Methods - We conducted a retrospective analysis of 1238 patients diagnosed with nmCRPC in 2000–2015 in the SEARCH database. Multivariable Cox proportional hazards and competing risks regression were used to determine the hazards of overall, prostate cancer-specific (PCSM), and other-cause mortality (OCM) across age, Charlson comorbidity index (CCI), and PSADT subgroups.
Results - Men with nmCRPC were elderly (median age 77) and had substantial comorbidity burdens (CCI > 1 n = 701, 57%). Multivariable Cox analysis showed higher CCI was associated with higher hazard of OCM, while slower PSADT was associated with lower hazard of PCSM across all age subgroups. Among those with CCI ≥ 3 (vs. CCI0), the hazard ratio of OCM was 2.7 (95% CI 1.1–6.3), 2.0 (95% CI 1.1–3.6), and 2.5 (95% CI 1.5–4.0) for those aged <70, 70–79, and ≥80, respectively. Among those with PSADT ≥ 9 months (vs. < 9 months), the hazard ratios for PCSM were 0.5 (95% CI 0.3–0.9), 0.6 (95% CI 0.4–0.9), and 0.6 (95% CI 0.4–0.9) for those aged <70, 70–79, and ≥80. Competing risks curves revealed PCSM was the predominant cause of death for those with PSADT < 9 months across all age and comorbidity groups. PCSM and OCM were relatively equal competitors for mortality among those with PSADT≥9 months except those aged > 80 with CCI ≥ 3, in whom OCM was the predominant cause of death.
Conclusions - Among men with nmCRPC, age, comorbidity, and PSADT are associated with risk and cause of death and may assist clinicians in counseling patients regarding cancer prognosis.
Introduction
Among men diagnosed with prostate cancer (PC), an estimated 10–20% develop castration-resistance (CRPC) within five years of diagnosis1. While the majority of men who progress to CRPC have metastases and a generally poor cancer prognosis with median overall survival of only 27– 33 months1-3, those with non-metastatic CRPC (nmCRPC) generally have a more protracted disease course4 with substantial variability in cancer prognosis. Our group recently found that PSA doubling time (PSADT) at the time of nmCRPC strongly predicts risk of metastasis and prostate cancer-specific mortality (PCSM); among 441 men with nmCRPC, median cancer-specific survival for patients with PSADT < 3, 3–8.9, 9–14.9, and ≥ 15 months was 15, 40, 60, and > 60 months, respectively5. These data show that cancer prognosis among men with nmCRPC is highly variable, with some men having low rates of cancer mortality at long-term endpoints.
Given that nmCRPC can be indolent and generally affects older men (median age at diagnosis in mid-70s1), another important consideration in assessment of prognosis is the competing risk of other-cause mortality (OCM)6. The impact of life expectancy— vis-à-vis its main predictors, age and comorbidity7— on competing risks for mortality has been explored at other points during the prostate cancer disease course, most notably in early-stage disease8,9. However, the competing risk of OCM may also be relevant in advanced disease, when patients are either older and/or when the disease is indolent. For example, in a study of 225 men with biochemical recurrence (BCR) after radical prostatectomy, the cumulative incidence of cancer mortality was only 3%, 11%, and 21% at 5, 10, and 15 years, respectively10. This would suggest that for older and sicker men experiencing BCR after surgery, the risk of OCM is a major factor in determining prognosis. It is unclear whether competing risks would have a similar impact on an even more advanced (and potentially more lethal) stage of disease such as nmCRPC.
Competing risks for mortality are also particularly relevant to treatment decision making for men with nmCRPC, since novel androgen deprivation therapies (ADTs) previously reserved for patients with metastatic disease have now shown efficacy in the non-metastatic setting. The reported results of the SPARTAN and PROSPER trials recently reported showed that both apalutamide and enzalutamide improved time to progression and metastasis-free survival when added to traditional androgen deprivation with luteinizing hormone releasing hormone (LHRH) agonists11,12. However, these therapies were also shown to increase risks of major side effects compared with placebo. Understanding whether a man is highly likely to die of prostate cancer or other causes may inform decisions regarding whether to pursue more aggressive treatment or to defer these additional therapies in order to minimize harms and maximize benefits.
In this study, we sought to examine the impact of age, comorbidity, and PSADT on all-cause mortality (ACM), PCSM, and OCM in a large sample of men with nmCRPC. Because the cancer prognosis associated with nmCRPC is variable and mainly affects older men, we hypothesized that these risk factors would help to predict competing risks of OCM and PCSM to help patients better select treatments commensurate with the aggressiveness of their disease.
Materials and Methods
Study Cohort: After obtaining IRB approval, we sampled 1292 men diagnosed with non-metastatic (stage M0/Mx) CRPC regardless of primary treatment modality between 2000 and 2015 from eight Veterans Affairs (VA) medical centers in the SEARCH database (Palo Alto, San Diego, San Francisco, and West LA, CA; Augusta, GA; Asheville and Durham, NC; Portland, OR). We previously reported methods used to identify these patients13. CRPC was defined as a PSA rise of ≥ 2 and ≥ 25% from the post-ADT nadir while being castrate14. Castration was defined as testosterone < 50 ng/dL, bilateral orchiectomy, or continuous receipt of LHRH agonist or antagonist. nmCRPC was defined as the absence of a positive imaging test (bone scan, MRI, CT, X-ray) for distant metastases before CRPC diagnosis. We excluded 4% (n = 54) who developed metastases within one month of nmCRPC diagnosis. Details on the abstraction of imaging data have been described previously15,16. Our final analytic sample included 1238 men.
Primary Predictors:
- Charlson comorbidity index (CCI): The CCI is a weighted comorbidity index that estimates life expectancy based on the presence or absence of specific comorbidities. We assessed comorbidities using claims based codes as defined by Deyo et al. prior to the date of nmCRPC diagnosis17. We omitted prostate cancer and metastatic prostate cancer from CCI calculations.
