20 years of taxane therapy in prostate cancer — the past, present and future

Abstract

2024 marked the twentieth anniversary of FDA approval for taxane therapy, namely docetaxel, in the treatment of advanced prostate cancer. Approval of this treatment modality signified a breakthrough in the field, as it demonstrated for the first time that a non-antiandrogen targeting agent could significantly improve survival of patients with prostate cancer. New evidence has further solidified the role of taxanes, demonstrating that it improves survival when used in combination with androgen-targeting therapy in early-stage metastatic hormone-sensitive prostate cancer. Consequently, taxane therapy is currently considered a gold standard and mainstay in the treatment of prostate cancer. In this Review, we present an up-to-date analysis of the pivotal role of taxanes in the treatment of prostate cancer, discuss reasons why taxanes are effective and discuss how the mechanisms that confer resistance to taxanes in prostate cancer might be exploited to develop more effective therapeutic strategies.

Introduction

Prostate cancer is the second most commonly diagnosed cancer worldwide and the fifth leading cause of cancer death among men¹. Cases of prostate cancer that are localized to the prostate are usually curable when treated with surgery or radiation. When prostate cancer disseminates to distant organs, however, the disease becomes lethal; in this situation, several systemic therapeutic strategies are used, including androgen-deprivation therapy (ADT), chemotherapy, targeted therapy and immunotherapy.

Taxanes such as docetaxel² and paclitaxel³, which are semi-synthetic derivatives from Taxus brevifolia and Taxus baccata trees, respectively, are widely used chemotherapy agents in multiple tumour types. Indeed, in the early 2000s docetaxel revolutionized the therapeutic landscape of prostate cancer by demonstrating for the first time that a non-androgen targeting agent could improve survival of patients with advanced prostate cancer.

The TAX327 and SWOG9916 trials^4,5 marked a turning point in the management of metastatic castration-resistant prostate cancer (mCRPC) by establishing docetaxel as the preferred first-line, disease-modifying therapy, leading to its approval by the FDA in 2004 (Fig. 1). Prior to these studies, treatment of prostate cancer in the setting of resistance to ADT was limited to use of the cytotoxic agent mitoxantrone (a topoisomerase II inhibitor)⁶, radiotherapy for bone pain caused by metastases⁷ and corticosteroids⁸. Although these treatments provided palliative benefits, particularly in terms of pain reduction and improved quality of life, none demonstrated a survival advantage in patients with CRPC.

Fig. 1 |. Timeline of milestone taxane clinical trials in prostate cancer.

Timeline of taxane (docetaxel and cabazitaxel) clinical trials in prostate cancer, showing the year of trial completion, treatment and disease stage. ADT, androgen-deprivation therapy; mCRPC, metastatic castration-resistant prostate cancer; mHSPC, metastatic hormone-sensitive prostate cancer.

Following impressive results from phase II studies showing that docetaxel had promising antitumour activity, the TAX 327 study published in 2004 randomly assigned patients with mCRPC to prednisone along with one of two docetaxel dosing schedules (either 75 mg/m² every 3 weeks or 30 mg/m² weekly for 5 of every 6 weeks) or prednisone plus mitoxantrone (12 mg/m² every 3 weeks)⁴. High-dose docetaxel (75 mg/m² every 3 weeks) was associated with improved overall survival (the primary end point) compared with mitoxantrone, and was also associated with improved rates of response in terms of serum PSA level and pain control (secondary endpoints). Also published in 2004, following success of the phase II trial CALGB9780 (ref. 9), which investigated use of docetaxel alongside estramustine and prednisone, the phase III trial SWOG9916 compared docetaxel (60 mg/m² every 3 weeks) given alongside estramustine with mitoxantrone (12 mg/m² every 3 weeks) given alongside prednisone in men with metastatic CRPC⁵. Docetaxel plus estramustine was associated with a significant overall survival benefit compared with mitoxantrone plus prednisone (17.5 months versus 15.6 months; hazard ratio (HR) for death = 0.80, 95% CI 0.67–0.97; P = 0.02]. Preclinical studies had previously shown that estramustine, a nitrogen mustard–oestrogen hybrid, synergized with docetaxel by promoting microtubule depolymerization and inhibiting P-glycoprotein^10,11. However, owing to the notable cardiovascular toxicity associated with estramustine¹² and the findings of the parallel TAX 327 trial demonstrating the efficacy of docetaxel alone, docetaxel plus prednisone became the new standard of care, offering a superior risk–benefit profile to other options. Of note, the addition of zoledronic acid (a bisphosphonate used to treat and prevent bone-related problems) to docetaxel in patients with mCRPC with radiological evidence of bone metastases was found to decrease skeletal-related events and received FDA approval in 2002 (ref. 13); these findings were later corroborated by the TRAPEZE clinical trial, published in 2016 (ref. 14).

Given the established efficacy of docetaxel in mCRPC, clinical trials that followed were aimed at testing its efficacy at earlier disease stages. In 2013, the GETUG-AFU15 trial, which stratified patients with metastatic hormone-sensitive prostate cancer (mHSPC) by metastatic burden (using CHAARTED definitions), demonstrated improvements in progression-free survival (PFS) with ADT plus docetaxel versus ADT alone but did not achieve a statistically significant overall survival benefit in high-volume disease¹⁵. In 2015, the CHAARTED study showed that adding docetaxel to ADT improved overall survival (57.6 versus 47.2 months, HR = 0.72, 95% CI 0.59–0.89) in patients with mHSPC, with the survival benefit most pronounced in patients with high-volume disease (51.2 in those who received ADT plus docetaxel versus 34.4 months in those who received ADT alone, HR = 0.63, 95% CI 0.50–0.79), and delayed progression to castration-resistant disease (20.2 versus 11.7 months, HR = 0.61, 95% CI 0.51–0.72)^16,17. Also, in 2018, results from the STAMPEDE trial, a multi-arm, multistage study, reinforced the overall survival benefit of docetaxel plus ADT in mHSPC, demonstrating a significant survival advantage over ADT alone (81 months versus 71 months, HR = 0.78, 95% CI 0.66–0.93)¹⁸. Importantly, STAMPEDE did not find evidence that the benefit of docetaxel was restricted to high-volume disease, suggesting that its efficacy extends across the broader metastatic population, or even in the non-metastatic setting¹⁸. Indeed, these results have been further supported by the PEACE-1 (ref. 19) and ARASENS²⁰ trials, which demonstrated that concomitant administration of ADT with docetaxel, abiraterone or darolutamide significantly improves survival of patients with mHSPC. Together, these findings resulted in chemo-hormonal therapy being established as a standard approach in patients with mHSPC.

