Advancing age within established Gleason score categories and the risk of prostate cancer-specific mortality (PCSM)

Authors


Andrea Russo, Brigham and Women's Hospital, Department of Radiation Oncology, 75 Francis St, ASB1-L2, Boston, MA 02215, USA. e-mail: alrusso@partners.org

Abstract

Study Type – Prognosis (case series)

Level of Evidence 4

What's known on the subject? and What does the study add?

There is limited data that suggests that men aged >70 years have a higher proportion of Gleason 8–10 prostate cancer than men aged <70 years, as well as a higher risk of PSA recurrence, distant metastases, and disease-specific death on univariate analysis.

The present study shows that older as compared with younger men with Gleason score 6 and 7 prostate cancer have an increased risk of prostate cancer-specific mortality. This may be due to the presence of occult high-grade disease and suggests further diagnostic studies, e.g. multiparametric MRI, may be indicated in these men to reduce biopsy sampling error.

OBJECTIVE

  • • To determine if advancing age is a risk factor for high-grade prostate cancer due to occult high-grade disease in elderly men with Gleason score 6 or 7 prostate cancer. We investigated whether advancing age is associated with the risk of prostate cancer-specific mortality (PCSM) within established Gleason score categories adjusting for known predictors of PCSM.

PATIENTS AND METHODS

  • • Using data from the Surveillance, Epidemiology and End Results database between 1 January 2004 to 31 December 2007, 166 104 men with non-metastatic prostate cancer were identified and formed the study cohort.
  • • Within established Gleason score categories, Fine and Gray's multivariable competing risk regressions were used to evaluate whether increasing age at diagnosis was significantly associated with an increased risk of PCSM, adjusting for prostate-specific antigen level and T-category at diagnosis and whether treatment was curative or non-curative.

RESULTS

  • • After adjusting for treatment and prognostic factors, Gleason score 8–10 and 7 as compared with ≤6 was associated with an increased risk of PCSM (P < 0.001).
  • • Increasing age was associated with an increased risk of PCSM only in Gleason score 6 (adjusted hazard ratio [AHR] 1.06, 95% confidence interval [CI] 1.04–1.08, P < 0.001) and 7 (AHR 1.02, 95% CI 1.01–1.03, P < 0.001), but not with Gleason score 8–10 (AHR 0.999, 95% CI 0.995–1.003, P= 0.61).
  • • These risks were highest in men aged >70 years having Gleason score 6 (AHR 1.10, 95% CI 1.07–1.13, P < 0.001) and Gleason score 7 prostate cancer (AHR 1.04, 95% CI 1.02–1.06, P < 0.001).

CONCLUSIONS

  • • PCSM increases with advancing age in men with Gleason score 6 and 7 but not 8–10 prostate cancer.
  • • Techniques to reduce biopsy sampling error in men, particularly those aged >70 years and healthy with Gleason score 6 and 7 disease deserve further study.
Abbreviations
(A)HR

(adjusted) hazard ratio

AJCC

American Joint Committee on Cancer

AS

active surveillance

NIH

National Institutes of Health

PCA3

prostate cancer antigen 3

RP

radical prostatectomy

SEER

Surveillance, Epidemiology and End Results (database).

INTRODUCTION

Currently in the USA, 58% of men with prostate cancer are aged >75 years at diagnosis, and nearly half of those men have low-risk disease at diagnosis [1,2]. Recently, the National Institutes of Health (NIH) adopted active surveillance (AS) as a reasonable and recommended treatment approach for older men with low-risk prostate cancer [3]. This precedent was based on over diagnosis of prostate cancer in the PSA era and treatment-related toxicities, e.g. sexual, urinary and bowel dysfunction, which can occur after treatment in men with low-risk disease.

Of importance, however, the Prostate Cancer Prevention Trial (PCPT) showed that each 1-year increment in age is associated with a 3% increase in the occurrence of Gleason score 8–10 prostate cancer after adjusting for known predictors of high-grade prostate cancer [4]. It is also known that men of advancing age have increasing gland size due to BPH, which is highly prevalent, in up to 70% of men aged >70 years [5]. Furthermore, in men with a normal prostate gland volume of 25–30 mL, a standard 12-core biopsy samples a very small proportion (0.04%) of the prostate gland, and only a portion of the peripheral zone where 75% of cancers arise [6,7]. Given these facts, there is a significant risk of missing occult high-grade prostate cancer in this population of older men due to sampling error associated with prostate biopsy.