- PSA doubling time: PSADT at the time of initial nmCRPC diagnosis was calculated as the natural log of two divided by the slope of the Impact of age, comorbidity, and PSA doubling time on long-term competing risks for mortality among men. . . 253 linear regression of the natural log of PSA over time in months. Subjects with a calculated PSADT < 0 or >120 were assigned 120 months for ease of analysis. All PSA values 2 years before CRPC diagnosis but after the post ADT PSA nadir were used to calculate PSADT. This calculation required at least two PSA values over at least 3 months. PSADT cutoffs of < 9 months vs. ≥ 9 months were determined a priori based on previous work5.
Covariates
Covariates collected at the time of nmCRPC diagnosis included age at CRPC, race, year of CRPC diagnosis, VA treatment center, biopsy Gleason score, primary treatment, months from ADT to CRPC, PSA at CRPC, and PSA velocity at CRPC. XRT + ADT was defined as radiation therapy with concomitant ADT within 12 months of diagnosis.
Primary outcomes
Baseline for all survival outcomes was the date of CRPC. Cause of death was defined as PCSM or OCM based on electronic medical record chart review. Men were considered to have died from PC if they had progressive prostate cancer metastases resistant to castration and death without another probable cause, also taking into account if they had renal failure with ureteral obstruction and hydronephrosis, had locally advanced disease, and/or had entered hospice for PC. Non-deceased patients were followed to date of last contact with the VA Health System.
Statistical analysis
Sample characteristics were compared across age groups (<70, 70–79, and ≥80) using Kruskal Wallis tests for continuous variables and chi-square tests for categorical variables. Within each age group, men were divided into subgroups based on CCI (0, 1, 2, and ≥3) and PSADT (<9 months, ≥9 months). Multivariable Cox proportional hazards analysis was used to determine the hazard of ACM associated with age, CCI, and PSADT. Competing risk regression analysis was used to determine the hazards of PCSM and OCM associated with age, CCI, and PSADT18. Cumulative incidence curves depicting PCSM vs. OCM over time for each age, CCI, and PSADT group were created based on competing risks regression. Stata v14.0 (Stata Corp., College Station, TX) was used for all statistical analyses. P < 0.05 was the threshold for statistical significance.
Results
Baseline characteristics
Men in our sample were generally elderly (median age 77, IQR 69–82) and had substantial comorbidity burdens (CCI > 1 n = 701, 57%) (Table 1). Overall median PSADT was 8.0 months, with older men having slightly higher PSADT (P < 0.001). Median follow-up after CRPC was 33.5 months (IQR 19.5, 54.1). Over the period of observation, half of the patients (56%) developed distant metastases, and over two-thirds (72%) died, 46% from cancer and 26% from other causes.
All-cause mortality, prostate cancer-specific mortality, and other-cause mortality
Cox proportional hazards and competing risk regression analyses revealed that higher PSADT was strongly associated with lower risk of PCSM across all age groups and ACM in those older than 70 (Table 2). Among menwith PSADT ≥ 9 months (vs. <9 months), the hazards of ACM were 0.6 (P = 0.003; 95% CI 0.4–0.8) and 0.6 (P = 0.007; 95% CI 0.5–0.9) for those aged 70–79 and ≥ 80, respectively. Among men with PSADT ≥ 9 months (vs. <9 months), the hazards of PCSM were 0.5 (P = 0.02; 95% CI 0.3–0.9), 0.6 (P = 0.01; 95% CI 0.4–0.9), and 0.6 (P = 0.016; 95% CI 0.4–0.9) for those aged <70, 70–79, and ≥ 80, respectively. PSADT was not associated with OCM.
Cox proportional hazards and competing risks regression analysis showed that higher CCI was associated with elevated risk of OCM across all age groups and ACM in those older than 70 (Table 2). Multivariable analysis showed that among those with CCI ≥ 3 (vs. 0), hazard ratios for OCM were 2.7 (95% CI 1.1–6.6), 2.0 (95% CI 1.1–3.6), and 2.5 (95% CI 1.5–4.0) for those aged <70, 70–79, and ≥ 80, respectively. Among men with CCI ≥ 3 (vs. 0), the hazards of ACM were 1.4 (P = 0.03; 95% CI 1.1–1.9) and 2.1 (P < 0.001; 95% CI 1.6–2.9) for those aged 70–79 and ≥80, respectively. CCI was not associated with PCSM.
Competing risks regression curves depict the competing risks of mortality by age, CCI, and PSADT (Fig. 1). PCSM was by far the predominant cause of death for men with PSADT < 9 months across all age groups. OCM and PCSM were relatively equal competitors for mortality in nearly all patients with PSADT ≥ 9 months; OCM was the predominant cause of death only in men older than 80 with CCI ≥ 3. Five-year cumulative incidence estimates for OCM and PCSM are shown in Table 3.
Fig. 1 Cumulative incidence of prostate cancer-specific and other-cause mortality by age, comorbidity, and PSADT for men (a) < 70, (b) 70–79, and (c) ≥ 80 years
Discussion
Our study suggests that a significant minority of men with nmCRPC do not die of prostate cancer and that simple clinical indicator can predict risk of death from prostate cancer vs. other causes. Among our sample of 1238 men diagnosed with nmCRPC, mortality was due to causes other than prostate cancer for 36% of the patients who died during the period of follow-up. We found that a combination of age, comorbidity, and PSADT could strongly risk stratify cause of death among these men. Specifically, those with a PSADT < 9 months showed markedly increased risk of PCSM regardless of age or comorbidity, and among those with PSADT ≥ 9 months, risk of OCM increased with higher comorbid disease burden, and thus OCM was a relatively equal competitor for mortality with PCSM for the vast majority of patients and the predominant cause of death in men over 80 with CCI scores ≥3. Understanding the competing risks of PCSM and OCM is a critical consideration when counseling patients regarding prognosis and treatment of advanced PC, especially when balancing risks and benefits of additional androgen axis targeting agents such as apalutamide and enzalutamide in this setting.
