The Influence of Prior Abiraterone Treatment on the Clinical Activity of Docetaxel in Men with Metastatic Castration-resistant Prostate Cancer

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.

In this exploratory study, we analyzed progression data after docetaxel treatment in men with metastatic CRPC who either did or did not receive prior abiraterone treatment. While our analysis is somewhat limited by its small sample size, our findings do support our initial hypothesis that abiraterone pretreatment leads to decreased responsiveness to docetaxel. Of note, a prior, single-arm, retrospective report by Mezynski and colleagues reached the same conclusion [18] . This report builds on that study by directly comparing outcomes after docetaxel treatment between contemporary cohorts of abiraterone-naïve and abiraterone-pretreated patients. It is notable that in our analysis, as well as in the one by Mezynski et al., that the observed median time to PSA progression (ie, PSA-PFS) was quite similar at 4.1 mo and 4.6 mo, respectively; this speaks to the validity of our PSA-PFS estimate in abiraterone-pretreated patients. Importantly, we observed PSA responses to docetaxel in almost one-third of subjects who previously had no PSA response to abiraterone, suggesting that a proportion of abiraterone-unresponsive patients may still derive a benefit from docetaxel.

While ideally this type of analysis would evaluate which sequence of administering abiraterone (ie, before or after docetaxel) leads to the longest OS, this was not feasible due to the small number of death events and the relatively short follow-up. In addition, during the period that this study was conducted, a number of new agents (eg, sipuleucel-T, enzalutamide, cabazitaxel) were approved for use after docetaxel treatment for men with CRPC on the basis of demonstrable survival gains[10], [19], and [20]. Controlling for these agents’ effect on OS independent of the sequence with which abiraterone was given was not feasible. It should be noted that we did assess for differences in the predocetaxel treatment with ketoconazole or antiandrogens (agents commonly used before docetaxel in this population) and found no significant difference between the two groups, indicating that therapies received prechemotherapy were unlikely to affect our results.

Additionally, given the imbalances in baseline characteristics between the two groups, it remains possible that differences in disease severity may have influenced the time to progression. One would expect, however, that if time to progression were only a function of baseline characteristics, and not influenced by cross-resistance between abiraterone and docetaxel, that the effect of docetaxel on PSA characteristics would be similar between groups. The fact that PSA-PFS was significantly different between the docetaxel-only and AD groups supports our initial hypothesis. When considered in the context of the aforementioned work showing that docetaxel may, at least partly, exert its antitumor effect in CRPC through inhibition of AR trafficking via microtubule inhibition, these clinical data support the argument for cross-resistance between taxanes and AR-targeting agents[11], [12], [13], [14], and [15]. Furthermore, these differences in PSA progression and clinical/radiographic progression on docetaxel persisted after differences between groups were accounted for through multivariable analysis.

This analysis has several limitations. First, our abiraterone-pretreated cohort was small, with only 24 subjects (and 20 PSA progression events). This may have led to a low precision in estimating our HRs, as evidenced by wide CIs ( Table 2 ). Second, despite our efforts, there is still the possibility that our analysis suffered from lead-time bias, given that the AD group was almost certainly further along in their disease course upon inclusion in this study. Third, given that this was a retrospective analysis, allocation between cohorts was nonrandom. While the control group was contemporary to the time period that abiraterone was used at our institution, a recent trend toward using abiraterone before docetaxel has been seen at our center and elsewhere. This trend is most appreciated by the fact that the earliest date of docetaxel initiation in the AD group was November 2011, while the latest date of docetaxel initiation in the docetaxel-only group was September 2012. Given that there is <1 yr of overlap between the date of docetaxel initiation between these groups speaks to a shift in practice patterns occurring after abiraterone's US Food and Drug Administration approval in April 2011. Finally, given that the definition for radiographic progression was not prospectively defined, differences in treatment practices among the various physicians caring for those men included in this study may have affected the accuracy of our PFS estimates. In addition to nonuniform scanning intervals, confirmatory CT scans and/or bone scans were not generally ordered, and thus the true PCWG2 criteria for defining progression were not strictly adhered to[4] and [21]. Nonetheless, these data reflect the real-world experience in managing men with metastatic CRPC on docetaxel chemotherapy.