Therefore, we investigated whether advancing age was associated with the risk of prostate cancer-specific mortality (PCSM) within established Gleason score categories adjusting for known predictors of PCSM.

PATIENTS AND METHODS

Data were obtained from the Surveillance, Epidemiology and End Results (SEER) database, which includes patients with cancer reported by 17 tumour registries since 2000 [8]. SEER captures ≈97% of incident cancers and the 17 tumour registries cover ≈26% of the USA population [9]. Information collected includes demographics, date of diagnosis, tumour characteristics, surgical treatment, radiation therapy, vital signs, follow-up and cause of death. From 2004 to 2007, pre-treatment PSA levels (ng/mL) were also collected and defined the study period for the current analysis.

Specifically, between 1 January 2004 to 31 December 2007, 166 104 men with non-metastatic prostate cancer were diagnosed, when PSA data were available and formed the study cohort. All patients had prostate cancer as their first cancer diagnosis. Gleason score was obtained from the biopsy report if the patient did not undergo radical prostatectomy (RP), and was obtained from the surgical pathology report if the patient had had a RP.

Initial management approaches were classified as either curative or non-curative. Curative therapies included those considered to represent standard initial treatment options, e.g. RP, with or without the use of radiation therapy, and radiation therapy (either external beam or brachytherapy), with or without the use of hormonal therapy. Non-curative initial management approaches included coding from SEER that would encompass AS, watchful waiting, or a diagnosis at autopsy, TURP, local tumour excision, subtotal or segmental prostatectomy, hyperthermia as monotherapy, laser ablation, and cryosurgery. Cryosurgery was deemed non-curative given that it is not considered to be a standard initial therapy option by the National Comprehensive Cancer Network [10].

DETERMINATION OF THE CAUSE OF DEATH

Demographic, clinical and pathological factors, along with patient follow-up and determination of cause of death were extracted from the database. A standard decision algorithm that uses International Classification of Diseases was used to process causes of death from death certificates. This method of determining cause of death has been validated by means of chart review with up to a 97% concordance rate [11] and by comparison of death certificate to cancer registry with an 83% confirmation rate [12]. This study was determined to be exempt from review by the Institutional Review Board of the Dana-Farber Cancer Institute/Brigham and Women's Hospital.

STATISTICAL METHODS

Clinical description of the study cohort stratified by Gleason score and age at diagnosis

Descriptive statistics were used to characterise the study population at diagnosis, stratified by Gleason score and age (Table 1). For categorical covariates including American Joint Committee on Cancer (AJCC) 6 clinical Tumour (T)-category and curative vs non-curative treatment, a comparison of the distribution of these factors stratified by age decade for each Gleason score (6, 7, or 8–10) were performed using the chi-square test [13]. For the continuous variables of age in years, PSA level in ng/mL and year of diagnosis, a Wilcoxon rank-sum test was used to compare the medians and their distributions [14].

Table 1.  Distribution of the clinical characteristics for the 166 104 men in the study cohort stratified by Gleason score and age at diagnosis
Gleason score 6 (n= 78 385)Age, years P
≤60 (n= 24 377)61 to ≤70 (n= 30 628)>70 (n= 23 380)
  1. IQR, interquartile range; AJCC 6 category, AJCC 6 tumour (T) category.

Median (IQR) PSA level, ng/mL5.1 (4.0, 7.0)5.8 (4.5, 8.1)7.0 (4.9, 10.4)<0.001
Treatment, n (%):
Curative20 411 (84)23 890 (78)13 164 (56)<0.001
Non-curative3 966 (16)6 738 (12)10 216 (44) 
AJCC 6 category, n (%):
T1c7 977 (33)14 586 (48)13 783 (59) 
T215 644 (64)15 292 (50)9 342 (40)<0.001
T3/T4756 (3)750 (2)255 (1) 
Gleason score 7 (n= 62 545)Age (years) P
≤60 (n= 17 636)61 to ≤70 (n= 24 188)>70 (n= 20 721)
Median (IQR) PSA level, ng/mL5.9 (4.5, 9.0)6.4 (4.8, 9.8)8.2 (5.6, 14)<0.001
Treatment, n (%):
Curative16 266 (92)21 289 (88)13 351 (64)<0.001
Non-curative1 370 (8)2 899 (12)7 370 (36) 
AJCC 6 category, n (%):
T1c2 727 (16)6 331 (26%)9 341 (45) 
T211 836 (67)14 209 (59%)10 125 (49)<0.001
T3/T43 073 (17)3 648 (15%)1 255 (6) 
Gleason score 8–10 (n= 25 174)Age (years) P
≤60 (n= 4593)61 to ≤70 (n= 8064)>70 (n= 12 517)
Median (IQR) PSA level, ng/mL9.9 (5.7, 26.1)9.4 (5.7, 22.4)13.0 (7.1, 32.1)<0.001
Treatment, n (%):
Curative3 657 (80)6 170 (77)6 430 (51%)<0.001
Non-curative936 (20)1 894 (23)6 087 (49%) 
AJCC 6 category, n (%):
T1c853 (18)2 103 (26)4 704 (37) 
T22 008 (44)3 646 (45)5 962 (48)<0.001
T3/T41 732 (38)2 315 (29)1 851 (15) 

Competing risks regression for PCSM

Fine and Gray's multivariable competing risk regressions [15] and Cox regression multivariable analysis [16] were used to evaluate whether increasing age at diagnosis was significantly associated with an increased risk of PCSM and all-cause mortality, respectively, adjusting for PSA level and AJCC 6 T-category at diagnosis and whether treatment was curative or non-curative. This regression was performed for men with Gleason score categories of 6, 7, and 8–10 separately, and again for men aged ≤60, 61–70, >70 years within each Gleason score category.

Age at diagnosis and PSA level were considered as continuous variables and the PSA level was log transformed to ensure it followed a normal distribution. Curative vs non-curative treatment and clinical AJCC 6 T-category were defined as categorical variables, with curative therapy and T1c as the baseline variables, respectively. Unadjusted and adjusted hazard ratios (AHRs) were calculated and reported with associated 95% CIs and a two-sided P < 0.05 was considered to indicate statistical significance.

Estimates of PCSM

PCSM was estimated using a cumulative incidence metric and these estimates were compared using a Gray's k-mean test [17]. These estimates were performed after diagnosis for men with Gleason score 6, 7, and 8–10 prostate cancer, stratified by age ≤60, 61–70, and >70 years. A significant P-value using a Bonferroni correction, given the nine comparison groups (Gleason score 6, 7, and 8–10, within the three age categories of ≤60, 61–70, and >70 years) was considered to be <0.006. Three-year point estimates and 95% CIs were calculated and reported. R version 2.12.0 software (R Foundation for Statistical Computing, Vienna, Austria) was used for all calculations pertaining to Gray's k-mean P-value and Fine and Gray's regression. SAS version 9.3 software (SAS Institute, Cary, NC) was used for all remaining calculations.

RESULTS

CLINICAL DESCRIPTION OF THE STUDY COHORT STRATIFIED BY GLEASON SCORE AND AGE AT DIAGNOSIS

As shown in Table 1, of the 166 104 men in the study cohort, 78 385 (47%), 62 545 (38%) and 25 174 (15%) had Gleason score 6, 7, and 8–10 prostate cancer, respectively. For age ≤60, 61–70, and >70 years at diagnosis, there were 46 606 (28%), 62 880 (38%), and 56 618 (34%) men, respectively.

As men advanced in age there was a significant increase (P < 0.001) in the median PSA level and use of non-curative treatment (P < 0.001), as well as a significant decrease (P < 0.001) in palpable or locally advanced disease as per DRE. These differences remained significant across all subgroups defined by Gleason score 6, 7, and 8–10.

COMPETING RISKS REGRESSION FOR PCSM

After a median (interquartile range) follow-up of 2.75 (1.83–3.92) years, there were a total of 12 453 deaths of which 3809 (31%) were from prostate cancer. Within Gleason score groups 6, 7, and 8–10, 334 (9.4%), 834 (21%), and 2641 (55%) men died from prostate cancer, respectively. As shown in Table 2, for all men Gleason score 8–10 and 7 as compared with ≤6 was associated with an increased risk of PCSM (P < 0.001; AHR 9.4, 95% CI 8.2–10.7 and AHR 2.6, 95% CI 2.2–2.9, respectively) after adjusting for age, PSA level, T-category and whether treatment was curative or non-curative. For men with Gleason score 6 or 7, increasing age was significantly associated with an increased risk of PCSM (Gleason 6: AHR 1.06, 95% CI 1.04–1.08, P < 0.001; Gleason 7: AHR 1.02, 95% CI 1.01–1.03, P < 0.001) after adjusting for PSA level at diagnosis, AJCC 6 T category per DRE, and whether treatment intent was curative or non-curative. However, increasing age was not associated with increased risk of PCSM in men with Gleason score 8–10 prostate cancer (AHR 0.999, 95% CI 0.995–1.003, P= 0.61). In addition, within all Gleason score groups, increasing PSA level, use of non-curative as compared with curative treatment and locally advanced as compared with clinically localised prostate cancer per DRE were all significantly associated with increased risk of PCSM (P < 0.001).

Table 2.  Unadjusted HRs and AHRs for PCSM from the Fine and Gray's competing risks regression for the 166 104 men with Gleason score 6, 7 and 8–10, respectively*
Clinical characteristicsNumber of menNumber of eventsUnivariable analysisMultivariable analysis
HR (95% CI) P AHR (95% CI) P
  • *

    For all men with Gleason score 8–10 and 7 as compared with ≤6 (as reference with AHR = 1.0) was associated with an increased risk of PCSM (P < 0.001, AHR 9.4, 95% CI 8.2–10.7 and AHR 2.6, 95% CI 2.2–2.9), respectively, after adjusting for age, PSA level, T-category and whether treatment was curative or non-curative. T category, tumour category; ref, reference.

Gleason score 678 385334    
 Age at diagnosis (years)78 3853341.10 (1.09–1.12)<0.0011.06 (1.04–1.08)<0.001
 PSA (ng/mL)78 3853342.88 (2.50–3.32)<0.0012.24 (1.91–2.62)<0.001
  Treatment:      
  Curative57 4651211.0 (ref)1.0 (ref)
  Non-curative20 9202135.01 (4.01–6.26)<0.0013.03 (2.32–3.96)<0.001
 AJCC T category:      
  T1c36 3461461.0 (ref)1.0 (ref)
  T240 2781640.98 (0.79–1.23)0.881.46 (1.16–1.82)0.001
  T3/T41 761243.19 (2.07–4.92)<0.0015.66 (3.65–8.78)<0.001
Gleason score 762 545834    
 Age at diagnosis (years)62 5458341.08 (1.07–1.09)<0.0011.02 (1.01–1.03)<0.001
 PSA (ng/mL)62 5458343.69 (3.43–3.98)<0.0012.73 (2.51–2.97)<0.001
 Treatment:      
  Curative50 9062541.0 (ref)1.0 (ref)
  Non-curative11 63958010.13 (8.74–11.73)<0.0014.55 (3.76–5.50)<0.001
 AJCC T category:      
  T1c18 3993441.0 (ref)1.0 (ref)
  T236 1703630.52 (0.45–0.61)<0.0010.87 (0.75–1.01)0.07
  T3/T47 9761270.84 (0.68–1.03)0.091.55 (1.25–1.93)<0.001
Gleason score 8–1025 1742641    
 Age at diagnosis (years)25 17426411.02 (1.02–1.03)<0.0010.999 (0.995–1.003)0.61
 PSA (ng/mL)25 17426412.26 (2.16–2.36)<0.0011.80 (1.71–1.89)<0.001
 Treatment:      
  Curative16 2577671.0 (ref)1.0 (ref)
  Non-curative8 91718744.91 (4.52–5.34)<0.0013.11 (2.78–3.48)<0.001
 AJCC T category      
  T1c7 6608511.0 (ref)1.0 (ref)
  T211 61610570.78 (0.71–0.86)<0.0010.94 (0.86–1.03)0.18
  T3/T45 8987331.11 (1.00–1.22)0.041.47 (1.32–1.63)<0.001

Table 3 enumerates the AHRs for PCSM and associated P-values as a function of age group and Gleason score. Notably, the highest risk of PCSM was in men aged >70 years. Specifically, men aged >70 years with Gleason score 6 and 7 had an increased risk of PCSM (AHR 1.10, 95% CI 1.07–1.13, P < 0.001 for Gleason score 6 and AHR 1.04, 95% CI 1.02–1.06, P < 0.001 for Gleason score 7). This increased risk in PCSM extended to include men aged 61–70 years if they had Gleason score 6 prostate cancer (AHR 1.11, 95% CI 1.02–1.20, P= 0.01). For men aged >70 years with Gleason score 8–10, there was not a significant association (P= 0.08) with advancing age and risk of PCSM. However, men who were aged ≤60 years had a higher risk of PCSM with decreasing age (AHR 0.97, 95% CI 0.95–0.99, P= 0.004), which may reflect better overall health and a concomitant decrease in competing risks with younger age.

Table 3.  AHRs and associated P-values for PCSM with increasing age stratified by Gleason score and age subgroups among the 166 104 men in the study cohort
Gleason score:AHR (P-value)
All ages≤60 years61–70 years>70 years
6 1.06 (<0.001) 1.04 (0.21) 1.11 (0.01) 1.10 (<0.001)
7 1.02 (<0.001) 0.99 (0.72)1.02 (0.38) 1.04 (<0.001)
8–101.00 (0.61) 0.97 (0.004) 0.98 (0.08)1.01 (0.08)

The association of increasing age with an increased risk of PCSM after adjusting for prostate cancer prognostic factors was also noted for the endpoint of all-cause mortality using Cox regression multivariable analysis for men with Gleason score ≤6 (P < 0.001) and 7 prostate cancer (P < 0.001) across all age subgroups. For men with Gleason score 8–10, increasing age was significantly associated with an increased risk of death (P < 0.001) only in the subgroup of men aged >70 years. This can be explained by a significantly higher proportion of Gleason score 9 or 10 as compared with 8 prostate cancer (P < 0.001) in men with a median age of >78 years as compared with those with a median age of ≤78, being 48.2% vs 43.4%, respectively.

ESTIMATES OF PCSM

As shown in Fig. 1A,B, men with Gleason score 6 and 7 prostate cancer, respectively, and aged >70 years were significantly (P < 0.001) more likely to experience PCSM than men aged ≤60 years or men aged 61–70 years. Figure 1C shows the lack of significance of advancing age on the risk of PCSM for men with Gleason score 8–10 similar to that shown in Table 2 (AHR 0.999, 95% CI 0.995–1.003, P= 0.61). This lack of significance is explained by the fact that the highest estimates of PCSM were in men aged >70 years, followed by men aged ≤60 years, with men aged 61–70 years having the lowest risk of PCSM. Importantly while these differences by age subgroup in estimates of PCSM were not in ascending order by age they were all significantly different from one another as shown in Fig. 1C.

Figure 1.

Cumulative incidence estimates of PCSM for men stratified by age and Gleason score. Pairwise P-values are all <0.001 except for the comparison of men aged ≤60 vs those aged 61–70 years being 0.06, 0.03 and 0.003 for men with Gleason score 6, 7, and 8–10, respectively. A significant P-value after adjustment for multiple comparisons (Bonferroni correction) is <0.006.

DISCUSSION

In the present study, we found using the SEER population database registry that advancing age was associated with an increased risk of PCSM in men with Gleason score 6 and 7 disease, but not Gleason score 8–10 prostate cancer after adjusting for PSA level, clinical T-category and whether the initial treatment approach was curative or non-curative. One hypothesis for this finding may be explained by an increasing occurrence of occult high-grade disease in men as they age that is missed on initial biopsy. The clinical significance of this finding is that for healthy men of advancing age with Gleason score 6 or 7 prostate cancer, where AS is considered an option [3] additional approaches to reducing biopsy sampling error, e.g. multiparametric MRI or image-guided biopsy [18], are warranted. Recent data suggests men aged >70 years who meet AS criteria are found to have unfavourable disease at RP in 40% of cases [19]. Given that the community-based data from the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE) data registry shows in the USA a median of 10 biopsy cores were obtained during the time period that the present study was conducted [20], simply using extended needle core biopsies in these men may not adequately address the under-grading issue. Therefore, these data raise the issue of the need for improved grade ascertainment through the use of perhaps more sophisticated imaged-guided technologies [21] and possibly further testing using biomarkers, e.g. prostate cancer antigen 3 (PCA3) and TMPRSS2-ERG [22], before placing elderly men on AS, which is typically done after a 12-core needle biopsy.

There are several points that require further discussion. First, AS is often recommended and now supported by a recent statement issued by the NIH [3]; however, as men age BPH increases in prevalence, raising the question as to whether the routine 12-core needle biopsy is sufficient to exclude occult high-grade prostate cancer before putting elderly men, particularly those in good health and with no or minimal competing risk and with ‘low-risk’ disease on AS. The present study suggests these men may harbour occult high-grade disease, and perhaps in this particular group of men we should consider further evaluation, e.g. using multiparametric MRI or image-guided biopsy. Specifically, recent data has shown that findings on 3 T multiparametic MRI are significantly associated with the results provided by clinical risk stratification based on the PSA level, Gleason score and clinical T-category [23], suggesting that the use of this technology may be able to identify those that may be at higher risk of progression compared with what can be ascertained using a standard 12-core biopsy. Furthermore, MRI-targeted biopsy has been shown to detect significantly more high-grade cancers than an extended systematic biopsy [24], and can increase the detection of clinically significant prostate cancer in men with elevated PSA levels and negative TRUS-guided biopsy [18]. In particular, MRI-guided biopsy can detect 38–59% [25] of prostate cancer compared with 22–29% with a systematic core biopsy [21]. Furthermore, 3 T MRI can also be used to determine prostate cancer aggressiveness with the use of apparent diffusion coefficient [21]. One study reported a discriminatory function of 0.90 using receiver-operating characteristic analysis in discerning low- from intermediate- and high-grade lesions [26]. Recent data suggests both PCA3 and TMRPSS2-ERG biomarkers are significantly associated with the presence of higher grade disease (P= 0.02 and P= 0.001) and higher volume disease (P= 0.003 and P < 0.001), which can be useful to further risk stratify men on AS [27]. Second, the median follow-up in the present study was 2.5 years, which while short was still powered based on the number of PCSM events to provide significant associations between clinical factors such as age and the risk of increased PCSM. Therefore, while absolute differences in PCSM estimates as shown in the plots are small, the relative effect over 20 years (e.g. age 50–70 years) are large. Specifically, the respective AHR for PCSM for men with Gleason score 6 and 7 is 1.06 and 1.02 or a 6% and 1.8% increase in risk per year of age. This translates into a 120% and 36% increase in the risk of PCSM when comparing men aged 70 years to men aged 50 years with Gleason score 6 and 7 prostate cancer respectively. Finally, there are several factors that were not available in the SEER database for analysis that are associated with the risk of PCSM including pre-treatment PSA velocity [28], presence of tertiary Gleason 5 disease [29], percentage of positive biopsies [30], and in some series the presence of perineural invasion [31]. Therefore, further studies are indicated that include these factors in addition to those in the present study to have an even more robust prediction for the risk of PCSM.

In conclusion, despite these potential limitations discussed, the risk of PCSM increases with advancing age in men with Gleason score 6 and 7, but not 8–10, prostate cancer. This may be consistent with the observation that the risk of occult high-grade disease increases as men age. Additional information using information from tests, such as multiparametric MRI, PC3 and TMPRSS2-ERG, to reduce biopsy sampling error in men, particularly those aged >70 years and healthy with Gleason score 6 and 7 disease deserves further study.

ACKNOWLEDGEMENTS

Dr. Leon Sun for his contribution to the paper.

CONFLICT OF INTEREST

None declared. Source of funding: Statistical support was through a generous gift from Dr. David Marrus.

Ancillary