Surprisingly little data exist regarding the epidemiology and treatment of men with nmCRPC, in part because definitions of the castration-resistant disease have been inconsistent over time and rely on PSA data that is not routinely captured in large secondary datasets19. Furthermore, men with nmCRPC have been historically excluded from clinical trials testing the efficacy of advanced PC drugs, which have been aimed primarily at men with metastatic CRPC19,20. For these reasons, there is a lack of longitudinal epidemiological data on men with nmCRPC19-21. This study leverages the unique advantages of the SEARCH database, which captures PSA data on a retrospective cohort of men with nmCRPC to provide more definitive estimates of long-term prognosis of men with nmCRPC. Of note, our data represent long-term follow-up of men with nmCRPC treated with the standard of care before widespread use of immunotherapies, so the survival estimates can serve as a baseline to compare with future populations.
While not directly addressing treatment, our results raise questions about whether information on competing risks may be used to identify patients who are most and least likely to benefit from addition of apalutamide and enzalutamide therapies to traditional androgen deprivation with LHRH agonists. The recently reported results from the SPARTAN and PROSPER trials show that apalutamide and enzalutamide improve metastasis-free survival over ADT alone, though both incurred significant morbidity over placebo11, 12, 21. In both cases, the key management issue may be how to best identify those most at risk for progression and, consequently, those who will likely benefit the most from treatment19. Our data may be useful in identifying patients who may benefit from a less aggressive approach to treatment. For example, it is conceivable that among patients with long PSADT and heavy comorbid disease burdens, the cardiovascular and metabolic risks of continuous ADT (or morbidity associated with the addition of enzalutamide) may outweigh the benefits23-25. Conversely, competing risks data may also help to identify patients at high risk for PCSM in whom benefits of adding these therapies would certainly outweigh the risks; interestingly, even among healthy men older than 80, the risk of PCSM appears to be substantial enough to warrant a more aggressive treatment approach. Future clinical trials targeted at these subgroups (or subgroup analysis of existing clinical trials) may provide definitive answers to these questions.
Our study shows that age, comorbidity, and tumor risk (vis-à-vis PSADT) are all key factors in predicting prognosis and cause of death in men with nmCRPC, similar to other time points in the prostate cancer disease course26. While the risk of PCSM persisted regardless of age and comorbidity—as we would expect in this group of patients with advanced disease6 —the relative risks of OCM vs. PCSM were nonetheless clearly influenced by each of these factors. In our cohort, for men with PSADT ≥ 9 months, older age, and higher CCI were indicators of higher risk of OCM. On the other hand, PCSM was the predominant cause of death for all patients with short PSADT (<9 months). These findings are consistent with previous studies showing that PSA kinetics is the most powerful indicator of PC aggressiveness among men diagnosed with nmCRPC20. However, by including factors that influence OCM (i.e., age, comorbidity), our comprehensive risk stratification paradigm improves discrimination of cause of death over risk stratification based on tumor risk factors alone.
Our study has several limitations. First, due to the retrospective nature of our study, several important predictors of prognosis such as functional status and frailty were not available. Second, our cohort only included patients from the VA, which may affect external generalizability to other patient populations. Third, metastatic workup at the time of diagnosis of CRPC was not uniform, and was left to the discretion of the treating physician, which may affect the homogeneity of our cohort (i.e., compared with a cohort of men who underwent requisite imaging prior to enrollment). It is also important to note that changing practice patterns with newer imaging modalities, in particular, PSMA PET, are being increasingly utilized in the nmCRPC setting and may reduce heterogeneity in prognosis by improving identification of men with distant metastases. Fourth, we did Table 3 Competing risks regression 5-year cumulative incidence estimates for prostate cancer-specific, and other-cause mortality by age, comorbidity, and PSA doubling time CCI 0 CCI 1–2 CCI 3+ OCM PCSM OCM PCSM OCM PCSM Age < 70 PSADT < 9 months 7% 29% 0% 58% 12% 47% PSADT ≥ 9 months 0% 0% 4% 4% 14% 18% Age 70–79 PSADT < 9 months 6% 52% 5% 63% 24% 54% PSADT ≥ 9 months 0% 4% 18% 12% 10% 27% Age ≥ 80 PSADT < 9 months 19% 30% 12% 56% 38% 47% PSADT ≥ 9 months 4% 11% 30% 29% 51% 21% All models are adjusted for age at CRPC, race, year of CRPC, treatment center, biopsy Gleason score, PSA at CRPC, primary treatment, months from ADT to CRPC, PSADT, and PSAV 258 C. A. Whitney et al. not have any information on whether patients elected hospice or palliative care rather than active management of disease. Fifth, care outside of the VA was not routinely captured in the dataset, which may affect the ascertainment of metastatic disease status. Lastly, we did not account for secondary treatments that could affect survival, although most novel medications were not approved until after most patients had died or been censored as lost to follow-up.
Conclusions
The current literature reveals the need for a more accurate assessment of nmCRPC prognosis since the natural history of prostate cancer within this disease state is highly variable19,21. Our study suggests that age, comorbidity, and PSADT are all critical in determining the risk of long-term cancer and non-cancer mortality in these men. These data may also be relevant for patients and physicians trying to decide whether or not to add potentially morbid additional androgen axis inhibitors to traditional ADT, particularly the oldest and sickest patients who are at high risk of morbidity from these medications. Integrating simple clinical variables such as these into decision making represents a key component of precision medicine that is too often ignored. We strongly believe that these factors should be taken into account when counseling patients about prognosis, and they may be important in selecting subgroups for future clinical trials in men with nmCRPC.
Acknowledgments: Dr. Daskivich would like to acknowledge funding from K08 CA230155. Dr. Aronson would like to acknowledge funding from RO1 CA231219.
Compliance with ethical standards: Conflict of interest Dr. Daskivich is a consultant for Janssen Research and Development, LLC. Dr. Freedland is a consultant for Astellas Pharmaceuticals and Janssen Research and Development. Other authors declare that they have no conflict of interest.
