Long-term Survival of Extremely Advanced Prostate Cancer Patients Diagnosed with Prostate-specific Antigen over 500 ng/ml

Transcription

1 Jpn J Clin Oncol 2014;44(12) doi: /jjco/hyu142 Advance Access Publication 19 September 2014 Long-term Survival of Extremely Advanced Prostate Cancer Patients Diagnosed with Prostate-specific Antigen over 500 ng/ml Toru Sugihara 1,*, Changhong Yu 1, Michael W. Kattan 1, Hideo Yasunaga 2, Hiroyuki Ihara 3, Mizuki Onozawa 4, Shiro Hinotsu 5 and Hideyuki Akaza 6 1 Department of Quantitative Health Sciences, Cleveland Clinic, Cleveland, OH, USA, 2 Department of Clinical Epidemiology and Health Economics, School of Public Health, The University of Tokyo, Tokyo, 3 Department of Urology, Shintoshi Hospital, Iwata, 4 Department of Urology, Tokyo-kita Medical Center, Tokyo, 5 Center for Innovative Clinical Medicine, Okayama University Hospital, Okayama and 6 Department of Strategic Investigation on Comprehensive Cancer Network, Research Center for Advanced Science and Technology, The University of Tokyo, Tokyo, Japan *For reprints and all correspondence: Toru Sugihara, Department of Quantitative Health Sciences, Cleveland Clinic Main Campus, Mail Code JJN3-01, 9500 Euclid Avenue, Cleveland, OH 44195, USA. ezy04707@nifty.com Received June 7, 2014; accepted August 23, 2014 Objective: To investigate survival of hormone-naïve prostate cancer patients diagnosed with prostate-specific antigen 500 ng/ml, stratified according to the prostate-specific antigen level and type of therapy. Methods: Data of prostate cancer patients with prostate-specific antigen 500 ng/ml diagnosed between 2001 and 2003 and receiving primary androgen deprivation therapy were extracted from the Japan Study Group of Prostate Cancer database. Cancer-specific survival and overall survival were assessed according to the prostate-specific antigen level ( , and 5000 ng/ml) and type of therapy using Kaplan Meier analyses and multivariate Cox proportional hazards models including age, Gleason score, oncological stage and comorbidity. Results: The median follow-up was 27 months (interquartile range, 13 51) and a total of 1961 patients were included. Five-year cancer-specific and overall mortalities were 39.0 and 33.0%, respectively. There was a significant inverse relationship between overall survival and prostatespecific antigen magnitude among combination therapy patients, but not monotherapy patients (log-rank test, P ¼ and 0.558, respectively). The median overall survival in combination therapy patients with low-, intermediate- and high prostate-specific antigen and monotherapy patients with any prostate-specific antigen were 79, 59, 45 and 43 months, respectively. Multivariate analysis showed that combination therapy in patients with low- and intermediate prostate-specific antigen was significantly associated with a favorable overall survival compared with monotherapy (hazard ratios 0.66 and 0.75, respectively, both P, 0.001). Similar results were obtained for cancer-specific survival. Conclusions: There are major survival differences in extremely high prostate-specific antigen cases according to the prostate-specific antigen level and hormone therapy type and those patients would benefit notably from combination androgen blockade. Key words: androgen antagonists neoplasm metastasis prostate neoplasms prostate-specific antigen survival analysis # The Author Published by Oxford University Press. All rights reserved. For Permissions, please journals.permissions@oup.com

2 1228 Extremely advanced prostate cancer survival INTRODUCTION Since the introduction of prostate-specific antigen (PSA) measurement into clinical practice in the 1980s, PSA has become recognized as the most reliable biomarker of prostate cancer (PCa) and is now widely used as a screening and diagnostic tool for PCa (1,2). In clinical practice, a PSA level of 20 ng/ml is used as the cutoff value for advanced PCa and PSA does not usually exceed 30 ng/ml at PCa diagnosis (3). However, though rare, it is also known that PSA can be several orders of magnitude higher and can exceed According to the Cancer of the Prostate Strategic Urologic Research Endeavor database, a large, community-based registry of PCa in the USA, 92.3% of PCa patients at diagnosis