Gleason score = 8 prostate cancer: much more like Gleason score 9?

Transcription

1 5 + 3 = 8 prostate cancer: much more like 9? Brandon A. Mahal, Vinayak Muralidhar*, Yu-Wei Chen, Toni K. Choueiri, Karen E. Hoffman, Jim C. Hu, Christopher J. Sweeney, James B. Yu**, Felix Y. Feng, Quoc-Dien Trinh and Paul L. Nguyen Department of Internal Medicine, Brigham and Women s Hospital, *Harvard Medical School, Harvard School of Public Health, Department of Medical Oncology, Dana-Farber Cancer Institute and Brigham and Women s Hospital, Harvard Medical School, Department of Radiation Oncology, MD Anderson Cancer Center, The University of Texas, Department of Urology, UCLA Medical Center, Los Angeles, CA, **Department of Therapeutic Radiology/Radiation Oncology, Yale, New Haven, CT, Department of Radiation Oncology, University of Michigan Health System, Ann Arbor, MI, Division of Urology, Brigham and Women s Hospital, and Department of Radiation Oncology, Dana- Farber Cancer Institute and Brigham and Women s Hospital, Harvard Medical School, Boston, MA, USA Objective To determine whether patients with = 8 prostate cancer have outcomes more similar to other patients with 8 disease or to patients with Gleason score 9 disease. Patients and Methods The Surveillance, Epidemiology and End Results (SEER) database was used to study men diagnosed with N0M0 8 or 9 prostate cancer from 2004 to Using = 8 as the referent, Fine and Gray competing risks regression analyses modelled the association between and prostate cancerspecific mortality (PCSM). Results The 5-year PCSM rates for patients with = 8, = 8, = 8, and 9 disease were 6.3%, 6.6%, 13.5%, and 13.9%, respectively (P < 0.001). Patients with = 8 or 9 disease had up to a two-fold increased risk of PCSM (adjusted hazard ratio [AHR] 1.89, 95% confidence interval [CI] , P < 0.001; and AHR 2.17, 95% CI , P < 0.001, respectively) compared with the referent group of patients ( = 8). There was no difference in PCSM between patients with = 8 vs 9 disease (P = 0.25). Conclusions 8 disease represents a heterogeneous entity with PCSM outcomes distinguishable by the primary Gleason pattern. The PCSM of = 8 and Gleason = 8 disease are similar, but patients with = 8 have a risk of PCSM that is twice as high as other patients with 8 disease and should be considered to have a similar poor prognosis as patients with 9 disease. Such patients should be allowed onto trials seeking the highest-risk patients in which to test novel aggressive treatment strategies. Keywords prostatic neoplasms,, cancer mortality, outcomes research Introduction In 2015, there will be an estimated new cases of prostate cancer and deaths from prostate cancer in the USA alone [1]. Prostate cancer risk groups have been developed based on PSA levels,, and clinical stage [1 3]. These risk groups are both predictive of prostate cancer outcome and guide treatment guidelines [2]. Although there are ongoing efforts to improve risk stratification based on molecular risk stratification, remains the most prognostic clinical risk factor in prostate cancer [4,5]. It is well-established that for 7 prostate cancer, clinical outcomes differ significantly by whether a patient has 4 + 3or3+ 4 disease [6 9]. However, despite the fact that 8 prostate cancer can be comprised of any of three different patterns (4 + 4 = 8, = 8, or = 8), these are all traditionally prognostically grouped together as a composite of 8 and all treated similarly. To maximise the potential to risk-stratify patients to personalise care, as well as to inform future clinical trial design, we performed a large-scale examination of whether differences in prostate cancer outcome exist by BJU International 2015 BJU International doi: /bju BJU Int 2016; 118: Published by John Wiley & Sons Ltd. wileyonlinelibrary.com

