- Email: [email protected]
Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death
EURURO-6411; No. of Pages 4 EUROPEAN UROLOGY XXX (2015) XXX–XXX
available at www.sciencedirect.com journal homepage: www.europeanurology.com
Brief Correspondence
Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death Mai Anh Huynh a,*, Ming-Hui Chen b, Jing Wu b, Michelle H. Braccioforte c, Brian J. Moran c, Anthony V. D’Amico d a
Harvard Radiation Oncology Program, Brigham and Women’s Hospital, Boston, MA, USA; b Department of Statistics, University of Connecticut, Storrs, CT,
USA;
c
Chicago Prostate Cancer Center, Westmont, IL, USA;
d
Department of Radiation Oncology, Brigham and Women’s Hospital-Dana Farber Cancer
Institute, Boston, MA, USA
Article info
Abstract
Article history: Accepted August 29, 2015
The International Society of Urological Pathology recommends that Gleason score (GS) 8 prostate cancer (PC) is one prognostic category, yet heterogeneity in cancer control potentially exists amongst men with GS 3 + 5/5 + 3 versus GS 4 + 4 PC. We compared PCspecific mortality (PCSM) and all-cause mortality (ACM) risk among men with GS 3 + 5/ 5 + 3 versus GS 4 + 4 PC using competing-risks and Cox regression analyses, adjusting for age, known PC prognostic factors, treatment, and a treatment propensity score. Between 1998 and 2012, 462 men with GS 8 PC were treated using brachytherapy with supplemental external-beam radiation therapy and/or androgen deprivation therapy at the Chicago Prostate Cancer Center. After a median follow-up of 7.6 yr, 118 men died, 26 of PC. PCSM (adjusted hazard ratio [AHR] 2.77, 95% confidence interval [CI] 1.13–6.80; p = 0.026) and ACM (AHR 1.75, 95% CI 1.06–2.87; p = 0.028) were significantly higher for men with GS 3 + 5/5 + 3 PC than for men with GS 4 + 4 PC. Subcategorizing GS 8 into PC with or without grade 5 should be considered as a stratification factor in randomized trials. Patient summary: Long-term success rates for men with Gleason score 8 prostate cancer vary depending on whether the most aggressive type of cancer (grade 5) is present at biopsy. # 2015 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Associate Editor: Giacomo Novara Keywords: Gleason score Prostate cancer Mortality
* Corresponding author. Harvard Radiation Oncology Program, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115, USA. Tel. +1 617 7326310; Fax: +1 617 6324247. E-mail address: [email protected] (M.A. Huynh).
Multiple studies have shown that higher Gleason score (GS) is an important prognostic factor across all treatments for prostate cancer (PC) [1]. Using data on prostate-specific antigen (PSA) recurrence, the International Society of Urological Pathology (ISUP) has recommended a five-tiered prognostic staging system [2], including: group 1, GS 6; group 2, GS 3 + 4 = 7; group 3, GS 4 + 3 = 7; group 4, GS 8; and group 5, GS 9–10 [3]. However, the known negative prognostic significance of Gleason grade 5 compared to Gleason grade 4 [4] raises the concern that prognostic group 4 (GS 8) is subject to heterogeneity with respect to longterm PC outcomes, including PC-specific mortality (PCSM) and all-cause mortality (ACM). Specifically, men with
GS 3 + 5, GS 5 + 3, and GS 4 + 4 would be grouped into a single category but may have distinct outcomes given the known negative prognostic significance of Gleason grade 5 compared to grade 4 PC [4]. Therefore, the purpose of this study was to use a prospectively assembled database to ascertain whether men with GS 3 + 5 or 5 + 3 had a higher risk of PCSM and ACM compared to men with GS 4 + 4 after adjusting for age, known prognostic PC factors, treatment, and a treatment propensity score. Between January 6, 1998 and May 18, 2012, 462 men (median age 72.08 yr) with 2002 American Joint Commission on Cancer (AJCC) tumor (T) category5 1c-3 and high-risk PC