had PSA levels,30 ng/ml, while patients with PSA.500,.1000 and.5000 ng/ml accounted for 0.20, 0.18 and 0.03%, respectively (4). Patients with very high PSA frequently had a distant metastasis and were empirically considered to have an extremely high risk of PCa death. Nevertheless, detailed data of survival in such patients are lacking because of the small numbers. Therefore, although patients may be strongly concerned about their survival time when advised of their extremely high PSA, the lack of data in such patients hinders physicians from providing an appropriate prognosis which may affect the quality of their patients end-stage of life. According to a provisional clinical opinion issued by the American Society of Clinical Oncology, a frank discussion of prognosis with a reasonable survival estimate should be included in a working list for integrated palliative care in advanced cancer (5,6). To foster academic attention on the long-term outcome of primary androgen deprivation therapy (PADT) for PCa, the Japan Study Group of Prostate Cancer (J-CaP) launched a nationwide prospective cohort study in Japan As a result, the Japan Cancer of the Prostate Risk Assessment (J-CAPRA) score was established to estimate the survival of PCa patients receiving PADT. Several studies revealed that survival estimates based on J-CAPRA were applicable and relevant to a wide range of PCa cases, from localized to highly advanced PCa (4,7 9). However, detailed data on the longterm survival among men with PSA 500 ng/ml have been lacking because all patients with PSA over 500 ng/ml are assigned the same risk score in the J-CAPRA scoring system. In the current study, we hypothesized that differences in survival remained in PCa patients with different levels of PSA.500 ng/ml, and performed a detailed analysis of cancerspecific mortality and overall mortality, taking advantage of the J-CaP database and the J-CAPRA score. PATIENTS AND METHODS In the present study, the J-CaP registry database was used as a data source. The study group focused on gathering and analyzing information about hormone therapy for PCa in Japan. The J-CaP database is a nationwide multicenter registry which has collected data about men with any stage of PCa who newly started PADT therapy between 2001 and 2003 in Japan. A total of men were enrolled by 384 institutions participating in J-CaP and included 50% of all men with PCa diagnosed during the period, and 95% of those treated with PADT in Japan (8,9). J-CaP data were collected and analyzed with the approval of the local ethics committees or institutional review boards. When an institution did not have its own local board, approval was obtained from the person with responsibility for the institution. We extracted the data of men with PCa diagnosed between January 2001 and December 2003 and who had a PSA level 500 ng/ml. As the baseline patient characteristics, we extracted information on age, PSA level at first diagnosis, oncological T, N and M stages (1997), Gleason score, type of PADT, comorbidities at diagnosis, cancer-specific survival (CSS) and overall survival (OS). Type of PADT was classified into combined androgen blockade (CAB) or non-cab. CAB was defined as a combination of an antiandrogen plus either a luteinizing hormone-releasing hormone analog or bilateral orchiectomy. Other PADT, such as monotherapy with an antiandrogen or a luteinizing hormone-releasing hormone analog, or surgical castration alone were designated non-cab. The endpoints in the present study were CSS and OS. Survival information was followed up every 3 months on an ongoing basis and the dataset as of April 2013 was used. Comorbidity was classified into three categories, which were knownfromapreviousstudytoimpactoncssandos: (i) hypertension, heart disease or stroke, (ii) diabetes and (iii) other cancer (8). TNM stage and the Gleason score were determined by each participating institution. As a risk adjustment for oncological background, TNM stage and the Gleason score were converted into a numeric J-CAPRA score according to an established method previously published (4,8). The PSA level was categorized into three groups: , and 5000 ng/ml with reference to a preceding study (4). Some missing data were found in the categories of