Local Therapy Improves Survival in Metastatic Prostate Cancer

Local Therapy Improves Survival in Metastatic Prostate Cancer

EURURO-7302; No. of Pages 7 EUROPEAN UROLOGY XXX (2017) XXX–XXX available at www.sciencedirect.com journal homepage: www.europeanurology.com Prostat...

265KB Sizes 0 Downloads 23 Views

EURURO-7302; No. of Pages 7 EUROPEAN UROLOGY XXX (2017) XXX–XXX

available at www.sciencedirect.com journal homepage: www.europeanurology.com

Prostate Cancer

Local Therapy Improves Survival in Metastatic Prostate Cancer Sami-Ramzi Leyh-Bannurah a,b,c,1,*, Ste´phanie Gazdovich a,d,1, Lars Buda¨us b, Emanuele Zaffuto a,e, Alberto Briganti e, Firas Abdollah e,f, Francesco Montorsi e, Jonas Schiffmann g, Mani Menon f, Shahrokh F. Shariat h, Margit Fisch c, Felix Chun c, Thomas Steuber b, Hartwig Huland b, Markus Graefen b, Pierre I. Karakiewicz a,d a

Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, Canada; b Martini-Clinic, Prostate Cancer Center Hamburg-

Eppendorf, Hamburg, Germany; c Department of Urology, University Medical Center Hamburg-Eppendorf, Hamburg, Germany;

d

Department of Urology,

University of Montreal Health Center, Montreal, Canada; e Department of Urology and Division of Experimental Oncology, URI, Urological Research Institute, IRCCS San Raffaele Scientific Institute, Milan, Italy; f Vattikuti Urology Institute and VUI Center for Outcomes Research Analytics and Evaluation (VCORE), Henry Ford Hospital, Henry Ford Health System, Detroit, MI, USA; h

g

Department of Urology, Academic Hospital Braunschweig, Braunschweig, Germany;

Department of Urology Medical University of Vienna, Vienna, Austria

Article info

Abstract

Article history: Accepted March 14, 2017

Background: Treatment of the primary, termed local therapy (LT), may improve survival in metastatic prostate cancer (mPCa) versus no local therapy (NLT). Objective: To assess cancer-specific mortality (CSM) after LT versus NLT in mPCa. Design, setting, and participants: Within the Surveillance, Epidemiology and End Results database (2004–2013), 13 692 mPCa patients were treated with LT (radical prostatectomy [RP] or radiation therapy [RT]) or NLT. Outcome measurements and statistical analysis: Multivariable competing risk regression analyses (MVA CRR) tested CSM after propensity score matching (PSM) in two analyses, (1) NLT versus LT and (2) RP versus RT, and were complemented with interaction, sensitivity, unmeasured confounder, and landmark analyses. Results and limitations: Of 13 692 mPCa patients, 474 received LT: 313 underwent RP and 161 RT. In MVA CRR, after PSM, LT (n = 474) results in lower CSM (subhazard ratio [SHR] 0.40, 95% confidence interval [CI] 0.32–0.50) versus NLT (n = 1896). In MVA CRR after PSM, RP (n = 161) results in lower CSM (SHR 0.59, 95% CI 0.35–0.99) versus RT (n = 161). Invariably, lowest CSM rates were recorded for Gleason 7, cT3, and M1a substage. Interaction and sensitivity analyses confirmed the robustness of results, and landmark analyses rejected the bias favouring LT. A strong unmeasured confounder (HR = 5), affecting 30% of NLT patients, could obliterate LT benefit. Data were retrospective. Conclusions: In mPCa, LT results in lower mortality relative to NLT. Within LT, lower mortality is recorded after RP than RT. Patients with most favourable grade, local stage, and metastatic substage derive most benefit from LT. They also derive most benefit from RP, when LT types are compared (RP vs RT). It is important to consider study limitations until ongoing clinical trials confirm the proposed benefits. Patient summary: Individuals with prostate cancer that spreads outside of the prostate might still benefit from prostate-directed treatments, such as radiation or surgery, in addition to receiving androgen deprivation therapy. # 2017 European Association of Urology. Published by Elsevier B.V. All rights reserved.

Associate Editor: James Catto Keywords: Standard of care Local radiotherapy Radical prostatectomy Cytoreductive prostatectomy Brachytherapy

1

These authors shared the first authorship. * Corresponding author. Cancer Prognostics and Health Outcomes Unit, 264 Blvd. Rene-Levesque E. Room 228, Montreal QC H2X 1P1, Canada. Tel. +1 514 806 5044; Fax: +1 514 227 5103. E-mail address: [email protected] (S.-R. Leyh-Bannurah).

http://dx.doi.org/10.1016/j.eururo.2017.03.020 0302-2838/# 2017 European Association of Urology. Published by Elsevier B.V. All rights reserved.

Please cite this article in press as: Leyh-Bannurah S-R, et al. Local Therapy Improves Survival in Metastatic Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.03.020

EURURO-7302; No. of Pages 7 2

EUROPEAN UROLOGY XXX (2017) XXX–XXX

1.

Introduction

sufficient matching. Propensity score matching (1:1 ratio due to sample size) was repeated for RT versus RP tests.

Treatment of the primary tumour, termed local treatment (LT), reduces mortality rates in several malignancies (renal, colon, and ovarian cancer), despite established metastatic spread [1–6]. LT may also improve survival in metastatic prostate cancer (mPCa), relative to standard of care: androgen deprivation therapy (ADT) with no LT (NLT) [7]. Evidence stems from six studies within four data repositories: the Surveillance, Epidemiology and End Results (SEER) database [8–10], the SEER-Medicare database [11], the National Cancer Database (NCDB) [12], and the Munich Cancer Registry [13]. Four studies reported improved survival with radical prostatectomy (RP) versus NLT [8,9,11,13]. However, none directly examined RP against radiotherapy (RT). Additionally, three institutional studies confirmed RP safety in select mPCa patients [14–16]. Methodological limitations apply to all six reports: four failed to account for other-cause mortality that may irreversibly confound all-cause mortality rates and lacked competing risk regression (CRR) [8,10,12,13]. Three failed to fully adjust for patient characteristics with propensity score matching [9,13]. One combined CRR and propensity score matching but had a limited sample size: 47 RP patients [11]. We combined CRR with propensity score matching, within the largest possible patient sample: the SEER database to test for differences in CSM according to LT versus NLT. Moreover, unlike previously, we tested for CSM differences according to LT type: RP versus RT. 2.

Patients and methods

2.1.

Patient selection

Within the SEER database (18 cancer registries, accounting for 26% of the US population), we identified patients diagnosed with adenocarcinoma of the prostate (International Classification of Disease for Oncology [61.9]; histological code: 8140) with metastatic disease at diagnosis (ie, SEER field ‘‘CS Mets at DX’’) and stages M1a–c (sixth edition of American Joint Committee on Cancer [AJCC] Cancer Staging Manual). All underwent LT: (1) RP (surgery site codes 50 and 70) with or without RT or (2) RT (ie, brachytherapy) with or without external beam radiation therapy (EBRT) or (3) NLT, between 2004 and 2013. Prostate-specific antigen (PSA) values were available for patients diagnosed between 2010 and 2013, and were included in subgroup analysis. Patients were stratified according to RP versus RT versus NLT status, as described earlier [8,9]. EBRT was excluded; it lacked target site information distinguishing local from extraprostatic treatment. Other surgical treatments than RP, for example, transurethral resection of the prostate, were also excluded. These selection criteria yielded 13 692 patients.

2.2.

Propensity score matching

Propensity score matching (4:1 ratio, with nearest-neighbour matching or calliper width of 0.1 of the standard deviation of the logit) yielded similar patient characteristics between LT (n = 474) and NLT cohorts (n = 13 218), emulated randomised trial design, minimised residual bias, and increased precision [17]. Adjustment variables consisted of age, race, biopsy Gleason score, clinical tumour stage, nodal stage, and metastatic substages. Standardised mean difference measurements were performed to confirm

2.3.