Authors: Colette A. Whitney,1 Lauren E. Howard,1,2 Stephen J. Freedland,1,3 Amanda M. DeHoedt,1 Christopher L. Amling,4 William J. Aronson,5 Matthew R. Cooperberg,6 Christopher J. Kane,7 Martha K. Terris,8,9 Timothy J. Daskivich3
1. Division of Urology, Durham Veterans Affairs Medical Center, Durham, NC, USA
2. Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham, NC, USA
3. Division of Urology, Cedars-Sinai Medical Center, Los Angeles, CA, USA
4. Department of Urology, Oregon Health Sciences University, Portland, OR, USA
5. Division of Urology, West Los Angeles Veterans Affairs Medical Center, Los Angeles, CA, USA
6. Department of Urology, University of California, San Francisco, CA, USA
7. Department of Urology, University of California, San Diego, CA, USA
8. Division of Urology, Charlie Norwood Veterans Affairs Medical Center, Augusta, GA, USA
9. Division of Urology, Medical College of Georgia, Augusta, GA, USA
References
1. Kirby M, Hirst C, Crawford ED. Characterising the castration resistant prostate cancer population: a systematic review. Int J Clin Pract. 2011;65:1180–92.
2. Ryan CJ, Smith MR, de Bono JS, Molina A, Logothetis CJ, de Souza P, et al. Abiraterone in metastatic prostate cancer without previous chemotherapy. N Engl J Med. 2013;368:138–48.
3. Beer TM, Armstrong AJ, Rathkopf DE, Loriot Y, Sternberg CN, Higano CS, et al. Enzalutamide in metastatic prostate cancer before chemotherapy. N Engl J Med. 2014;371:424–33.
4. D’Amico AV, Schultz D, Schneider L, Hurwitz M, Kantoff PW, Richie JP. Comparing prostate specific antigen outcomes after different types of radiotherapy management of clinically localized prostate cancer highlights the importance of controlling for established prognostic factors. J Urol. 2000;163:1797–801.
5. Howard LE, Moreira D, De Hoedt A, Aronson WJ, Kane CJ, Amling CL, et al. Thresholds for PSA doubling time in men with non-metastatic castration-resistant prostate cancer. BJU Int. 2017;120(5B):E80–E86.
6. Read WL, Tierney RM, Page NC, Costas I, Govindan R, Spitznagel EL, et al. Differential prognostic impact of comorbidity. J Clin Oncol. 2004;22:3099–103.
7. Daskivich TJ, Fan KH, Koyama T, Albertsen PC, Goodman M, Hamilton AS, et al. Effect of age, tumor risk, and comorbidity on competing risks for survival in a U.S. population-based cohort of men with prostate cancer. Ann Intern Med. 2013;158:709–17.
8. Daskivich TJ, Chamie K, Kwan L, Labo J, Dash A, Greenfield S, et al. Comorbidity and competing risks for mortality in men with prostate cancer. Cancer. 2011;117:4642–50.
9. Albertsen PC, Moore DF, Shih W, Lin Y, Li H, Lu-Yao GL. Impact of comorbidity on survival among men with localized prostate cancer. J Clin Oncol. 2011;29:1335–41.
10. Uchio EM, Aslan M, Wells CK, Calderone J, Concato J. Impact of biochemical recurrence in prostate cancer among US veterans. Arch Intern Med. 2010;170:1390–5.
11. Small EJ, Saad F, Chowdhury S, Hadaschik BA, Graff JN, Olmos D, et al. SPARTAN, a phase 3 double-blind, randomized study of apalutamide (APA) versus placebo (PBO) in patients (pts) with nonmetastatic castration-resistant prostate cancer (nmCRPC). J Clin Oncol. 2018;36:161.
12. Hussain M, Fizazi K, Saad F, Rathenborg P, Shore ND, Demirhan E, et al. PROSPER: a phase 3, randomized, double-blind, placebo (PBO)-controlled study of enzalutamide (ENZA) in men with nonmetastatic castration-resistant prostate cancer (M0 CRPC). J Clin Oncol. 2018;36.
13. Moreira DM, Howard LE, Sourbeer KN, Amarasekara HS, Chow LC, Cockrell DC, et al. Predicting bone scan positivity in nonmetastatic castration-resistant prostate cancer. Prostate Cancer Prostatic Dis. 2015;18:333–7.
14. Scher HI, Morris MJ, Basch E, Heller G. End points and outcomes in castration-resistant prostate cancer: from clinical trials to clinical practice. J Clin Oncol. 2011;29:3695–704.
15. Howard LE, Moreira DM, De Hoedt A, Aronson WJ, Kane CJ, Amling CL, et al. Thresholds for PSA doubling time in men with non-metastatic castration-resistant prostate cancer. BJU Int. 2017;120(5b):E80–e6.
16. Hanyok BT, Howard LE, Amling CL, Aronson WJ, Cooperberg MR, Kane CJ, et al. Is computed tomography a necessary part of a metastatic evaluation for castration-resistant prostate cancer? Results from the Shared Equal Access Regional Cancer Hospital Database. Cancer. 2016;122:222–9.
17. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–9.
18. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496– 509.
19. Rozet F, Roumeguere T, Spahn M, Beyersdorff D, Hammerer P. Non-metastatic castrate-resistant prostate cancer: a call for improved guidance on clinical management. World J Urol. 2016;34:1505–13.
20. Tombal B. Non-metastatic CRPC and asymptomatic metastatic CRPC: which treatment for which patient? Ann Oncol. 2012;23 (Suppl 10):x251–8.
21. Gillessen S, Omlin A, Attard G, de Bono JS, Efstathiou E, Fizazi K, et al. Management of patients with advanced prostate cancer: recommendations of the St Gallen Advanced Prostate Cancer Consensus Conference (APCCC) 2015. Ann Oncol. 2016. Impact of age, comorbidity, and PSA doubling time on long-term competing risks for mortality among men. . . 259
22. Pfizer and Astellas announce positive top-line results from phase 3 PROSPER Trial of XTANDI (enzalutamide) in patients with non-metastatic castration-resistant prostate cancer 2017. Available from: http://press.pfizer.com/press-release/pfizer-and-a-stellas-announce-positive-top-line-results-phase-3-prosper-trialxtandi-en.
23. Saigal CS, Gore JL, Krupski TL, Hanley J, Schonlau M, Litwin MS, et al. Androgen deprivation therapy increases cardiovascular morbidity in men with prostate cancer. Cancer. 2007;110: 1493–500.