http://dx.doi.org/10.1016/j.eururo.2015.08.054 0302-2838/# 2015 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Huynh MA, et al. Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death. Eur Urol (2015), http://dx.doi.org/10.1016/j.eururo.2015.08.054
EURURO-6411; No. of Pages 4 2
EUROPEAN UROLOGY XXX (2015) XXX–XXX
based on the highest biopsy GS being 8 (3 + 5 or 4 + 4 or 5 + 3) from at least one core formed the prospectively assembled study cohort. The biopsy GS was assigned by a pathologist with expertise in genitourinary cancers. Men were treated using prostate brachytherapy and additional supplemental therapies, including neoadjuvant external-beam radiation therapy (EBRT, 45 Gy; n = 63) or androgen deprivation therapy (ADT; n = 117) or both (n = 157), at the Chicago Prostate Cancer Center. At the time of PSA failure, defined as PSA nadir +2 ng/ml, salvage ADT was administered. Univariate and multivariate Cox [6] and Fine-Gray [7] regression models were used to assess whether men with GS 3 + 5/5 + 3 versus GS 4 + 4 were at higher risk of ACM or PCSM respectively, adjusting for age, PSA, T category, treatment, and treatment propensity score. Cox regression [6] was used for ACM given all deaths are considered events, whereas competing-risks regression [7] was used for PCSM and other-cause mortality (OCM) given that 92 of the 118 deaths were from other causes. Age-adjusted ACM, defined as 1 – [Kaplan-Meier estimates [8] of overall survival (OS)], and cumulative incidence [9] estimates of PCSM and OCM were calculated for men with GS 3 + 5/5 + 3 versus GS 4 + 4. SAS version 9.3 (SAS Institute, Cary, NC, USA) was used for all calculations apart from the Fine-Gray regression and cumulative incidence estimates, for which R version 3.0.1 (R Project for Statistical Computing, Vienna, Austria) was used. After a median follow-up of 7.6 yr (interquartile range [IQR] 5.6–10.5 yr), 118 of the 462 men died, 26 of PC. There was significantly higher risk of PCSM (adjusted hazard ratio [AHR] 2.77, 95% confidence interval [CI] 1.13–6.80; p = 0.026) and ACM (AHR 1.75, 95% CI 1.06–2.87; p = 0.028) but not OCM (AHR 1.37, 95% CI 0.79–2.40; p = 0.26) among men with GS 3 + 5/5 + 3 compared to GS
4 + 4 PC after adjusting for treatment and known PC prognostic factors (Table 1). When looking at the individual terms of GS 5 + 3/3 + 5 versus GS 4 + 4, the higher risk for both PCSM and ACM remained, although this was only significant for PCSM among men with GS 5 + 3 compared to GS 4 + 4 (Table 1), justifying collapse of these two GS subgroups into one cohort for illustration in Figure 1. Increasing age was also significantly associated with higher ACM risk (AHR 1.07, 95% CI 1.02–1.12; p = 0.004) but did not reach statistical significance for PCSM (AHR 1.08, 95% CI 0.999–1.17; p = 0.054). As shown in Figure 1A–C, cumulative incidence estimates of PCSM (p = 0.02), OCM (p = 0.18) and age-adjusted ACM (p = 0.01), were significantly higher, not significantly different, and significantly higher respectively for GS 3 + 5/ 5 + 3 compared to GS 4 + 4. We observed that the risks of both PCSM and ACM were significantly higher in men with GS 3 + 5/5 + 3 compared to those with GS 4 + 4 PC. This observation was noted after adjusting for treatment received, a treatment propensity score, known PC prognostic factors, and age. The clinical significance of this finding is that the ISUP proposal [3] to classify men with GS 8 in a single category (prognostic group 4) may not be optimal given the difference in longterm PCSM and ACM outcomes we observed between GS 5 + 3/3 + 5 and GS 4 + 4 groups. Specifically, our results provide evidence supporting subdivision of men with GS 8 PC into two prognostic groups: those with and without Gleason grade 5 PC. Prospective and adequately powered studies are warranted to validate our results in the contemporary era of 12-core prostate needle biopsy and post-2005 ISUP recommendations [10] and within individual treatment modalities to ensure generalizability across treatment modalities for patients presenting today.
