The current clinical states of CRPC remain defined by patterns of spread, but are now heavily dependent on prior therapy exposure (Fig. 2.1), an issue that was not emphasized in PCWG2 given the lack of approved therapies in 2008. For purposes of this book chapter, we will not discuss localized disease, rising PSA after local therapy (biochemical recurrence), or metastatic disease in hormone-naïve PCa. Instead, we will focus on sites of metastatic spread in CRPC and prior therapy exposure. In addition, within each CRPC clinical state, we will discuss the importance of independent prognostic factors including histologic subtype, serum and blood-based prognostic biomarkers, symptoms, and molecular alterations linked to CRPC progression.

As more systemic treatments become available for men with CRPC, it is important to keep in mind prior therapy exposure for each patient. Prior therapy will dictate prognosis and response to the next treatment strategy, given that cross-resistance to at least several newly approved agents is expected and has been observed in the clinic. The promotion of resistance to subsequent therapy with exposure to novel hormonal agents may be a result of clonal selection, tumor evolution through mutation, or epigenetically regulated cellular plasticity and adaptation. Thus, understanding the prior exposure of a patient to a range of systemic therapies will facilitate the rational sequencing of therapies in the clinic and anticipated clinical benefit from a further line of treatment.

Both of the novel therapeutic agents abiraterone acetate and enzalutamide were tested initially in the post-chemotherapy state, after patients had progressed on docetaxel. However, it is now clear that these agents are likely more active in the pre-docetaxel mCRPC setting [14, 15], based on greater magnitudes and durability of PSA decline and disease control. For example, the PSA response rate and progression free survival for abiraterone acetate in the post-docetaxel CRPC setting is about 50 % and 8 months, while in the pre-docetaxel it is 70 % and 16 months, respectively. In the phase I–II trial of enzalutamide, 65 patients were chemotherapy-naïve and had a better PSA progression free survival [PFS] (greater than 25 % increase in PSA from baseline) compared to the 75 patients who were post-chemotherapy (median PFS not reached for pre-chemotherapy group vs. 27 weeks for post-chemotherapy group) [15]. The reasons for this may relate to lesser disease burden in the pre-docetaxel mCRPC disease state, but may also relate to cross-resistance of hormonal therapy and docetaxel. Recent studies suggest that docetaxel may have an underlying hormonal mechanism, through disruption of AR nuclear transport on microtubule shuttles. Thus, a biologic rationale has emerged that may explain clinical cross-resistance with docetaxel and hormonal therapies. Thus, appreciating prior therapy exposure is essential in predicting clinical benefit to subsequent therapy.

The phase I trial of abiraterone acetate included 19 patients who had prior ketoconazole exposure, and PSA declines of more than 50 % were noted in 9 of these 19 patients (47 %) [16]. Therefore, it appeared as though abiraterone had disease activity in CRPC even in patients who had received ketoconazole previously. Many of these men were not truly ketoconazole-resistant, and thus, it is not clear how active abiraterone would be in truly ketoconazole-resistant men. Studies of enzalutamide have been conducted in men without prior exposure to ketoconazole or abiraterone, and thus it is currently unclear whether cross-resistance exists. However, current retrospective series indicate clear evidence for a lower response rate and duration of response when enzalutamide or abiraterone are used after each other [17–19]. Therefore, the timing and sequencing of treatments can certainly determine disease response to treatment and is essential in determining clinical disease state.

Retrospective studies suggest clinical cross-resistance between enzalutamide and abiraterone acetate. One retrospective study examined 35 patients with CRPC in the post-docetaxel state, who had received abiraterone followed by enzalutamide [19]. In this study, 40 % of patients had a rising PSA as their best response to treatment with primary resistance to enzalutamide. In patients who had some response to enzalutamide (defined by at least one declining PSA value), median time to progression was only 4 months before secondary resistance to enzalutamide developed. Two other retrospective studies examined abiraterone acetate treatment after docetaxel and enzalutamide. One studied 30 patients with CRPC treated with abiraterone acetate after enzalutamide [17] and found that patients had a median time to progression of 15.4 weeks. Only three of the patients had a PSA response with abiraterone. The other retrospective study looked at 38 patients with CRPC who were treated with abiraterone after progression with both docetaxel and enzalutamide [18]. Only ten of these patients (26 %) had any PSA response, and median PFS was only 2.7 months. These data suggest that prior therapy with either abiraterone acetate or enzalutamide may promote cross-resistance with the other agent. While the mechanisms for this cross-resistance are speculative, they may involve the development of AR mutations or ligand binding domain deletion variants that become ligand independent and do not respond to either treatment.

Given the novelty of the systemic therapeutic agents, insufficient prospective data have been generated regarding cross-resistance and underlying mechanisms; biomarkers prospectively defining these mechanisms in the clinic are currently lacking. Until biomarkers of AR and androgen activity are available from clinical specimens, our current disease states remain defined by prior therapy exposure and responses.