For patients with mCRPC progressing after docetaxel, options were limited until 2010, when the TROPIC trial demonstrated that cabazitaxel (25 mg/m² given every 3 weeks) plus prednisone was associated with significantly improved overall survival compared with mitoxantrone plus prednisone (15.1 months versus 12.7 months; HR = 0.70, 95% CI 0.59–0.83), and was also associated with improved PSA response and PFS²¹. Cabazitaxel is a semi-synthetic taxane derived from natural precursors that has been engineered to overcome the resistance mechanisms seen with conventional taxanes, offering superior antitumour activity in preclinical models of taxane resistance²². In 2017, the PROSELICA trial demonstrated that a lower dose of cabazitaxel (20 mg/m² every 3 weeks) was non-inferior in terms of overall survival to the standard dose, with a better safety profile in terms of grade 3–4 neutropenia²³. Importantly, the randomized phase III trial FIRSTANA²⁴, which compared cabazitaxel (20 mg/m² or 25 mg/m²) with docetaxel (75 mg/m²) administered every 3 weeks to chemotherapy-naive patients with mCRPC demonstrated that these two taxanes showed similar clinical efficacy, but had distinct toxicity profiles, which could help to prioritize either docetaxel or cabazitaxel administration based on patients’ comorbidities and anticipated toxicities. The use of granulocyte colony-stimulating factor (G-CSF) in patients receiving chemotherapy can provide a strategy to further reduce the risk of chemotherapy-induced febrile neutropenia, and can help to optimize tolerability, although regional variability in its use has been observed, with G-CSF often underutilized in regimens associated with a high risk of febrile neutropenia^25,26.

Consequently, despite evidence that cabazitaxel has comparable efficacy and a more favourable toxicity profile at lower doses than docetaxel, cabazitaxel remains predominantly recommended in the post-docetaxel setting alongside other agents such as abiraterone or enzalutamide. Indeed, the CARD trial showed that among patients with mCRPC who had been previously treated with docetaxel and an androgen-signalling–targeted agent (abiraterone or enzalutamide), cabazitaxel monotherapy was superior to giving the other androgen-signalling inhibitor (abiraterone or enzalutamide), with significant benefits in the primary outcome of radiographic PFS (8 months versus 3.7 months; HR = 0.54, 95% CI 0.40–0.73) as well as in the secondary outcomes of overall survival, PSA response rate and pain reduction²⁷. Moreover, the CABADOC trial evaluated patient preference between cabazitaxel and docetaxel in chemotherapy-naive patients with CRPC, and found that 43% of patients preferred cabazitaxel, citing less fatigue, better quality of life, and fewer adverse effects, whereas 27% preferred docetaxel and 30% expressed no preference²⁸. Finally, in addition to standard dosing schedules, alternative regimens such as biweekly docetaxel or low-dose docetaxel strategies, as well as cabazitaxel 16 mg/m² every 2 weeks, have been explored to improve tolerability while maintaining clinical benefit^29,30.

In contrast to the stark success of combining taxanes with anti-androgen therapy, studies investigating the synergistic combination of taxanes with other targeted agents have largely been disappointing. Trials have failed to demonstrate survival benefits with the addition of targeted agents and often reported increased toxicity in experimental arms. Notable examples of these disappointing results include CALGB 90401 (docetaxel plus bevacizumab), MAINSAIL (docetaxel plus lenalidomide), ENTHUSE (docetaxel plus zibotentan) and VENICE (docetaxel plus aflibercept)^31–34. These negative findings highlight the importance of obtaining strong preclinical and phase II evidence before advancing to large-scale phase III studies to ensure that combination strategies are both biologically rational and clinically viable.

In the following sections, we examine the mechanisms of action of taxanes (Fig. 2) and the biological pathways of resistance that consequently arise (Fig. 3), focusing on potential strategies to overcome resistance and enhance therapeutic efficacy.

Fig. 2 |. Primary mechanism of action of taxanes.

Taxanes bind β-tubulin and block microtubule dynamic instability, which leads to prolonged mitotic arrest and/or mitotic errors, ultimately resulting in cancer cell death. MTOC, microtubule organizing centre.

Fig. 3 |. Taxane resistance mechanisms.

Taxane therapy resistance includes alterations in tubulin subunits, canonical genetic drivers, apoptosis drug efflux pumps, developmental and undifferentiated phenotypes, chromosomal stability and mitotic fidelity, metabolic reprogramming and nucleocytoplasmic transport. DDR, DNA damage repair; GLS1, glutaminase 1; MCL-1, myeloid cell leukemia 1; NF-κB, nuclear factor-κB; ROS, reactive oxygen species.

Mechanisms of action of taxane agents

Actions on tubulin

Microtubules are hollow cylindrical polymers with a diameter of approximately 22 nm, typically composed of 13 protofilaments assembled around a hollow core. The cylindrical polymers are assembled from heterodimers of α-tubulin and β-tubulin and exhibit intrinsic polarity: the minus end, enriched in α-tubulin, is anchored at the centrosome (the microtubule-organizing centre (MTOC)), whereas the plus end, exposing β-tubulin, is the site of dynamic growth and shrinkage (Fig. 2). This dynamic instability, governed by the regulated addition and removal of tubulin dimers at the plus end, is essential for vital cellular processes such as division, intracellular transport and motility^35,36. Upon incorporation into microtubules, the GTP bound to β-tubulin is hydrolysed^37,38, triggering a conformational change that favours depolymerization³⁹. The delay between dimer addition and GTP hydrolysis creates a stabilizing cap of GTP-bound β-tubulin at the growing end^35,40. The regulation of these processes is further refined by a host of microtubule-associated proteins (MAPs), including polymerases and depolymerases that sculpt microtubules along their lengths^41,42, molecular motors that move cargo along the microtubules^41,43,44, microtubule plus-end-tracking proteins (+TIPs) that provide stability and favour polymerization at the growing end^45–47, and other structural proteins that do not have specific enzymatic actions but that provide stability, such as MAP2 (ref. 48) or tau^49,50.

Taxanes bind with high affinity to β-tubulin within assembled microtubules. Their binding occurs specifically within the lumen of the microtubule, between the M-loop (connecting S7 and S9) and the S9–S10 loop. For example, paclitaxel binds to a hydrophobic cleft on β-tubulin, a site that has been precisely mapped by electron crystallography and photoaffinity labelling to encompass amino acid residues 217–233 (ref. 51). Docetaxel shares the same binding site as paclitaxel but exhibits an even greater affinity, thereby enhancing its microtubule-stabilizing effect. By binding to and stabilizing microtubules, taxanes suppress dynamic instability, which is essential for microtubule disassembly⁵² and is crucial for cell processes such as mitosis and cell division. In particular, taxanes hamper remodelling of the cytoskeleton during cell division. The binding of taxanes to microtubules stabilizes the microtubules and leads to disruption of the mitotic spindle, leading to aberrant chromosome attachment and segregation^53,54. Disruption of the mitotic spindle leads to activation of the spindle assembly checkpoint (SAC), which in turn induces prolonged mitotic arrest, which can culminate in either apoptosis or mitotic slippage — a process that can result in chromosome missegregation, formation of micronuclei, subsequent genomic instability and cell death^55,56.

Alterations in β-tubulin structure, occurring as a result of either changes in isoform expression or point mutations in tubulin genes, have a pivotal role in mediating resistance to taxanes. Multiple β-tubulin isoforms exist and are characterized by differences in their C-terminal tails and post-translational modifications; these isoforms thus have different affinities with MAPs and have distinct roles in microtubule stability^57–59. In particular, the class III β-tubulin isoform (βIII-tubulin) promotes microtubule plasticity through regulation of microtubule dynamics^60,61, which is associated with paclitaxel resistance⁶². Indeed, paclitaxel preferentially binds to βI-tubulin rather than the βIII isoform, in which the Ser277Ala substitution disrupts key hydrogen bond interactions surrounding the M-loop⁶³. In the clinical setting, overexpression of βIII-tubulin occurs in aggressive tumours after androgen ablation as well as after exposure to docetaxel treatment⁶⁴. In turn, high βIII isoform expression has been reported to be an independent predictor of reduced survival^65,66, although it is unclear if high βIII isoform expression is linked to a poorer response to therapy.