2 Mahal et al. primary and secondary pattern among men with Gleason score 8 disease and whether any of these patterns may be more prognostically similar to 9. Patients and Methods Sponsored by the National Cancer Institute, the Surveillance, Epidemiology and End Results (SEER) programme collects and publishes cancer incidence, survival, and treatment data from population-based cancer registries; the 17 tumour registries encompass nearly 28% of the USA population and capture 97% of incident cancers [10,11]. The SEER programme was used to identify a national cohort of men with N0M0 8 or 9 prostate cancer, diagnosed between 2004 and Patients were excluded if PSA data or if information on clinical T stage was missing (6 626 patients), leaving patients eligible for this study. The study period did not include patients diagnosed with prostate cancer before 2004, as PSA information and s were not available in SEER before then [12]. The primary independent variable of interest in this study was and patients were stratified by predetermined strata as follows: = 8, = 8, = 8, and 9., as provided by the SEER programme from 2004 to 2009 [12], represents the highest score identified at surgery or at biopsy for nonsurgically managed patients. From 2010 to 2011, both biopsy and pathological s are provided in SEER [12]; however, we categorised scores according to the highest at either biopsy or surgery to be consistent with grading patterns from 2004 to T stage was determined using the AJCC sixth edition as provided by the SEER programme [10]. Treatment received was classified according to the National Comprehensive Cancer Network (NCCN) guidelines and included radical prostatectomy (RP), radiation therapy (external beam radiation therapy and/or brachytherapy), or non-definitive treatment (including observation, expectant management, or androgen-deprivation therapy alone) [3]. Race was categorised as Black vs non- Black. Both age at diagnosis and PSA level were included as continuous variables. The SEER registries conduct follow-up to determine vital status and cause of death via various methods such as contact with treating physicians, links to state vital records, and links to the National Death Index. Deaths not coded due to cancer are considered censored data. Baseline patient characteristics across groups (4 + 4 = 8, = 8, = 8, and 9) were compared with the Wilcoxon test or chi-squared test, as appropriate. Multivariable Fine and Gray competing-risks regression analysis was used to determine prostate cancer-specific mortality (PCSM) as a function of (4 + 4 = 8, = 8, = 8, and 9) [13]. For the main study analyses, = 8 was pre-specified as the baseline referent group. Multivariable competing risks regression analyses were adjusted for the demographics of age, race, type of treatment received, PSA level, and T-stage. Cumulative incidence plots were generated from the Fine and Gray analysis described above. Fine and Gray competing risks regression analyses were repeated adjusting for year of diagnosis in addition to the aforementioned variables to account for any differences in Gleason scoring methods by year. Competing risks were also repeated among patients with 5 years of follow-up diagnosed from 2004 to Furthermore, Fine-Gray competing risks regression analyses were repeated for patients who received RP (whose Gleason scores represent pathological scores) and for patients who did not receive RP (whose s represent scores at biopsy). Lastly, all competing risks models described above were repeated using 9 as the referent to determine whether there were any differences in PCSM between = 8 and 9 disease. P values were two-sided and a threshold of 0.05 was used to determine statistical significance. All statistical analyses were performed using STATA 13.1 (StataCorp, College Station, TX, USA). This study was approved by the Dana-Farber/Harvard Cancer Center Institutional Review Board; a waiver for informed consent was obtained. Results Baseline Patient Characteristics In all, patients with N0M0 8 or 9 disease were included in this study. Baseline characteristics including age at diagnosis, race, T stage, PSA management type, tumour size, percentage of positive cores, margin status, and seminal vesicle invasion were clinically similar, but statistically different when compared across the four groups (4 + 4 = 8, = 8, = 8, and 9) as shown in Table 1. Notably, patients with 9 disease had the highest rate of T3 T4 disease (12.7%) and patients with = 8 presented with the highest PSA levels (median [interquartile range] PSA level of 8.8 [ ] ng/ml). There were (66.4%) patients who had a from biopsy alone and patients (33.6%) patients who had a from RP. Among patients who received RP, those with of had the highest rate of positive margins (22.6%) and those with of 9 had the highest rate of seminal vesicle invasion (30.5%). Association between Gleason Score and PCSM The median follow-up was 36 months. The 5-year PCSM rates for patients with = 8, = 8, = 8, and 9 disease were 6.3%, 6.6%, 13.5%, and 13.9%, respectively (P < 0.001). Cumulative incidence estimates of PCSM projected out to 95 months were significantly higher 96 BJU International 2015 BJU International