N stage, M stage and Gleason score. To make the most of the available data, we filled in any missing values by the multiple imputation method by generating five complementary copies. Multiple imputation is a procedure used to replace missing values with other plausible values by creating multiple filling-in patterns to avert bias caused by missing data (10). For imputation, a polytomous regression method was applied to the missing data in N stage, M stage and Gleason score. It is known that the benefit of PADT varies by PSA level. After the multiple imputation process, survival curves stratified according to the PSA level and type of PADT were determined by the Kaplan Meier method,and were tested by the log-rank test. Multivariate analyses using a Cox proportional hazard model for CSS and OS were performed with the covariates of J-CAPRA scores of TNM stage and Gleason score, age and comorbidities at diagnosis. The threshold for significance was set at P, Univariable comparisons were performed by the x 2 test and the Mann Whitney U-test as

3 Jpn J Clin Oncol 2014;44(12) 1229 appropriate. Statistical analyses were conducted using R version (R Foundation for Statistical Computing, Vienna, Austria) with survival and mice 2.18 packages (11,12). RESULTS Of enrolled men with PCa, 1961 (7.4%) were selected based on the criterion of having PSA 500 ng/ml at diagnosis. The median follow-up period was 27 months [interquartile range (IQR), months]. Cancer-specific and overall mortalities were 30.3 and 40.4%, respectively. Table 1 shows the baseline patient characteristics. There was no substantial background difference according to the type of PADT. In the non-cab group, monotherapy with an antiandrogen, a luteinizing hormone-releasing hormone analog, surgical castration alone and other PADT were selected in 15% (n ¼ 85), 41% (n ¼ 237), 21% (n ¼ 121) and 23% (n ¼ 135), respectively. The type of therapy among the non-cab patients were not associated with CSS and OS in the multivariable Cox proportional hazard models (data not shown). Data were missing for 14.3% of Gleason score, 8.9% of N stage and 1.8% of M stage. Those missing data were filled by the multiple imputation procedure. Figure 1 shows Kaplan Meier curves of CCS and OS stratified by PSA level and type of PADT. In total, the median CCS and OS were 84 and 55 months, respectively, and survival at 1-, 2-, 5- and 10-year follow-ups were 83.8, 71.3, 39.0 and 7.6%, respectively, for CCS, and 81.6, 66.7, 33.0 and 5.4%, respectively, for OS. In Fig. 1A and B, the three survival curves of non-cab cases with different PSA levels markedly overlapped with each other (log-rank test, P ¼ for CSS and P ¼ for OS). Figure 1C and 1D shows the survival curves of CAB cases with the three PSA levels, and a curve of all non-cab cases combined. CAB cases were well stratified by PSA level, while the non-cab curve overlapped with that of CAB cases with PSA 5000 ng/ml (overall log-rank test among three lines of CAB cases, P ¼ 0.33 for CSS and P ¼ for OS). Table 2 presents the results of multivariable analysis using the Cox proportional hazards model. CSS and OS among non-cab cases were comparably poor regardless of PSA magnitude, and did not differ significantly from those of CAB patients with PSA 5000 ng/ml, as was expected from the Kaplan Meier analyses. Conversely, significantly lower hazard ratios (HRs) compared with non-cab cases were observed in CAB cases with PSA and ng/ml for CSS and OS (HR 0.62 and 0.72, respectively, for CSS; and HR, 0.66 and 0.75, respectively, for OS; all P, 0.05) Similar results were confirmed when stratified analyses were performed according to the PSA levels and the type of PADT (data not shown). Older age, high Gleason score and positive metastatic status were also significantly associated with increased HRs for CSS and OS. Table 1. Baseline characteristics of patients with prostate cancer and PSA 500 ng/ml receiving primary androgen deprivation therapy extracted from the J-CaP database Variables (J-CAPRA score) CAB for PADT, n (%) or median (IQR) Non-CAB for PADT, n (%) or median (IQR) Total 1383 (100.0) 578 (100.0) Age at diagnosis 73 (67 78) 73 (68 79) (range), years PSA at diagnosis, 1280 ( ) 1400 ( ) ng/ml (3 pt.) 532 (38.5) 210 (36.3) (3 pt.) 