Despite extensive research on the binding site of taxanes to tubulin, whether mutations in β-tubulin can lead to taxane resistance is still unclear. For example, a docetaxel-resistant cell line (LNDCr) that was developed was found to have an acquired Phe270Ile mutation in class I β-tubulin⁶⁷, a mutation that has also been reported as a predictor of paclitaxel resistance in ovarian cancer^68,69. Although mutations in the predominant class I β-tubulin subunit can arise rapidly with paclitaxel treatment, their clinical relevance remains under debate^70,71. A systematic review of 149 clinically observed missense mutations in β-tubulin highlighted that taxane resistance might result from loss of non-bonded interactions, steric hindrance or other conformational changes⁷². A particular Thr274Ile mutation disrupts the M-loop, being particularly noteworthy as it parallels the architectural disruption observed in the βIII-tubulin isoform⁷³.

Pro-apoptotic effect

Although taxane-induced mitotic arrest is a critical event, the ultimate fate of the cell depends on taxanes engaging the activation of pro-apoptotic pathways (or inhibiting pro-survival signals). One such mechanism is the activation of the intrinsic apoptotic pathway, which occurs in response to DNA damage or cell-cycle deregulation, and involves increased mitochondrial membrane permeability and release of cytochrome C, which eventually activates downstream effector caspases that seal the death of the cell through apoptosis⁷⁴.

A safeguard of this response is the anti-apoptotic protein BCL-2 (refs. 75–77). Although BCL-2 is expressed at low levels in normal prostate epithelium, it is frequently overexpressed during prostate-cancer progression^78,79. Taxane treatment induces phosphorylation of BCL-2 on key serine and threonine residues by kinases such as CDK and Raf-1, which impairs the anti-apoptotic function of BCL-2 and promotes cell death⁸⁰. Downregulation of the anti-apoptotic BCL-2 is a mechanism by which cells become sensitive to taxanes^81–83, whereas transcriptomic analysis of taxane-resistant mCRPC tumours showed upregulation of BCL-2 (ref. 84). Despite these findings establishing BCL-2 as a key player in taxane-mediated resistance, efforts to enhance docetaxel therapy through direct inhibition of BCL-2, using agents such as oblimersen or AT101, have not improved outcomes in phase II clinical trials^85,86.

The expression of other anti-apoptotic proteins, such as BCL-X, myeloid cell leukemia 1 (MCL-1) and clusterin, is also reportedly increased in aggressive prostate cancer^81,87. However, inhibition of clusterin using custirsen (OGX-011) in combination with cabazitaxel and prednisone has failed to show survival benefit⁸⁸ despite promising preclinical data⁸⁷. More data are needed to establish why targeting anti-apoptotic pathways has failed to overcome taxane resistance in the clinical setting. It is plausible that, in the setting of taxane exposure, cellular rewiring⁸⁹ might upregulate other compensatory pathways that mediate resistance to taxanes despite inhibition of the anti-apoptotic pathways.

Canonical prostate-cancer drivers and resistance to taxane therapy

Androgen-receptor signalling

Many preclinical studies have focused on the role of taxane therapy and androgen-receptor dynamics in the prostate-cancer cell, with data suggesting that paclitaxel can impair androgen-receptor localization to the nucleus by either a direct interaction with tubulin or as an indirect consequence of mitotic catastrophe (a form of irreversible cell-growth arrest resulting from defective mitosis)^90–93. It is important to note that paclitaxel is not approved for mCRPC and that the concentrations used in these studies far surpass those used at a clinical level⁹⁴; as such, whether the prevention of androgen-receptor translocation is achievable in the clinic is unknown^18,95,96. When docetaxel was tested at 1 nmol/l, these findings on androgen-receptor translocation were not reproducible in CRPC cell-line C4–2 cells⁹⁷. By contrast, a study examined circulating tumour cells (CTCs) obtained from patients diagnosed with mCRPC treated with docetaxel and found that androgen-receptor subcellular localization was associated with taxane-resistance status, showing that impaired androgen-receptor nuclear translocation was associated with increased sensitivity to docetaxel⁹⁰.

Aside from its role in impairing androgen-receptor transport, there are data suggesting that taxanes contribute to downregulation of androgen-receptor levels in vitro⁹⁸. Other studies have validated these findings, showing that both docetaxel treatment and paclitaxel treatment lead to decreased expression of androgen-receptor target genes such as KLK3 (which encodes PSA) and NKX3.1, while increasing expression of androgen-receptor-repressed genes such as maspin⁹⁹. Paclitaxel treatment also leads to upregulation of FOXO1 protein and promotes its nuclear translocation, where it has a role as an androgen-receptor repressor⁹⁹.

These mechanistic findings highlight two key clinical questions. First, is there synergy between androgen-signalling inhibitors and taxanes, and if so, what is the optimal combination or sequence? And second, do androgen-signalling inhibitors and taxanes share any mechanisms of resistance? In one retrospective cohort¹⁰⁰, only 26% of patients who received docetaxel after abiraterone achieved a >50% PSA decline and median overall survival was 12.5 months, results that are inferior to those observed in the TAX-327 trial, which showed a 45% rate of >50% PSA decline and a median overall survival of 18.9 months without the use of androgen-signalling inhibitors. On the other hand, enzalutamide can still improve survival in patients with mCRPC that has progressed after docetaxel, as seen in the phase III AFFIRM clinical trial¹⁰¹, indicating that AR signalling remains a driver in mCRPC, even with prior taxane therapy.

Furthermore, cabazitaxel might act through independent mechanisms that bypass the taxane–androgen receptor cross-resistance interplay. In this regard, a study showed that treatment with cabazitaxel led to a reduction of cell viability in both enzalutamide-sensitive and enzalutamide-resistant cells, as well in androgen-receptor-negative PC3 cells, even though cabazitaxel did not seem to block the nuclear localization of the androgen receptor¹⁰². The same study showed that a high dose of cabazitaxel (administered at 25 mg/m² every 3 weeks) in patients with mCRPC whose disease had progressed on docetaxel and abiraterone had an overall median survival of 10.9 months and a >50% PSA decline occurred in 35% of patients¹⁰². Findings from the TROPIC phase III clinical trial also suggest that cabazitaxel is efficacious in patients whose disease has progressed during or after docetaxel treatment²¹.

The CARD trial²⁷ enrolled 255 patients with mCRPC who had previously received docetaxel and then progressed within 12 months on an androgen-signalling inhibitor (abiraterone or enzalutamide). Cabazitaxel significantly improved imaging-based PFS (HR = 0.54, 95% CI 0.40–0.73; P < 0.001) and overall survival (HR = 0.64, 95% CI 0.46–0.89; P = 0.008) compared with the use of the other androgen-signalling inhibitor.