3 8 prostate cancer Table 1 Baseline clinical and demographic characteristics by among all patients with N0M0 Gleason 8 or 9 disease ( patients). Characteristic = = = 8 9 P Number of patients Median (IQR) age, years 69 (62 76) 65 (59 72) 67 (61 74) 68 (62 76) <0.001 N (%*) Race Non-Black (84.7) (81.7) 738 (82.7) (86.6) Black (15.3) 489 (18.3) 154 (17.3) (13.5) Management type No definitive therapy (22.8) 482 (18.1) 198 (22.2) (27.1) <0.001 Radiation therapy (45.9) 757 (28.4) 321 (36.0) (39.9) RP (26.8) (48.8) 311 (34.9) (25.4) Combined therapy 954 (4.4) 126 (4.7) 62 (7.0) (7.6) Median (IQR) PSA level, ng/ml 8.6 ( ) 7.6 ( ) 8.8 ( ) 8.6 ( ) <0.001 T Stage, n (%) T (51.2) (52.3) 447 (50.1) (41.9) <0.001 T (41.8) (42.3) 394 (44.2) (45.4) T3 T (7.0) 144 (5.4) 51 (5.7) (12.7) Median (IQR) % positive cores 50.0 ( ) 58.3 ( ) 62.5 ( ) 75.0 ( ) <0.001 Median (IQR) tumour size, mm 19 (13 26) 18 (12 27) 20 (14 26) 20 (15 30) <0.001 Positive margins, n (%) Yes 946 (14.2) 296 (20.6) 85 (22.6) 911 (17.7) <0.001 No (85.8) (79.4) 291 (77.4) (82.3) Seminal vesical invasion, n (%) Yes (17.1) 200 (13.9) 77 (20.5) (30.5) <0.001 No (82.9) (86.1) 299 (79.5) (69.5) IQR, interquartile range. *Percentages may not add to 100 due to rounding. Among patients with on biopsy only and only available from 2010 to 2011 (4 344 patients). Among patients who underwent RP ( patients). Baseline patient characteristics across groups (4 + 4 = 8, = 8, = 8, and 9) were compared with the Wilcoxon test or chi-squared test as appropriate. among men with 9 or = 8 disease than for those with = 8or3+ 5 = 8 disease (P < 0.001; Fig. 1). On multivariable analysis, patients with = 8 or 9 disease had up to a two-fold increased risk of PCSM [adjusted hazard ratio (AHR) 1.89, 95% CI ; and AHR 2.17, 95% CI , both P < 0.001, respectively] compared with the referent group of patients with = 8 disease (Table 2). Fig. 1 Cumulative incidence plot for PCSM by (GS) among all patients with N0M0 8 or 9 disease ( patients). % Prostate Cancer-Specific Mortality No. at Risk GS 9 GS 5+3=8 GS 3+5=8 GS 4+4= Time, Months P< There was no difference in PCSM adjusted hazard or incidence between patients with = 8vs9 disease (P = 0.25) when the competing risks model was repeated using 9 patients as the referent group. When adjusting for year of diagnosis or when repeating the analysis for patients with 5 years of follow-up (diagnosed from 2004 to 2006), the aforementioned results remained nearly unchanged (Table 2). When limiting the cohort to patients with pathological s who received RP, patients with = 8 or 9 disease still had an over two-fold increased risk of PCSM (AHR 2.42, 95% CI , P = 0.002; and AHR 2.90, 95% CI ; P < 0.001, respectively) compared with the referent group (Table 3); there was no difference in PCSM between patients with = 8 vs 9 disease (P = 0.53). Similarly, when limiting the cohort to patients with clinical/biopsy confirmed Gleason scores (did not receive RP), patients with = 8 or 9 disease were at an up to two-fold increased hazard of PCSM (AHR 1.81, 95% CI , P < 0.001; and AHR 2.05; 95% CI ; P < 0.001, respectively) compared with patients with = 8 disease (Table 3) even after adjusting for receipt of radiation therapy; again there was no difference between patients with = 8 vs 9 disease, even after adjusting for receipt of radiation therapy (P = 0.35). BJU International 2015 BJU International 97

