The use of PSA as a prognostic tool in prostate cancer is not novel and has been applied at different disease stages. Multiple studies have described the prognostic value of early PSA decline after starting abiraterone [15–17]. Rescigno et al found that ≥ 30% decline in PSA at 4 weeks was associated with longer overall survival (25.8 vs 15.1 months, HR = 0.47, p < 0.001) in univariate and multivariate analyses. Similarly, Facchini et al concluded that ≥ 50% decline in PSA at 15 days after initiating abiraterone was associated with prolonged rPFS (HR = 0.28, p < 0.01) and longer OS (HR = 0.21, p < 0.01). These findings remained significant with PSA values at 90 days [15]. Finally, Schiff et al investigated the prognostic value of PSA decline post-abiraterone therapy and found a significant association between PSA reductions of both 30% and 50% with improved OS at both the 4-week and 12-week time points [17]. For enzalutamide, Alumkal et al reported that PSA decline ≥ 50% at 12 weeks in patients with mCRPC was significantly associated with longer OS (HR = 4.41, p < 0.001) [18]. However, the studies by Rescigno et al and Schiff et al included both docetaxel-naïve as well as docetaxel-refractory patients, while Alumkal et al specifically studied patients with docetaxel-naïve mCRPC.
The current study adds to the previous literature, suggesting that a ≥ 50% decline in PSA within 3 months of initiating second-generation hormone therapy (2nd HT), specifically in the post-taxane setting, is associated with significant improvements in rPFS and CSS in patients with mCRPC. While previous studies evaluated OS and investigated disease progression using conventional imaging, our study represents the first study to investigate cancer specific survival and disease progression using metabolic imaging (C-11 choline PET scan), which is known have higher sensitivity in detecting disease relapse. In addition, we investigated the prognostic value of disease volume on rPFS and CSM, and as expected found that patients with high disease burden (> 5 metastases) may have poorer oncological outcomes compared to patients with pre-treatment low burden metastatic disease. Specifically, patients with pre-treatment high-disease burden had a nearly 4-fold and 5-fold increased risk for radiographic disease progression and cancer specific mortality, respectively, compared to patients with pre-treatment low-disease burden.
Verzoni et al. examined predictors of long-term response to abiraterone acetate in patients with mCRPC in the post-docetaxel setting and found that only PSA and Gleason score were predictive of long-term treatment response in the multivariate analysis [19]. Specifically, the authors did not find any statistically significant correlation with age, M stage, or site of metastases (bone or visceral). They did not specifically examine the volume of metastases. Chang et al also found that high-volume metastases (defined as the presence of visceral metastases, or ≥ 4 bone lesions with ≥ 1 location outside the vertebral bodies and pelvis) were an independent risk factor for overall survival in post-docetaxel mCRPC in multivariate analysis. PSA decline was treated as an outcome measure in this study [20].
In a previous study by our group, Al-Amiri et al described a radiographic paradoxical response to 2nd HT in which patients demonstrated rDP with either downtrending or stable PSA. In a univariate analysis of predictors of this phenomenon, pre-treatment volume of disease, presence of bony metastases, and local disease were statically significant predictors of this paradoxical response. On the other hand, PSA flare phenomenon has been reported with the use of abiraterone and enzalutamide [21, 22]. Ueda et al examined 83 patients undergoing treatment with AA for mCRPC and found that the incidence ranged from 6.0-10.8%, depending on the definition of PSA flare used [23]. PFS was not different among patients with PSA flare as compared to those patients without. Armstrong et al noted a significantly lesser incidence of PSA flare associated with enzalutamide at < 1% [22].
These data may have implications in both monitoring and treatment selection in mCRPC. In settings where regular radiographic monitoring is not feasible, a PSA decline ≥ 50% at 3 months could serve as a stratification criterion, wherein patients who do not meet this PSA decline cutoff are selected for imaging. Indeed, a prospective genomic and transcriptomic study of patients with mCRPC treated initially with enzalutamide showed that “non-responders” (defined as PSA decline < 50%) had a different tumor profile on pre-treatment biopsy [18]. Specifically, nonresponding tumors demonstrated lower androgen receptor transcriptional activity and a more stem-cell-like profile, implying that mCRPC non-responsive to 2nd HT may represent a divergent tumor phenotype. Further research is warranted regarding whether the absence of early PSA decline should change management, and if so, which subsequent therapeutic agent is optimal.
While this study demonstrates profound differences in outcomes between patients with and without a PSA decline > 50%, this study was not without limitations. The retrospective nature of the study comes with inherent limitations regarding the comparability of the two groups. Additionally, the sample size was relatively small. The study was limited to a single institution. Finally, only patients who underwent PSA testing within 3 months after initiating 2nd HT were included in the study.
Finally, future work is needed to determine the optimal PSA decline cutoff for stratification. PSA declines of 30% and 50% are frequently studied as prognostic factors, but most of these studies are too limited in sample size to study a range of cutoffs. Larger studies would provide the investigation of an optimal time point to monitor PSA decline. A prospective, multi-institutional study investigating outcomes in relation to the magnitude and timing of PSA decline could help validate our findings and define the optimal PSA monitoring regimen in post-docetaxel mCRPC treated with 2nd HT.