The optimal sequencing of taxanes and androgen-signalling inhibitors is still under exploration¹⁰³. In mHSPC, stratification of patients based on prior docetaxel use showed that the addition of enzalutamide (ENZAMET trial¹⁰⁴; AARCHES trial¹⁰⁵) or apalutamide (TITAN¹⁰⁶) led to increased time to progression, but without a clear overall survival advantage. On the other hand, two phase III clinical trials evaluating the addition of abiraterone (PEACE-1 (ref. 19)) or darolutamide (ARASENS²⁰) concomitantly to ADT + docetaxel therapy showed improved overall survival in mHSPC. It is not immediately clear why some triplet studies, such as PEACE-1 and ARASENS, were positive while many doublet strategies failed, although this likely reflects the biological synergy achieved by targeting both hormone-dependent and hormone-independent pathways from the outset. By contrast, sequential or incomplete doublet regimens might allow time for clonal adaptation and cellular rewiring, diminishing long-term efficacy. Indeed, a 2022 meta-analysis confirmed that although sequential therapy with androgen-signalling inhibitors after docetaxel did not result in an improvement of overall survival, the risk of death was reduced by 28% when androgen-signalling inhibitors were used concomitantly with docetaxel¹⁰⁷. Thus, the combination of ADT, anti-androgen therapy and docetaxel is currently the gold-standard treatment strategy for patients with mHSPC who are fit enough to tolerate triplet therapy. However, despite level 1 evidence from these trials, analysis of real-world data reflects a clear underutilization of triple therapy in routine clinical practice, possibly owing to concerns of a ‘one-size-fits-all’ approach, overtreatment-related toxicity and a historical preference for doublet regimens^108,109.

Androgen-receptor variants

The expression of androgen-receptor variants, such as androgen receptor splice variant 7 (AR-V7) and ARv567, can influence the cellular response to taxanes and dictate clinical outcome. AR-V7 arises from mRNA splicing in which exons 4 to 8 are lost and a cryptic exon is included^110,111. ARv567, which was identified in a panel of 25 different prostate-cancer xenografts, is generated by the alternative splicing and skipping of exons 5 to 7 (ref. 112). Taxanes can sequester ARv567 (ref. 113) in a process that is mediated by the retention of a portion of the bipartite nuclear localization signal (AA629–AA634) necessary for androgen receptor–importin interaction^114–116. The minimal microtubule-binding domain of the androgen receptor appears to be close to the hinge region in the C terminus (AA559-AA663)¹¹³; this region is lacking in the AR-V7, which appears to confer the ability of escaping to taxane-induced sequestration.

Supporting these mechanistic findings, prostate-cancer cell lines that are resistant to taxanes — 22Rv1 and LNCaP95 — express increased levels of AR-V7, whereas transfection of AR-V7 but not the full-length androgen receptor (AR-FL) leads to increased resistance to paclitaxel, docetaxel and cabazitaxel¹¹⁷. At the clinical level, AR-V7 transcript levels increase after ADT and even more after chemotherapy¹¹⁸. Notably, AR-FL requires androgen supplementation for nuclear translocation, whereas EGFP-tagged AR-V7 variants translocate spontaneously, a mechanism that remains intact despite taxane therapy^113,119. Co-expression of AR-V7 with AR-FL influences how AR-FL responds to taxanes through both impairment of taxane-mediated cytosolic sequestration and through co-occupation of promoters of androgen-receptor target genes^117,119. By contrast, cabazitaxel seems to function via androgen-receptor-independent mechanisms; several studies have shown that its growth inhibition is not impacted by androgen receptor knockdown and that its clinical response in mCRPC is independent of AR-V7 status^{102,113,119,120}.

Although clinical data have repeatedly shown an association between AR-V7 and resistance to androgen-signalling inhibitors¹²¹, the relationship between AR-V7 expression and taxane resistance in clinical trials is less strong. In a prospective analysis of 37 patients with CRPC initiating docetaxel or cabazitaxel therapy, AR-V7 detection in CTCs via reverse transcriptase polymerase chain reaction (RT-PCR) was not associated with primary resistance to taxanes: AR-V7-positive patients had a 41% PSA response (compared with 65% in AR-V7-negative patients), but this difference was not statistically significant¹²². In fact, several patients with detectable AR-V7 at baseline converted to AR-V7-negative status during taxane therapy in the same study, suggesting that AR-V7 positivity might indicate that a patient is suitable for taxane treatment, especially considering that these patients might be resistant to androgen inhibition. The PROPHECY study showed that AR-V7-positive status was associated with shorter PFS and overall survival with androgen-signalling inhibitors and still showed clinical benefits with taxane therapy^122,123. Indeed, in a larger study of 161 patients with mCRPC¹²¹, AR-V7 positivity was strongly associated with resistance to androgen-signalling inhibitors (enzalutamide and abiraterone) and poorer clinical outcomes, including lower PSA responses, shorter PFS, and decreased overall survival. However, these AR-V7-positive patients demonstrated superior overall survival when treated with taxanes (8.9 months versus 4.6 months). Taken together, the results from these studies indicate that in the context of AR-V7-positive tumours where limited taxane or anti-androgen therapy options exist, AR-V7-positive patients might benefit from taxane administration. However, besides this prioritization strategy, clinical studies have also shown that AR-V7-positive tumours might correlate with worse response to taxane therapy^124,125, highlighting the urgent need to develop specific AR-V7 targeting strategies. For example, the TAXYNERGY study showed that patients with AR-V7-positive status had a decreased PSA₅₀ response rate (>50% PSA decline) when compared with AR-V7-negative patients, and argued that although it was a non-significant reduction, it still could indicate modest taxane resistance in the AR-V7-positive group¹²⁶.

TMPRSS2–ERG fusion

The TMPRSS2–ERG gene fusion is a hallmark genomic rearrangement seen exclusively in prostate cancer, where it is found in around ~50% of cases, and leads to high expression of a truncated functional ERG protein¹²⁷. Preclinical studies have demonstrated that prostate-cancer cells engineered to overexpress the TMPRSS2–ERG gene fusion exhibit decreased sensitivity to docetaxel^128,129, either via binding to soluble tubulin, thereby reducing the binding affinity between taxanes and tubulin¹²⁸, or via induction of epithelial–mesenchymal transition (EMT) via upregulation of ZEB1 (ref. 129).

The predictive role of ERG status for taxane therapy seems to vary by disease stage. In a retrospective cohort of 55 patients with mHSPC treated with a combination of docetaxel and ADT, ERG-positive patients (42% of the total) exhibited a longer PFS and a lower frequency of early progression than did ERG-negative carriers¹³⁰. Similarly, a pooled analysis of the phase III trials GETUG12 (in high-risk localized prostate cancer) and GETUG15 (in mHSPC) found that the addition of docetaxel to ADT improved failure-free survival in patients with ERG-positive tumours but not in those with ERG-negative tumours¹³¹. However, a study analysing TMPRSS2–ERG fusion transcripts in the peripheral blood mononuclear cell (PBMC) fraction (used as a surrogate to capture CTC-derived material) in 34 patients with mCRPC found that ERG positivity was associated with poor taxane response, reduced PFS and reduced overall survival¹³². The authors of this study chose to evaluate PBMC fraction rather than isolated CTCs, citing concerns that CTCs undergoing EMT might evade detection by epithelial-marker-based approaches, even though CTCs are commonly used and newer technologies are available to discriminate based on epithelial identity. In doing so, they only characterized 16% of patients as being positive for the TMPRSS2–ERG gene fusion, a surprisingly low percentage that might indicate patient selection bias in their small population, or an unconventional isolation technique¹³². Of note, the same study reported no association between detection of TMPRSS2–ERG transcripts in the PBMC fraction and taxane response, PFS and overall survival in patients treated with cabazitaxel¹³².

To further complicate this topic, a 2024 biomarker-driven, adaptive platform trial using isolated circulating tumour DNA (ctDNA) from patients with mCRPC reported that patients with TMPRSS2–ERG gene fusions had better outcomes with androgen-signalling inhibitors than they did with taxanes¹³³. It is possible that ERG might sensitize tumours to taxanes in patients with HSPC but confer resistance in patients with CRPC, a finding that could be the result of hormonal adaptations, tubulin remodelling or alterations in androgen signalling; however, more data are needed to reach a solid conclusion.

PTEN status

PTEN is a well-established tumour suppressor gene in prostate cancer, with loss-of-function mutations affecting 15–20% of primary prostate-cancer cases and increasing in frequency to 40–60% cases in CRPC^134–136. PTEN loss leads to activation of the PI3K–AKT–mTOR pathway, which promotes tumour growth, survival and downregulation of androgen-receptor activity, which contributes to resistance to androgen-signalling inhibitors such as abiraterone or enzalutamide^134,137. Activation of the PI3K–AKT–mTOR pathway contributes to a more aggressive phenotype that could potentially contribute to taxane resistance¹³⁸, for example, via upregulation of ABCB1 (ref. 139) expression or through stabilization of microtubules¹⁴⁰. In an androgen-deprived prostate-cancer-cell model (C4–2AT6), docetaxel treatment resulted in an increase in phosphorylated AKT (pAKT) levels in a dose-dependent manner¹⁴¹. Combining docetaxel with the PI3K–AKT inhibitor LY294002 increased apoptosis in these resistant cells¹⁴¹. Similar sensitizing effects have been reported using other inhibitors such as NVP-BEZ235 (a dual PI3K–mTOR inhibitor)¹⁴², AZD5356 (a pan-AKT inhibitor)¹⁴³, AZD8186 (a PI3Kβ/δ inhibitor)¹⁴⁴ or capivasertib (a pan-AKT inhibitor)¹⁴⁵.

Similar to findings at the preclinical level, administering AKT inhibitors in combination with taxanes might produce enhanced anti-tumour effects. In the phase II ProCAID trial^146,147, the combination of capivasertib and docetaxel improved survival in patients with mCRPC compared with placebo and docetaxel (31.15 versus 20.27 months; HR = 0.54, 95% CI 0.34–0.88). Notably, the benefit of capivasertib did not vary based on PI3–AKT–PTEN pathway activation status. In another study¹⁴⁸, 39% of patients with mCRPC exhibited PTEN loss, which was significantly associated with shorter overall survival, but did not correlate with reduced efficacy of docetaxel. Importantly, however, the phase III clinical trial CAPitello-280 evaluating capivasertib in combination with docetaxel and ADT was recently discontinued after an interim analysis determined that it was unlikely to meet its primary endpoints of PFS and overall survival¹⁴⁹. These findings contrast with the earlier ProCAID results, although capivasertib remains under investigation for PTEN-deficient, hormone-sensitive prostate cancer (NCT04493853).

RB1 loss

Loss of retinoblastoma tumour suppressor protein (RB1) occurs with high frequency in prostate cancer¹⁵⁰, and correlates with a more aggressive phenotype that leads to enhanced migration and invasion¹⁵¹ and increased resistance to ADT^152,153. Tumours that become CRPC through RB1 loss might be particularly vulnerable to taxane therapy¹⁵³, especially cabazitaxel, as shown in isogenic RB1-knockdown models⁹⁷. By contrast, according to the GETUG-12 trial in high-risk localized prostate cancer, RB1 loss was not significantly associated with worse recurrence-free survival, nor did it show a significant correlation with overall disease outcomes¹³¹. Another strategy for cancer treatment involves identifying RB1-intact tumours and blocking D-cyclin-dependent kinases CDK4 and CDK6, for example, through the use of palbociclib, which is approved for oestrogen-receptor-positive advanced breast cancer¹⁵⁴. Under usual circumstances, CDK4 and CDK6 hyperphosphorylate RB1, leading to cell-cycle progression through E2F-mediated transcription¹⁵⁵. However, in a 2021 phase II study in patients with RB1-intact mHSPC, the addition of palbociclib to ADT did not improve PSA response rate or radiographic response rate¹⁵⁶.

MYC

The oncogene MYC, located at chromosome 8q24, is frequently gained or amplified in advanced and recurrent prostate cancer^157,158, leading to elevated c-MYC levels^159–161. c-MYC regulates a broad transcriptional network, including NKX3–1, PIM1, TMPRSS2, SPARC and EGF, influencing cell-cycle progression and stem-cell differentiation¹⁶¹. The role of c-MYC in taxane resistance is controversial, and distinct mechanisms probably exist depending on disease stage. In high-risk localized prostate cancer, a stage at which c-MYC overexpression is highly prevalent¹⁶¹, c-MYC overexpression can lead to docetaxel sensitization via upregulation of ID-1 (ref. 162). Consistent with this finding, an immunohistochemical study of high-risk localized prostate cancer found that c-MYC overexpression sensitized tumours to docetaxel and correlated with improved PFS¹⁶³. Conversely, docetaxel-resistant residual tumour cells at the mCRPC stage show increased MYC expression owing to stabilization via a CXCR4–ERK signalling loop¹⁶⁴. At this same stage, paclitaxel-resistant cells release damage-associated protein HMGB1, which activates c-MYC signalling, suggesting a potential therapeutic target for re-sensitization to paclitaxel¹⁶⁵. This conflicting role of MYC in conferring taxane resistance at later disease stages might be explained by its role in transition towards a more aggressive, neuroendocrine phenotype via upregulation of EZH2, methylation of H3K27me3 and co-expression of AURKA^166,167. Interestingly, although a phase II trial that used alisertib (an AURKA inhibitor) in patients with mCRPC with neuroendocrine features failed to meet its primary end point¹⁶⁸, a subset of patients with Aurora-A and N-MYC activation achieved significant clinical benefit from alisertib, supporting the rationale of combined testing of alisertib with docetaxel in this patient subgroup.

Mutations in DNA repair genes

When the homologous recombination pathway is impaired, such as occurs with BRCA gene defects, inhibition of poly(ADP-ribose) polymerase can lead to the accumulation of double-stranded DNA breaks and eventual cancer-cell death, a therapeutic strategy that has shown success in ovarian cancer¹⁶⁹, breast cancer¹⁷⁰, as well as in other tumour types^171–173. In patients with mCRPC whose disease had progressed on enzalutamide or abiraterone, the PROfound trial established the poly(adenosine diphosphate–ribose) polymerase (PARP) inhibitor olaparib as the standard of care in patients harbouring DNA repair alterations — including mutations in ATM, BRCA1 and BRCA2 — as olaparib was associated with improved PFS^174,175. Notably, subgroup analyses revealed that olaparib provided a benefit irrespective of prior taxane exposure¹⁷⁶, and exploratory post hoc analyses showed that patients with BRCA alterations who received olaparib experienced longer radiographic PFS and overall survival than those treated with androgen-signalling inhibitors¹⁷⁷.