4 Mahal et al. Table 2 Multivariable Fine and Gray competing risks regression for PCSM among all patients with N0M0 8 or 9 disease ( patients). Characteristic Multivariable analysis (1) Multivariable analysis (2) Multivariable analysis (3) AHR (95% CI) P AHR (95% CI) P AHR (95% CI) P = (ref) 1.0 (ref) 1.0 (ref) = ( ) ( ) ( ) = ( ) < ( ) < ( ) < ( ) < ( ) < ( ) <0.001 PSA (per ng/ml increase) 1.00 ( ) < ( ) < ( ) <0.001 T stage T1 1.0 (ref) 1.0 (ref) 1.0 (ref) T ( ) < ( ) < ( ) <0.001 T3 T ( ) < ( ) < ( ) <0.001 Treatment No definitive therapy 1.0 (ref) 1.0 (ref) 1.0 (ref) Radiation therapy 0.43 ( ) < ( ) < ( ) <0.001 RP 0.29 ( ) < ( ) < ( ) <0.001 Age (per year increase) 1.00 ( ) ( ) 0.73 Race Non-Black 1.0 (ref) 1.0 (ref) 1.0 (ref) Black 1.06 ( ) ( ) ( ) (1) Multivariable analysis adjusted for all variables presented in the table; (2) Multivariable analysis adjusted for all variables presented in the table in addition to year of diagnosis; (3) Multivariable analysis adjusted for all variables presented in the table and only including patients with at least 5 years of follow-up, i.e. those diagnosed from 2004 to 2006 ( patients). Table 3 Multivariable Fine and Gray competing risks regression for PCSM among patients with pathological ( patients) or biopsy confirmed ( patients) 8 or 9 disease. Characteristic Multivariable analysis* for patients with pathologically confirmed Gleason score 8 or 9 disease ( RP patients) Multivariable analysis* for patients with biopsy confirmed 8 or 9 disease ( non-rp patients) AHR (95% CI) P AHR (95% CI) P = (ref) 1.0 (ref) = ( ) ( ) = ( ) ( ) < ( ) < ( ) <0.001 PSA (per ng/ml increase) 1.00 ( ) < ( ) <0.001 T stage T1 1.0 (ref) 1.0 (ref) T ( ) < ( ) <0.001 T3 T ( ) < ( ) <0.001 Treatment No radiation therapy 1.0 (ref) 1.0 (ref) Radiation therapy 1.40 ( ) ( ) <0.001 Age (per year increase) 1.00 ( ) ( ) 0.21 Race Non-Black 1.0 (ref) 1.0 (ref) Black 0.87 ( ) ( ) 0.19 *Multivariable analyses are adjusted for all variables presented in the table. Referent/ No Radiation Therapy among RP patients includes patients who received RP as monotherapy without combined radiation. Referent/ No Radiation Therapy among non-rp patients includes patients who did not receive definitive therapy. Radiation among RP patients includes patients who received combined RP and radiation therapy. Radiation among non-rp patients includes patients who received radiation monotherapy without RP. Discussion In the present study of a large contemporary cohort of patients in the USA with high-risk 8 or 9 prostate cancer, we found that patients with = 8 or 9 disease have an up to two-fold increased hazard and incidence of PCSM when compared with patients with = 8 disease. The 5-year PCSM with = 8 was indistinguishable from those with 9 disease (13.5% and 13.9%, respectively), while 5-year PCSM for those with = 8 and = 8 disease were significantly lower (6.6% and 98 BJU International 2015 BJU International

5 8 prostate cancer 6.3%, respectively). All findings remained significant and similar in magnitude when limiting the analyses to patients with pathological s (patients who received RP) and when limiting analyses to patients with only biopsy s (patients who did not receive RP). These results strongly suggest that 8 disease represents a heterogeneous entity with PCSM outcomes and risk driven by the primary Gleason pattern. Our present study has significant clinical implications that should guide and potentially alter clinical nomograms, future treatment, and study approaches to high-risk prostate cancer. Namely, clinical nomograms, management approaches, and clinical trials should incorporate the heterogeneity of Gleason score 8 disease based on the primary Gleason pattern. In current clinical practice, 8 and 9 scores are grouped together, along with Gleason 10 disease, into a single high-risk clinical group that are all treated according to the same clinical guidelines [2]. Although treatment algorithms group 8 10 disease together, clinicians generally recognise that 9 10 disease is associated with worse prostate cancer outcome than Gleason score 8 disease [14], and may factor this prognostic difference into treatment decisions where guidelines allow room for variation. For example, NCCN guidelines recommend that men receiving radiation for 8 10 prostate cancer should receive 2 3 years of androgen-deprivation therapy [15,16], and some clinicians may lean towards 3 years for 9 or 10 vs 2 years for 8. Our present data suggests that clinicians who have adopted this treatment algorithm should treat 5 + 3as aggressively as they treat 9 disease. Our present data also have implications for the design of clinical trials. Investigators designing clinical trials of novel agents that seek to enrol the highest risk patients to maximise events and thereby maximise the ability to detect differences might consider enrolling patients with 9, 10, and = 8 disease, and should recognise that their event rate would be significantly lower if the trial allowed all patients with 8 10 and that had a large proportion of patients with 3 +5 = 8or4+ 4 = 8 disease. Although current validated predictive prostate cancer clinical nomograms, such as the University of California, San Francisco-Cancer of the Prostate Risk Assessment (UCSF- CAPRA) score, Partin tables, and Kattan nomogram, attempt to capture heterogeneity within prostate cancer risk groups, they are not able to accurately capture the heterogeneity that probably exists within high-risk 8 disease [17 19]. For example, when calculating a UCSF-CAPRA score, a patient with = 8 disease would be assigned a higher risk score (4 total points for Gleason score risk = 3 points for primary Gleason pattern 4 or 5 disease + 1 point for secondary Gleason pattern 4 or 5 disease) than a patient with = 8 disease (3 total points for risk = 3 points for primary Gleason pattern 4 or 5 disease + 0 points for secondary Gleason pattern 3 disease); of note, patients with Gleason score = 8 and 9 disease receive the same UCSF-CAPRA Gleason risk score [17]. Similarly, when calculating the 15- year prostate cancer-specific survival for men with prostate cancer treated with RP using the Kattan nomogram, 15-year disease-specific survival estimates are lower for patients with = 8 disease than for those with Gleason score = 8 disease; the Kattan nomogram also predicts similar outcome for patients with = 8 disease as for those with 9 disease [19]. Lastly, Partin tables, used to predict extension of tumour beyond the prostate, do not differentiate by primary pattern within Gleason 8 disease [18]. In both the UCSF-CAPRA and Kattan nomograms described above, patients with = 8 disease are assigned a worse risk score for adverse prostate-cancer specific outcomes than those with = 8 disease. However, our present data suggest that nomograms should consider a of = 8 to be much more unfavourable than = 8. Similarly, the UCSF- CAPRA score considers a of = 8tobeas unfavourable as that of 9 disease, but our present results suggest that = 8 disease is much less likely to result in PCSM than 9 disease. Given our present findings, we suggest that clinical nomograms be updated with a heavier emphasis on the primary pattern. Specifically, nomograms should dictate that patients with = 8 disease have the same risk of PCSM as those with 9 disease, and patients with = 8 disease have the same risk for PCSM as those with = 8 disease. Of note, the methods for Gleason scoring of biopsies changed with the 2005 International Society of Urological Pathology (ISUP) guidelines, as described by Epstein et al. [20]. Specifically, these guidelines suggested that tertiary grades no longer be reported for biopsy specimens and only be reported for pathological specimens and that Gleason scoring of biopsies be based on the primary grade plus the highest grade (instead of the primary plus secondary); e.g with tertiary grade 5 became renamed in the biopsies after Given the new scoring guidelines from the ISUP it makes sense, therefore, that the patients with a of have better outcomes than those with a of 5 + 3, as many of the patients with disease probably had primary 3 and secondary 4 with tertiary grade 5 ( 10% grade 5). Similarly, the 9 could have been 4 + 5or5+ 4, and it makes sense that the primary pattern of 4 or 5 does much BJU International 2015 BJU International 99

