgen synthesis inhibitors, primarily abiraterone, block the enzyme CYP17 that produces androgens in the testes, adrenal glands, and tumor cells themselves, suppressing androgen production more comprehensively than traditional castration.
Androgen receptor antagonists (ARAs) block the binding site on AR itself, preventing androgens from activating it. Second-generation ARAs including enzalutamide, apalutamide, and darolutamide are substantially more effective than older first-generation drugs because they block AR activity more completely and have no partial agonist activity that could paradoxically stimulate the receptor.
Major phase III clinical trials have demonstrated that these drugs significantly improve survival across multiple disease stages. For example, the LATITUDE trial showed abiraterone extended median overall survival to 53.3 months versus 36.5 months with ADT alone in newly diagnosed metastatic hormone-sensitive disease. The AFFIRM trial showed enzalutamide extended overall survival to 18.4 months versus 13.6 months in patients who had progressed after chemotherapy.
A critical clinical finding is that cross-resistance develops between drugs of the same class. When cancer progresses on abiraterone and is then treated with enzalutamide (or vice versa), responses are typically short-lived, with median time to PSA progression of only 2.8 to 3.9 months. This cross-resistance means that sequential use of different AR inhibitors provides limited additional benefit and new strategies are urgently needed.
One major resistance mechanism involves the production of AR splice variants (AR-V), which are abnormal, truncated versions of the AR protein. These variants lack the ligand-binding domain where AR antagonists bind and where abiraterone exerts its indirect effects. Without this domain, the truncated receptor can still activate cancer cell growth, but none of the available drugs can block it.
AR-V7 is the most clinically relevant splice variant, present in only 1% of early-stage cancers but rising to 75% of metastatic CRPC cases. The PROPHECY trial demonstrated that patients with detectable AR-V7 in circulating tumor cells had significantly shorter progression-free and overall survival on abiraterone or enzalutamide, with PSA responses seen in only 0 to 11% compared to 26 to 28% in AR-V7-negative patients.
Critically, AR-V7 does not predict resistance to taxane chemotherapy. In patients with AR-V7-positive tumors, chemotherapy with docetaxel or cabazitaxel appears more effective than continuing ARSI treatment. This creates a clinically actionable decision: detecting AR-V7 in circulating tumor cells can guide physicians to switch from hormonal agents to chemotherapy at an earlier point in treatment.
The AKR1C3 enzyme adds further complexity to splice variant resistance. This enzyme converts weak androgens into more potent testosterone and DHT while also stabilizing both AR-V7 and full-length AR. This creates cross-resistance to all four major ARSI drugs through a single enzyme. Indomethacin, an AKR1C3 inhibitor, is being evaluated in combination with enzalutamide as a potential strategy to reverse this resistance.
Point mutations in the ligand-binding domain of AR represent another major resistance mechanism. Rather than simply reducing the effectiveness of AR antagonists, some mutations actively convert them into agonists, drugs that bind to AR and stimulate rather than block it. The result is that a treatment meant to shut off cancer growth instead becomes a growth signal.
The F877L mutation is the most clinically relevant example. This mutation converts enzalutamide and apalutamide from potent antagonists into partial agonists of AR. CTC DNA analysis in patients with resistance to these drugs has confirmed the presence of this mutation in some cases, and the double mutant F877L/T878A makes enzalutamide an even stronger partial agonist.
Different mutations confer resistance to different drugs, and this creates both a challenge and an opportunity. Darolutamide has been shown in laboratory studies to retain antagonistic activity against a broader range of clinically relevant AR mutations compared to enzalutamide, including mutations that convert enzalutamide into an agonist. However, whether this laboratory advantage translates into better clinical outcomes remains uncertain.
Novel approaches targeting AR from a completely different angle are in development. PROTACs (proteolysis-targeting chimeric molecules) such as ARV-110 and ARCC-4 do not block AR but instead tag it for destruction by the cell's own protein degradation machinery. This approach can target not only wild-type AR but also amplified AR and multiple mutant forms, potentially bypassing resistance mutations that neutralize conventional antagonists.
AR overexpression, meaning an increase in the total amount of AR protein in cancer cells, occurs in approximately 80% of castration-resistant prostate cancer cases. When AR is overexpressed, even the very low androgen levels present after castration may be sufficient to activate enough receptors to drive tumor growth, bypassing the purpose of androgen deprivation.
AR gene amplification, where the cell acquires multiple copies of the AR gene, accounts for approximately 30% of cases of overexpression and is significantly more common in CRPC than in hormone-sensitive disease. AR amplification is particularly associated with progression after enzalutamide and carries a worse prognosis for both progression-free and overall survival.
AR amplification has an important interaction with a newer class of treatments: PSMA-targeted radioligand therapies such as 177Lu-PSMA-617. AR inhibition normally upregulates PSMA expression, making cancer cells better targets for PSMA-based therapies. However, AR amplification reverses this relationship by downregulating PSMA, making cells less visible and less sensitive to these treatments. Patients with AR gain were 2.4 times less likely to respond to PSMA-ligand therapy.