24. Keating NL, O’Malley AJ, Smith MR. Diabetes and cardiovascular disease during androgen deprivation therapy for prostate cancer. J Clin Oncol. 2006;24:4448–56.
25. Keating NL, O’Malley AJ, Freedland SJ, Smith MR. Diabetes and cardiovascular disease during androgen deprivation therapy: observational study of veterans with prostate cancer. J Natl Cancer Inst. 2010;102:39–46.
26. Daskivich TJ, Litwin MS, Penson DF. Effect of age, tumor risk, and comorbidity in a U.S. population-based cohort of men with prostate cancer. Ann Intern Med. 2013;159:371.
Whitney, C., Howard, L., Freedland, S., DeHoedt, A., Amling, C., & Aronson, W. et al. (2018). Impact of age, comorbidity, and PSA doubling time on long-term competing risks for mortality among men with non-metastatic castration-resistant prostate cancer. Prostate Cancer And Prostatic Diseases, 22(2), 252-260. doi: 10.1038/s41391-018-0095-0
Received: 4 June 2018 / Revised: 31 July 2018 / Accepted: 26 August 2018 / Published online: 2 October 2018 © Springer Nature Limited 2018
Read More: A Commentary from the Associate Editor of PCAN
Given that nmCRPC can be indolent and generally affects older men (median age at diagnosis in mid-70s1), another important consideration in assessment of prognosis is the competing risk of other-cause mortality (OCM)6. The impact of life expectancy— vis-à-vis its main predictors, age and comorbidity7— on competing risks for mortality has been explored at other points during the prostate cancer disease course, most notably in early-stage disease8,9. However, the competing risk of OCM may also be relevant in advanced disease, when patients are either older and/or when the disease is indolent. For example, in a study of 225 men with biochemical recurrence (BCR) after radical prostatectomy, the cumulative incidence of cancer mortality was only 3%, 11%, and 21% at 5, 10, and 15 years, respectively10. This would suggest that for older and sicker men experiencing BCR after surgery, the risk of OCM is a major factor in determining prognosis. It is unclear whether competing risks would have a similar impact on an even more advanced (and potentially more lethal) stage of disease such as nmCRPC.
Competing risks for mortality are also particularly relevant to treatment decision making for men with nmCRPC, since novel androgen deprivation therapies (ADTs) previously reserved for patients with metastatic disease have now shown efficacy in the non-metastatic setting. The reported results of the SPARTAN and PROSPER trials recently reported showed that both apalutamide and enzalutamide improved time to progression and metastasis-free survival when added to traditional androgen deprivation with luteinizing hormone releasing hormone (LHRH) agonists11,12. However, these therapies were also shown to increase risks of major side effects compared with placebo. Understanding whether a man is highly likely to die of prostate cancer or other causes may inform decisions regarding whether to pursue more aggressive treatment or to defer these additional therapies in order to minimize harms and maximize benefits.
In this study, we sought to examine the impact of age, comorbidity, and PSADT on all-cause mortality (ACM), PCSM, and OCM in a large sample of men with nmCRPC. Because the cancer prognosis associated with nmCRPC is variable and mainly affects older men, we hypothesized that these risk factors would help to predict competing risks of OCM and PCSM to help patients better select treatments commensurate with the aggressiveness of their disease.
Materials and Methods
Study Cohort: After obtaining IRB approval, we sampled 1292 men diagnosed with non-metastatic (stage M0/Mx) CRPC regardless of primary treatment modality between 2000 and 2015 from eight Veterans Affairs (VA) medical centers in the SEARCH database (Palo Alto, San Diego, San Francisco, and West LA, CA; Augusta, GA; Asheville and Durham, NC; Portland, OR). We previously reported methods used to identify these patients13. CRPC was defined as a PSA rise of ≥ 2 and ≥ 25% from the post-ADT nadir while being castrate14. Castration was defined as testosterone < 50 ng/dL, bilateral orchiectomy, or continuous receipt of LHRH agonist or antagonist. nmCRPC was defined as the absence of a positive imaging test (bone scan, MRI, CT, X-ray) for distant metastases before CRPC diagnosis. We excluded 4% (n = 54) who developed metastases within one month of nmCRPC diagnosis. Details on the abstraction of imaging data have been described previously15,16. Our final analytic sample included 1238 men.
Primary Predictors:
- Charlson comorbidity index (CCI): The CCI is a weighted comorbidity index that estimates life expectancy based on the presence or absence of specific comorbidities. We assessed comorbidities using claims based codes as defined by Deyo et al. prior to the date of nmCRPC diagnosis17. We omitted prostate cancer and metastatic prostate cancer from CCI calculations.
- PSA doubling time: PSADT at the time of initial nmCRPC diagnosis was calculated as the natural log of two divided by the slope of the Impact of age, comorbidity, and PSA doubling time on long-term competing risks for mortality among men. . . 253 linear regression of the natural log of PSA over time in months. Subjects with a calculated PSADT < 0 or >120 were assigned 120 months for ease of analysis. All PSA values 2 years before CRPC diagnosis but after the post ADT PSA nadir were used to calculate PSADT. This calculation required at least two PSA values over at least 3 months. PSADT cutoffs of < 9 months vs. ≥ 9 months were determined a priori based on previous work5.
Covariates
Covariates collected at the time of nmCRPC diagnosis included age at CRPC, race, year of CRPC diagnosis, VA treatment center, biopsy Gleason score, primary treatment, months from ADT to CRPC, PSA at CRPC, and PSA velocity at CRPC. XRT + ADT was defined as radiation therapy with concomitant ADT within 12 months of diagnosis.
Primary outcomes
Baseline for all survival outcomes was the date of CRPC. Cause of death was defined as PCSM or OCM based on electronic medical record chart review. Men were considered to have died from PC if they had progressive prostate cancer metastases resistant to castration and death without another probable cause, also taking into account if they had renal failure with ureteral obstruction and hydronephrosis, had locally advanced disease, and/or had entered hospice for PC. Non-deceased patients were followed to date of last contact with the VA Health System.