6 Mahal et al. worse than the primary pattern of 3 among the patients with disease. This is the first study, in a large national cohort, to show large PCSM differences among patients with 8 disease. Our present study results, as detailed above, show the importance of primary pattern 5 disease among patients with 8 prostate cancer. Prior studies have suggested that pattern 5 disease is a powerful and significant predictor of prostate cancer outcome; however, studies either tend to focus on secondary or tertiary pattern 5 disease or do not examine or demonstrate differences within 8 disease [21 25]. Furthermore, although it has been recognised that primary Gleason pattern is associated with prostate cancer outcome, most of the developed data has focused on 7 disease (4 + 3vs3+ 4) or has lumped primary 4 and 5 together without making a distinction between the two and so there is little to no data on the differences in cancer outcome within 8 disease [6 9]. Given the paucity of literature focusing on primary pattern within 8 disease, our present study represents a novel and important contribution to the literature that is illustrative of the heterogeneity of Gleason score 8 disease. Our present results must be viewed within the limitations of the study. First, SEER replaced the biopsy with the RP for surgically managed patients from 2004 to However, we repeated Fine and Gray competing risks regressions among patients who received RP (pathological ) and among patients who did not receive RP (biopsy/clinical ) and PCSM hazard associations by remained unchanged (Table 3). Second, the median follow-up of our present study was only 36 months. Nevertheless, this was still enough time to see differences in PCSM among patients with 8 disease. Future studies with longer follow-up could inform whether the associations seen in the present study hold over time. Lastly, SEER does not provide information on androgen-deprivation therapy; however, it is assumed that most of the high-risk patients in this cohort who received radiation therapy received guideline concordant care and therefore also received androgen-deprivation therapy. Despite the potential limitations of our present study, this study in a large national contemporary cohort reveals the clinical heterogeneity of 8 disease. Specifically, we found that patients with = 8 disease, similar to patients with 9 disease, have a nearly two-fold increased risk of PCSM compared with other patients with 8 disease. Our present findings challenge current prostate cancer nomograms for outcomes among patients with 8 disease and suggest that PCSM outcomes and risk are driven by the primary Gleason pattern. Patients with 8 and primary Gleason 5 disease should be counselled differently about risks of recurrence and death from prostate cancer than a patient with 8 and primary 4 or 3 disease, and should be enrolled on clinical trials testing novel aggressive methods and new therapeutic agents for advanced prostate cancer. Acknowledgements This work is supported by David and Cynthia Chapin, the Prostate Cancer Foundation, Fitz s Cancer Warriors, Hugh Simons in Honour of Frank and Anne Simons, the Scott Forbes/Gina Ventre Fund, and a grant from an Anonymous Family Foundation. Conflicts of Interest Paul L. Nguyen consulted for Medivation/Astellas and Genome DX. Felix Y. Feng has consulted for Medivation/ Astellas, Celgene, and Genome Dx, and receives grant funding from Varian and Medivation/Astellas. James B. Yu has funding from 21st century oncology. The remaining authors have no conflict of interest disclosures to report. References 1 Boorjian SA, Karnes RJ, Rangel LJ, Bergstralh EJ, Blute ML. Mayo Clinic validation of the D amico risk group classification for predicting survival following radical prostatectomy. J Urol 2008; 179: National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology, Prostate Cancer Available at: Accessed December D Amico AV, Whittington R, Malkowicz SB et al. Biochemical outcome after radical prostatectomy, external beam radiation therapy, or interstitial radiation therapy for clinically localized prostate cancer. JAMA 1998; 280: Epstein JI. An update of the Gleason grading system. J Urol 2010; 183: Andren O, Fall K, Franzen L, Andersson SO, Johansson JE, Rubin MA. How well does the predict prostate cancer death? A 20-year followup of a population based cohort in Sweden. J Urol 2006; 175: Wright