Prostate cancer (PCa) remains a major health concern, resulting in nearly 260 000 annual deaths worldwide [1] . Since Charles Huggins first described the effects of surgical castration on men with metastatic PCa in the 1940s, PCa has been recognized as an androgen-responsive disease[2] and [3]. To date, androgen ablation, either through surgical or medical castration (ie, gonadotropin-releasing hormone [GnRH] agonist/antagonist therapy), has remained the cornerstone of advanced PCa management. Unfortunately, nearly all men progress in spite of castrate testosterone levels, at which point they are described as having castration-resistant prostate cancer (CRPC) ( Fig. 1 ) [4] . In 2004, docetaxel became the first agent shown to produce survival gains in the CRPC setting[5] and [6]. More recently, a number of additional, second-line, androgen-directed therapies (ie, abiraterone and enzalutamide) have been shown to prolong life in men with metastatic CRPC[7], [8], [9], and [10]. At this juncture, the question of how to best sequence these agents with one another and the issue of cross-resistance between them remain inadequately addressed.

Fig. 1 Prostate cancer clinical states model (adapted from Scher et al. [4] ).

The realization that a large proportion of men who develop CRPC remain dependent on androgen/androgen-receptor (AR) signaling has led to a renewed interest in the AR as a therapeutic target. As a result, a number of newer AR-directed therapies have been developed. Mechanistically, these agents primarily work through ligand depletion (eg, abiraterone) or through interference with AR trafficking and signaling (eg, enzalutamide) [7] . When combined with a GnRH agonist, abiraterone decreases testosterone from the castrate level (<50 ng/dl) to the undetectable level (<1 ng/dl) through CYP-17 enzyme inhibition. Abiraterone was initially approved in the postdocetaxel window; however, more recently it has also been approved for prechemotherapy use[8] and [9]. To this end, in the last 1–2 yr, abiraterone use has increased in the prechemotherapy space. However, because the original phase 3 docetaxel data were published in the preabiraterone era, it has been questioned whether the efficacy of docetaxel might be negatively affected by prior abiraterone treatment[5] and [6].

Furthermore, laboratory studies have shown that taxanes may be at least partly effective in the treatment of PCa by disrupting microtubule-mediated AR nuclear trafficking[11], [12], [13], [14], and [15]. In addition to in vitro and in vivo work supporting this purported mechanism of action, there are also microarray and circulating tumor-cell clinical data that support docetaxel's ability to inhibit AR nuclear translocation. In humans, docetaxel has been shown to significantly decrease nuclear localization of the AR compared to docetaxel-naïve controls. Furthermore, through analyzing the circulating tumor cells of subjects treated with docetaxel, Darshan and colleagues demonstrated that patients with PSA progression (>25% PSA increase) on docetaxel were significantly more likely to have nuclear AR localization compared to those with stable or docetaxel-responsive (≥30% PSA decline) disease [14] . These observations raise the possibility of cross-resistance between taxanes and androgen-directed agents such as abiraterone and enzalutamide.

In the case of ketoconazole, a less potent CYP-17 inhibitor than abiraterone, its effects on subsequent responses to docetaxel have been mixed. In subjects pretreated with ketoconazole, one group of investigators reported trends toward shortened progression-free survival (PFS) and overall survival (OS) on docetaxel, while another group found no difference in PFS or OS based on prior ketoconazole treatment status[16] and [17]. Finally, a small (n = 35), single-arm, retrospective analysis of patients treated with abiraterone followed by docetaxel found lower-than-expected docetaxel activity when compared to historic controls [18] . Therefore, the impact of prior abiraterone therapy on subsequent responsiveness to docetaxel remains poorly defined.

We hypothesized that in men with metastatic CRPC, the administration of abiraterone prior to docetaxel would lead to impaired docetaxel efficacy compared with patients that were abiraterone-naïve at the time of docetaxel initiation. Herein, we report the results of a retrospective analysis in patients from a single, large academic center who either had or had not received abiraterone prior to docetaxel treatment. By using data from the abiraterone era, we aim to provide a contemporary perspective on the interaction between these two important agents.