693 (50.1) 285 (49.3) (11.4) 83 (14.4) (3 pt.) Gleason score 6 (0 pt.) 118 (8.5) 60 (10.4) (1 pt.) 313 (22.6) 151 (26.1) 8 10 (2 pt.) 754 (54.5) 283 (49.0) Missing 198 (14.3) 84 (14.5) T stage T1a T2a (0 pt.) 33 (2.4) 15 (2.6) T2b T3a (1 pt.) 428 (30.9) 178 (30.8) T3b (2 pt.) 467 (33.8) 189 (32.7) T4 (3 pt.) 455 (32.9) 196 (33.9) N stage N0 (0 pt.) 554 (40.1) 217 (37.5) N1 (1 pt.) 709 (51.3) 305 (52.8) Missing 120 (8.7) 56 (9.7) M stage M0 (0 pt.) 134 (9.7) 61 (10.6) M1 (3 pt.) 1224 (88.5) 505 (87.4) Missing 25 (1.8) 12 (2.1) Comorbidity Hypertension, heart disease or stroke 498 (36.0) 216 (37.4) Diabetes 136 (9.8) 44 (7.6) Other cancer 105 (7.6) 43 (7.4) Follow-up period 28 (14 53) 23 (9 44),0.001 (months) Cancer-specific 406 (29.4) 190 (32.9) mortality Overall mortality 532 (38.5) 262 (45.3) Missing data were filled in by multiple imputation methods for further analyses. J-CAPRA, Japan Cancer of the Prostate Risk Assessment; CAB, combined androgen blockade; J-CaP, Japan Study Group of Prostate Cancer; PADT, primary androgen deprivation therapy; IQR, interquartile range; PSA, prostate-specific antigen; pt., point of J-CAPRA score. P

4 1230 Extremely advanced prostate cancer survival Figure 1. Kaplan Meier survival curves. (A) Cancer-specific survival among non-cab cases. (B) Overall survival among non-cab cases. (C) Cancer-specific survival among all cases. (D) Overall survival among all cases. CAB, combined androgen blockade. DISCUSSION In the present study, we analyzed CSS and OS of 1961 PCa cases with extremely high PSA levels over 500 ng/ml, and showed that when CAB was not given as PADT, CSS and OS exhibited no difference regardless of the PSA value. CSS and OS in these patients were as poor as those in the CAB patients with PSA 5000 ng/ml, while PCa patients undergoing CAB for PADT presented significantly better CSS and OS compared with non-cab patients at lower PSA levels. In other words, PSA magnitude could be an independent predictor in CAB patients and not in non-cab patients. The large median differences in OS between CAB and non-cab cases of 34 months for PSA ng/ml, and 19 months for PSA ng/ml, implies that CAB should be strongly recommended for PADT in these groups. In general, the survival benefit of CAB over PADT monotherapy is well demonstrated by several randomized controlled trials (RCTs) (13,14). A meta-analysis of RCTs by the Prostate Cancer Trialists Collaborative Group in 2000 showed that CAB with a non-steroidal antiandrogen provided

5 Jpn J Clin Oncol 2014;44(12) 1231 Table 2. Cox proportional hazards model analysis of cancer-specific survival and overall survival of men with prostate cancer and PSA 500 ng/ml receiving primary androgen deprivation therapy extracted from the J-CaP database Cancer-specific survival Overall survival HR (95% CI) P HR (95% CI) P CAB, PSA ng/ml 0.62 ( ) ( ) CAB, PSA ng/ml 0.72 ( ) ( ) CAB, PSA 5000 ng/ml 0.81 ( ) ( ) Non-CAB, PSA ng/ml Reference Reference Non-CAB, PSA ng/ml 0.94 ( ) ( ) Non-CAB, PSA 5000 ng/ml 0.91 ( ) ( ) Age (by 1 year) 1.01 ( ) ( ),0.001 Gleason score (by one J-CAPRA point) 1.16 ( ) ( ) T stage (by one J-CAPRA point) 1.07 ( ) ( ) N1 (vs. N0) 0.96 ( ) ( ) M1 (vs. M0) 3.57 ( ), ( ),0.001 Hypertension, heart disease or stroke 1.05 ( ) ( ) Diabetes 1.12 ( ) ( ) Other cancer 1.23 ( ) ( ) HR, hazard ratio; CI, confidence interval. 8% risk reduction in overall death compared with castration alone (14).AnRCTledbyAkazaetal.(13) demonstrated substantial OS benefit in favor of CAB with bicalutamide over PADT monotherapy in PCa patients with Stages C and D1 (HR 0.78) but not with Stage D2. However, academic research focusing on the field of extremely high PSA has not been well undertaken because of the rarity of such cases, and the current study has filled the gap in the evidence in the field. Several attempts have been made to develop a prediction tool for PCa prognosis. The J-CAPRA score is a wellvalidated numeric indicator to estimate survival in a wide variety of PCa patients (4,7 9). For example, Akaza et al. (8) showed that CSS and OS were also well stratified by low (0 2), intermediate (3 7) and high (8) J-CAPRA risk score groups. Intriguingly, although the risk accumulation by the J-CAPRA scoring system was saturated at PSA over 500 