JL, Salinas CA, Lin DW et al. Prostate cancer specific mortality and Gleason 7 disease differences in prostate cancer outcomes between cases with Gleason and Gleason tumors in a population based cohort. J Urol 2009; 182: Lavery HJ, Droller MJ. Do Gleason patterns 3 and 4 prostate cancer represent separate disease states? J Urol 2012; 188: Stark JR, Perner S, Stampfer MJ et al. and lethal prostate cancer: does = 4 + 3? J Clin Oncol 2009; 27: Turker P, Bas E, Bozkurt S et al. Presence of high grade tertiary Gleason pattern upgrades the Gleason sum score and is inversely associated with biochemical recurrence-free survival. Urol Oncol 2013; 31: Surveillance, Epidemiology, and End Results (SEER) Program. Research Data ( ), National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, based on November 2013 SEER data submission, posted to the SEER web site, April Available at: Accessed April Surveillance, Epidemiology, and End Results (SEER) Program. SEER Cancer Statistics Review, , National Cancer Institute. Bethesda, MD, based on November 2013 SEER data submission, posted to the SEER 100 BJU International 2015 BJU International

7 8 prostate cancer web site, April Available at: Accessed April Surveillance, Epidemiology, and End Results (SEER) Program. Appendix C: Site Specific Coding Modules SEER Coding and Staging Manual. Available at: Accessed May Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999; 94: Tsao CK, Gray KP, Nakabayashi M et al. Patients with Biopsy Gleason 9 and 10 prostate cancer have significantly worse outcomes compared with Gleason 8 disease. J Urol 2015; 194: Bolla M, de Reijke TM, Van Tienhoven G et al. Duration of androgen suppression in the treatment of prostate cancer. N Engl J Med 2009; 360: Horwitz EM, Bae K, Hanks GE et al. Ten-year follow-up of radiation therapy oncology group protocol 92-02: a phase III trial of the duration of elective androgen deprivation in locally advanced prostate cancer. J Clin Oncol 2008; 26: Cooperberg MR, Pasta DJ, Elkin EP et al. The University of California, San Francisco Cancer of the Prostate Risk Assessment score: a straightforward and reliable preoperative predictor of disease recurrence after radical prostatectomy. J Urol 2005; 173: Partin AW, Kattan MW, Subong EN et al. Combination of prostatespecific antigen, clinical stage, and to predict pathological stage of localized prostate cancer. A multi-institutional update. JAMA 1997; 277: Kattan MW, Eastham JA, Stapleton AM, Wheeler TM, Scardino PT. A preoperative nomogram for disease recurrence following radical prostatectomy for prostate cancer. J Natl Cancer Inst 1998; 90: 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: Hattab EM, Koch MO, Eble JN, Lin H, Cheng L. Tertiary Gleason pattern 5 is a powerful predictor of biochemical relapse in patients with 7 prostatic adenocarcinoma. J Urol 2006; 175: Whittemore DE, Hick EJ, Carter MR, Moul JW, Miranda-Sousa AJ, Sexton WJ. Significance of tertiary Gleason pattern 5 in 7 radical prostatectomy specimens. J Urol 2008; 179: Trpkov K, Zhang J, Chan M, Eigl BJ, Yilmaz A. Prostate cancer with tertiary Gleason pattern 5 in prostate needle biopsy: clinicopathologic findings and disease progression. Am J Surg Pathol 2009; 33: van Oort IM, Schout BM, Kiemeney LA, Hulsbergen CA, Witjes JA. Does the tertiary Gleason pattern influence the PSA progression-free interval after retropubic radical prostatectomy for organ-confined prostate cancer? Eur Urol 2005; 48: Patel AA, Chen MH, Renshaw AA, D Amico AV. PSA failure following definitive treatment of prostate cancer having biopsy 7 with tertiary grade 5. JAMA 2007; 298: Correspondence: Paul L. Nguyen, Dana Farber Cancer Institute/Brigham and Women s Hospital, Harvard Medical School, 75 Francis Street, Boston, MA 02115, USA. pnguyen@lroc.harvard.edu Abbreviations: AHR, adjusted hazard ratio; ISUP, International Society of Urological Pathology; NCCN, National Comprehensive Cancer Network; PCSM, prostate cancer-specific mortality; RP, radical prostatectomy; SEER, Surveillance, Epidemiology and End Results; UCSF-CAPRA, the University of California, San Francisco-Cancer of the Prostate Risk Assessment. BJU International 2015 BJU International 101