Table 1 – Unadjusted and adjusted hazard ratios for overall mortality [6] and prostate cancer–specific mortality [7] for each clinical factor Clinical factor
Men (n)
Gleason score a 4+4 3 + 5 or 5 + 3 Log PSA (ng/ml) Age (yr) Clinical tumor category b T1 T2–3 Treatment Other BT + ADT + EBRT Treatment propensity score
ACM Deaths
Univariable
PCSM Multivariable
(n)
HR (95% CI)
p value
AHR (95% CI)
421 41 462 462
99 19 118 118
1.0 (reference) 1.91 (1.17–3.13) 1.20 (0.98–1.48) 1.06 (1.04–1.09)
– 0.01 0.08 <0.0001
1.0 (reference) 1.75 (1.06–2.87) 1.19 (0.84–1.68) 1.07 (1.02–1.12)
– 0.028 0.33 0.004
223 239
43 75
1.0 (reference) 1.74 (1.20–2.54)
– 0.004
1.0 (reference) 1.53 (0.88–2.66)
305 157 462
80 38 118
1.0 (reference) 0.86 (0.58–1.26) 1.0 (0.99–1.01)
– 0.43 0.68
1.0 (reference) 0.86 (0.56–1.33) 1.01 (0.98–1.03)
PC
p value deaths (n)
Univariable
Multivariable
HR (95% CI)
p value
AHR (95% CI)
p value
20 6 26 26
1.0 (reference) 2.92 (1.17–7.27) 1.45 (0.84–2.50) 1.04 (0.97–1.10)
– 0.021 0.18 0.26
1.0 (reference) 2.77 (1.13–6.80) 1.11 (0.48–2.57) 1.08 (0.999–1.17)
– 0.026 0.80 0.054
– 0.13
5 21
1.0 (reference) 4.05 (1.53–10.70)
– 0.005
1.0 (reference) 2.60 (0.73–9.19)
– 0.14
– 0.50 0.56
14 12 26
1.0 (reference) 1.60 (0.74–3.45) 1.02 (0.998–1.04)
– 0.23 0.08
1.0 (reference) 1.22 (0.53–2.79) 1.02 (0.98–1.06)
– 0.64 0.28
ACM = all-cause mortality; PC = prostate cancer; PCSM = PC-specific mortality; HR = hazard ratio; CI = confidence interval; AHR = adjusted HR; GS = Gleason score; PSA = prostate-specific antigen; BT = brachytherapy; ADT = androgen deprivation therapy; EBRT, external beam radiation therapy. a GS 5 + 3 versus GS 4 + 4: AHR 3.39 (95% CI 1.15–9.99; p = 0.027) for PCSM; AHR 1.76 (95% CI 0.80–3.83; p = 0.16) for ACM. GS 3 + 5 versus GS 4 + 4: AHR 2.36 (95% CI 0.67–8.32; p = 0.18) for PCSM; AHR 1.74 (95% CI 0.95–3.22; p = 0.07) for ACM. GS 3 + 5/5 + 3 versus GS 4 + 4: AHR 1.37 (95% CI 0.79–2.40; p = 0.26) for other-cause mortality (OCM). None of the other covariates in the multivariable model reached statistical significance; however, the result for increasing age was AHR 1.05 (95% CI 0.996–1.11; p = 0.067). Patient characteristics at study entry are listed in Supplementary Table 1. Results for all covariates from the univariate and multivariate analyses for OCM are reported in Supplementary Table 2. b According to the American Joint Cancer Center 2002 [5].
Please cite this article in press as: Huynh MA, et al. Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death. Eur Urol (2015), http://dx.doi.org/10.1016/j.eururo.2015.08.054
EURURO-6411; No. of Pages 4 EUROPEAN UROLOGY XXX (2015) XXX–XXX
3
Moreover, investigation of the impact of the percentage of Gleason grade 5 tissue in biopsy cores would be informative. Despite these potential limitations, it appears that men with GS 3 + 5/5 + 3 PC have a higher risk of both PCSM and ACM compared to men with GS 4 + 4 PC after adjusting for age, treatment, and known PC prognostic factors. Therefore, this subdivision could be considered as a prerandomization stratification factor in planned and future randomized trials evaluating the impact of adding novel treatments to current standards of care on PCSM and ACM and in postrandomization analyses of ongoing or completed randomized studies that include men with GS 8 PC. Author contributions: Mai Anh Huynh had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: D’Amico. Acquisition of data: Braccioforte, Moran. Analysis and interpretation of data: Huynh, Chen, Wu, D’Amico. Drafting of the manuscript: Huynh, Chen, D’Amico. Critical revision of the manuscript for important intellectual content: Huynh, Chen, Wu, Braccioforte, Moran, D’Amico. Statistical analysis: Chen, Wu. Obtaining funding: None. Administrative, technical, or material support: Braccioforte. Supervision: D’Amico. Other: None. Financial disclosures: Mai Anh Huynh certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None. Funding/Support and role of the sponsor: None. Acknowledgments: The authors would like to thank and recognize Dr. Andrew Renshaw for his suggestion that men with Gleason score 8 may not all have the same prognosis due to the distinct possibilities of Gleason score 4 + 4, 3 + 5 and 5 + 3.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. eururo.2015.08.054.