ng/ml, the current study revealed there were still large survival differences in CSS and OS according to the level of PSA.500 ng/ml. More definitive survival prediction would aid physicians and patients in planning for end of life, including palliative care, because men diagnosed with PCa and an extremely high PSA suffer from a great deal of concern about their remaining life span even though the disease is still of hormone naïve. Today, the idea has emerged from the oncological field that palliative care is not limited to end-of-life care, but must be focused on the relief of suffering, in all of its dimensions, throughout the course of a patient s illness. The expert panel of the American Society of Clinical Oncology included, in a working list of measures, a component of appropriate discussion with patients about their prognosis andcurability(5). A study of patients with metastatic nonsmall-cell lung cancer by Temel et al. (6) concludedthat understanding their prognosis aided patients with decisionmaking about care near the end of life. Hence, patients understanding of the length of time they have left and the benefits and risks of treatment is quite important for their subsequent quality of life. Despite the severe prognosis, the survival of patients with metastatic PCa keeps improving over time. Tangen et al. (15) reported changes in survival of metastatic PCa with use of hormone therapy in three Phase III trials conducted by the Southwest Oncology Group in the USA: S8494 (enrolled , PSA measurement not available), S8894 [ , median PSA ¼126 ng/ml (IQR, )] and S9346. [accrued , median PSA ¼ 51 ng/ml [IQR, )]. They found an overall improvement in survival over the years, with median survival times of 30, 33 and 49 months, respectively. In the current study, the median survival in CAB cases was much better than in the previous studies even though the study would include cases with much higher PSA levels, which implies that more improvement may be expected in future. Multivariable analyses showed similar results for CSS and OS. In addition to type of PADT and PSA level, older age, high Gleason score and metastatic status were associated with an increased HR, but T and N stages were not. It is likely that in cases of extremely high PSA, PCa local status as determinedbytandnstageshaslessimpactonsurvivalthan

6 1232 Extremely advanced prostate cancer survival factors representing overall PCa aggressiveness such as the Gleason score and M stage. In contrast with a previous study, comorbidity did not appear to influence survival in the current study, which may have arisen because of the small proportion of other causes of death, with PCa accounting for 80% of all deaths (8). In terms of PSA 5000 ng/ml, no significant differences in CSS and OS were observed between CAB and non-cab cases. It could partially be relatedtothesmallsampleinthis category, but another explanation is that the survival benefit of CAB diminishes as PCa advances. Sylvester et al. (16) compared survival between a PCa group with castration alone and CAB and concluded there was a smaller survival benefit of CAB among patients with more advanced PCa than those with mild bone metastasis. The similar result was shown in the RCT by Akaza et al. (13), which said the survival benefit of CAB over a luteinizing hormone-releasing hormone analog monotherapy in patients with Stage D1 was not observed in those with Stage D2. There are a number of limitations in the present study. First, the present study was not an RCT. Patients were assigned to the CAB and non-cab sides on a clinical-practice basis. A hidden confounder may influence on the observed outcome. For example, a patient with limited income might be likely to select surgical castration alone instead of CAB. Therefore, the results of the current study must be interpreted cautiously. Secondly, the study did not reflect the benefit of several newly emerging drugs for metastatic castration-resistant PCa. The enrollment period of the study was and docetaxel wasapprovedin2008injapan(7). Therefore, the result mainly reflects the pre-docetaxel era. Thirdly, the J-Cap database lacked detailed information which may affect survival such as medications other than CAB, areas of metastasis, bone fractures and