The histology of PCa is commonly based on their origin from glandular-forming epithelial cells. While the cell of origin of PC is debated and may be the basal cell of the prostate, a subset of a luminal cell population, or both depending on context, typical prostate cancer is adenocarcinoma and consists of glandular architecture with varying degrees of loss of differentiation. Cells in the normal male prostate do not normally proliferate, but are able to survive repeat bouts of castration and regeneration. However, over time, with cumulative mutations and epigenetic lesions, PCa cells acquire the ability to continuously proliferate despite AR ablation, and acquire an invasive phenotype [3]. The Gleason grading system was initially described in 1966 [47], and was validated as a prognostic measure at the Minneapolis Veterans Administration in 1974 [48]. This grading system was updated in 2005 to correspond more closely with patient outcome [49]. As a score of nuclear polymorphism, glandular disruption, disease heterogeneity, basement membrane disruption, and de-differentiation of prostate cancer, the Gleason score reflects the aggressiveness of the individual PCa. The Gleason sum continues to be independently prognostic for survival in the CRPC setting, as demonstrated repeatedly in multiple nomograms for prostate cancer survival [44, 50]. The Gleason sum has also been found recently to be potentially predictive for treatment response and sensitivity to treatment with docetaxel in the TAX-327 trial [51]. In this analysis, higher tumor grades (Gleason score ≥7) had a more pronounced survival benefit from treatment with docetaxel than with mitoxantrone [51]. Thus, higher Gleason grading can be both prognostic for survival and potentially predictive for a greater magnitude of treatment benefit from docetaxel. While this requires confirmation, tumor grade may provide a surrogate biomarker for genomic instability, proliferation, and de-differentiation that may be exploitable in the clinic and provides independent prognostic information.

Adenocarcinoma is the main histologic form of PCa, accounting for 95 % of all PCa [52]. The remaining 5 % is made up of ductal, mucinous, small cell, and anaplastic carcinoma, each of which comprises a more aggressive variant that may progress to CRPC more rapidly and behave as high grade tumors. In the national SEER database, there is an incidence of 61 cases per 10,000 people per year for mucinous carcinoma, 49 cases per 10,000 people per year for ductal carcinoma, and 35 cases per 10,000 people for anaplastic or neuroendocrine carcinoma [53]. Mucinous carcinoma has similar 5-year OS to adenocarcinoma (75.1 vs. 76.5 % respectively), and ductal carcinoma has a slightly more aggressive disease course, with a 5-year OS rate of 61.7 % [53].

Contrastingly, neuroendocrine prostate cancer (NEPC) is much more aggressive and has a 5-year OS rate of only 12.6 % [53]. NEPC can arise either de novo with a low serum PSA level, obstructive symptoms, and often distant metastases at the time of diagnosis, or more commonly, NEPC emerges as a secondarily resistant subtype after prostate adenocarcinoma treatment. NEPC is independent of AR signaling and can be a mechanism of castration resistance (Table 2.2). NEPC often overexpresses chromogranin A and synaptophysin, biomarkers which are detectable in serum [54]. Many molecular alterations accumulate in NEPC, including amplification of Aurora Kinase A and N-myc [55], overexpression of EZH2 [55], loss of Rb [56], and activation of the PI3 kinase pathway [57]; these represent potential therapeutic targets. During therapy with an Aurora Kinase A inhibitor, these tumors revert to adenocarcinoma features, indicating histologic plasticity [55]. This phenomenon mimics what is observed in the reversible transitions between small cell and non-small cell lung carcinomas [58]. This toggle effect raises the possibility that prostate cancer cells have an inherent plasticity to change histological subtypes when evading treatment pressures and imply that the current or prior histologic appearance of a CRPC patient’s tumor may be important clinically.

NEPC correlates with poor clinical prognosis [53]. A series of 21 patients with NEPC were treated with cisplatin-based chemotherapy active in small cell carcinoma of the lung, with a resulting median OS of 9.4 months (range 1–25 months) [59]. A phase II trial of cytotoxic chemotherapy in 120 neuroendocrine prostate cancer patients subsequently treated patients first with carboplatin and docetaxel (CD), followed by cisplatin and etoposide (EP) [60]. Primary endpoints included response rates and time to progression. Only 74 patients were able to undergo treatment with both regimens. Of these men, 50 % had benefit from both regimens, 34 % responded to CD but not to EP, 9 % responded to EP but not to CD, and 7 % did not respond to either regimen. Median OS was 16 months [60]. Therefore, NEPC carries a poor prognosis despite multi-agent platinum chemotherapy. The emerging biologic differences inherent in NEPC transformation however suggest that molecularly targeted therapies, likely in combination, may have specific activity in this disease. The Aurora Kinase inhibitor MLN-8237 is currently in a phase II trial which is actively enrolling patients with NEPC to evaluate drug efficacy and predictive biomarkers (ClinicalTrials.gov number NCT01799278). Thus, NEPC represents a distinct CRPC histologic and molecularly defined entity. Current barriers, however, include defining the clinical characteristics of men with CRPC who have this NEPC genotype.

Multiple serum or blood-based biomarkers have prognostic significance in the progression of clinical states of CRPC, including PSA, alkaline phosphatase (AP), lactate dehydrogenase (LDH), albumin, and CTCs [74]. With the exception of CTCs, most of these biomarkers have been shown in nomograms with multivariate analyses to be independently prognostic of patient survival from CRPC as well as have implications for treatment selection, and CTCs have also been independently associated with mortality in men with metastatic CRPC [75]. For example, AR signaling regulates PSA production, and increasing PSA levels may indicate AR pathway dependence, while low PSA levels despite metastatic progression such as in NEPC indicates androgen receptor independent prostate cancer (ARIPC). Patients with intact wild-type AR are dependent on AR signaling and appropriate for AR-targeting therapies; however, those patients who develop splice variants of AR with C-terminal deletions would not benefit from AR-targeting agents. In addition, patients can further develop ARIPC, where cellular proliferation is no longer dependent on AR signaling. In these cases of ARIPC, cytotoxic agents are potentially more beneficial. Bone biomarkers such as AP may reflect the tumor burden within bone and potential benefit with bone-targeted therapies.