For patients with mCRPC with BRCA1 and BRCA2 mutations, olaparib is favoured over cabazitaxel in terms of radiographic PFS according to data from an indirect comparison between PROfound and CARD¹⁷⁸. Conversely, in patients with non-BRCA mutations in DNA damage repair (DDR) genes, cabazitaxel seems to be more beneficial than olaparib¹⁷⁸. Similar data emerged from the TRITON-3 phase III trial¹⁷⁹ comparing the PARP inhibitor rucaparib with the physician’s choice of treatment (including enzalutamide, docetaxel or abiraterone), which showed that rucaparib significantly extended radiographic PFS over docetaxel in patients harbouring BRCA mutations, and that this benefit was not observed in patients with ATM mutations. Although ATM-deficient and BRCA-deficient tumours share certain molecular features, growing evidence suggests that BRCA mutations might confer resistance to taxanes, whereas ATM mutations seem to be associated with taxane sensitivity^180,181. A multicentre retrospective review¹⁸² investigated the optimal sequence of PARP inhibitors and taxanes in patients with mCRPC, and hinted at a potential advantage when taxanes are administered prior to PARP inhibitors in patients with ATM mutations, although these findings were not statistically significant.

Considering the question of whether mutations on DDR genes can predict response to taxanes, a study using liquid biopsies of cell-free (cf)DNA from 375 patients with mCRPC revealed that although individual DDR alterations did not significantly alter treatment outcomes of patients on taxanes, the presence of ≥1 DDR alteration was associated with worsened PFS on taxanes¹⁸³. By contrast, individual DDR mutations were associated with significantly worsened PFS, overall survival and PSA response rates in patients treated with androgen-signalling inhibitors, providing further evidence that patients with DDR mutations should be given PARP inhibitors or taxanes, but not androgen-signalling inhibitors. Opposing these findings, a study involving 53 patients with mCRPC treated with docetaxel found that patients with BRCA2-positive tumours had a significantly lower response rate to docetaxel than patients with wild type BRCA2 (ref. 184). This study used tumour tissue sequencing (instead of cfDNA) and defined treatment response as PSA decline, rather than radiographic PFS and overall survival, which might explain the different findings.

Despite these discordant clinical findings regarding DDR mutations and taxane response, it is clear that BRCA mutations correlate with a more aggressive disease course and are enriched in advanced prostate cancer^185,186. Preclinical data indicate that loss of BRCA1 can lead to increased microtubule dynamicity, potentially through mechanisms involving gamma-tubulin ubiquitination, thereby reducing the efficacy of taxane-induced microtubule stabilization and subsequent apoptosis of cancer cells¹⁸⁷. These findings imply that BRCA-mutant tumours might be less sensitive to taxanes and support the p rioritization of PARP inhibitors in this molecular subset of tumours.

Drug efflux mechanisms in taxane resistance

One mechanism that the prostate-cancer cell can use to overcome the action of taxanes is regulating how much of the drug is maintained intracellularly at a given time. The most well-known member of the drug efflux pump category is P-glycoprotein, also known as ABCB1 (ATP-binding cassette subfamily B member 1), encoded by the MDR1 gene, which is part of a family of drug exporters that compromises another 49 members¹⁸⁸. ABCB1 is overexpressed in over 50% of cancers and is a well-studied cause of multidrug resistance¹⁸⁸. ABCB1 has been extensively implicated in reducing taxane efficacy by actively pumping out docetaxel and cabazitaxel^189–191. For instance, ABCB1 expression is significantly upregulated in docetaxel-resistant cell lines, which require an IC50 that is 70-fold higher than that of parental C4–2B cells¹³⁹, a phenomenon also observed in taxane-resistant DU145-DR and 22Rv1-DR cell lines¹⁹². Pharmacological inhibition of ABCB1 via knockdown or using the small molecular inhibitor elacridar restores sensitivity to cabazitaxel^191,193.

Systemic inhibition of ABCB1 has, however, proven challenging in the clinical setting for various reasons, such as its polyspecificity (its ability to bind to multiple different substrates), its high conformational flexibility and the fact that it is expressed in normal tissues such as the blood–brain barrier, where it performs protective functions^194,195. Semi-synthetic taxanes such as cabazitaxel or ortataxel were developed to overcome, in part, the susceptibility of taxanes to ABCB1-mediated efflux^196,197. As we have discussed, however, cabazitaxel-treated prostate-cancer cells are able to upregulate ABCB1 to levels that still confer resistance^193,198. Furthermore, variability in basal ABCB1 levels across prostate-cancer cell lines suggests that cells with lower or undetectable ABCB1 expression might rely on alternative mechanisms of drug efflux to confer taxane resistance^199–201.

In cases where ABCB1 is not the predominant mechanism of resistance, overexpression of other transporters from the ABC family, such as ββ (also known as MRP4), can contribute to taxane resistance. C4–2/D cells that are resistant to docetaxel show increased expression of ABCC4 in comparison with their docetaxel-sensitive counterparts^202,203, and anti-androgen therapy with bicalutamide has been shown to lead to downregulation of ABCC4 and re-sensitization to docetaxel²⁰². In the clinical setting, ABCC4 inhibitors such as ceefourin-1 (refs. 204,205) could be promising candidates owing to their low cellular toxicity, high stability and the fact that ABCC4 has multiple drug-binding sites.

Emerging mechanisms of taxane resistance as potential therapeutic targets

As the long list of failed clinical trials indicates, it is unlikely that targeting a specific molecular pathway will lead to notable clinical benefit, particularly in later disease stages where extensive transcriptional rewiring has already occurred, enabling tumour cells to adapt and withstand therapeutic stress through multiple pathways. Instead, we advocate for exploring key regulatory mechanisms that influence and govern multiple cellular pathways, thereby increasing the likelihood of clinical benefit upon their exploitation. In this section we outline some of the emerging targets that might be worthy of therapeutical exploitation in upcoming clinical trials, including the following: undifferentiated and developmental phenotypes; increased nuclear transport; development of chromosomal instability adaptation mechanisms; metabolic reprogramming; and the antitumour immune response (Fig. 3).

Developmental and undifferentiated phenotypes

Notch and Hedgehog are two conserved signalling cascades that are critical for cell-fate determination, proliferation, angiogenesis and apoptosis. Abnormal Notch and Hedgehog activity leads to increased tumour growth, migration and invasion in multiple cancers^206–208, and normal function of these cascades is essential for correct embryological development of prostate tissue²⁰⁷. Our group reported the existence of a small subpopulation of mCRPC cells that survive docetaxel therapy by upregulating Notch, Hedgehog and other markers associated with stemness properties¹⁹² while downregulating epithelial differentiation markers and HLA class I antigens²⁰⁹. In parallel, therapeutic pressure exerted by taxane therapy also activates the EMT process by which prostate-cancer cells acquire stemness markers and downregulate epithelial-differentiation markers^210–213. Indeed, in several studies, cells undergoing taxane treatment ensure survival by increasing levels of transcription factors ZEB1/ZEB2 (refs. 213,214), which suppress E-cadherin expression, upregulate expression of mesenchymal markers such as vimentin (encoded by VIM), N-cadherin (encoded by CDH2) and fibronectin (encoded by FN1), and upregulate stemness markers such as CD44 (ref. 212).