Statistical analysis
Sample characteristics were compared across age groups (<70, 70–79, and ≥80) using Kruskal Wallis tests for continuous variables and chi-square tests for categorical variables. Within each age group, men were divided into subgroups based on CCI (0, 1, 2, and ≥3) and PSADT (<9 months, ≥9 months). Multivariable Cox proportional hazards analysis was used to determine the hazard of ACM associated with age, CCI, and PSADT. Competing risk regression analysis was used to determine the hazards of PCSM and OCM associated with age, CCI, and PSADT18. Cumulative incidence curves depicting PCSM vs. OCM over time for each age, CCI, and PSADT group were created based on competing risks regression. Stata v14.0 (Stata Corp., College Station, TX) was used for all statistical analyses. P < 0.05 was the threshold for statistical significance.
Results
Baseline characteristics
Men in our sample were generally elderly (median age 77, IQR 69–82) and had substantial comorbidity burdens (CCI > 1 n = 701, 57%) (Table 1). Overall median PSADT was 8.0 months, with older men having slightly higher PSADT (P < 0.001). Median follow-up after CRPC was 33.5 months (IQR 19.5, 54.1). Over the period of observation, half of the patients (56%) developed distant metastases, and over two-thirds (72%) died, 46% from cancer and 26% from other causes.
All-cause mortality, prostate cancer-specific mortality, and other-cause mortality
Cox proportional hazards and competing risk regression analyses revealed that higher PSADT was strongly associated with lower risk of PCSM across all age groups and ACM in those older than 70 (Table 2). Among menwith PSADT ≥ 9 months (vs. <9 months), the hazards of ACM were 0.6 (P = 0.003; 95% CI 0.4–0.8) and 0.6 (P = 0.007; 95% CI 0.5–0.9) for those aged 70–79 and ≥ 80, respectively. Among men with PSADT ≥ 9 months (vs. <9 months), the hazards of PCSM were 0.5 (P = 0.02; 95% CI 0.3–0.9), 0.6 (P = 0.01; 95% CI 0.4–0.9), and 0.6 (P = 0.016; 95% CI 0.4–0.9) for those aged <70, 70–79, and ≥ 80, respectively. PSADT was not associated with OCM.
Cox proportional hazards and competing risks regression analysis showed that higher CCI was associated with elevated risk of OCM across all age groups and ACM in those older than 70 (Table 2). Multivariable analysis showed that among those with CCI ≥ 3 (vs. 0), hazard ratios for OCM were 2.7 (95% CI 1.1–6.6), 2.0 (95% CI 1.1–3.6), and 2.5 (95% CI 1.5–4.0) for those aged <70, 70–79, and ≥ 80, respectively. Among men with CCI ≥ 3 (vs. 0), the hazards of ACM were 1.4 (P = 0.03; 95% CI 1.1–1.9) and 2.1 (P < 0.001; 95% CI 1.6–2.9) for those aged 70–79 and ≥80, respectively. CCI was not associated with PCSM.
Competing risks regression curves depict the competing risks of mortality by age, CCI, and PSADT (Fig. 1). PCSM was by far the predominant cause of death for men with PSADT < 9 months across all age groups. OCM and PCSM were relatively equal competitors for mortality in nearly all patients with PSADT ≥ 9 months; OCM was the predominant cause of death only in men older than 80 with CCI ≥ 3. Five-year cumulative incidence estimates for OCM and PCSM are shown in Table 3.
Fig. 1 Cumulative incidence of prostate cancer-specific and other-cause mortality by age, comorbidity, and PSADT for men (a) < 70, (b) 70–79, and (c) ≥ 80 years
Discussion
Our study suggests that a significant minority of men with nmCRPC do not die of prostate cancer and that simple clinical indicator can predict risk of death from prostate cancer vs. other causes. Among our sample of 1238 men diagnosed with nmCRPC, mortality was due to causes other than prostate cancer for 36% of the patients who died during the period of follow-up. We found that a combination of age, comorbidity, and PSADT could strongly risk stratify cause of death among these men. Specifically, those with a PSADT < 9 months showed markedly increased risk of PCSM regardless of age or comorbidity, and among those with PSADT ≥ 9 months, risk of OCM increased with higher comorbid disease burden, and thus OCM was a relatively equal competitor for mortality with PCSM for the vast majority of patients and the predominant cause of death in men over 80 with CCI scores ≥3. Understanding the competing risks of PCSM and OCM is a critical consideration when counseling patients regarding prognosis and treatment of advanced PC, especially when balancing risks and benefits of additional androgen axis targeting agents such as apalutamide and enzalutamide in this setting.
Surprisingly little data exist regarding the epidemiology and treatment of men with nmCRPC, in part because definitions of the castration-resistant disease have been inconsistent over time and rely on PSA data that is not routinely captured in large secondary datasets19. Furthermore, men with nmCRPC have been historically excluded from clinical trials testing the efficacy of advanced PC drugs, which have been aimed primarily at men with metastatic CRPC19,20. For these reasons, there is a lack of longitudinal epidemiological data on men with nmCRPC19-21. This study leverages the unique advantages of the SEARCH database, which captures PSA data on a retrospective cohort of men with nmCRPC to provide more definitive estimates of long-term prognosis of men with nmCRPC. Of note, our data represent long-term follow-up of men with nmCRPC treated with the standard of care before widespread use of immunotherapies, so the survival estimates can serve as a baseline to compare with future populations.
While not directly addressing treatment, our results raise questions about whether information on competing risks may be used to identify patients who are most and least likely to benefit from addition of apalutamide and enzalutamide therapies to traditional androgen deprivation with LHRH agonists. The recently reported results from the SPARTAN and PROSPER trials show that apalutamide and enzalutamide improve metastasis-free survival over ADT alone, though both incurred significant morbidity over placebo11, 12, 21. In both cases, the key management issue may be how to best identify those most at risk for progression and, consequently, those who will likely benefit the most from treatment19. Our data may be useful in identifying patients who may benefit from a less aggressive approach to treatment. For example, it is conceivable that among patients with long PSADT and heavy comorbid disease burdens, the cardiovascular and metabolic risks of continuous ADT (or morbidity associated with the addition of enzalutamide) may outweigh the benefits23-25. Conversely, competing risks data may also help to identify patients at high risk for PCSM in whom benefits of adding these therapies would certainly outweigh the risks; interestingly, even among healthy men older than 80, the risk of PCSM appears to be substantial enough to warrant a more aggressive treatment approach. Future clinical trials targeted at these subgroups (or subgroup analysis of existing clinical trials) may provide definitive answers to these questions.