References [1] Humphrey PA. Gleason grading and prognostic factors in carcinoma of the prostate. Mod Pathol 2004;17:292–306. [2] Humphrey P, Epstein J, Amin M, Delahunt B, Egevad L, Srigley J. Paper presented at: International Society of Urological Pathology (ISUP) Gleason Grading Consensus Meeting; January 19, 2015, 2014; Chicago, IL.
Fig. 1 – (a) Cumulative incidence of prostate cancer–specific mortality (PCSM) stratified by Gleason score (GS) 3 + 5/5 + 3 versus GS 4 + 4. Gray’s k-sample p = 0.02; 8-yr point estimates of PCSM: GS 3 + 5/5 + 3, 13% (95% confidence interval [CI] 4.7–27%); GS 4 + 4, 5.9% (95% CI 3.5–9.1%). (b) Cumulative incidence of other-cause mortality (OCM) stratified by GS 3 + 5/5 + 3 versus GS 4 + 4. Gray’s k-sample p = 0.18; 8-yr point estimates
of OCM: GS 3 + 5/5 + 3, 26% (95% CI 13–42%); GS 4 + 4, 23% (95% CI 18–28%). (c) Age-adjusted [8] estimates of all-cause mortality (ACM = 1 – Kaplan-Meier estimate of overall survival) stratified by GS 3 + 5/5 + 3 versus GS 4 + 4. Age adjusted log-rank p = 0.01; 8-yr point estimates of ACM: GS 3 + 5/5 + 3, 39% (95% CI 26–57%); GS 4 + 4, 29% (95% CI 24–34%).
Please cite this article in press as: Huynh MA, et al. Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death. Eur Urol (2015), http://dx.doi.org/10.1016/j.eururo.2015.08.054
EURURO-6411; No. of Pages 4 4
EUROPEAN UROLOGY XXX (2015) XXX–XXX
[3] Epstein JI, Zelefsky MJ, Sjoberg DD, et al. A contemporary prostate cancer grading system: a validated alternative to the Gleason score. Eur Urol. In press. http://dx.doi.org/10.1016/j.eururo.2015.06.046 [4] Lim SK, Kim KH, Shin TY, et al. Gleason 5 + 4 has worse oncological and pathological outcomes compared with Gleason 4 + 5: significance of Gleason 5 pattern. Ann Surg Oncol 2013;20:3127–32. [5] Greene FL, Page DL, Fleming ID, et al. AJCC cancer staging manual. ed. 6. Philadelphia, PA: Lippincott; 2002. [6] Klein JP, Moeschberger ML. Semiparametric proportional hazards
[7] Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496–509. [8] Cupples LA, Gagnon DR, Ramaswamy R, D’Agostino RB. Age-adjusted survival curves with application in the Framingham Study. Stat Med 1995;14:1731–44. [9] Gaynor JJ, Fener EJ, Tan CC, et al. On the use of cause-specific failure and conditional failure probabilities: examples from clinical oncology data. J Am Stat Assoc 1993;88:400–9. [10] Epstein JI, Allsbrook Jr WC, Amin MB, Egevad LL, ISUP Grading
regression with fixed covariates. In: Survival analysis: techniques for
Committee. The 2005 International Society of Urological Pathology
censored and truncated data. ed. 2. New York, NY: Springer; 2010. p.
(ISUP) consensus conference on Gleason grading of prostatic carci-
243–93.
noma. Am J Surg Pathol 2005;29:1228–42.