Targeting the self-renewal pathways (via Notch and Hedgehog) or EMT pathways (via ZEB1) might be feasible strategies to hinder the taxane-resistant subpopulations that emerge during the course of therapy. Indeed, inhibition of Notch signalling via a gamma secretase inhibitor was shown to re-sensitize mCRPC cells to docetaxel while decreasing numbers of cells expressing markers of stemness^210,215. On the other hand, suppression of the Hedgehog pathway via vismodegib or sonidegib has thus far failed to show clinical benefit in mCRPC or high-risk localized prostate cancer as single-agent therapy^216,217, suggesting that future success might require patient enrichment (for example, focusing on tumours with strong stemness pathway activation) or combination strategies. Inhibition of ZEB1 via miR-27b and miR-34a is able to reverse the EMT process and restore epithelial characteristics to prostate-cancer cells, which in turn re-sensitizes them to docetaxel²¹⁸. Indeed, ZEB1 expression can be modulated pharmacologically by using histone deacetylase inhibitors to resensitize tumour cells to chemotherapy²¹⁹.

Notch is also an upstream regulator of master transcription factor GATA2, which has been implicated in taxane resistance in prostate cancer²²⁰, including mediation of docetaxel resistance via insulin-like growth factor 2 (IGF2), IGF1 receptor (IGF1R) and insulin-receptor pathways²²¹. Inhibition of IGF2–IGF1R signalling resensitizes cells to chemotherapy and improves survival in preclinical models²²², highlighting the role of IGF2 in protection from taxane-induced apoptosis²²¹. Clinical trials evaluating IGF pathway inhibitors in prostate cancer have investigated figitumumab (an anti-IGF1R antibody) combined with docetaxel in mCRPC²²³ and linsitinib (a dual inhibitor of IGF1R and insulin receptor tyrosine kinase activity) in mCRPC²²⁴, but both have shown severe toxic effects, limiting their clinical development.

Nucleocytoplasmic transport

Another key process that probably contributes to taxane resistance is nucleocytoplasmic transport. Many aggressive tumours across cancer types display a pronounced reliance on nuclear pore complexes for survival and become susceptible to cell death when genes encoding nucleoporins (for example, NUP160 and NUP93) are knocked down, an effect also observed in xenograft models²²⁵. Aggressive prostate-cancer tumours have been shown to overexpress the nucleoporin POM121, which partners with importin-β to shuttle critical oncogenic transcription factors (including androgen receptors, MYC and GATA2) into the nucleus^226,227. Inhibition of this axis can be achieved via use of importazole, a small molecule inhibitor of importin-β that leads to re-sensitization of mCRPC cells to docetaxel²²⁶. However, targeting nuclear import in the clinical setting remains to be investigated using clinical-grade compounds.

Metabolic reprogramming

Tumour progression to therapy depends on altered metabolism driven by increased energetic demands, increased oxidative stress and decreased nutrient availability^228,229. In particular, metabolic reprogramming through mesenchymal transition, dysregulated ferroptosis and acquisition of a senescent phenotype have been highlighted as processes influencing how prostate-cancer cells respond to taxane therapy. The mesenchymal phenotype, which, as we have outlined previously, is characterized by resistance to taxanes, presents with a distinct increase in metabolic reprogramming²³⁰ with a heavy reliance on oxidative phosphorylation to meet increased energy demands. These cells overexpress GLS1 (glutaminase 1), the enzyme that catalyses the conversion of glutamine to glutamate, which is then converted to alpha-ketoglutarate to feed the tricarboxylic-acid cycle and support oxidative phosphorylation²³¹. As such, depletion of glutamine or targeting of GLS1 enhances taxane sensitivity and reduces ATP production, and therefore could inhibit proliferation and survival of cancer cells²³¹.

Another interesting mechanism that mesenchymal cells use to bypass taxane resistance is adipose stromal cells, which suppress reactive oxygen species generation and therefore prevent oxidative-stress-induced apoptosis²³². Another metabolic reprogramming pathway that influences taxane resistance through reactive oxygen species inhibition is the suppression of ferroptosis — an iron-dependent form of cell death driven by lipid peroxidation and reactive oxygen species accumulation, which occur as a result of depletion of the anti-oxidant glutathione²³³. In this context, a 2024 study described a mechanism by which cancer-associated fibroblast (CAF)-derived exosomes promote docetaxel resistance by inhibiting ferroptosis²³⁴. These exosomes, enriched in miR-432–5p, suppress CHAC1 — a glutathione-degrading enzyme that normally promotes ferroptosis — thereby preserving glutathione levels, blocking ferroptotic cell death and enhancing resistance to docetaxel.

The acquisition of a senescent state is a common occurrence after docetaxel treatment and is associated with a worse prognosis^235,236. This senescent state is characterized by the upregulation of nuclear factor-κB (NF-κB), which in turn activates the secretion of pro-inflammatory cytokines such as IL-6, IL-8 and matrix metalloproteinases, a state also known as senescence-associated secretory phenotype (SASP)²³⁷. A 2024 study has shown that monoacylglycerol lipase is a critical regulator of docetaxel efficiency as it represses NF-κB transcriptional activity, which leads to a reduction in senescent markers and repression of the SASP phenotype²³⁸. These findings present another example of how senolytic therapy via targeting of NF-κB might provide an additional benefit when combined with chemotherapy.

Chromosomal instability and mitotic fidelity

Mitotic dysregulation, another hallmark of taxane resistance in prostate-cancer cells, is achieved via upregulation of myriad mitotic fidelity kinases²³⁹. Aurora kinase A (AURKA) and Aurora kinase B (AURKB) are markedly upregulated in docetaxel-resistant cells as a mechanism of tolerance to increased genomic chaos, and are linked with resistance to paclitaxel^166,240. MPS1 kinase is involved in the SAC, and its inhibition forces premature exit from taxane-induced mitotic arrest, driving chromosome mis-segregation and restoring chemosensitivity²⁴¹. BUB1 is another component of the SAC that localizes to kinetochores and enables recruitment of checkpoint proteins. BUB1 expression is upregulated in mCRPC and taxane-resistant tumours²⁴². Finally, MASTL, another mitotic kinase that promotes mitotic entry and progression by sustaining CDK1 substrate phosphorylation during mitosis, has been identified as a top survival kinase in aggressive prostate-cancer cells, and acts as a safeguard that enables rapidly dividing, chromosomally unstable tumour cells to survive²⁴³. Taken together, mitotic fidelity kinases could influence taxane resistance via enhanced mitotic checkpoint signalling and increased overall fidelity of cell division, which might enable them to withstand the activity of taxanes until metabolized to sub-toxic concentrations. Importantly, these mechanisms could be targeted in the context of prostate cancer to re-sensitize cells to taxane therapy. For example, targeting MASTL (genetically or with the first-generation inhibitor GKI-1 (ref. 244)) has been shown to slow growth of CRPC cells and enhance docetaxel efficacy²⁴³.

Taxanes and immunotherapy

Combining chemotherapy with immunotherapy has been approved for the treatment of multiple tumour types²⁴⁵, including lung, breast, gastric and bladder tumours. For example, taxanes in combination with anti-programmed cell death 1 (anti-PD1) immune-checkpoint inhibitors are a standard of care for treatment of non-small-cell lung carcinoma^246–248 and triple-negative breast cancer^249–251.

Chemotherapy-induced cell death can stimulate antitumour immunity by presenting antigens that activate cytotoxic lymphocytes, inducing a process known as immunogenic cell death^252,253. Additional mechanisms by which chemotherapy can promote antitumour immunity include decreasing immunosuppressive cells, such as myeloid-derived suppressor cells^254–256 and regulatory T cells²⁵⁷, as well as activating natural-killer cells^258,259.