Our study shows that age, comorbidity, and tumor risk (vis-à-vis PSADT) are all key factors in predicting prognosis and cause of death in men with nmCRPC, similar to other time points in the prostate cancer disease course26. While the risk of PCSM persisted regardless of age and comorbidity—as we would expect in this group of patients with advanced disease6 —the relative risks of OCM vs. PCSM were nonetheless clearly influenced by each of these factors. In our cohort, for men with PSADT ≥ 9 months, older age, and higher CCI were indicators of higher risk of OCM. On the other hand, PCSM was the predominant cause of death for all patients with short PSADT (<9 months). These findings are consistent with previous studies showing that PSA kinetics is the most powerful indicator of PC aggressiveness among men diagnosed with nmCRPC20. However, by including factors that influence OCM (i.e., age, comorbidity), our comprehensive risk stratification paradigm improves discrimination of cause of death over risk stratification based on tumor risk factors alone.
Our study has several limitations. First, due to the retrospective nature of our study, several important predictors of prognosis such as functional status and frailty were not available. Second, our cohort only included patients from the VA, which may affect external generalizability to other patient populations. Third, metastatic workup at the time of diagnosis of CRPC was not uniform, and was left to the discretion of the treating physician, which may affect the homogeneity of our cohort (i.e., compared with a cohort of men who underwent requisite imaging prior to enrollment). It is also important to note that changing practice patterns with newer imaging modalities, in particular, PSMA PET, are being increasingly utilized in the nmCRPC setting and may reduce heterogeneity in prognosis by improving identification of men with distant metastases. Fourth, we did Table 3 Competing risks regression 5-year cumulative incidence estimates for prostate cancer-specific, and other-cause mortality by age, comorbidity, and PSA doubling time CCI 0 CCI 1–2 CCI 3+ OCM PCSM OCM PCSM OCM PCSM Age < 70 PSADT < 9 months 7% 29% 0% 58% 12% 47% PSADT ≥ 9 months 0% 0% 4% 4% 14% 18% Age 70–79 PSADT < 9 months 6% 52% 5% 63% 24% 54% PSADT ≥ 9 months 0% 4% 18% 12% 10% 27% Age ≥ 80 PSADT < 9 months 19% 30% 12% 56% 38% 47% PSADT ≥ 9 months 4% 11% 30% 29% 51% 21% All models are adjusted for age at CRPC, race, year of CRPC, treatment center, biopsy Gleason score, PSA at CRPC, primary treatment, months from ADT to CRPC, PSADT, and PSAV 258 C. A. Whitney et al. not have any information on whether patients elected hospice or palliative care rather than active management of disease. Fifth, care outside of the VA was not routinely captured in the dataset, which may affect the ascertainment of metastatic disease status. Lastly, we did not account for secondary treatments that could affect survival, although most novel medications were not approved until after most patients had died or been censored as lost to follow-up.
Conclusions
The current literature reveals the need for a more accurate assessment of nmCRPC prognosis since the natural history of prostate cancer within this disease state is highly variable19,21. Our study suggests that age, comorbidity, and PSADT are all critical in determining the risk of long-term cancer and non-cancer mortality in these men. These data may also be relevant for patients and physicians trying to decide whether or not to add potentially morbid additional androgen axis inhibitors to traditional ADT, particularly the oldest and sickest patients who are at high risk of morbidity from these medications. Integrating simple clinical variables such as these into decision making represents a key component of precision medicine that is too often ignored. We strongly believe that these factors should be taken into account when counseling patients about prognosis, and they may be important in selecting subgroups for future clinical trials in men with nmCRPC.
Acknowledgments: Dr. Daskivich would like to acknowledge funding from K08 CA230155. Dr. Aronson would like to acknowledge funding from RO1 CA231219.
Compliance with ethical standards: Conflict of interest Dr. Daskivich is a consultant for Janssen Research and Development, LLC. Dr. Freedland is a consultant for Astellas Pharmaceuticals and Janssen Research and Development. Other authors declare that they have no conflict of interest.
Authors: Colette A. Whitney,1 Lauren E. Howard,1,2 Stephen J. Freedland,1,3 Amanda M. DeHoedt,1 Christopher L. Amling,4 William J. Aronson,5 Matthew R. Cooperberg,6 Christopher J. Kane,7 Martha K. Terris,8,9 Timothy J. Daskivich3
1. Division of Urology, Durham Veterans Affairs Medical Center, Durham, NC, USA
2. Department of Biostatistics and Bioinformatics, Duke University School of Medicine, Durham, NC, USA
3. Division of Urology, Cedars-Sinai Medical Center, Los Angeles, CA, USA
4. Department of Urology, Oregon Health Sciences University, Portland, OR, USA
5. Division of Urology, West Los Angeles Veterans Affairs Medical Center, Los Angeles, CA, USA
6. Department of Urology, University of California, San Francisco, CA, USA
7. Department of Urology, University of California, San Diego, CA, USA
8. Division of Urology, Charlie Norwood Veterans Affairs Medical Center, Augusta, GA, USA
9. Division of Urology, Medical College of Georgia, Augusta, GA, USA
References
1. Kirby M, Hirst C, Crawford ED. Characterising the castration resistant prostate cancer population: a systematic review. Int J Clin Pract. 2011;65:1180–92.
2. Ryan CJ, Smith MR, de Bono JS, Molina A, Logothetis CJ, de Souza P, et al. Abiraterone in metastatic prostate cancer without previous chemotherapy. N Engl J Med. 2013;368:138–48.
3. Beer TM, Armstrong AJ, Rathkopf DE, Loriot Y, Sternberg CN, Higano CS, et al. Enzalutamide in metastatic prostate cancer before chemotherapy. N Engl J Med. 2014;371:424–33.