An unconventional strategy to address AR overexpression is bipolar androgen therapy (BAT), which alternates between near-castrate testosterone levels and episodic supraphysiologic doses. The rationale is that flooding cells with high testosterone can paradoxically downregulate AR and resensitize cancer to subsequent ADT. The TRANSFORMER trial showed that BAT followed by enzalutamide produced better progression-free survival (median 28.2 months) compared to enzalutamide followed by BAT (19.6 months).
When AR is blocked effectively by modern drugs, prostate cancer cells can activate entirely different molecular pathways to sustain survival. The PI3K/AKT/mTOR pathway, which controls cell growth and survival, is altered in approximately 60% of CRPC cases through PTEN deletion or PIK3CA mutations. In preclinical models with PTEN loss, the AR and PI3K/AKT pathways reinforce each other: blocking one activates the other through reciprocal feedback. Dual inhibition of both pathways with AKT inhibitors combined with ARSI is showing promising results in patients with PTEN loss or AKT pathway mutations.
The Wnt-beta-catenin pathway, which regulates cell fate decisions, was found to harbor activating mutations in 11% of mCRPC patients who developed resistance to enzalutamide or abiraterone. These mutations, in genes including CTNNB1 and APC, conferred shorter overall survival. Wnt signaling also drives angiogenesis through downstream upregulation of VEGF, providing yet another route for tumor progression independent of androgens.
The glucocorticoid receptor (GR) is structurally related to AR and shares many of the same DNA binding sites. GR expression increases dramatically after ADT treatment, and in some patients, activated GR effectively substitutes for AR by regulating the same cancer-promoting genes. This GR-bypass mechanism is a documented mechanism of enzalutamide resistance, and combinations of AR inhibitors with GR antagonists are under investigation.
DNA repair pathway alterations, particularly mutations in BRCA1, BRCA2, and related homologous recombination repair genes, present in approximately 15 to 27% of mCRPC patients, create a separate vulnerability. PARP inhibitors such as olaparib and rucaparib exploit this weakness and have received FDA approval for BRCA-mutated CRPC. When combined with abiraterone, olaparib improved radiographic progression-free survival from 8.2 to 13.8 months even in patients without confirmed HRR mutations.
One of the most dramatic resistance mechanisms is treatment-related neuroendocrine differentiation, where prostate cancer cells under sustained hormonal pressure lose their identity as prostate gland cells and transform into a completely different cell type: neuroendocrine cells. These cells do not depend on AR signaling and are therefore inherently resistant to all androgen-targeting therapies.
This transformation occurs in approximately 17% of patients who progress after ADT, and the transformed tumors behave like small cell cancer with a substantially worse prognosis: median overall survival of 36.6 months compared to 44.5 months for conventional adenocarcinoma. The transformation is driven by loss of tumor suppressors TP53 and RB1, which normally prevent this cell fate switch.
The oncoproteins N-Myc and Aurora kinase A drive androgen-independent progression through lineage plasticity, enabling cells to adopt neuroendocrine characteristics. Aurora kinase A inhibitors have been investigated in neuroendocrine prostate cancer and showed responses in patients with MYCN and AURKA gene amplification, though overall clinical benefit has been limited.
Epigenetic regulators, specifically EZH2 (a histone methyltransferase) and bromodomain proteins, play key roles in neuroendocrine transformation and AR gene regulation. EZH2 inhibitors and BET bromodomain inhibitors are being evaluated to prevent or reverse neuroendocrine differentiation and resensitize resistant tumors to hormonal therapy, with early-phase trials showing modest PSA responses particularly in AR-V7-negative patients.
A consistent theme across all resistance mechanisms is that multiple alterations typically coexist in each patient's tumor. A single patient may simultaneously have AR amplification, an AR point mutation, activation of the PI3K pathway, and partial neuroendocrine differentiation. This polyclonal resistance landscape means that drugs designed to target one mechanism will rarely produce durable responses on their own.
The cross-resistance observed when patients are treated sequentially with different ARSI drugs reflects this reality. When cancer has already evolved to survive under hormonal pressure, it has often activated multiple backup pathways simultaneously. Sequential single-agent approaches simply apply further selective pressure, accelerating the outgrowth of already-present resistant clones.
The most promising clinical results are emerging from rational combination strategies that simultaneously target two complementary pathways. PARP inhibitors with AR inhibitors are showing particular promise in DNA-repair-deficient tumors. AKT inhibitors combined with AR inhibitors show benefit specifically in PTEN-loss tumors. These combinations are most effective when patients are selected by the specific molecular alteration the combination targets.
The future of prostate cancer treatment will require comprehensive molecular profiling of each patient's tumor at multiple timepoints to identify which resistance mechanisms are active, followed by dynamic treatment adaptation. Liquid biopsy, which detects cancer DNA or cells in the bloodstream, makes serial profiling feasible and allows treatment decisions to be updated as the tumor evolves without requiring repeated invasive biopsies.