Although retrospective patient tumour sample analysis showed that tumour-infiltrating T cells can increase after chemohormonal therapy²⁶⁰, unfortunately, to date, the combination of docetaxel and immune-checkpoint therapy has not shown clinical benefit in patients with prostate cancer. The initial phase Ib/II KEYNOTE-365 study demonstrated the safety and antitumour efficacy of pembrolizumab plus docetaxel and prednisone in patients with metastatic CRPC²⁶¹. However, the large, randomized phase III KEYNOTE-921 study (n = 1,030 patients) showed that the addition of pembrolizumab to docetaxel did not significantly improve either radiographic PFS or overall survival of patients with mCRPC²⁶². These results add to the list of studies demonstrating that standard therapies (for example, enzalutamide, abiraterone and apalutamide) have limited efficacy when combined with anti-PD1 inhibitors (such as pembrolizumab and nivolumab) in prostate cancer. Future studies investigating the combination of taxanes with other immune-checkpoint inhibitors (for example, B7-H3)²⁶³ or T cell engager molecules (for example, STEAP1 × CD3 antibodies)²⁶⁴ may succeed in adding immunotherapy to the armamentarium for treating prostate cancer.

Taxane response and liquid biopsies

As discussed, advanced prostate cancer is a heterogeneous disease that evolves dynamically in response to treatment, influenced by myriad genetic factors. PSA has been the primary biomarker for monitoring prostate-cancer progression, but its predictive value for therapeutic response is limited²⁶⁵. Moreover, tissue biopsies, although crucial for initial diagnosis, pose procedural risks, potential comorbidities and sampling bias owing to intratumoural heterogeneity. To address these challenges, liquid biopsies have emerged as a minimally invasive alternative for prognostic and predictive insights into prostate-cancer progression²⁶⁶.

CTCs are an emerging method that has demonstrated utility in patient stratification for taxane-based chemotherapy. Whole-exome sequencing of CTCs can identify somatic single-nucleotide variants and copy-number alterations that can be informative in patient selection and stratification²⁶⁷, and variations in CTC burden correlate with overall survival in patients with mCRPC²⁶⁸. The IMMC-38 trial demonstrated that a ≥30% decline in CTC count after taxane-based chemotherapy was associated with significantly improved overall survival, with early responses evident as soon as 4 weeks after treatment²⁶⁸; high baseline CTC counts were a strong predictor of increased risk of death²⁶⁹. The MAINSAIL phase III clinical trial of patients with mCRPC treated with docetaxel showed that a CTC count of ≥5 cells per 7.5 ml blood at baseline was associated with lower overall survival, and that a decrease in CTC count was linked with improved prognosis^34,270. The same criterion was used in a 2023 study combining CTC characterization with serum biomarkers such as PSA and alkaline phosphatase²⁷¹, supporting the potential use of CTCs to triage patients for docetaxel therapy. Indeed, ≥5 pre-docetaxel CTCs per 7.5 ml of blood was associated with worse PFS in metastatic hormone-sensitive prostate cancer treated with first-line chemo-hormonal therapy²⁷¹. However, CTC count was not well correlated with PSA response or RECIST-defined tumour s hrinkage in another study²⁷².

A notable limitation of CTCs is the reliance on epithelial epitope-dependent isolation methods, which might fail to detect tumour cells undergoing EMT^273–275, which can serve as biomarkers of increased metastatic burden, aggressivity and taxane resistance^212,276,277. Platforms capable of identifying epithelial, mesenchymal and intermediate phenotypes are currently under development and should be prioritized to provide a more comprehensive assessment of disease burden and treatment response^277–279.

Beyond CTCs, assessment of ctDNA has gained attention as an additional liquid-biopsy marker with prognostic implications for prostate cancer. Persistent detection of ctDNA after androgen-receptor signalling inhibitory therapy was associated with reduced overall survival in two studies in patients with mCRPC^280,281. Likewise, a cohort of 491 patients who initially showed a PSA response to systemic therapy, those starting first-line mCRPC therapy with ctDNA fractions exceeding 30% had a significantly higher risk of death than patients with ctDNA fractions below 2%²⁸². In the PRESIDE trial, which demonstrated prolonged PFS with continuing enzalutamide beyond mCRPC progression and initiation of docetaxel²⁸³, persistence of or increased ctDNA at the start of the second docetaxel cycle was associated with rapid disease progression, suggesting early resistance to docetaxel²⁸⁴.

Conclusions

Over two decades after its FDA approval, taxane therapy remains the gold-standard treatment for patients with metastatic prostate cancer, and its use in combination with androgen-targeting therapy has extended to earlier, hormone-sensitive disease stages. Furthermore, the extended use of cabazitaxel has emerged as an attractive therapeutic option in the post-docetaxel setting, particularly in patients whose disease progresses after androgen-receptor signalling inhibitors.

Despite these advances, the therapeutic ceiling of taxanes is increasingly evident. Attempts to enhance efficacy through combination strategies have largely failed, owing to multiple factors, including overlapping toxicities (especially in elderly patients with comorbidities), empiric drug development based on the proven activity of these drug combinations in other tumour types and the inherent adaptability of advanced tumours through cellular rewiring.

Mechanistic studies on taxane resistance have largely focused on classical pathways of resistance, including tubulin isoform switching, pump-efflux regulation, apoptotic pathways, among others. Additionally, genetic prostate-cancer drivers such as PTEN loss, TMPRSS2–ERG fusion and BRCA mutations further influence taxane sensitivity. We argue that targeting a single mechanism is unlikely to lead to notable clinical benefit, particularly in advanced-disease stages where multiple pathways are activated during disease progression and confer therapy resistance. In this context, our vision is that targeting key regulatory mechanisms that control multiple cellular pathways might prove to be more beneficial. These regulatory mechanisms include emerging mechanisms of resistance such as stemness and EMT pathways, nucleocytoplasmic transport, tumour-dependent metabolic processes and mitotic fidelity kinases. These biologically important processes are still being studied at the preclinical or early clinical level, but have the potential to synergize with taxane therapy to disrupt multiple avenues of cellular rewiring simultaneously.

Finally, although docetaxel has been approved for the treatment of prostate cancer for two decades, no biomarker panel has been demonstrated to appropriately select patients that should respond to this therapy. This lack of biomarker panel might be mainly attributed to the limited capacity to obtain biopsies from patients with metastatic disease (because of invasiveness and risk), which impedes the comprehensive molecular analysis needed to develop effective biomarker panels. This challenge will continue as current guidelines prioritize early docetaxel administration in patients with low-volume hormone-sensitive prostate cancer, which complicates future biopsies because tumour burden is low. Consequently, the development of predictive biomarkers using liquid-biopsy approaches, as discussed, will be critical to selecting patients who are most likely to respond to taxane therapy. Biomarker-driven patient stratification and rational trial design will also enable more careful evaluation of emerging therapies that might be useful at specific disease stages and also could drive a move into personalized treatment paradigms.

Key points.

• Taxane therapy — docetaxel and cabazitaxel — is currently considered the mainstay of treatment for prostate cancer.

• The enhanced efficacy when taxane therapy is combined with anti-androgen therapy provides a strong premise to continue developing innovative combination therapeutic strategies.

• Canonical prostate cancer drivers reduce taxane therapy efficacy.

• Targeting mechanisms that regulate multiple signalling pathways might improve taxane therapy efficacy.

• Development of predictive biomarkers to indicate which patients taxanes will be effective in requires longitudinal liquid-biopsy approaches.