4. D’Amico AV, Schultz D, Schneider L, Hurwitz M, Kantoff PW, Richie JP. Comparing prostate specific antigen outcomes after different types of radiotherapy management of clinically localized prostate cancer highlights the importance of controlling for established prognostic factors. J Urol. 2000;163:1797–801.
5. Howard LE, Moreira D, De Hoedt A, Aronson WJ, Kane CJ, Amling CL, et al. Thresholds for PSA doubling time in men with non-metastatic castration-resistant prostate cancer. BJU Int. 2017;120(5B):E80–E86.
6. Read WL, Tierney RM, Page NC, Costas I, Govindan R, Spitznagel EL, et al. Differential prognostic impact of comorbidity. J Clin Oncol. 2004;22:3099–103.
7. Daskivich TJ, Fan KH, Koyama T, Albertsen PC, Goodman M, Hamilton AS, et al. Effect of age, tumor risk, and comorbidity on competing risks for survival in a U.S. population-based cohort of men with prostate cancer. Ann Intern Med. 2013;158:709–17.
8. Daskivich TJ, Chamie K, Kwan L, Labo J, Dash A, Greenfield S, et al. Comorbidity and competing risks for mortality in men with prostate cancer. Cancer. 2011;117:4642–50.
9. Albertsen PC, Moore DF, Shih W, Lin Y, Li H, Lu-Yao GL. Impact of comorbidity on survival among men with localized prostate cancer. J Clin Oncol. 2011;29:1335–41.
10. Uchio EM, Aslan M, Wells CK, Calderone J, Concato J. Impact of biochemical recurrence in prostate cancer among US veterans. Arch Intern Med. 2010;170:1390–5.
11. Small EJ, Saad F, Chowdhury S, Hadaschik BA, Graff JN, Olmos D, et al. SPARTAN, a phase 3 double-blind, randomized study of apalutamide (APA) versus placebo (PBO) in patients (pts) with nonmetastatic castration-resistant prostate cancer (nmCRPC). J Clin Oncol. 2018;36:161.
12. Hussain M, Fizazi K, Saad F, Rathenborg P, Shore ND, Demirhan E, et al. PROSPER: a phase 3, randomized, double-blind, placebo (PBO)-controlled study of enzalutamide (ENZA) in men with nonmetastatic castration-resistant prostate cancer (M0 CRPC). J Clin Oncol. 2018;36.
13. Moreira DM, Howard LE, Sourbeer KN, Amarasekara HS, Chow LC, Cockrell DC, et al. Predicting bone scan positivity in nonmetastatic castration-resistant prostate cancer. Prostate Cancer Prostatic Dis. 2015;18:333–7.
14. Scher HI, Morris MJ, Basch E, Heller G. End points and outcomes in castration-resistant prostate cancer: from clinical trials to clinical practice. J Clin Oncol. 2011;29:3695–704.
15. Howard LE, Moreira DM, De Hoedt A, Aronson WJ, Kane CJ, Amling CL, et al. Thresholds for PSA doubling time in men with non-metastatic castration-resistant prostate cancer. BJU Int. 2017;120(5b):E80–e6.
16. Hanyok BT, Howard LE, Amling CL, Aronson WJ, Cooperberg MR, Kane CJ, et al. Is computed tomography a necessary part of a metastatic evaluation for castration-resistant prostate cancer? Results from the Shared Equal Access Regional Cancer Hospital Database. Cancer. 2016;122:222–9.
17. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. J Clin Epidemiol. 1992;45:613–9.
18. Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496– 509.
19. Rozet F, Roumeguere T, Spahn M, Beyersdorff D, Hammerer P. Non-metastatic castrate-resistant prostate cancer: a call for improved guidance on clinical management. World J Urol. 2016;34:1505–13.
20. Tombal B. Non-metastatic CRPC and asymptomatic metastatic CRPC: which treatment for which patient? Ann Oncol. 2012;23 (Suppl 10):x251–8.
21. Gillessen S, Omlin A, Attard G, de Bono JS, Efstathiou E, Fizazi K, et al. Management of patients with advanced prostate cancer: recommendations of the St Gallen Advanced Prostate Cancer Consensus Conference (APCCC) 2015. Ann Oncol. 2016. Impact of age, comorbidity, and PSA doubling time on long-term competing risks for mortality among men. . . 259
22. Pfizer and Astellas announce positive top-line results from phase 3 PROSPER Trial of XTANDI (enzalutamide) in patients with non-metastatic castration-resistant prostate cancer 2017. Available from: http://press.pfizer.com/press-release/pfizer-and-a-stellas-announce-positive-top-line-results-phase-3-prosper-trialxtandi-en.
23. Saigal CS, Gore JL, Krupski TL, Hanley J, Schonlau M, Litwin MS, et al. Androgen deprivation therapy increases cardiovascular morbidity in men with prostate cancer. Cancer. 2007;110: 1493–500.
24. Keating NL, O’Malley AJ, Smith MR. Diabetes and cardiovascular disease during androgen deprivation therapy for prostate cancer. J Clin Oncol. 2006;24:4448–56.
25. Keating NL, O’Malley AJ, Freedland SJ, Smith MR. Diabetes and cardiovascular disease during androgen deprivation therapy: observational study of veterans with prostate cancer. J Natl Cancer Inst. 2010;102:39–46.
26. Daskivich TJ, Litwin MS, Penson DF. Effect of age, tumor risk, and comorbidity in a U.S. population-based cohort of men with prostate cancer. Ann Intern Med. 2013;159:371.
Whitney, C., Howard, L., Freedland, S., DeHoedt, A., Amling, C., & Aronson, W. et al. (2018). Impact of age, comorbidity, and PSA doubling time on long-term competing risks for mortality among men with non-metastatic castration-resistant prostate cancer. Prostate Cancer And Prostatic Diseases, 22(2), 252-260. doi: 10.1038/s41391-018-0095-0
Received: 4 June 2018 / Revised: 31 July 2018 / Accepted: 26 August 2018 / Published online: 2 October 2018 © Springer Nature Limited 2018
Read More: A Commentary from the Associate Editor of PCAN