Statistical analyses

To ensure intergroup comparability, we exclusively relied on metrics applicable to all patients, regardless of LT versus NLT status or of LT type (RP vs RT): biopsy Gleason score, clinical tumour, nodes, and metastatic substages were used in propensity score matching and in all analyses. Covariates consisted of age, race, marital status, biopsy Gleason score, clinical tumour, nodes, and metastatic substages. Subsequently, clinical variables that qualified as independent cancer-specific mortality (CSM) predictors were used in a risk stratification scheme of 1 versus 2 risk factors. To further examine the effect of risk factors (1 vs 2 risk factors), we refitted the Cox model by adding an interaction term: risk stratification scheme and treatment status. Landmark analyses were performed at 6, 12, 18, and 24 mo after the time of diagnosis, to address the potential effect of immortal time bias, which may favourably affect patients treated with either RP or RT, relative to NLT patients [18]. Sensitivity analyses tested the effect of a potential unmeasured confounder by (1) computing the prevalence required to render our result statistically insignificant assuming that such a confounder has a moderate subhazard ratio (SHR; eg, 2), and (2) computing the SHR required to render our results statistically insignificant assuming a moderate prevalence ratio (eg, 30% in LT and 10% in NLT) [19,20]. All tests were two sided with a statistical significance set at p < 0.05. Analyses were performed with the statistical package for R (version 3.2.2; the R foundation for Statistical Computing, Vienna, Austria).

3.

Results

Prior to propensity score matching, NLT patients were oldest (72 yr), relative to RT (68 yr) and RP patients (63 yr, Tables 1 and 2). The rate for biopsy Gleason score 7 was lowest in NLT patients (18%), followed by RP (47%) and RT patients (48%). The rate for clinical stage T3 was highest in RP (97%), followed by in RT (94%) and NLT (89%) patients. Node stage N0/NX was virtually the same in RP (92%) and RT (91%) patients and lower in NLT patients (80%). Finally, metastatic substage M1a was also virtually the same in RP (11%) and RT (12%) patients and lower in NLT group (6.1%). After propensity score matching (NLT vs LT and RP vs RT), residual statistically significant differences remained only for year of diagnosis in RP versus RT comparisons: 17% and 6.2% of the population in year 2013, respectively, in RP and RT arms. After propensity score matching, the median follow-up of NLT versus LT and RP versus RT patients without CSM or other-cause mortality was 31.0 mo (interquartile range [IQR] 12.0–58.0) versus 43.5 (IQR 18.0–80.0) and 39.0 mo (IQR 16–72) versus 56.0 mo (IQR 28.0–86.0), respectively. In propensity score–matched multivariable competing risk regression analyses (MVA CRR), both LT types, RP and RT, yielded lower CSM rates (65% and 52%, respectively) relative to NLT (SHR 0.35, 95% CI 0.26–0.46 and SHR 0.48, 95% CI 0.35–0.66; both p < 0.001; Table 3). Additionally, CSM was also lower with Gleason 7 (vs GS 8; SHR 1.84, 95% CI 1.59–2.13; p < 0.001), cT3 (vs T4; SHR 1.85, 95% CI 1.39–2.46; p < 0.001), and substage M1a (vs M1c; SHR 1.98, 95% CI 1.52–2.58; p < 0.001) and in married men (vs divorced/widowed; SHR 1.25, 95% CI 1.03–1.51; p < 0.024).

Please cite this article in press as: Leyh-Bannurah S-R, et al. Local Therapy Improves Survival in Metastatic Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.03.020

EURURO-7302; No. of Pages 7 3

EUROPEAN UROLOGY XXX (2017) XXX–XXX

Table 1 – Descriptive characteristics of 13 692 patients diagnosed with metastatic prostate cancer between 2004 and 2013 from the Surveillance Epidemiology and End Results database, stratified according to treatment type with and without propensity score matching (ratio 4:1) Variables

Median age, yr (IQR) Race, n (%) Caucasian African American Other Unknown Marital status, n (%) Married Divorced/widowed Single Unknown Year of diagnosis, n (%) 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Biopsy Gleason score, n (%) 6 7 8 Unknown Clinical T stage, n (%) T1c T2 T3 T4 Clinical N stage, n (%) N0/Nx N1 AJCC M Stage, n (%) M1a M1b M1c PSA at biopsy, ng/ml, n (%) c 20 >20 Cancer-specific death, n (%)

No local treatment (n = 13 218; %)

Local treatment (n = 474; %)

p value

72 (63–80)

64 (58–69)

0.01 0.6

9765 (74) 2521 (19) 799 (6.0) 133 (1.0)

352 (74) 85 (18) 34 (7.2) 3 (0.6)

7347 2799 2049 1023

(56) (21) (16) (7.7)

348 (73) 63 (13) 39 (8.2) 24 (5.1)

1190 1187 1215 1250 1307 1323 1384 1337 1462 1563

(9.0) (9.0) (9.2) (9.5) (9.9) (10) (11) (10) (11) (12)

40 43 57 44 59 44 38 49 47 53

a

Propensity score-adjusted no local treatment (n = 1896; %)

Propensity score-adjusted local treatment (n = 474; %)

64 (57–70)

64 (58–69)

1430 (75) 341 (18) 106 (5.6) 19 (1.0)

352 (74) 85 (18) 34 (7.2) 3 (0.6)

1379 (73) 266 (14) 149 (7.9) 102 (5.4)

348 (73) 63 (13) 39 (8.2) 24 (5.1)

199 180 184 192 197 193 188 185 184 194

(11) (9.5) (9.7) (10) (10) (10) (9.9) (9.8) (9.7) (10)

40 43 57 44 59 44 38 49 47 53

247 611 894 144

(13) (32) (47) (7.6)

71 (15) 152 (32) 217 (46) 34 (7.2)

959 (51) 721 (38) 137 (7.2) 79 (4.2)

246 (52) 171 (36) 36 (7.6) 21 (4.4)

1749 (92) 147 (7.8)

435 (92) 39 (8.2)

180 (9.5) 1341 (71) 375 (20)

54 (11) 325 (69) 95 (20)

124 (71) 50 (29) 779 NA

128 (74) 46 (26) 94 NA

<0.001

71 (15) 152 (32) 217 (46) 34 (7.2)

6279 4318 1199 1422

246 (52) 171 (36) 36 (7.6) 21 (4.4)

0.6

0.7

0.9

<0.001 (48) (33) (9.1) (11)

0.8

<0.001 10596 (80) 2622 (20)

435 (92) 39 (8.2)

800 (6.1) 9478 (72) 2940 (22)

54 (11) 325 (69) 95 (20)

864 (17) 4249 (83) 5920 NA

128 (74) 46 (26) 94 NA

0.4

<0.001

0.7

<0.001

NA

0.059 0.5

(8.4) (9.1) (12) (9.3) (13) (9.3) (8.0) (10) (9.9) (11)

<0.001 453 (3.4) 1888 (14) 7270 (55) 3607 (27)

b

>0.9

0.3 (8.4) (9.1) (12) (9.3) (13) (9.3) (8.0) (10) (9.9) (11)

p value

NA

IQR = interquartile range; AJCC = American Joint Committee on Cancer; NLT = no local treatment; LT = local treatment; NA = not applicable; PSA = prostatespecific antigen. a Comparing NLT versus LT (unmatched). b Comparing NLT versus LT (propensity score-adjusted cohorts). c PSA values were available only for patients diagnosed between 2010 and 2013; proportions presented are of the corresponding subgroups.