Please cite this article in press as: Huynh MA, et al. Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death. Eur Urol (2015), http://dx.doi.org/10.1016/j.eururo.2015.08.054
available at www.sciencedirect.com journal homepage: www.europeanurology.com
Brief Correspondence
Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death Mai Anh Huynh a,*, Ming-Hui Chen b, Jing Wu b, Michelle H. Braccioforte c, Brian J. Moran c, Anthony V. D’Amico d a
Harvard Radiation Oncology Program, Brigham and Women’s Hospital, Boston, MA, USA; b Department of Statistics, University of Connecticut, Storrs, CT,
USA;
c
Chicago Prostate Cancer Center, Westmont, IL, USA;
d
Department of Radiation Oncology, Brigham and Women’s Hospital-Dana Farber Cancer
Institute, Boston, MA, USA
Article info
Abstract
Article history: Accepted August 29, 2015
The International Society of Urological Pathology recommends that Gleason score (GS) 8 prostate cancer (PC) is one prognostic category, yet heterogeneity in cancer control potentially exists amongst men with GS 3 + 5/5 + 3 versus GS 4 + 4 PC. We compared PCspecific mortality (PCSM) and all-cause mortality (ACM) risk among men with GS 3 + 5/ 5 + 3 versus GS 4 + 4 PC using competing-risks and Cox regression analyses, adjusting for age, known PC prognostic factors, treatment, and a treatment propensity score. Between 1998 and 2012, 462 men with GS 8 PC were treated using brachytherapy with supplemental external-beam radiation therapy and/or androgen deprivation therapy at the Chicago Prostate Cancer Center. After a median follow-up of 7.6 yr, 118 men died, 26 of PC. PCSM (adjusted hazard ratio [AHR] 2.77, 95% confidence interval [CI] 1.13–6.80; p = 0.026) and ACM (AHR 1.75, 95% CI 1.06–2.87; p = 0.028) were significantly higher for men with GS 3 + 5/5 + 3 PC than for men with GS 4 + 4 PC. Subcategorizing GS 8 into PC with or without grade 5 should be considered as a stratification factor in randomized trials. Patient summary: Long-term success rates for men with Gleason score 8 prostate cancer vary depending on whether the most aggressive type of cancer (grade 5) is present at biopsy. # 2015 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Associate Editor: Giacomo Novara Keywords: Gleason score Prostate cancer Mortality
* Corresponding author. Harvard Radiation Oncology Program, Brigham and Women’s Hospital, 75 Francis Street, Boston, MA 02115, USA. Tel. +1 617 7326310; Fax: +1 617 6324247. E-mail address: [email protected] (M.A. Huynh).
Multiple studies have shown that higher Gleason score (GS) is an important prognostic factor across all treatments for prostate cancer (PC) [1]. Using data on prostate-specific antigen (PSA) recurrence, the International Society of Urological Pathology (ISUP) has recommended a five-tiered prognostic staging system [2], including: group 1, GS 6; group 2, GS 3 + 4 = 7; group 3, GS 4 + 3 = 7; group 4, GS 8; and group 5, GS 9–10 [3]. However, the known negative prognostic significance of Gleason grade 5 compared to Gleason grade 4 [4] raises the concern that prognostic group 4 (GS 8) is subject to heterogeneity with respect to longterm PC outcomes, including PC-specific mortality (PCSM) and all-cause mortality (ACM). Specifically, men with
GS 3 + 5, GS 5 + 3, and GS 4 + 4 would be grouped into a single category but may have distinct outcomes given the known negative prognostic significance of Gleason grade 5 compared to grade 4 PC [4]. Therefore, the purpose of this study was to use a prospectively assembled database to ascertain whether men with GS 3 + 5 or 5 + 3 had a higher risk of PCSM and ACM compared to men with GS 4 + 4 after adjusting for age, known prognostic PC factors, treatment, and a treatment propensity score. Between January 6, 1998 and May 18, 2012, 462 men (median age 72.08 yr) with 2002 American Joint Commission on Cancer (AJCC) tumor (T) category5 1c-3 and high-risk PC
http://dx.doi.org/10.1016/j.eururo.2015.08.054 0302-2838/# 2015 European Association of Urology. Published by Elsevier B.V. All rights reserved.