pain. Fourthly, the clinical and pathological diagnosis was dependent on the information from each facility. Detailed pathological data apart from the Gleason score was lacking and the Gleason grading rule was different from the current system proposed by the International Society of Urological Pathology 2005 (17). Fifthly, some patients may have been censored because of lost to follow-up as a result of aggressive progression of PCa. Detailed reasons for loss to follow-up were unavailable and this possibility would cause underestimation of mortality. Despite these limitations, our analyses cast a spotlight on an area of extremely high PSA where few research attempts have been made and produced worthwhile survival information for patients with advanced PCa. In conclusion, the study based on multicenter registry data presented that there were significant differences in CSS and OS in PCa cases with PSA 500 ng/ml according to the PSA level and type of PADT. CAB for PADT showed a significant survival gain in this population. Funding This study was supported by the ITO Genboku and SAGARA Chian Memorial Scholarship from Saga Prefecture, Japan. Conflict of interest statement None declared. References 1. Carter HB, Albertsen PC, Barry MJ, et al. Early detection of prostate cancer: AUA guideline. JUrol2013;190: Papsidero LD, Wang MC, Valenzuela LA, Murphy GP, Chu TM. A prostate antigen in sera of prostatic cancer patients. Cancer Res 1980;40: Cooperberg MR, Broering JM, Litwin MS, et al. The contemporary management of prostate cancer in the United States: lessons from the cancer of the prostate strategic urologic research endeavor (CapSURE), a national disease registry. JUrol2004;171: Cooperberg MR, Hinotsu S, Namiki M, et al. Risk assessment among prostate cancer patients receiving primary androgen deprivation therapy. J Clin Oncol 2009;27: Smith TJ, Temin S, Alesi ER, et al. American Society of Clinical Oncology provisional clinical opinion: the integration of palliative care into standard oncology care. J Clin Oncol 2012;30: Temel JS, Greer JA, Admane S, et al. Longitudinal perceptions of prognosis and goals of therapy in patients with metastatic non-small-cell lung cancer: results of a randomized study of early palliative care. JClin Oncol 2011;29: Kimura T, Onozawa M, Miyazaki J, et al. Prognostic impact of young age on stage IV prostate cancer treated with primary androgen deprivation therapy. Int J Urol 2014;21: Akaza H, Hinotsu S, Usami M, et al. Evaluation of primary androgen deprivation therapy in prostate cancer patients using the J-CAPRA risk score. Prostate Int 2013;1: Kimura T, Onozawa M, Miyazaki J, et al. Validation of the prognostic grouping of the seventh edition of the tumor-nodes-metastasis classification using a large-scale prospective cohort study database of prostate cancer treated with primary androgen deprivation therapy. Int J Urol 2013;20: Rubin DB, Schenker N. Multiple imputation in health-care databases: an overview and some applications. Stat Med 1991;10: Buuren S, Groothuis-Oudshoorn K. MICE: multivariate imputation by chained equations in R. J Stat Softw 2011;45: R Core Team. R: a language and environment for statistical computing (14 March 2014, date last accessed). 13. Akaza H, Hinotsu S, Usami M, et al. Combined androgen blockade with bicalutamide for advanced prostate cancer: long-term follow-up of a phase 3, double-blind, randomized study for survival. Cancer 2009;115: Prostate Cancer Trialists Collaborative Group. Maximum androgen blockade in advanced prostate cancer: an overview of the randomised trials. Lancet 2000;355: Tangen CM, Hussain MH, Higano CS, et al. Improved overall survival trends of men with newly diagnosed M1 prostate cancer: a SWOG phase III trial experience (S8494, S8894 and S9346). JUrol2012;188: Sylvester RJ, Denis L, de Voogt H. The importance of prognostic factors in the interpretation of two EORTC metastatic prostate cancer trials. European Organization for Research and Treatment of Cancer (EORTC) Genito-Urinary Tract Cancer Cooperative Group. Eur Urol 1998;33: Epstein JI, Allsbrook WC, Jr, Amin MB, Egevad LL. The 2005 International Society of Urological Pathology (ISUP) Consensus Conference on Gleason Grading of Prostatic Carcinoma. Am J Surg Pathol 2005;29: