- Email: [email protected]
Age at diagnosis and prostate cancer treatment and prognosis: a population-based cohort study
Annals of Oncology 0: 1–9, 2017 doi:10.1093/annonc/mdx742 Published online 17 November 2017
ORIGINAL ARTICLE
Age at diagnosis and prostate cancer treatment and prognosis: a population-based cohort study A. Pettersson1*, D. Robinson2, H. Garmo3,4, L. Holmberg4,5 & P. Stattin5,6 1 Clinical Epidemiology Unit, Department of Medicine Solna, Karolinska Institutet, Stockholm; 2Department of Urology, Ryhov Hospital, Jo¨nko¨ping; 3Regional Cancer Centre Uppsala O¨rebro, Uppsala University Hospital, Uppsala, Sweden; 4Division of Cancer Studies, Faculty of Life Sciences and Medicine, King’s College London, London, UK; 5Department of Surgical Sciences, Uppsala University, Uppsala; 6Department of Surgical and Perioperative Sciences, Urology and Andrology, Umea˚ University Hospital, Umea˚, Sweden
*Correspondence to: Prof. Andreas Pettersson, Department of Medicine Solna, Karolinska Institutet, Clinical Epidemiology Unit T2, Karolinska University Hospital, SE-171 76 Stockholm, Sweden. Tel: þ46-709-111630; Fax: þ46-8-517-793-04; E-mail: [email protected]
Background: Old age at prostate cancer diagnosis has been associated with poor prognosis in several studies. We aimed to investigate the association between age at diagnosis and prognosis, and if it is independent of tumor characteristics, primary treatment, year of diagnosis, mode of detection and comorbidity. Patients and methods: We conducted a nation-wide cohort study including 121 392 Swedish men aged 55–95 years in Prostate Cancer data Base Sweden 3.0 diagnosed with prostate cancer in 1998–2012 and followed for prostate cancer death through 2014. Data were available on age, stage, grade, prostate-specific antigen (PSA)-level, mode of detection, comorbidity, educational level and primary treatment. We used Cox regression to calculate hazard ratios (HRs) and 95% confidence intervals (CIs). Results: With increasing age at diagnosis, men had more comorbidity, fewer PSA-detected cancers, more advanced cancers and were less often treated with curative intent. Among men with high-risk or regionally metastatic disease, the proportion of men with unknown M stage was higher among old men versus young men. During a follow-up of 751 000 person-years, 23 649 men died of prostate cancer. In multivariable Cox-regression analyses stratified by treatment, old age at diagnosis was associated with poorer prognosis among men treated with deferred treatment (HRage 85þ versus 60–64: 7.19; 95% CI: 5.61– 9.20), androgen deprivation therapy (HRage 85þ versus 60–64: 1.72; 95% CI: 1.61–1.84) or radical prostatectomy (HRage 75þ versus 60–64: 2.20; 95% CI: 1.01–4.77), but not radiotherapy (HRage 75þ versus 60–64: 1.08; 95% CI: 0.76–1.53). Conclusion: Our findings argue against a strong inherent effect of age on risk of prostate cancer death, but indicate that in current clinical practice, old men with prostate cancer receive insufficient diagnostic workup and subsequent curative treatment. Key words: cohort study, prostate cancer, age at diagnosis, treatment, prognosis
Introduction Management of old men with prostate cancer is a growing clinical dilemma. Prostate cancer primarily affects old men, life expectancy is rapidly improving and old patients are often excluded from clinical trials [1, 2]. Indeed, despite that almost a quarter of men diagnosed with prostate cancer are 75 years or older [3], none of the major clinical randomized trials comparing radical to conservative treatment included men older than 75 years [4–6]. Old age at prostate cancer diagnosis has been associated with poor prognosis in several observational studies [7, 8].
The association is likely largely explained by less prostate-specific antigen (PSA) screening, worse cancer characteristics at diagnosis and less treatment with curative intent in old compared with young men [8–10]. The association between old age at diagnosis and poor prognosis is however difficult to assess in observational studies as most population-based studies lack detailed data on clinicopathologic factors and comorbidity, whereas most institutional based studies include select patient populations. It is moreover possible that prostate cancer in old men is inherently different from that in younger men. Old age at diagnosis has been
C The Author 2017. Published by Oxford University Press on behalf of the European Society for Medical Oncology. V
All rights reserved. For Permissions, please email: [email protected].
Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article associated with increased risk of upstaging and upgrading at radical prostatectomy [11–13], indicating that prostate cancer in old men is more aggressive. The aim of this study was to investigate the association between age at diagnosis and prostate cancer death, and to assess if it is independent of tumor characteristics, primary treatment, year of diagnosis, mode of detection and comorbidity, in a populationbased cohort with detailed data on clinicopathologic factors and comorbidity. To this end, we conducted a nation-wide cohort study including 121 392 Swedish men aged 55–95 years in Prostate Cancer data Base Sweden (PCBaSe) 3.0 diagnosed with prostate cancer in 1998–2012 and followed for prostate cancer death through 2014.
Patients and methods Data sources We used The National Prostate Cancer Register (NPCR) to assemble the study cohort [14]. NPCR includes 98% of all men diagnosed with prostate cancer in Sweden since 1998. Data on mode of detection (PSAscreening, lower urinary tract symptoms, other), clinical TNM stage, biopsy tumor differentiation (Gleason score or WHO grade), serum PSA level (ng/ml) at diagnosis and planned primary treatment within 6 months of diagnosis [radical prostatectomy, radiotherapy, deferred treatment (i.e. watchful waiting or active surveillance), primary androgen depravation therapy] have been reported to NPCR since 1998. Detailed radiotherapy data (total dose, fractionation, etc.) have been registered in NPCR since 2008. For the period 1998–2007, detailed radiotherapy data have been retrieved from RETRORAD, a retrospective collection of radiotherapy data from the Oncology Information Systems and local databases at the Oncology Departments throughout Sweden [14]. Data on pathological stage and grade, the total number of diagnostic biopsy cores and the number of diagnostic biopsy cores containing cancer have been registered in NPCR since 2007. Additional covariate and outcome data were obtained through linkage to other nationwide registries in PCBaSe 3.0 [14]. Data on comorbidity and surgical procedures were obtained from The National Patient Register, which includes all in-patient medical and procedural discharge diagnoses according to International Classification of Disease (ICD) codes. Migration data were obtained from the Register of Population and Population Changes, and causes of death data from the Cause of Death Registry; agreement between cause of death according to the Cause of Death Registry and medical charts is estimated to be 86% [15]. Education level data were obtained from the LISA database.
Annals of Oncology Statistical analyses We calculated the proportion of men undergoing different treatments stratified by risk category as well as by age at diagnosis. To further assess differences in baseline cancer characteristics by age at diagnosis among men undergoing curative treatment, we calculated tumor characteristics stratified by treatment (radical prostatectomy versus radiotherapy) and by age at diagnosis. To assess if pretreatment radiological workup differed among men treated with radical prostatectomy versus radiotherapy, we calculated the proportion of men with unknown M stage (i.e. Mx according to UICC 6th edition 2002) at the time of treatment by risk category and age at diagnosis. Finally, to assess if age at diagnosis was associated with upstaging or upgrading at radical prostatectomy, we calculated the proportion of men in each pathological stage and grade category stratified by preoperative stage and grade as well as by age at diagnosis. In survival analyses, men were followed from the date of diagnosis to the date of death due to prostate cancer (end point), death due to other causes, emigration or 31 December 2014, whichever occurred first. The cumulative incidence of prostate cancer death was calculated in 1-year classes and smoothed by locally weighted regression [20]. Relative risks were estimated by calculating hazard ratios (HRs) and 95% confidence intervals (CIs) using Cox proportional hazard regression. In Cox regression models, age at diagnosis was modeled as a categorical variable in 5-year age groups (age 60–64 reference category), with the exception of those aged 85–95 years who were combined due to small numbers. We ran unadjusted models and models adjusted for year of diagnosis, comorbidity, education level, clinical TNM stage, Gleason score, PSA level at diagnosis, mode of detection and treatment. For variables containing missing values (11% for mode of detection, <5% for all other covariates; Table 1), we imputed the missing values using chained equations with five imputation data sets per value [21]. To investigate if the association between age at diagnosis and prognosis differs by treatment, we ran the fully adjusted model stratified by treatment. For men treated with radical prostatectomy or radiotherapy, we combined all age groups 75 years as few men >79 years received either of those treatments (Table 1). In these stratified analyses of curative treatment, we excluded 2653 men for whom we could not validate receipt of treatment, or who had metastatic disease, defined by a positive bone imaging or PSA 100 ng/ml. Among men treated with radical prostatectomy or radiotherapy in 2007–2012, we also conducted a sensitivity analysis additionally adjusting for the proportion of biopsy cores positive for cancer. Among men treated with radical prostatectomy in 2007–2012, we also ran sensitivity analyses adjusting for pathological rather than preoperative stage and grade. Statistical analyses were carried out using R version 2.15.0. All P-values were two sided. P-values < 0.05 were considered statistically significant. The study was approved by the Ethical Review Board in Umea˚, Sweden.
Results Study cohort We restricted the study cohort to men aged 55–95 years at diagnosis [16, 17]. The men were divided into five risk categories based on a modified version of the National Comprehensive Cancer Network (NCCN) guidelines: low risk (cT1-T2, PSA < 10 and Gleason score 6/WHO grade 1), intermediate risk (cT1-T2 and PSA 10–<20 or Gleason score 7/WHO grade 2), high risk (cT3 or PSA 20–<50 or Gleason score 8/WHO grade 3), regionally metastatic (cT4 or N1 or PSA 50–<100) and metastatic (M1 or PSA 100). Men with unknown risk category were excluded [18]. Comorbidity at diagnosis was categorized according to the Charlson Comorbidity Index [19] based on ICD discharge codes up to 10 years before the prostate cancer diagnosis. Education level was categorized as low (9 school years), middle (10–12 school years) or high (13 school years).
The study cohort included 121 392 men aged 55–95 years diagnosed with prostate cancer in 1998–2012 and followed for outcome through 2014. In total 24 949 men had been treated with radical prostatectomy, 15 893 with radiotherapy, 31 887 with deferred treatment and 44 676 with primary androgen deprivation therapy. The mean age at diagnosis was 71 years (SD: 8.5 years). During a follow-up of 751 000 person-years (mean 6.2 years, SD: 3.9 years), 23 649 (19%) men died of prostate cancer and 29 368 (24%) men died of other causes. Table 1 presents baseline characteristics overall and stratified by age at diagnosis. With increasing age at diagnosis, men had more comorbidity, lower education level, fewer PSA-detected cancer, more advanced cancers and were less often treated with
2 | Pettersson et al. Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology
Table 1. Characteristics at baseline for 121 392 men in Prostate Cancer data Base Sweden (PCBaSe) diagnosed with prostate cancer between 1998 and 2012, stratified by age Age at diagnosis 55–59 years (N 5 10 597)
60–64 years (N 5 19 981)
65–69 years (N 5 24 900)
70–74 years (N 5 23 076)
75–79 years (N 5 20 351)
80–84 years (N 5 14 411)
85 years 1 (N 5 8076)
All (N 5 121 392)
N
N
N
N
N
N
N
N
Year of diagnosis 1998–2002 2338 2003–2007 4304 2008–2012 3955 Comorbidity (CCI) 0 9536 1 663 2 279 3þ 119 Education level High 3352 Middle 4580 Low 2629 Missing 36 Risk categorya Low-risk 4960 Intermediate-risk 2902 High-risk 1453 Regionally metastatic 456 Metastatic 826 Clinical T stage T1c 5936 T1a 193 T1b 80 T2 2931 T3 1193 T4 193 Missing 71 N stage N0 1856 N1 312 NX 8348 Missing 81 M stage M0 4091 M1 606 MX 5832 Missing 68 Gleason/WHO 2–6 6272 WHO ¼ 1 119 3þ4 1736 4þ3 761 Gleason ¼ 7 335 WHO ¼ 2 180 8 619 9–10 449 WHO ¼ 3 84 Missing 42 PSA (ng/ml)
%
(22.1) (40.6) (37.3) (90.0) (6.3) (2.6) (1.1)
%
%
4009 (20.1) 5706 (22.9) 7777 (38.9) 8600 (34.5) 8195 (41.0) 10 593 (42.5)
%
6768 (29.3) 7940 (34.4) 8368 (36.3)
%
%
%
%
7185 (35.3) 7091 (34.8) 6075 (29.9)
4994 (34.7) 2674 5150 (35.7) 2721 4267 (29.6) 2681
(33.1) (33.7) (33.2)
33 674 (27.7) 43 583 (35.9) 44 134 (36.4)
17 096 (85.6) 19 874 (79.8) 16 661 (72.2) 13 167 (64.7) 1694 (8.5) 2882 (11.6) 3458 (15.0) 3648 (17.9) 856 (4.3) 1441 (5.8) 1870 (8.1) 2148 (10.6) 335 (1.7) 703 (2.8) 1087 (4.7) 1388 (6.8)
8206 (56.9) 4177 2958 (20.5) 1803 1872 (13.0) 1191 1375 (9.5) 905
(51.7) (22.3) (14.7) (11.2)
88 717 (73.1) 17 106 (14.1) 9657 (8.0) 5912 (4.9)
(31.6) (43.2) (24.8) (0.3)
5542 (27.7) 8223 (41.2) 6118 (30.6) 98 (0.5)
5875 (23.6) 4302 (18.6) 3097 (15.2) 9523 (38.2) 8097 (35.1) 6481 (31.8) 9317 (37.4) 10 482 (45.4) 10 482 (51.5) 185 (0.7) 195 (0.8) 291 (1.4)
1840 (12.8) 794 4235 (29.4) 2061 8087 (56.1) 4338 249 (1.7) 883
(9.8) (25.5) (53.7) (10.9)
24 802 (20.4) 43 200 (35.6) 51 453 (42.4) 1937 (1.6)
(46.8) (27.4) (13.7) (4.3) (7.8)
8064 (40.4) 6067 (30.4) 3277 (16.4) 960 (4.8) 1613 (8.1)
8245 (33.1) 7667 (30.8) 5125 (20.6) 1440 (5.8) 2423 (9.7)
1111 2494 5349 1869 3588
(7.7) 298 (17.3) 875 (37.1) 2874 (13.0) 1308 (24.9) 2721
(3.7) (10.8) (35.6) (16.2) (33.7)
30 505 31 324 31 229 9775 18 559
5152 6687 6145 1767 3325
(22.3) (29.0) (26.6) (7.7) (14.4)
8035 (34.8) 777 (3.4) 547 (2.4) 7638 (33.1) 4955 (21.5) 834 (3.6) 290 (1.3)
2675 4632 7006 1975 4063
(13.1) (22.8) (34.4) (9.7) (20.0)
(56.0) (1.8) (0.8) (27.7) (11.3) (1.8) (0.7)
10 611 (53.1) 11 497 (46.2) 436 (2.2) 657 (2.6) 192 (1.0) 369 (1.5) 5630 (28.2) 7605 (30.5) 2575 (12.9) 3990 (16.0) 394 (2.0) 583 (2.3) 143 (0.7) 199 (0.8)
2302 (16.0) 879 563 (3.9) 231 560 (3.9) 292 4864 (33.8) 2405 4834 (33.5) 3190 970 (6.7) 848 318 (2.2) 231
(10.9) (2.9) (3.6) (29.8) (39.5) (10.5) (2.9)
43 969 (36.2) 3627 (3.0) 2734 (2.3) 38 126 (31.4) 26 520 (21.8) 4833 (4.0) 1583 (1.3)
(17.5) (2.9) (78.8) (0.8)
3372 (16.9) 3972 (16.0) 2531 (11.0) 841 (4.1) 354 (2.5) 214 516 (2.6) 638 (2.6) 477 (2.1) 247 (1.2) 132 (0.9) 74 15 947 (79.8) 20 111 (80.8) 19 887 (86.2) 19 081 (93.8) 13 808 (95.8) 7734 146 (0.7) 179 (0.7) 181 (0.8) 182 (0.9) 117 (0.8) 54
(2.6) (0.9) (95.8) (0.7)
13 140 (10.8) 2396 (2.0) 104 916 (86.4) 940 (0.8)
(38.6) (5.7) (55.0) (0.6)
7972 (39.9) 10 700 (43.0) 9121 (39.5) 6288 (30.9) 1163 (5.8) 1721 (6.9) 2188 (9.5) 2442 (12.0) 10 727 (53.7) 12 326 (49.5) 11 617 (50.3) 11 459 (56.3) 119 (0.6) 153 (0.6) 150 (0.7) 162 (0.8)
3543 (24.6) 1593 2007 (13.9) 1247 8757 (60.8) 5190 104 (0.7) 46
(19.7) (15.4) (64.3) (0.6)
43 308 (35.7) 11 374 (9.4) 65 908 (54.3) 802 (0.7)
(59.2) (1.1) (16.4) (7.2) (3.2) (1.7) (5.8) (4.2) (0.8) (0.4)
10 903 (54.6) 11 989 (48.1) 252 (1.3) 468 (1.9) 3495 (17.5) 4529 (18.2) 1714 (8.6) 2416 (9.7) 620 (3.1) 858 (3.4) 332 (1.7) 633 (2.5) 1390 (7.0) 2112 (8.5) 973 (4.9) 1453 (5.8) 191 (1.0) 290 (1.2) 111 (0.6) 152 (0.6)
3113 586 1841 1640 820 1049 2192 2054 607 509
(14.3) (3.5) (10.5) (11.2) (4.9) (9.1) (17.4) (16.8) (5.4) (7.0)
48 243 3125 19 334 12 162 5178 5310 12 696 10 492 2784 2068
8966 622 3841 2502 1025 979 2373 1972 511 285
(38.9) (2.7) (16.6) (10.8) (4.4) (4.2) (10.3) (8.5) (2.2) (1.2)
4709 (23.1) 770 (3.8) 694 (3.4) 7053 (34.7) 5783 (28.4) 1011 (5.0) 331 (1.6)
(25.1) (25.8) (25.7) (8.1) (15.3)
5845 796 3048 2226 1128 1399 2602 2236 664 407
(28.7) (3.9) (15.0) (10.9) (5.5) (6.9) (12.8) (11.0) (3.3) (2.0)
(21.6) 1155 (4.1) 282 (12.8) 844 (11.4) 903 (5.7) 392 (7.3) 738 (15.2) 1408 (14.3) 1355 (4.2) 437 (3.5) 562
(39.7) (2.6) (15.9) (10.0) (4.3) (4.4) (10.5) (8.6) (2.3) (1.7) Continued
doi:10.1093/annonc/mdx742 | 3 Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology
Table 1. Continued Age at diagnosis
0–4 4.1–10 10.1–20 20.1–50 50.1–100 100þ Missing Treatmentb Radical prostatectomy Radiotherapy Deferred treatmentc Anti-androgen GnRH agonist Missingd Mode of detection PSA screening LUTSe Other symptoms Missing
55–59 years (N 5 10 597)
60–64 years (N 5 19 981)
65–69 years (N 5 24 900)
70–74 years (N 5 23 076)
75–79 years (N 5 20 351)
80–84 years (N 5 14 411)
85 years 1 (N 5 8076)
All (N 5 121 392)
N
%
N
N
N
N
N
%
N
%
N
1560 5541 1684 886 305 576 45
(14.7) (52.3) (15.9) (8.4) (2.9) (5.4) (0.4)
2428 (12.2) 2470 (9.9) 10 161 (50.9) 11 636 (46.7) 3693 (18.5) 5041 (20.2) 1808 (9.0) 2856 (11.5) 703 (3.5) 1127 (4.5) 1110 (5.6) 1647 (6.6) 78 (0.4) 123 (0.5)
5475
(51.7)
8847 (44.3)
1783 1726 206 972 435
(16.8) (16.3) (1.9) (9.2) (4.1)
4870 2261 2536 930
(46.0) (21.3) (23.9) (8.8)
%
%
1512 8492 5302 3657 1546 2370 197
%
%
(6.6) (36.8) (23.0) (15.8) (6.7) (10.3) (0.9)
981 4903 4716 4439 1936 3060 316
(4.8) (24.1) (23.2) (21.8) (9.5) (15.0) (1.6)
554 2237 2998 3680 1851 2809 282
(3.8) (15.5) (20.8) (25.5) (12.8) (19.5) (2.0)
220 716 1283 2106 1205 2280 266
(2.7) (8.9) (15.9) (26.1) (14.9) (28.2) (3.3)
7919 (31.8)
2529 (11.0)
151
(0.7)
24
(0.2)
4
(0.0)
24 949 (20.6)
3860 (19.3) 3903 (19.5) 498 (2.5) 2158 (10.8) 715 (3.6)
5314 (21.3) 5990 (24.1) 906 (3.6) 3902 (15.7) 869 (3.5)
4032 (17.5) 7380 (32.0) 1591 (6.9) 6779 (29.4) 765 (3.3)
880 (4.3) 7062 (34.7) 1892 (9.3) 9798 (48.1) 568 (2.8)
22 (0.2) 2 4093 (28.4) 1723 1141 (7.9) 446 8746 (60.7) 5641 385 (2.7) 260
(0.0) (21.3) (5.5) (69.8) (3.2)
15 893 (13.1) 31 877 (26.3) 6680 (5.5) 37 996 (31.3) 3997 (3.3)
8515 (42.6) 5036 (25.2) 4729 (23.7) 1701 (8.5)
9295 (37.3) 6897 (27.7) 6271 (25.2) 2437 (9.8)
6215 7154 6907 2800
3531 6839 7172 2809
1674 5217 5691 1829
(7.7) (37.9) (41.5) (12.9)
34 724 36 464 36 660 13 544
(26.9) (31.0) (29.9) (12.1)
(17.4) (33.6) (35.2) (13.8)
(11.6) 624 (36.2) 3060 (39.5) 3354 (12.7) 1038
9725 43 686 24 717 19 432 8673 13 852 1307
% (8.0) (36.0) (20.4) (16.0) (7.1) (11.4) (1.1)
(28.6) (30.0) (30.2) (11.2)
a Low risk (cT1-T2, PSA <10 ng/ml, and Gleason score 6/WHO grade 1), intermediate risk (cT1-T2, and PSA 10–<20 ng/ml or Gleason score 7/WHO grade 2), high risk (cT3 or PSA 20–<50 ng/ml or Gleason score 8/WHO grade 3), regionally metastatic (cT4 or N1 or PSA 50–<100 ng/ml) and metastatic (M1 or PSA 100 ng/ml). b According to primary treatment in NPCR. c Deferred treatment includes watchful waiting and active surveillance. d Including 0.6% registered as unspecified curative treatment in NPCR. e Lower urinary tract symptoms. CCI, Charlson Comorbidity Index.
curative intent (Table 1). The proportion of men treated with curative intent started to decrease more steeply at approximately age 65, especially for men with intermediate- or high-risk disease (Figure 1). In all age groups, men treated with radiotherapy had worse tumor characteristics than men treated with radical prostatectomy (supplementary Table S1, available at Annals of Oncology online). With increasing age at diagnosis, however, cancer characteristics generally worsened more among men treated with radical prostatectomy than among men treated with radiotherapy. For example, among men treated with radical prostatectomy, the proportion of men with T3 tumors increased from 3% among those aged 55–59 years at diagnosis to 10% among those aged 75þ years. The corresponding proportions among men treated with radiotherapy were stable at 20%. The proportion of men with unknown M stage at the time of treatment was higher among men treated with radical prostatectomy than among men treated with radiotherapy across risk categories and age groups (Figure 2). Among men with low- or intermediate-risk cancer, the proportion of men with unknown M stage decreased with increasing age at diagnosis both among
men treated with radical prostatectomy and among men treated with radiotherapy. Among men with high-risk or regionally metastatic disease the proportion of men with unknown M stage was rather stable across age groups among men treated with radiotherapy, whereas it generally increased by age group among men treated with radical prostatectomy. Old age at diagnosis was associated with both higher risk of upstaging and upgrading at radical prostatectomy (supplemen tary Table S2, available at Annals of Oncology online). Figure 3 shows the 5-year cumulative incidence of prostate cancer death by age at diagnosis. The risk increased steeply after age 65; it was <10% at <65 years and rose to >30% at age 85. Table 2 presents relative risks for prostate cancer death by 5-year age groups. Among men aged 85 or more, adjusting for year of diagnosis had a weak effect on the risk estimates, whereas adjusting for cancer characteristics strongly decreased the risk; the HR was 8.59 (95% CI: 8.10–9.10) in the unadjusted model, 8.18 (95% CI: 7.72–8.67) after adjustment for year of diagnosis and 3.35 (95% CI: 3.16–3.55) after adjustment for cancer characteristics. After additional adjustment for mode of detection and treatment, the HR decreased to 2.14 (95% CI: 2.02–2.28).
4 | Pettersson et al. Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology All men
Proportion treatment
100 %
Missing data Androgen deprivation Deferred treatment Radiotherapy Radical prostatectomy 0% 55
60
65
70 75 80 Age at diagnosis
85
90
95
Low risk
Intermediate risk
Proportion treatment
100 %
0% 55
60
65
70
75
80
85
55
60
High risk
70
75
80
85
80
85
Regionally metastatic
Proportion treatment
100 %
65
0% 55
60
65 70 75 Age at diagnosis
80
85
55
60
65 70 75 Age at diagnosis
Figure 1. Proportion of men treated with radical prostatectomy, radiotherapy, deferred treatment (i.e. watchful waiting or active surveillance) or androgen deprivation therapy by age at diagnosis among 121 392 men in PCBaSe diagnosed with prostate cancer between 1998 and 2012. Proportions were calculated in 1-year classes and smoothed by locally weighted regression.
Proportion of men with unknown M status (i.e., MX) at the time of treatment by treatment, age and risk group 90 80
Proportion (%)
70 60 50 40 30 20 10
Low risk
High risk
Intermediate risk
Regionally metastatic -5 60 9 -6 65 4 -6 70 9 -7 4 75 +
55
-5 60 9 -6 65 4 -6 70 9 -7 4 75 +
55
-5 60 9 -6 65 4 -6 70 9 -7 4 75 +
55
55
-5 60 9 -6 65 4 -6 70 9 -7 4 75 +
0
Age at diagnosis Radical prostatectomy
Radiotherapy
Figure 2. The proportion of men with unknown M status in PCBaSe diagnosed with prostate cancer between 1998 and 2012.
doi:10.1093/annonc/mdx742 | 5 Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology
Cumulative incidence proportion
0.5
0.4
0.3
0.2
0.1
0 55
60
65
70
75 80 Age at diagnosis
85
90
95
Figure 3. Five-year cumulative incidence of prostate cancer death by age at diagnosis among 121 392 men in PCBaSe diagnosed with prostate cancer between 1998 and 2012.
Table 2. Univariable and multivariable hazard ratios (HRs) and 95% confidence intervals (CIs) of prostate cancer–specific death by 5-year age groups Model 1a
Age at diagnosis 55–59 60–64 65–69 70–74 75–79 80–84 85þ
Model 2b
Model 3c
Model 4d
Model 5e
Model 6f
N
Pyrs
Events HR
(95% CI)
HR
(95% CI)
HR
(95% CI)
HR
(95% CI)
HR
(95% CI)
HR
(95% CI)
10 597 19 981 24 900 23 076 20 351 14 411 8076
81 692 145 496 171 259 149 755 114 001 63 578 25 014
995 2027 3122 4617 5582 4463 2843
(0.80–0.94) Reference (1.24–1.38) (2.11–2.35) (3.40–3.76) (4.95–5.50) (8.10–9.10)
0.86 1.00 1.29 2.15 3.37 4.95 8.18
(0.80–0.93) Reference (1.22–1.37) (2.04–2.26) (3.20–3.55) (4.69–5.22) (7.72–8.67)
0.94 1.00 1.18 1.52 1.99 2.39 3.35
(0.87–1.01) Reference (1.12–1.25) (1.44–1.60) (1.89–2.09) (2.26–2.53) (3.16–3.55)
0.95 1.00 1.17 1.47 1.90 2.26 3.12
(0.88–1.03) Reference (1.10–1.23) (1.39–1.55) (1.80–2.00) (2.14–2.39) (2.94–3.32)
0.96 1.00 1.16 1.43 1.82 2.15 2.94
(0.89–1.04) Reference (1.09–1.22) (1.35–1.50) (1.73–1.92) (2.03–2.27) (2.77–3.12)
1.01 1.00 1.06 1.13 1.30 1.54 2.14
(0.94–1.10) Reference (1.01–1.13) (1.07–1.19) (1.23–1.37) (1.45–1.63) (2.02–2.28)
0.87 1.00 1.31 2.23 3.58 5.22 8.59
a
Model 1 is unadjusted. Model 2 is adjusted for year of diagnosis (continuous). c Model 3 is adjusted for the same variable as in Model 2, plus additionally for Gleason score (categorical: 2–6, 3 þ 4, 4 þ 3, 8, 9–10), PSA level at diagnosis (categorical: 0–4, 4, 10, 10–20, 20–50, 50–100, >100), clinical T stage (categorical: cT1a, cT1b, cT1c, cT2, cT3, cT4), N stage (categorical: N0, N1, NX), M stage (categorical: M0, M1, MX). d Model 4 is adjusted for the same variables as in Model 3, plus additionally for comorbidity/CCI (categorical: 0, 1, 2, 3) and education level (categorical: low, middle, high). e Model 5 is adjusted for the same variables as in Model 4, plus additionally for mode of detection (categorical: PSA screening, lower urinary tract symptoms, other). f Model 6 is adjusted for the same variables as in Model 5, plus additionally for treatment (categorical: radical prostatectomy, radiotherapy, deferred treatment, androgen deprivation therapy). b
In general, adjustments had less influence on the risk estimates among younger men. Table 3 presents relative risks from the fully adjusted model stratified by treatment. Among men treated with radical prostatectomy, the risk increased stepwise from HR 0.78 (95% CI: 0.59– 1.02) among men age 55–59 to HR 2.20 (95% CI: 1.01–4.77) among men above age 75. In contrast, there was no association between age and risk of prostate cancer death after radiotherapy, HR 1.03 (95% CI: 0.81–1.30) among men age 55–59 and HR1.08 (95% CI: 0.76–1.53) among men above 75. Among men who received deferred treatment or androgen deprivation therapy, old men had a much higher risk of prostate cancer death.
In sensitivity analyses, among men treated with radical prostatectomy, additional adjustment for number of biopsy cores positive for cancer (supplementary Table S3, available at Annals of Oncology online) or for pathological stage and grade rather than preoperative stage and grade did not materially alter the risk estimates (supple mentary Table S4, available at Annals of Oncology online).
Discussion In this large cohort study, old age at diagnosis was associated with higher risk of prostate cancer death even after adjustment for
6 | Pettersson et al. Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology
Table 3. Multivariablea hazard ratios (HRs) and 95% confidence intervals (CIs) of prostate cancer-specific death stratified by treatment Radical prostatectomyb
Radiotherapyc
Deferred treatmentd
Androgen depravation therapye
Age at diagnosis HR
(95% CI)
HR
(95% CI)
Age at diagnosis HR
(95% CI)
HR
(95% CI)
55–59 60–64 65–69 70–74 75þ
(0.59–1.02) Reference (0.87–1.34) (1.01–1.77) (1.01–4.77)
1.03 1.00 1.06 0.98 1.08
(0.81–1.30) Reference (0.89–1.26) (0.81–1.19) (0.76–1.53)
55–59 60–64 65–69 70–74 75–79 80–84 85þ
(0.49–1.17) Reference (1.20–1.97) (2.03–3.20) (2.75–4.32) (3.74–5.96) (5.61–9.20)
1.08 1.00 1.01 0.99 1.09 1.26 1.72
(0.98–1.18) Reference (0.94–1.08) (0.93–1.05) (1.03–1.16) (1.18–1.34) (1.61–1.84)
0.78 1.00 1.08 1.34 2.20
0.76 1.00 1.54 2.55 3.45 4.72 7.19
a
Adjusted for year of diagnosis (continuous), Gleason score (categorical: 2–6, 3 þ 4, 4 þ 3, 8, 9–10), PSA level at diagnosis (categorical: 0–4, 4, 10, 10–20, 20–50, 50–100, >100), clinical T stage (categorical: cT1a, cT1b, cT1c, cT2, cT3, cT4), N stage (categorical: N0, N1, NX), M stage (categorical: M0, M1, MX), comorbidity/CCI (categorical: 0, 1, 2, 3), education level (categorical: low, middle, high) and mode of detection (categorical: PSA screening, lower urinary tract symptoms, other). b Analysis based on 24 047 men and 500 events. c Analysis based on 14 776 men and 860 events. d Deferred treatment includes men treated with watchful waiting or active surveillance. Analysis based on 31 887 men and 3003 events. e Analysis based on 44 676 men and 18 492 events.
cancer characteristics, primary treatment, year of diagnosis, mode of detection and comorbidity. In analyses stratified by treatment, old age was associated with a higher risk of prostate cancer death among men treated with radical prostatectomy, deferred treatment or androgen depravation therapy, but not among men treated with radiotherapy. A main concern in this and other observational studies investigating the association between age at prostate cancer diagnosis and prognosis is selection bias and confounding by indication. Ideally, treatment selection is based on cancer characteristics, life expectancy based on chronological age, comorbidity and patient preference. Several studies have, however, reported undertreatment of old men with prostate cancer [8–10]. Likewise, in this study, we found substantial differences in baseline characteristics, treatment patterns and pretreatment workup by age at diagnosis. Taken together, biases stemming from these sources are difficult to fully control for. Hence, although we had highvalidity covariate data in this study, residual confounding remains a concern. Few contemporary studies have specifically studied age at diagnosis as a prognostic factor for men with prostate cancer. Bechis et al. found poorer prognosis with increasing age at diagnosis in 11 790 men in CaPSURE treated with radical prostatectomy, radiotherapy, watchful waiting or primary androgen deprivation therapy for localized prostate cancer [8]. Contrasting to our finding, however, the association disappeared after adjustment for treatment. There are differences between the CaPSURE study and our study that may explain the contrasting findings, including patient selection (the CaPSURE study only included localized disease and is not population based) and statistical power (our study was about 10 times larger). Importantly, the CaPSURE study did not present results stratified by treatment. In our cohort, the risk of prostate cancer death increased stepwise with increasing age at diagnosis in men treated with radical
prostatectomy. Results from prior studies are mixed and appear to differ between population-based studies and institutional series. In a SEER study including 160 787 men treated with radical prostatectomy, old age at diagnosis was associated with higher prostate cancer death rates [7]. In most [13, 22–25], though not all [11, 26], institutional series of men treated with radical prostatectomy, older age at diagnosis was not associated with poorer prognosis. A possible explanation for these divergent findings is that institutional series, in contrast to registry-based studies, typically adjust for pathological rather than preoperative tumor characteristics, thereby indirectly adjusting for upstaging and upgrading. We did not find support for this notion in our study; in sensitivity analyses, adjusting for pathological instead of preoperative stage and grade, the results remained. Another possible explanation is that older men treated with radical prostatectomy at major institutions undergo more extensive preoperative workup and/or are a select and homogenous group of men with a more favorable prognosis. We observed no association between age at diagnosis and prostate cancer death among men treated with curatively intended radiotherapy. These results argue against the notion that tumors in old men are inherently more aggressive. Results from prior studies are mixed. In a SEER study including 149 967 men treated with radiotherapy, older age at diagnosis was associated poorer prognosis among men with low- or intermediate-risk disease, but less so among men with high-risk disease [7]. Institutional radiotherapy series have reported older age at diagnosis to be associated with poorer prognosis [27], unassociated [28, 29] or better prognosis [30]. The divergent finding in our cohort between men treated with radical prostatectomy versus radiotherapy can, aside from chance, be explained by at least two different, nonmutually exclusive mechanisms: (i) older versus younger men treated with radical prostatectomy undergo less extensive preoperative workup
doi:10.1093/annonc/mdx742 | 7 Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article and/or are a select group of men with more advanced and aggressive cancer and (ii) radical prostatectomy is less effective in older versus younger men. In support of the first explanation, our data indicate that a rather large proportion of older men treated with radical prostatectomy, especially men with high-risk tumors, did not undergo appropriate preoperative workup/imaging. Moreover, our data indicate that older versus younger men treated with radical prostatectomy generally have worse cancer characteristics. The second explanation is line with the results from the SPCG4 trial [4], which found less treatment effect of radical prostatectomy in men aged 65 versus <65 years. The recently published updated results from the PIVOT trial did however not replicate this finding, arguing against the hypothesis that radical prostatectomy is less effective in older versus younger men [6]. The ProtecT trial did also not find any treatment differences by age, but was limited to men aged 50–69 years and had low power [5]. Among men undergoing deferred treatment in our cohort, old age at diagnosis was associated with substantially poorer prognosis. Speculatively, this association could be caused a high proportion of young men on active surveillance subsequently receiving curative treatment [30]. Alternatively, a larger proportion of old men with aggressive cancer receive watchful waiting compared with younger men with less aggressive cancer [30]. Other contributing factors may be that old men receive less aggressive treatment at disease progression. Among men treated with androgen deprivation therapy, we also observed higher risk of prostate cancer death among men above age 75. As for men undergoing deferred treatment, this may be explained by older men being frailer and/or receiving less aggressive secondary treatment. Strengths of this study include its population-based design, virtually complete follow-up, large sample size and well-annotated covariate data. The main limitation lies in the observational study design, as discussed above. Additional limitations include the lack of data before 2008 distinguishing active surveillance from watchful waiting, and lack of data on secondary treatment and pathological stage and grade.
Conclusion In conclusion, the possible selection mechanisms among older versus younger men in the radical prostatectomy and deferred treatment groups discussed above, combined with the nullassociation between age at diagnosis and prognosis after radiotherapy, argue against a strong inherent effect of age on risk of prostate cancer death as the explanation for our findings. Rather, our findings indicate that in current clinical practice, old men with prostate cancer receive insufficient diagnostic workup and subsequent curative treatment. Stronger adherence to guidelines on workup and treatment of prostate cancer is warranted also in old men, in particular given the strong increase in life expectancy.
Acknowledgements This project was made possible by the continuous work of the National Prostate Cancer Register of Sweden (NPCR) steering group: Pa¨r Stattin (chairman), Anders Widmark, Camilla Thellenberg, Ove Andre´n, Eva Johansson, Ann-Sofi Fransson, Magnus To¨rnblom, Stefan Carlsson, Marie Hja¨lm Eriksson,
Annals of Oncology David Robinson, Mats Ande´n, Johan Stranne, Jonas Hugosson, Ingela Franck Lissbrant, Maria Nyberg, Go¨ran Ahlgren, Rene´ Blom, Calle Walller, Per Fransson, Fredrik Sandin and Karin Hellstro¨m.
Funding The Swedish Research Council (825-2012-5047); The Swedish Cancer Society (16 0700); Uppsala County Council (no grant number applies).
Disclosure The authors have declared no conflicts of interest.
References 1. Droz JP, Aapro M, Balducci L et al. Management of prostate cancer in older patients: updated recommendations of a working group of the International Society of Geriatric Oncology. Lancet Oncol 2014; 15: e404–e414. 2. Grummet JP, Plass K, N’Dow J. Prostate cancer management in an ageing population. Eur Urol 2017; 72(4): 532–533. 3. Heinzer H, Steuber T. Prostate cancer in the elderly. Urol Oncol 2009; 27(6): 668–672. 4. Bill-Axelson A, Holmberg L, Garmo H et al. Radical prostatectomy or watchful waiting in early prostate cancer. N Engl J Med 2014; 370(10): 932–942. 5. Hamdy FC, Donovan JL, Lane JA et al. 10-Year outcomes after monitoring, surgery, or radiotherapy for localized prostate cancer. N Engl J Med 2016; 375(15): 1415–1424. 6. Wilt TJ, Jones KM, Barry MJ et al. Follow-up of prostatectomy versus observation for early prostate cancer. N Engl J Med 2017; 377(2): 132–142. 7. Abdollah F, Sun M, Thuret R et al. A competing-risks analysis of survival after alternative treatment modalities for prostate cancer patients: 1988– 2006. Eur Urol 2011; 59(1): 88–95. 8. Bechis SK, Carroll PR, Cooperberg MR. Impact of age at diagnosis on prostate cancer treatment and survival. J Clin Oncol 2011; 29(2): 235–241. 9. Bratt O, Folkvaljon Y, Hjalm Eriksson M et al. Undertreatment of men in their seventies with high-risk nonmetastatic prostate cancer. Eur Urol 2015; 68(1): 53–58. 10. Hoffman KE. Management of older men with clinically localized prostate cancer: the significance of advanced age and comorbidity. Semin Radiat Oncol 2012; 22(4): 284–294. 11. Brassell SA, Rice KR, Parker PM et al. Prostate cancer in men 70 years old or older, indolent or aggressive: clinicopathological analysis and outcomes. J Urol 2011; 185(1): 132–137. 12. Busch J, Magheli A, Leva N et al. Higher rates of upgrading and upstaging in older patients undergoing radical prostatectomy and qualifying for active surveillance. BJU Int 2014; 114(4): 517–521. 13. Richstone L, Bianco FJ, Shah HH et al. Radical prostatectomy in men aged >or¼70 years: effect of age on upgrading, upstaging, and the accuracy of a preoperative nomogram. BJU Int 2008; 101(5): 541–546. 14. Van Hemelrijck M, Garmo H, Wigertz A et al. Cohort profile update: the National Prostate Cancer Register of Sweden and Prostate Cancer data Base—a refined prostate cancer trajectory. Int J Epidemiol 2016; 45(1): 73–82. 15. Fall K, Stromberg F, Rosell J et al. Reliability of death certificates in prostate cancer patients. Scand J Urol Nephrol 2008; 42(4): 352–357. 16. Salinas CA, Tsodikov A, Ishak-Howard M, Cooney KA. Prostate cancer in young men: an important clinical entity. Nat Rev Urol 2014; 11(6): 317–323.
8 | Pettersson et al. Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology 17. Thorstenson A, Garmo H, Adolfsson J, Bratt O. Cancer specific mortality in men diagnosed with prostate cancer before age 50 years: a nationwide population based study. J Urol 2017; 197(1): 61–66. 18. Tomic K, Westerberg M, Robinson D et al. Proportion and characteristics of men with unknown risk category in the National Prostate Cancer Register of Sweden. Acta Oncol 2016; 55(12): 1461–1466. 19. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis1987; 40(5): 373–383. 20. Cleveland WS. Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 1979; 74(368): 829–836. 21. Buuren S, Flexible Imputation of Missing Data. Boca Raton, FL: CRC Press, 2012. 22. Briganti A, Spahn M, Joniau S et al. Impact of age and comorbidities on long-term survival of patients with high-risk prostate cancer treated with radical prostatectomy: a multi-institutional competing-risks analysis. Eur Urol 2013; 63(4): 693–701. 23. Sanchez-Salas R, Prapotnich D, Rozet F et al. Laparoscopic radical prostatectomy is feasible and effective in 0 fit0 senior men with localized prostate cancer. BJU Int 2010; 106(10): 1530–1536.
24. Siddiqui SA, Sengupta S, Slezak JM et al. Impact of patient age at treatment on outcome following radical retropubic prostatectomy for prostate cancer. J Urol 2006; 175(3): 952–957. 25. Sun L, Caire AA, Robertson CN et al. Men older than 70 years have higher risk prostate cancer and poorer survival in the early and late prostate specific antigen eras. J Urol 2009; 182(5): 2242–2248. 26. Freedland SJ, Presti JC Jr, Kane CJ et al. Do younger men have better biochemical outcomes after radical prostatectomy? Urology 2004; 63: 518–522. 27. D’Amico AV, Cote K, Loffredo M et al. Advanced age at diagnosis is an independent predictor of time to death from prostate carcinoma for patients undergoing external beam radiation therapy for clinically localized prostate carcinoma. Cancer 2003; 97: 56–62. 28. Aaronson D, Cowan J, Carroll P, Konety B. Association of age and response to androgen-deprivation therapy with or without radiotherapy for prostate cancer: data from CaPSURE. BJU Int 2010; 105(7): 951–955. 29. Hoffman KE, Chen MH, Moran BJ et al. Prostate cancer-specific mortality and the extent of therapy in healthy elderly men with high-risk prostate cancer. Cancer 2010; 116(11): 2590–2595. 30. Loeb S, Berglund A, Stattin P. Population based study of use and determinants of active surveillance and watchful waiting for low and intermediate risk prostate cancer. J Urol 2013; 190(5): 1742–1749.
doi:10.1093/annonc/mdx742 | 9 Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
ORIGINAL ARTICLE
Age at diagnosis and prostate cancer treatment and prognosis: a population-based cohort study A. Pettersson1*, D. Robinson2, H. Garmo3,4, L. Holmberg4,5 & P. Stattin5,6 1 Clinical Epidemiology Unit, Department of Medicine Solna, Karolinska Institutet, Stockholm; 2Department of Urology, Ryhov Hospital, Jo¨nko¨ping; 3Regional Cancer Centre Uppsala O¨rebro, Uppsala University Hospital, Uppsala, Sweden; 4Division of Cancer Studies, Faculty of Life Sciences and Medicine, King’s College London, London, UK; 5Department of Surgical Sciences, Uppsala University, Uppsala; 6Department of Surgical and Perioperative Sciences, Urology and Andrology, Umea˚ University Hospital, Umea˚, Sweden
*Correspondence to: Prof. Andreas Pettersson, Department of Medicine Solna, Karolinska Institutet, Clinical Epidemiology Unit T2, Karolinska University Hospital, SE-171 76 Stockholm, Sweden. Tel: þ46-709-111630; Fax: þ46-8-517-793-04; E-mail: [email protected]
Background: Old age at prostate cancer diagnosis has been associated with poor prognosis in several studies. We aimed to investigate the association between age at diagnosis and prognosis, and if it is independent of tumor characteristics, primary treatment, year of diagnosis, mode of detection and comorbidity. Patients and methods: We conducted a nation-wide cohort study including 121 392 Swedish men aged 55–95 years in Prostate Cancer data Base Sweden 3.0 diagnosed with prostate cancer in 1998–2012 and followed for prostate cancer death through 2014. Data were available on age, stage, grade, prostate-specific antigen (PSA)-level, mode of detection, comorbidity, educational level and primary treatment. We used Cox regression to calculate hazard ratios (HRs) and 95% confidence intervals (CIs). Results: With increasing age at diagnosis, men had more comorbidity, fewer PSA-detected cancers, more advanced cancers and were less often treated with curative intent. Among men with high-risk or regionally metastatic disease, the proportion of men with unknown M stage was higher among old men versus young men. During a follow-up of 751 000 person-years, 23 649 men died of prostate cancer. In multivariable Cox-regression analyses stratified by treatment, old age at diagnosis was associated with poorer prognosis among men treated with deferred treatment (HRage 85þ versus 60–64: 7.19; 95% CI: 5.61– 9.20), androgen deprivation therapy (HRage 85þ versus 60–64: 1.72; 95% CI: 1.61–1.84) or radical prostatectomy (HRage 75þ versus 60–64: 2.20; 95% CI: 1.01–4.77), but not radiotherapy (HRage 75þ versus 60–64: 1.08; 95% CI: 0.76–1.53). Conclusion: Our findings argue against a strong inherent effect of age on risk of prostate cancer death, but indicate that in current clinical practice, old men with prostate cancer receive insufficient diagnostic workup and subsequent curative treatment. Key words: cohort study, prostate cancer, age at diagnosis, treatment, prognosis
Introduction Management of old men with prostate cancer is a growing clinical dilemma. Prostate cancer primarily affects old men, life expectancy is rapidly improving and old patients are often excluded from clinical trials [1, 2]. Indeed, despite that almost a quarter of men diagnosed with prostate cancer are 75 years or older [3], none of the major clinical randomized trials comparing radical to conservative treatment included men older than 75 years [4–6]. Old age at prostate cancer diagnosis has been associated with poor prognosis in several observational studies [7, 8].
The association is likely largely explained by less prostate-specific antigen (PSA) screening, worse cancer characteristics at diagnosis and less treatment with curative intent in old compared with young men [8–10]. The association between old age at diagnosis and poor prognosis is however difficult to assess in observational studies as most population-based studies lack detailed data on clinicopathologic factors and comorbidity, whereas most institutional based studies include select patient populations. It is moreover possible that prostate cancer in old men is inherently different from that in younger men. Old age at diagnosis has been
C The Author 2017. Published by Oxford University Press on behalf of the European Society for Medical Oncology. V
All rights reserved. For Permissions, please email: [email protected].
Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article associated with increased risk of upstaging and upgrading at radical prostatectomy [11–13], indicating that prostate cancer in old men is more aggressive. The aim of this study was to investigate the association between age at diagnosis and prostate cancer death, and to assess if it is independent of tumor characteristics, primary treatment, year of diagnosis, mode of detection and comorbidity, in a populationbased cohort with detailed data on clinicopathologic factors and comorbidity. To this end, we conducted a nation-wide cohort study including 121 392 Swedish men aged 55–95 years in Prostate Cancer data Base Sweden (PCBaSe) 3.0 diagnosed with prostate cancer in 1998–2012 and followed for prostate cancer death through 2014.
Patients and methods Data sources We used The National Prostate Cancer Register (NPCR) to assemble the study cohort [14]. NPCR includes 98% of all men diagnosed with prostate cancer in Sweden since 1998. Data on mode of detection (PSAscreening, lower urinary tract symptoms, other), clinical TNM stage, biopsy tumor differentiation (Gleason score or WHO grade), serum PSA level (ng/ml) at diagnosis and planned primary treatment within 6 months of diagnosis [radical prostatectomy, radiotherapy, deferred treatment (i.e. watchful waiting or active surveillance), primary androgen depravation therapy] have been reported to NPCR since 1998. Detailed radiotherapy data (total dose, fractionation, etc.) have been registered in NPCR since 2008. For the period 1998–2007, detailed radiotherapy data have been retrieved from RETRORAD, a retrospective collection of radiotherapy data from the Oncology Information Systems and local databases at the Oncology Departments throughout Sweden [14]. Data on pathological stage and grade, the total number of diagnostic biopsy cores and the number of diagnostic biopsy cores containing cancer have been registered in NPCR since 2007. Additional covariate and outcome data were obtained through linkage to other nationwide registries in PCBaSe 3.0 [14]. Data on comorbidity and surgical procedures were obtained from The National Patient Register, which includes all in-patient medical and procedural discharge diagnoses according to International Classification of Disease (ICD) codes. Migration data were obtained from the Register of Population and Population Changes, and causes of death data from the Cause of Death Registry; agreement between cause of death according to the Cause of Death Registry and medical charts is estimated to be 86% [15]. Education level data were obtained from the LISA database.
Annals of Oncology Statistical analyses We calculated the proportion of men undergoing different treatments stratified by risk category as well as by age at diagnosis. To further assess differences in baseline cancer characteristics by age at diagnosis among men undergoing curative treatment, we calculated tumor characteristics stratified by treatment (radical prostatectomy versus radiotherapy) and by age at diagnosis. To assess if pretreatment radiological workup differed among men treated with radical prostatectomy versus radiotherapy, we calculated the proportion of men with unknown M stage (i.e. Mx according to UICC 6th edition 2002) at the time of treatment by risk category and age at diagnosis. Finally, to assess if age at diagnosis was associated with upstaging or upgrading at radical prostatectomy, we calculated the proportion of men in each pathological stage and grade category stratified by preoperative stage and grade as well as by age at diagnosis. In survival analyses, men were followed from the date of diagnosis to the date of death due to prostate cancer (end point), death due to other causes, emigration or 31 December 2014, whichever occurred first. The cumulative incidence of prostate cancer death was calculated in 1-year classes and smoothed by locally weighted regression [20]. Relative risks were estimated by calculating hazard ratios (HRs) and 95% confidence intervals (CIs) using Cox proportional hazard regression. In Cox regression models, age at diagnosis was modeled as a categorical variable in 5-year age groups (age 60–64 reference category), with the exception of those aged 85–95 years who were combined due to small numbers. We ran unadjusted models and models adjusted for year of diagnosis, comorbidity, education level, clinical TNM stage, Gleason score, PSA level at diagnosis, mode of detection and treatment. For variables containing missing values (11% for mode of detection, <5% for all other covariates; Table 1), we imputed the missing values using chained equations with five imputation data sets per value [21]. To investigate if the association between age at diagnosis and prognosis differs by treatment, we ran the fully adjusted model stratified by treatment. For men treated with radical prostatectomy or radiotherapy, we combined all age groups 75 years as few men >79 years received either of those treatments (Table 1). In these stratified analyses of curative treatment, we excluded 2653 men for whom we could not validate receipt of treatment, or who had metastatic disease, defined by a positive bone imaging or PSA 100 ng/ml. Among men treated with radical prostatectomy or radiotherapy in 2007–2012, we also conducted a sensitivity analysis additionally adjusting for the proportion of biopsy cores positive for cancer. Among men treated with radical prostatectomy in 2007–2012, we also ran sensitivity analyses adjusting for pathological rather than preoperative stage and grade. Statistical analyses were carried out using R version 2.15.0. All P-values were two sided. P-values < 0.05 were considered statistically significant. The study was approved by the Ethical Review Board in Umea˚, Sweden.
Results Study cohort We restricted the study cohort to men aged 55–95 years at diagnosis [16, 17]. The men were divided into five risk categories based on a modified version of the National Comprehensive Cancer Network (NCCN) guidelines: low risk (cT1-T2, PSA < 10 and Gleason score 6/WHO grade 1), intermediate risk (cT1-T2 and PSA 10–<20 or Gleason score 7/WHO grade 2), high risk (cT3 or PSA 20–<50 or Gleason score 8/WHO grade 3), regionally metastatic (cT4 or N1 or PSA 50–<100) and metastatic (M1 or PSA 100). Men with unknown risk category were excluded [18]. Comorbidity at diagnosis was categorized according to the Charlson Comorbidity Index [19] based on ICD discharge codes up to 10 years before the prostate cancer diagnosis. Education level was categorized as low (9 school years), middle (10–12 school years) or high (13 school years).
The study cohort included 121 392 men aged 55–95 years diagnosed with prostate cancer in 1998–2012 and followed for outcome through 2014. In total 24 949 men had been treated with radical prostatectomy, 15 893 with radiotherapy, 31 887 with deferred treatment and 44 676 with primary androgen deprivation therapy. The mean age at diagnosis was 71 years (SD: 8.5 years). During a follow-up of 751 000 person-years (mean 6.2 years, SD: 3.9 years), 23 649 (19%) men died of prostate cancer and 29 368 (24%) men died of other causes. Table 1 presents baseline characteristics overall and stratified by age at diagnosis. With increasing age at diagnosis, men had more comorbidity, lower education level, fewer PSA-detected cancer, more advanced cancers and were less often treated with
2 | Pettersson et al. Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology
Table 1. Characteristics at baseline for 121 392 men in Prostate Cancer data Base Sweden (PCBaSe) diagnosed with prostate cancer between 1998 and 2012, stratified by age Age at diagnosis 55–59 years (N 5 10 597)
60–64 years (N 5 19 981)
65–69 years (N 5 24 900)
70–74 years (N 5 23 076)
75–79 years (N 5 20 351)
80–84 years (N 5 14 411)
85 years 1 (N 5 8076)
All (N 5 121 392)
N
N
N
N
N
N
N
N
Year of diagnosis 1998–2002 2338 2003–2007 4304 2008–2012 3955 Comorbidity (CCI) 0 9536 1 663 2 279 3þ 119 Education level High 3352 Middle 4580 Low 2629 Missing 36 Risk categorya Low-risk 4960 Intermediate-risk 2902 High-risk 1453 Regionally metastatic 456 Metastatic 826 Clinical T stage T1c 5936 T1a 193 T1b 80 T2 2931 T3 1193 T4 193 Missing 71 N stage N0 1856 N1 312 NX 8348 Missing 81 M stage M0 4091 M1 606 MX 5832 Missing 68 Gleason/WHO 2–6 6272 WHO ¼ 1 119 3þ4 1736 4þ3 761 Gleason ¼ 7 335 WHO ¼ 2 180 8 619 9–10 449 WHO ¼ 3 84 Missing 42 PSA (ng/ml)
%
(22.1) (40.6) (37.3) (90.0) (6.3) (2.6) (1.1)
%
%
4009 (20.1) 5706 (22.9) 7777 (38.9) 8600 (34.5) 8195 (41.0) 10 593 (42.5)
%
6768 (29.3) 7940 (34.4) 8368 (36.3)
%
%
%
%
7185 (35.3) 7091 (34.8) 6075 (29.9)
4994 (34.7) 2674 5150 (35.7) 2721 4267 (29.6) 2681
(33.1) (33.7) (33.2)
33 674 (27.7) 43 583 (35.9) 44 134 (36.4)
17 096 (85.6) 19 874 (79.8) 16 661 (72.2) 13 167 (64.7) 1694 (8.5) 2882 (11.6) 3458 (15.0) 3648 (17.9) 856 (4.3) 1441 (5.8) 1870 (8.1) 2148 (10.6) 335 (1.7) 703 (2.8) 1087 (4.7) 1388 (6.8)
8206 (56.9) 4177 2958 (20.5) 1803 1872 (13.0) 1191 1375 (9.5) 905
(51.7) (22.3) (14.7) (11.2)
88 717 (73.1) 17 106 (14.1) 9657 (8.0) 5912 (4.9)
(31.6) (43.2) (24.8) (0.3)
5542 (27.7) 8223 (41.2) 6118 (30.6) 98 (0.5)
5875 (23.6) 4302 (18.6) 3097 (15.2) 9523 (38.2) 8097 (35.1) 6481 (31.8) 9317 (37.4) 10 482 (45.4) 10 482 (51.5) 185 (0.7) 195 (0.8) 291 (1.4)
1840 (12.8) 794 4235 (29.4) 2061 8087 (56.1) 4338 249 (1.7) 883
(9.8) (25.5) (53.7) (10.9)
24 802 (20.4) 43 200 (35.6) 51 453 (42.4) 1937 (1.6)
(46.8) (27.4) (13.7) (4.3) (7.8)
8064 (40.4) 6067 (30.4) 3277 (16.4) 960 (4.8) 1613 (8.1)
8245 (33.1) 7667 (30.8) 5125 (20.6) 1440 (5.8) 2423 (9.7)
1111 2494 5349 1869 3588
(7.7) 298 (17.3) 875 (37.1) 2874 (13.0) 1308 (24.9) 2721
(3.7) (10.8) (35.6) (16.2) (33.7)
30 505 31 324 31 229 9775 18 559
5152 6687 6145 1767 3325
(22.3) (29.0) (26.6) (7.7) (14.4)
8035 (34.8) 777 (3.4) 547 (2.4) 7638 (33.1) 4955 (21.5) 834 (3.6) 290 (1.3)
2675 4632 7006 1975 4063
(13.1) (22.8) (34.4) (9.7) (20.0)
(56.0) (1.8) (0.8) (27.7) (11.3) (1.8) (0.7)
10 611 (53.1) 11 497 (46.2) 436 (2.2) 657 (2.6) 192 (1.0) 369 (1.5) 5630 (28.2) 7605 (30.5) 2575 (12.9) 3990 (16.0) 394 (2.0) 583 (2.3) 143 (0.7) 199 (0.8)
2302 (16.0) 879 563 (3.9) 231 560 (3.9) 292 4864 (33.8) 2405 4834 (33.5) 3190 970 (6.7) 848 318 (2.2) 231
(10.9) (2.9) (3.6) (29.8) (39.5) (10.5) (2.9)
43 969 (36.2) 3627 (3.0) 2734 (2.3) 38 126 (31.4) 26 520 (21.8) 4833 (4.0) 1583 (1.3)
(17.5) (2.9) (78.8) (0.8)
3372 (16.9) 3972 (16.0) 2531 (11.0) 841 (4.1) 354 (2.5) 214 516 (2.6) 638 (2.6) 477 (2.1) 247 (1.2) 132 (0.9) 74 15 947 (79.8) 20 111 (80.8) 19 887 (86.2) 19 081 (93.8) 13 808 (95.8) 7734 146 (0.7) 179 (0.7) 181 (0.8) 182 (0.9) 117 (0.8) 54
(2.6) (0.9) (95.8) (0.7)
13 140 (10.8) 2396 (2.0) 104 916 (86.4) 940 (0.8)
(38.6) (5.7) (55.0) (0.6)
7972 (39.9) 10 700 (43.0) 9121 (39.5) 6288 (30.9) 1163 (5.8) 1721 (6.9) 2188 (9.5) 2442 (12.0) 10 727 (53.7) 12 326 (49.5) 11 617 (50.3) 11 459 (56.3) 119 (0.6) 153 (0.6) 150 (0.7) 162 (0.8)
3543 (24.6) 1593 2007 (13.9) 1247 8757 (60.8) 5190 104 (0.7) 46
(19.7) (15.4) (64.3) (0.6)
43 308 (35.7) 11 374 (9.4) 65 908 (54.3) 802 (0.7)
(59.2) (1.1) (16.4) (7.2) (3.2) (1.7) (5.8) (4.2) (0.8) (0.4)
10 903 (54.6) 11 989 (48.1) 252 (1.3) 468 (1.9) 3495 (17.5) 4529 (18.2) 1714 (8.6) 2416 (9.7) 620 (3.1) 858 (3.4) 332 (1.7) 633 (2.5) 1390 (7.0) 2112 (8.5) 973 (4.9) 1453 (5.8) 191 (1.0) 290 (1.2) 111 (0.6) 152 (0.6)
3113 586 1841 1640 820 1049 2192 2054 607 509
(14.3) (3.5) (10.5) (11.2) (4.9) (9.1) (17.4) (16.8) (5.4) (7.0)
48 243 3125 19 334 12 162 5178 5310 12 696 10 492 2784 2068
8966 622 3841 2502 1025 979 2373 1972 511 285
(38.9) (2.7) (16.6) (10.8) (4.4) (4.2) (10.3) (8.5) (2.2) (1.2)
4709 (23.1) 770 (3.8) 694 (3.4) 7053 (34.7) 5783 (28.4) 1011 (5.0) 331 (1.6)
(25.1) (25.8) (25.7) (8.1) (15.3)
5845 796 3048 2226 1128 1399 2602 2236 664 407
(28.7) (3.9) (15.0) (10.9) (5.5) (6.9) (12.8) (11.0) (3.3) (2.0)
(21.6) 1155 (4.1) 282 (12.8) 844 (11.4) 903 (5.7) 392 (7.3) 738 (15.2) 1408 (14.3) 1355 (4.2) 437 (3.5) 562
(39.7) (2.6) (15.9) (10.0) (4.3) (4.4) (10.5) (8.6) (2.3) (1.7) Continued
doi:10.1093/annonc/mdx742 | 3 Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology
Table 1. Continued Age at diagnosis
0–4 4.1–10 10.1–20 20.1–50 50.1–100 100þ Missing Treatmentb Radical prostatectomy Radiotherapy Deferred treatmentc Anti-androgen GnRH agonist Missingd Mode of detection PSA screening LUTSe Other symptoms Missing
55–59 years (N 5 10 597)
60–64 years (N 5 19 981)
65–69 years (N 5 24 900)
70–74 years (N 5 23 076)
75–79 years (N 5 20 351)
80–84 years (N 5 14 411)
85 years 1 (N 5 8076)
All (N 5 121 392)
N
%
N
N
N
N
N
%
N
%
N
1560 5541 1684 886 305 576 45
(14.7) (52.3) (15.9) (8.4) (2.9) (5.4) (0.4)
2428 (12.2) 2470 (9.9) 10 161 (50.9) 11 636 (46.7) 3693 (18.5) 5041 (20.2) 1808 (9.0) 2856 (11.5) 703 (3.5) 1127 (4.5) 1110 (5.6) 1647 (6.6) 78 (0.4) 123 (0.5)
5475
(51.7)
8847 (44.3)
1783 1726 206 972 435
(16.8) (16.3) (1.9) (9.2) (4.1)
4870 2261 2536 930
(46.0) (21.3) (23.9) (8.8)
%
%
1512 8492 5302 3657 1546 2370 197
%
%
(6.6) (36.8) (23.0) (15.8) (6.7) (10.3) (0.9)
981 4903 4716 4439 1936 3060 316
(4.8) (24.1) (23.2) (21.8) (9.5) (15.0) (1.6)
554 2237 2998 3680 1851 2809 282
(3.8) (15.5) (20.8) (25.5) (12.8) (19.5) (2.0)
220 716 1283 2106 1205 2280 266
(2.7) (8.9) (15.9) (26.1) (14.9) (28.2) (3.3)
7919 (31.8)
2529 (11.0)
151
(0.7)
24
(0.2)
4
(0.0)
24 949 (20.6)
3860 (19.3) 3903 (19.5) 498 (2.5) 2158 (10.8) 715 (3.6)
5314 (21.3) 5990 (24.1) 906 (3.6) 3902 (15.7) 869 (3.5)
4032 (17.5) 7380 (32.0) 1591 (6.9) 6779 (29.4) 765 (3.3)
880 (4.3) 7062 (34.7) 1892 (9.3) 9798 (48.1) 568 (2.8)
22 (0.2) 2 4093 (28.4) 1723 1141 (7.9) 446 8746 (60.7) 5641 385 (2.7) 260
(0.0) (21.3) (5.5) (69.8) (3.2)
15 893 (13.1) 31 877 (26.3) 6680 (5.5) 37 996 (31.3) 3997 (3.3)
8515 (42.6) 5036 (25.2) 4729 (23.7) 1701 (8.5)
9295 (37.3) 6897 (27.7) 6271 (25.2) 2437 (9.8)
6215 7154 6907 2800
3531 6839 7172 2809
1674 5217 5691 1829
(7.7) (37.9) (41.5) (12.9)
34 724 36 464 36 660 13 544
(26.9) (31.0) (29.9) (12.1)
(17.4) (33.6) (35.2) (13.8)
(11.6) 624 (36.2) 3060 (39.5) 3354 (12.7) 1038
9725 43 686 24 717 19 432 8673 13 852 1307
% (8.0) (36.0) (20.4) (16.0) (7.1) (11.4) (1.1)
(28.6) (30.0) (30.2) (11.2)
a Low risk (cT1-T2, PSA <10 ng/ml, and Gleason score 6/WHO grade 1), intermediate risk (cT1-T2, and PSA 10–<20 ng/ml or Gleason score 7/WHO grade 2), high risk (cT3 or PSA 20–<50 ng/ml or Gleason score 8/WHO grade 3), regionally metastatic (cT4 or N1 or PSA 50–<100 ng/ml) and metastatic (M1 or PSA 100 ng/ml). b According to primary treatment in NPCR. c Deferred treatment includes watchful waiting and active surveillance. d Including 0.6% registered as unspecified curative treatment in NPCR. e Lower urinary tract symptoms. CCI, Charlson Comorbidity Index.
curative intent (Table 1). The proportion of men treated with curative intent started to decrease more steeply at approximately age 65, especially for men with intermediate- or high-risk disease (Figure 1). In all age groups, men treated with radiotherapy had worse tumor characteristics than men treated with radical prostatectomy (supplementary Table S1, available at Annals of Oncology online). With increasing age at diagnosis, however, cancer characteristics generally worsened more among men treated with radical prostatectomy than among men treated with radiotherapy. For example, among men treated with radical prostatectomy, the proportion of men with T3 tumors increased from 3% among those aged 55–59 years at diagnosis to 10% among those aged 75þ years. The corresponding proportions among men treated with radiotherapy were stable at 20%. The proportion of men with unknown M stage at the time of treatment was higher among men treated with radical prostatectomy than among men treated with radiotherapy across risk categories and age groups (Figure 2). Among men with low- or intermediate-risk cancer, the proportion of men with unknown M stage decreased with increasing age at diagnosis both among
men treated with radical prostatectomy and among men treated with radiotherapy. Among men with high-risk or regionally metastatic disease the proportion of men with unknown M stage was rather stable across age groups among men treated with radiotherapy, whereas it generally increased by age group among men treated with radical prostatectomy. Old age at diagnosis was associated with both higher risk of upstaging and upgrading at radical prostatectomy (supplemen tary Table S2, available at Annals of Oncology online). Figure 3 shows the 5-year cumulative incidence of prostate cancer death by age at diagnosis. The risk increased steeply after age 65; it was <10% at <65 years and rose to >30% at age 85. Table 2 presents relative risks for prostate cancer death by 5-year age groups. Among men aged 85 or more, adjusting for year of diagnosis had a weak effect on the risk estimates, whereas adjusting for cancer characteristics strongly decreased the risk; the HR was 8.59 (95% CI: 8.10–9.10) in the unadjusted model, 8.18 (95% CI: 7.72–8.67) after adjustment for year of diagnosis and 3.35 (95% CI: 3.16–3.55) after adjustment for cancer characteristics. After additional adjustment for mode of detection and treatment, the HR decreased to 2.14 (95% CI: 2.02–2.28).
4 | Pettersson et al. Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology All men
Proportion treatment
100 %
Missing data Androgen deprivation Deferred treatment Radiotherapy Radical prostatectomy 0% 55
60
65
70 75 80 Age at diagnosis
85
90
95
Low risk
Intermediate risk
Proportion treatment
100 %
0% 55
60
65
70
75
80
85
55
60
High risk
70
75
80
85
80
85
Regionally metastatic
Proportion treatment
100 %
65
0% 55
60
65 70 75 Age at diagnosis
80
85
55
60
65 70 75 Age at diagnosis
Figure 1. Proportion of men treated with radical prostatectomy, radiotherapy, deferred treatment (i.e. watchful waiting or active surveillance) or androgen deprivation therapy by age at diagnosis among 121 392 men in PCBaSe diagnosed with prostate cancer between 1998 and 2012. Proportions were calculated in 1-year classes and smoothed by locally weighted regression.
Proportion of men with unknown M status (i.e., MX) at the time of treatment by treatment, age and risk group 90 80
Proportion (%)
70 60 50 40 30 20 10
Low risk
High risk
Intermediate risk
Regionally metastatic -5 60 9 -6 65 4 -6 70 9 -7 4 75 +
55
-5 60 9 -6 65 4 -6 70 9 -7 4 75 +
55
-5 60 9 -6 65 4 -6 70 9 -7 4 75 +
55
55
-5 60 9 -6 65 4 -6 70 9 -7 4 75 +
0
Age at diagnosis Radical prostatectomy
Radiotherapy
Figure 2. The proportion of men with unknown M status in PCBaSe diagnosed with prostate cancer between 1998 and 2012.
doi:10.1093/annonc/mdx742 | 5 Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology
Cumulative incidence proportion
0.5
0.4
0.3
0.2
0.1
0 55
60
65
70
75 80 Age at diagnosis
85
90
95
Figure 3. Five-year cumulative incidence of prostate cancer death by age at diagnosis among 121 392 men in PCBaSe diagnosed with prostate cancer between 1998 and 2012.
Table 2. Univariable and multivariable hazard ratios (HRs) and 95% confidence intervals (CIs) of prostate cancer–specific death by 5-year age groups Model 1a
Age at diagnosis 55–59 60–64 65–69 70–74 75–79 80–84 85þ
Model 2b
Model 3c
Model 4d
Model 5e
Model 6f
N
Pyrs
Events HR
(95% CI)
HR
(95% CI)
HR
(95% CI)
HR
(95% CI)
HR
(95% CI)
HR
(95% CI)
10 597 19 981 24 900 23 076 20 351 14 411 8076
81 692 145 496 171 259 149 755 114 001 63 578 25 014
995 2027 3122 4617 5582 4463 2843
(0.80–0.94) Reference (1.24–1.38) (2.11–2.35) (3.40–3.76) (4.95–5.50) (8.10–9.10)
0.86 1.00 1.29 2.15 3.37 4.95 8.18
(0.80–0.93) Reference (1.22–1.37) (2.04–2.26) (3.20–3.55) (4.69–5.22) (7.72–8.67)
0.94 1.00 1.18 1.52 1.99 2.39 3.35
(0.87–1.01) Reference (1.12–1.25) (1.44–1.60) (1.89–2.09) (2.26–2.53) (3.16–3.55)
0.95 1.00 1.17 1.47 1.90 2.26 3.12
(0.88–1.03) Reference (1.10–1.23) (1.39–1.55) (1.80–2.00) (2.14–2.39) (2.94–3.32)
0.96 1.00 1.16 1.43 1.82 2.15 2.94
(0.89–1.04) Reference (1.09–1.22) (1.35–1.50) (1.73–1.92) (2.03–2.27) (2.77–3.12)
1.01 1.00 1.06 1.13 1.30 1.54 2.14
(0.94–1.10) Reference (1.01–1.13) (1.07–1.19) (1.23–1.37) (1.45–1.63) (2.02–2.28)
0.87 1.00 1.31 2.23 3.58 5.22 8.59
a
Model 1 is unadjusted. Model 2 is adjusted for year of diagnosis (continuous). c Model 3 is adjusted for the same variable as in Model 2, plus additionally for Gleason score (categorical: 2–6, 3 þ 4, 4 þ 3, 8, 9–10), PSA level at diagnosis (categorical: 0–4, 4, 10, 10–20, 20–50, 50–100, >100), clinical T stage (categorical: cT1a, cT1b, cT1c, cT2, cT3, cT4), N stage (categorical: N0, N1, NX), M stage (categorical: M0, M1, MX). d Model 4 is adjusted for the same variables as in Model 3, plus additionally for comorbidity/CCI (categorical: 0, 1, 2, 3) and education level (categorical: low, middle, high). e Model 5 is adjusted for the same variables as in Model 4, plus additionally for mode of detection (categorical: PSA screening, lower urinary tract symptoms, other). f Model 6 is adjusted for the same variables as in Model 5, plus additionally for treatment (categorical: radical prostatectomy, radiotherapy, deferred treatment, androgen deprivation therapy). b
In general, adjustments had less influence on the risk estimates among younger men. Table 3 presents relative risks from the fully adjusted model stratified by treatment. Among men treated with radical prostatectomy, the risk increased stepwise from HR 0.78 (95% CI: 0.59– 1.02) among men age 55–59 to HR 2.20 (95% CI: 1.01–4.77) among men above age 75. In contrast, there was no association between age and risk of prostate cancer death after radiotherapy, HR 1.03 (95% CI: 0.81–1.30) among men age 55–59 and HR1.08 (95% CI: 0.76–1.53) among men above 75. Among men who received deferred treatment or androgen deprivation therapy, old men had a much higher risk of prostate cancer death.
In sensitivity analyses, among men treated with radical prostatectomy, additional adjustment for number of biopsy cores positive for cancer (supplementary Table S3, available at Annals of Oncology online) or for pathological stage and grade rather than preoperative stage and grade did not materially alter the risk estimates (supple mentary Table S4, available at Annals of Oncology online).
Discussion In this large cohort study, old age at diagnosis was associated with higher risk of prostate cancer death even after adjustment for
6 | Pettersson et al. Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology
Table 3. Multivariablea hazard ratios (HRs) and 95% confidence intervals (CIs) of prostate cancer-specific death stratified by treatment Radical prostatectomyb
Radiotherapyc
Deferred treatmentd
Androgen depravation therapye
Age at diagnosis HR
(95% CI)
HR
(95% CI)
Age at diagnosis HR
(95% CI)
HR
(95% CI)
55–59 60–64 65–69 70–74 75þ
(0.59–1.02) Reference (0.87–1.34) (1.01–1.77) (1.01–4.77)
1.03 1.00 1.06 0.98 1.08
(0.81–1.30) Reference (0.89–1.26) (0.81–1.19) (0.76–1.53)
55–59 60–64 65–69 70–74 75–79 80–84 85þ
(0.49–1.17) Reference (1.20–1.97) (2.03–3.20) (2.75–4.32) (3.74–5.96) (5.61–9.20)
1.08 1.00 1.01 0.99 1.09 1.26 1.72
(0.98–1.18) Reference (0.94–1.08) (0.93–1.05) (1.03–1.16) (1.18–1.34) (1.61–1.84)
0.78 1.00 1.08 1.34 2.20
0.76 1.00 1.54 2.55 3.45 4.72 7.19
a
Adjusted for year of diagnosis (continuous), Gleason score (categorical: 2–6, 3 þ 4, 4 þ 3, 8, 9–10), PSA level at diagnosis (categorical: 0–4, 4, 10, 10–20, 20–50, 50–100, >100), clinical T stage (categorical: cT1a, cT1b, cT1c, cT2, cT3, cT4), N stage (categorical: N0, N1, NX), M stage (categorical: M0, M1, MX), comorbidity/CCI (categorical: 0, 1, 2, 3), education level (categorical: low, middle, high) and mode of detection (categorical: PSA screening, lower urinary tract symptoms, other). b Analysis based on 24 047 men and 500 events. c Analysis based on 14 776 men and 860 events. d Deferred treatment includes men treated with watchful waiting or active surveillance. Analysis based on 31 887 men and 3003 events. e Analysis based on 44 676 men and 18 492 events.
cancer characteristics, primary treatment, year of diagnosis, mode of detection and comorbidity. In analyses stratified by treatment, old age was associated with a higher risk of prostate cancer death among men treated with radical prostatectomy, deferred treatment or androgen depravation therapy, but not among men treated with radiotherapy. A main concern in this and other observational studies investigating the association between age at prostate cancer diagnosis and prognosis is selection bias and confounding by indication. Ideally, treatment selection is based on cancer characteristics, life expectancy based on chronological age, comorbidity and patient preference. Several studies have, however, reported undertreatment of old men with prostate cancer [8–10]. Likewise, in this study, we found substantial differences in baseline characteristics, treatment patterns and pretreatment workup by age at diagnosis. Taken together, biases stemming from these sources are difficult to fully control for. Hence, although we had highvalidity covariate data in this study, residual confounding remains a concern. Few contemporary studies have specifically studied age at diagnosis as a prognostic factor for men with prostate cancer. Bechis et al. found poorer prognosis with increasing age at diagnosis in 11 790 men in CaPSURE treated with radical prostatectomy, radiotherapy, watchful waiting or primary androgen deprivation therapy for localized prostate cancer [8]. Contrasting to our finding, however, the association disappeared after adjustment for treatment. There are differences between the CaPSURE study and our study that may explain the contrasting findings, including patient selection (the CaPSURE study only included localized disease and is not population based) and statistical power (our study was about 10 times larger). Importantly, the CaPSURE study did not present results stratified by treatment. In our cohort, the risk of prostate cancer death increased stepwise with increasing age at diagnosis in men treated with radical
prostatectomy. Results from prior studies are mixed and appear to differ between population-based studies and institutional series. In a SEER study including 160 787 men treated with radical prostatectomy, old age at diagnosis was associated with higher prostate cancer death rates [7]. In most [13, 22–25], though not all [11, 26], institutional series of men treated with radical prostatectomy, older age at diagnosis was not associated with poorer prognosis. A possible explanation for these divergent findings is that institutional series, in contrast to registry-based studies, typically adjust for pathological rather than preoperative tumor characteristics, thereby indirectly adjusting for upstaging and upgrading. We did not find support for this notion in our study; in sensitivity analyses, adjusting for pathological instead of preoperative stage and grade, the results remained. Another possible explanation is that older men treated with radical prostatectomy at major institutions undergo more extensive preoperative workup and/or are a select and homogenous group of men with a more favorable prognosis. We observed no association between age at diagnosis and prostate cancer death among men treated with curatively intended radiotherapy. These results argue against the notion that tumors in old men are inherently more aggressive. Results from prior studies are mixed. In a SEER study including 149 967 men treated with radiotherapy, older age at diagnosis was associated poorer prognosis among men with low- or intermediate-risk disease, but less so among men with high-risk disease [7]. Institutional radiotherapy series have reported older age at diagnosis to be associated with poorer prognosis [27], unassociated [28, 29] or better prognosis [30]. The divergent finding in our cohort between men treated with radical prostatectomy versus radiotherapy can, aside from chance, be explained by at least two different, nonmutually exclusive mechanisms: (i) older versus younger men treated with radical prostatectomy undergo less extensive preoperative workup
doi:10.1093/annonc/mdx742 | 7 Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article and/or are a select group of men with more advanced and aggressive cancer and (ii) radical prostatectomy is less effective in older versus younger men. In support of the first explanation, our data indicate that a rather large proportion of older men treated with radical prostatectomy, especially men with high-risk tumors, did not undergo appropriate preoperative workup/imaging. Moreover, our data indicate that older versus younger men treated with radical prostatectomy generally have worse cancer characteristics. The second explanation is line with the results from the SPCG4 trial [4], which found less treatment effect of radical prostatectomy in men aged 65 versus <65 years. The recently published updated results from the PIVOT trial did however not replicate this finding, arguing against the hypothesis that radical prostatectomy is less effective in older versus younger men [6]. The ProtecT trial did also not find any treatment differences by age, but was limited to men aged 50–69 years and had low power [5]. Among men undergoing deferred treatment in our cohort, old age at diagnosis was associated with substantially poorer prognosis. Speculatively, this association could be caused a high proportion of young men on active surveillance subsequently receiving curative treatment [30]. Alternatively, a larger proportion of old men with aggressive cancer receive watchful waiting compared with younger men with less aggressive cancer [30]. Other contributing factors may be that old men receive less aggressive treatment at disease progression. Among men treated with androgen deprivation therapy, we also observed higher risk of prostate cancer death among men above age 75. As for men undergoing deferred treatment, this may be explained by older men being frailer and/or receiving less aggressive secondary treatment. Strengths of this study include its population-based design, virtually complete follow-up, large sample size and well-annotated covariate data. The main limitation lies in the observational study design, as discussed above. Additional limitations include the lack of data before 2008 distinguishing active surveillance from watchful waiting, and lack of data on secondary treatment and pathological stage and grade.
Conclusion In conclusion, the possible selection mechanisms among older versus younger men in the radical prostatectomy and deferred treatment groups discussed above, combined with the nullassociation between age at diagnosis and prognosis after radiotherapy, argue against a strong inherent effect of age on risk of prostate cancer death as the explanation for our findings. Rather, our findings indicate that in current clinical practice, old men with prostate cancer receive insufficient diagnostic workup and subsequent curative treatment. Stronger adherence to guidelines on workup and treatment of prostate cancer is warranted also in old men, in particular given the strong increase in life expectancy.
Acknowledgements This project was made possible by the continuous work of the National Prostate Cancer Register of Sweden (NPCR) steering group: Pa¨r Stattin (chairman), Anders Widmark, Camilla Thellenberg, Ove Andre´n, Eva Johansson, Ann-Sofi Fransson, Magnus To¨rnblom, Stefan Carlsson, Marie Hja¨lm Eriksson,
Annals of Oncology David Robinson, Mats Ande´n, Johan Stranne, Jonas Hugosson, Ingela Franck Lissbrant, Maria Nyberg, Go¨ran Ahlgren, Rene´ Blom, Calle Walller, Per Fransson, Fredrik Sandin and Karin Hellstro¨m.
Funding The Swedish Research Council (825-2012-5047); The Swedish Cancer Society (16 0700); Uppsala County Council (no grant number applies).
Disclosure The authors have declared no conflicts of interest.
References 1. Droz JP, Aapro M, Balducci L et al. Management of prostate cancer in older patients: updated recommendations of a working group of the International Society of Geriatric Oncology. Lancet Oncol 2014; 15: e404–e414. 2. Grummet JP, Plass K, N’Dow J. Prostate cancer management in an ageing population. Eur Urol 2017; 72(4): 532–533. 3. Heinzer H, Steuber T. Prostate cancer in the elderly. Urol Oncol 2009; 27(6): 668–672. 4. Bill-Axelson A, Holmberg L, Garmo H et al. Radical prostatectomy or watchful waiting in early prostate cancer. N Engl J Med 2014; 370(10): 932–942. 5. Hamdy FC, Donovan JL, Lane JA et al. 10-Year outcomes after monitoring, surgery, or radiotherapy for localized prostate cancer. N Engl J Med 2016; 375(15): 1415–1424. 6. Wilt TJ, Jones KM, Barry MJ et al. Follow-up of prostatectomy versus observation for early prostate cancer. N Engl J Med 2017; 377(2): 132–142. 7. Abdollah F, Sun M, Thuret R et al. A competing-risks analysis of survival after alternative treatment modalities for prostate cancer patients: 1988– 2006. Eur Urol 2011; 59(1): 88–95. 8. Bechis SK, Carroll PR, Cooperberg MR. Impact of age at diagnosis on prostate cancer treatment and survival. J Clin Oncol 2011; 29(2): 235–241. 9. Bratt O, Folkvaljon Y, Hjalm Eriksson M et al. Undertreatment of men in their seventies with high-risk nonmetastatic prostate cancer. Eur Urol 2015; 68(1): 53–58. 10. Hoffman KE. Management of older men with clinically localized prostate cancer: the significance of advanced age and comorbidity. Semin Radiat Oncol 2012; 22(4): 284–294. 11. Brassell SA, Rice KR, Parker PM et al. Prostate cancer in men 70 years old or older, indolent or aggressive: clinicopathological analysis and outcomes. J Urol 2011; 185(1): 132–137. 12. Busch J, Magheli A, Leva N et al. Higher rates of upgrading and upstaging in older patients undergoing radical prostatectomy and qualifying for active surveillance. BJU Int 2014; 114(4): 517–521. 13. Richstone L, Bianco FJ, Shah HH et al. Radical prostatectomy in men aged >or¼70 years: effect of age on upgrading, upstaging, and the accuracy of a preoperative nomogram. BJU Int 2008; 101(5): 541–546. 14. Van Hemelrijck M, Garmo H, Wigertz A et al. Cohort profile update: the National Prostate Cancer Register of Sweden and Prostate Cancer data Base—a refined prostate cancer trajectory. Int J Epidemiol 2016; 45(1): 73–82. 15. Fall K, Stromberg F, Rosell J et al. Reliability of death certificates in prostate cancer patients. Scand J Urol Nephrol 2008; 42(4): 352–357. 16. Salinas CA, Tsodikov A, Ishak-Howard M, Cooney KA. Prostate cancer in young men: an important clinical entity. Nat Rev Urol 2014; 11(6): 317–323.
8 | Pettersson et al. Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018
Original article
Annals of Oncology 17. Thorstenson A, Garmo H, Adolfsson J, Bratt O. Cancer specific mortality in men diagnosed with prostate cancer before age 50 years: a nationwide population based study. J Urol 2017; 197(1): 61–66. 18. Tomic K, Westerberg M, Robinson D et al. Proportion and characteristics of men with unknown risk category in the National Prostate Cancer Register of Sweden. Acta Oncol 2016; 55(12): 1461–1466. 19. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis1987; 40(5): 373–383. 20. Cleveland WS. Robust locally weighted regression and smoothing scatterplots. J Am Stat Assoc 1979; 74(368): 829–836. 21. Buuren S, Flexible Imputation of Missing Data. Boca Raton, FL: CRC Press, 2012. 22. Briganti A, Spahn M, Joniau S et al. Impact of age and comorbidities on long-term survival of patients with high-risk prostate cancer treated with radical prostatectomy: a multi-institutional competing-risks analysis. Eur Urol 2013; 63(4): 693–701. 23. Sanchez-Salas R, Prapotnich D, Rozet F et al. Laparoscopic radical prostatectomy is feasible and effective in 0 fit0 senior men with localized prostate cancer. BJU Int 2010; 106(10): 1530–1536.
24. Siddiqui SA, Sengupta S, Slezak JM et al. Impact of patient age at treatment on outcome following radical retropubic prostatectomy for prostate cancer. J Urol 2006; 175(3): 952–957. 25. Sun L, Caire AA, Robertson CN et al. Men older than 70 years have higher risk prostate cancer and poorer survival in the early and late prostate specific antigen eras. J Urol 2009; 182(5): 2242–2248. 26. Freedland SJ, Presti JC Jr, Kane CJ et al. Do younger men have better biochemical outcomes after radical prostatectomy? Urology 2004; 63: 518–522. 27. D’Amico AV, Cote K, Loffredo M et al. Advanced age at diagnosis is an independent predictor of time to death from prostate carcinoma for patients undergoing external beam radiation therapy for clinically localized prostate carcinoma. Cancer 2003; 97: 56–62. 28. Aaronson D, Cowan J, Carroll P, Konety B. Association of age and response to androgen-deprivation therapy with or without radiotherapy for prostate cancer: data from CaPSURE. BJU Int 2010; 105(7): 951–955. 29. Hoffman KE, Chen MH, Moran BJ et al. Prostate cancer-specific mortality and the extent of therapy in healthy elderly men with high-risk prostate cancer. Cancer 2010; 116(11): 2590–2595. 30. Loeb S, Berglund A, Stattin P. Population based study of use and determinants of active surveillance and watchful waiting for low and intermediate risk prostate cancer. J Urol 2013; 190(5): 1742–1749.
doi:10.1093/annonc/mdx742 | 9 Downloaded from https://academic.oup.com/annonc/advance-article-abstract/doi/10.1093/annonc/mdx742/4638417 by University of Florida user on 09 January 2018