Please cite this article in press as: Huynh MA, et al. Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death. Eur Urol (2015), http://dx.doi.org/10.1016/j.eururo.2015.08.054
EURURO-6411; No. of Pages 4 2
EUROPEAN UROLOGY XXX (2015) XXX–XXX
based on the highest biopsy GS being 8 (3 + 5 or 4 + 4 or 5 + 3) from at least one core formed the prospectively assembled study cohort. The biopsy GS was assigned by a pathologist with expertise in genitourinary cancers. Men were treated using prostate brachytherapy and additional supplemental therapies, including neoadjuvant external-beam radiation therapy (EBRT, 45 Gy; n = 63) or androgen deprivation therapy (ADT; n = 117) or both (n = 157), at the Chicago Prostate Cancer Center. At the time of PSA failure, defined as PSA nadir +2 ng/ml, salvage ADT was administered. Univariate and multivariate Cox [6] and Fine-Gray [7] regression models were used to assess whether men with GS 3 + 5/5 + 3 versus GS 4 + 4 were at higher risk of ACM or PCSM respectively, adjusting for age, PSA, T category, treatment, and treatment propensity score. Cox regression [6] was used for ACM given all deaths are considered events, whereas competing-risks regression [7] was used for PCSM and other-cause mortality (OCM) given that 92 of the 118 deaths were from other causes. Age-adjusted ACM, defined as 1 – [Kaplan-Meier estimates [8] of overall survival (OS)], and cumulative incidence [9] estimates of PCSM and OCM were calculated for men with GS 3 + 5/5 + 3 versus GS 4 + 4. SAS version 9.3 (SAS Institute, Cary, NC, USA) was used for all calculations apart from the Fine-Gray regression and cumulative incidence estimates, for which R version 3.0.1 (R Project for Statistical Computing, Vienna, Austria) was used. After a median follow-up of 7.6 yr (interquartile range [IQR] 5.6–10.5 yr), 118 of the 462 men died, 26 of PC. There was significantly higher risk of PCSM (adjusted hazard ratio [AHR] 2.77, 95% confidence interval [CI] 1.13–6.80; p = 0.026) and ACM (AHR 1.75, 95% CI 1.06–2.87; p = 0.028) but not OCM (AHR 1.37, 95% CI 0.79–2.40; p = 0.26) among men with GS 3 + 5/5 + 3 compared to GS
4 + 4 PC after adjusting for treatment and known PC prognostic factors (Table 1). When looking at the individual terms of GS 5 + 3/3 + 5 versus GS 4 + 4, the higher risk for both PCSM and ACM remained, although this was only significant for PCSM among men with GS 5 + 3 compared to GS 4 + 4 (Table 1), justifying collapse of these two GS subgroups into one cohort for illustration in Figure 1. Increasing age was also significantly associated with higher ACM risk (AHR 1.07, 95% CI 1.02–1.12; p = 0.004) but did not reach statistical significance for PCSM (AHR 1.08, 95% CI 0.999–1.17; p = 0.054). As shown in Figure 1A–C, cumulative incidence estimates of PCSM (p = 0.02), OCM (p = 0.18) and age-adjusted ACM (p = 0.01), were significantly higher, not significantly different, and significantly higher respectively for GS 3 + 5/ 5 + 3 compared to GS 4 + 4. We observed that the risks of both PCSM and ACM were significantly higher in men with GS 3 + 5/5 + 3 compared to those with GS 4 + 4 PC. This observation was noted after adjusting for treatment received, a treatment propensity score, known PC prognostic factors, and age. The clinical significance of this finding is that the ISUP proposal [3] to classify men with GS 8 in a single category (prognostic group 4) may not be optimal given the difference in longterm PCSM and ACM outcomes we observed between GS 5 + 3/3 + 5 and GS 4 + 4 groups. Specifically, our results provide evidence supporting subdivision of men with GS 8 PC into two prognostic groups: those with and without Gleason grade 5 PC. Prospective and adequately powered studies are warranted to validate our results in the contemporary era of 12-core prostate needle biopsy and post-2005 ISUP recommendations [10] and within individual treatment modalities to ensure generalizability across treatment modalities for patients presenting today.
Table 1 – Unadjusted and adjusted hazard ratios for overall mortality [6] and prostate cancer–specific mortality [7] for each clinical factor Clinical factor
Men (n)
Gleason score a 4+4 3 + 5 or 5 + 3 Log PSA (ng/ml) Age (yr) Clinical tumor category b T1 T2–3 Treatment Other BT + ADT + EBRT Treatment propensity score
ACM Deaths
Univariable
PCSM Multivariable
(n)
HR (95% CI)
p value
AHR (95% CI)
421 41 462 462
99 19 118 118
1.0 (reference) 1.91 (1.17–3.13) 1.20 (0.98–1.48) 1.06 (1.04–1.09)
– 0.01 0.08 <0.0001
1.0 (reference) 1.75 (1.06–2.87) 1.19 (0.84–1.68) 1.07 (1.02–1.12)
– 0.028 0.33 0.004
223 239
43 75
1.0 (reference) 1.74 (1.20–2.54)
– 0.004
1.0 (reference) 1.53 (0.88–2.66)
305 157 462
80 38 118
1.0 (reference) 0.86 (0.58–1.26) 1.0 (0.99–1.01)
– 0.43 0.68
1.0 (reference) 0.86 (0.56–1.33) 1.01 (0.98–1.03)
PC
p value deaths (n)
Univariable
Multivariable
HR (95% CI)
p value
AHR (95% CI)
p value
20 6 26 26
1.0 (reference) 2.92 (1.17–7.27) 1.45 (0.84–2.50) 1.04 (0.97–1.10)
– 0.021 0.18 0.26
1.0 (reference) 2.77 (1.13–6.80) 1.11 (0.48–2.57) 1.08 (0.999–1.17)
– 0.026 0.80 0.054
– 0.13
5 21
1.0 (reference) 4.05 (1.53–10.70)
– 0.005
1.0 (reference) 2.60 (0.73–9.19)
– 0.14
– 0.50 0.56
14 12 26
1.0 (reference) 1.60 (0.74–3.45) 1.02 (0.998–1.04)
– 0.23 0.08
1.0 (reference) 1.22 (0.53–2.79) 1.02 (0.98–1.06)
– 0.64 0.28
ACM = all-cause mortality; PC = prostate cancer; PCSM = PC-specific mortality; HR = hazard ratio; CI = confidence interval; AHR = adjusted HR; GS = Gleason score; PSA = prostate-specific antigen; BT = brachytherapy; ADT = androgen deprivation therapy; EBRT, external beam radiation therapy. a GS 5 + 3 versus GS 4 + 4: AHR 3.39 (95% CI 1.15–9.99; p = 0.027) for PCSM; AHR 1.76 (95% CI 0.80–3.83; p = 0.16) for ACM. GS 3 + 5 versus GS 4 + 4: AHR 2.36 (95% CI 0.67–8.32; p = 0.18) for PCSM; AHR 1.74 (95% CI 0.95–3.22; p = 0.07) for ACM. GS 3 + 5/5 + 3 versus GS 4 + 4: AHR 1.37 (95% CI 0.79–2.40; p = 0.26) for other-cause mortality (OCM). None of the other covariates in the multivariable model reached statistical significance; however, the result for increasing age was AHR 1.05 (95% CI 0.996–1.11; p = 0.067). Patient characteristics at study entry are listed in Supplementary Table 1. Results for all covariates from the univariate and multivariate analyses for OCM are reported in Supplementary Table 2. b According to the American Joint Cancer Center 2002 [5].
Please cite this article in press as: Huynh MA, et al. Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death. Eur Urol (2015), http://dx.doi.org/10.1016/j.eururo.2015.08.054
EURURO-6411; No. of Pages 4 EUROPEAN UROLOGY XXX (2015) XXX–XXX
3
Moreover, investigation of the impact of the percentage of Gleason grade 5 tissue in biopsy cores would be informative. Despite these potential limitations, it appears that men with GS 3 + 5/5 + 3 PC have a higher risk of both PCSM and ACM compared to men with GS 4 + 4 PC after adjusting for age, treatment, and known PC prognostic factors. Therefore, this subdivision could be considered as a prerandomization stratification factor in planned and future randomized trials evaluating the impact of adding novel treatments to current standards of care on PCSM and ACM and in postrandomization analyses of ongoing or completed randomized studies that include men with GS 8 PC. Author contributions: Mai Anh Huynh had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: D’Amico. Acquisition of data: Braccioforte, Moran. Analysis and interpretation of data: Huynh, Chen, Wu, D’Amico. Drafting of the manuscript: Huynh, Chen, D’Amico. Critical revision of the manuscript for important intellectual content: Huynh, Chen, Wu, Braccioforte, Moran, D’Amico. Statistical analysis: Chen, Wu. Obtaining funding: None. Administrative, technical, or material support: Braccioforte. Supervision: D’Amico. Other: None. Financial disclosures: Mai Anh Huynh certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None. Funding/Support and role of the sponsor: None. Acknowledgments: The authors would like to thank and recognize Dr. Andrew Renshaw for his suggestion that men with Gleason score 8 may not all have the same prognosis due to the distinct possibilities of Gleason score 4 + 4, 3 + 5 and 5 + 3.
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. eururo.2015.08.054.
References [1] Humphrey PA. Gleason grading and prognostic factors in carcinoma of the prostate. Mod Pathol 2004;17:292–306. [2] Humphrey P, Epstein J, Amin M, Delahunt B, Egevad L, Srigley J. Paper presented at: International Society of Urological Pathology (ISUP) Gleason Grading Consensus Meeting; January 19, 2015, 2014; Chicago, IL.
Fig. 1 – (a) Cumulative incidence of prostate cancer–specific mortality (PCSM) stratified by Gleason score (GS) 3 + 5/5 + 3 versus GS 4 + 4. Gray’s k-sample p = 0.02; 8-yr point estimates of PCSM: GS 3 + 5/5 + 3, 13% (95% confidence interval [CI] 4.7–27%); GS 4 + 4, 5.9% (95% CI 3.5–9.1%). (b) Cumulative incidence of other-cause mortality (OCM) stratified by GS 3 + 5/5 + 3 versus GS 4 + 4. Gray’s k-sample p = 0.18; 8-yr point estimates
of OCM: GS 3 + 5/5 + 3, 26% (95% CI 13–42%); GS 4 + 4, 23% (95% CI 18–28%). (c) Age-adjusted [8] estimates of all-cause mortality (ACM = 1 – Kaplan-Meier estimate of overall survival) stratified by GS 3 + 5/5 + 3 versus GS 4 + 4. Age adjusted log-rank p = 0.01; 8-yr point estimates of ACM: GS 3 + 5/5 + 3, 39% (95% CI 26–57%); GS 4 + 4, 29% (95% CI 24–34%).
Please cite this article in press as: Huynh MA, et al. Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death. Eur Urol (2015), http://dx.doi.org/10.1016/j.eururo.2015.08.054
EURURO-6411; No. of Pages 4 4
EUROPEAN UROLOGY XXX (2015) XXX–XXX
[3] Epstein JI, Zelefsky MJ, Sjoberg DD, et al. A contemporary prostate cancer grading system: a validated alternative to the Gleason score. Eur Urol. In press. http://dx.doi.org/10.1016/j.eururo.2015.06.046 [4] Lim SK, Kim KH, Shin TY, et al. Gleason 5 + 4 has worse oncological and pathological outcomes compared with Gleason 4 + 5: significance of Gleason 5 pattern. Ann Surg Oncol 2013;20:3127–32. [5] Greene FL, Page DL, Fleming ID, et al. AJCC cancer staging manual. ed. 6. Philadelphia, PA: Lippincott; 2002. [6] Klein JP, Moeschberger ML. Semiparametric proportional hazards
[7] Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc 1999;94:496–509. [8] Cupples LA, Gagnon DR, Ramaswamy R, D’Agostino RB. Age-adjusted survival curves with application in the Framingham Study. Stat Med 1995;14:1731–44. [9] Gaynor JJ, Fener EJ, Tan CC, et al. On the use of cause-specific failure and conditional failure probabilities: examples from clinical oncology data. J Am Stat Assoc 1993;88:400–9. [10] Epstein JI, Allsbrook Jr WC, Amin MB, Egevad LL, ISUP Grading
regression with fixed covariates. In: Survival analysis: techniques for
Committee. The 2005 International Society of Urological Pathology
censored and truncated data. ed. 2. New York, NY: Springer; 2010. p.
(ISUP) consensus conference on Gleason grading of prostatic carci-
243–93.
noma. Am J Surg Pathol 2005;29:1228–42.
Please cite this article in press as: Huynh MA, et al. Gleason Score 3 + 5 or 5 + 3 versus 4 + 4 Prostate Cancer: The Risk of Death. Eur Urol (2015), http://dx.doi.org/10.1016/j.eururo.2015.08.054