Using interaction term analyses, we tested whether treatment type (RT or RP vs NLT) interacts with independent predictors of CSM (Gleason 8, cT4, M1b–c substages) that were stratified as 1 variable versus 2 variables, and are henceforth referred to as risk criteria. The interaction term analyses revealed the following: (1) both RP and RT have a strong protective effect in patient with 1 risk criterion (SHR of RP: 0.16, 95% CI 0.09–.28; SHR of RT: 0.33, 95% CI 0.19–0.56; Supplementary Table 1), and (2) RP and RT are less effective in patients with 2 risk criteria than those with 1 risk criterion

(SHR of interaction term between RP and 2 risk criteria: 3.95, 95% CI 2.02–7.67; SHR of interaction term between RT and 2 risk criteria: 2.08, 95% CI 1.07–4.04; Supplementary Table 1). Based on the result of the interaction term analyses, and to further investigate whether RP and RT have a protective effect in those with 2 risk criteria, we conducted a subgroup analysis only in patients with 2 risk factors. Here, the central tendency for RP and RT suggests that even in patients with 2 risk factors, LT is still protective (SHR of RP: 0.60; SHR of RT: 0.67).

Please cite this article in press as: Leyh-Bannurah S-R, et al. Local Therapy Improves Survival in Metastatic Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.03.020

EURURO-7302; No. of Pages 7 4

EUROPEAN UROLOGY XXX (2017) XXX–XXX

Table 2 – Descriptive characteristics of 474 locally treated patients diagnosed with metastatic prostate cancer between 2004 and 2013 from the Surveillance Epidemiology and End Results database, stratified according to treatment type with and without propensity score matching (ratio 1:1) Variables

Median age, yr (IQR) Race, n (%) Caucasian African American Other Unknown Marital status, n (%) Married Divorced/widowed Single Unknown Year of diagnosis, n (%) 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Biopsy Gleason score, n (%) 6 7 8 Unknown Clinical T stage, n (%) T1c T2 T3 T4 Clinical N stage, n (%) N0/Nx N1 AJCC M Stage, n (%) M1a M1b M1c Cancer-specific death, n (%)

Radical prostatectomy (n = 313; %)

Radiotherapy (n = 161; %)

63 (58–67)

68 (60–73)

237 (76) 48 (15) 26 (8.3) 2 (0.6)

115 (71) 37 (23) 8 (5.0) 1 (0.6)

237 (76) 38 (12) 24 (7.7) 14 (4.5)

111 (69) 25 (16) 15 (9.3) 10 (6.2)

23 25 29 27 39 30 28 31 38 43

17 (11) 18 (11) 28 (17) 17 (11) 20 (12) 14 (8.7) 10 (6.2) 18 (11) 9 (5.6) 10 (6.2)

p value

>0.9 0.15

a

Propensity score-adjusted radical prostatectomy (n = 161; %)

Propensity score-adjusted radiotherapy (n = 161; %)

66 (60–71)

68 (60–73)

119 (74) 24 (15) 16 (9.9) 2 (1.2)

115 (71) 37 (23) 8 (5.0) 1 (0.6)

110 (68) 31 (19) 13 (8.1) 7 (4.4)

111 (69) 25 (16) 15 (9.3) 10 (6.2)

15 12 16 10 18 13 15 19 15 28

(9.3) (7.5) (9.9) (6.2) (11) (8.1) (9.3) (12) (9.3) (17)

17 (11) 18 (11) 28 (17) 17 (11) 20 (12) 14 (8.7) 10 (6.2) 18 (11) 9 (5.6) 10 (6.2)

29 45 77 10

(18) (28) (48) (6.2)

30 47 72 12

(19) (29) (45) (7.5)

78 58 14 11

(49) (36) (8.7) (6.8)

83 54 14 10

(52) (34) (8.7) (6.2)

0.5

30 47 72 12

(19) (29) (45) (7.5)

163 (52) 117 (37) 22 (7.0) 11 (3.5)

83 54 14 10

(52) (34) (8.7) (6.2)

288 (92) 25 (8.0)

147 (91) 14 (8.7)

35 (11) 222 (71) 56 (18) 52 NA

19 (12) 103 (64) 39 (24) 42 NA

>0.9

0.5

>0.9

0.9

0.8 149 (93) 12 (7.5)

147 (91) 14 (8.7)

23 (14) 102 (63) 36 (22) 25 NA

19 (12) 103 (64) 39 (24) 42 NA

0.2

NA

0.4 0.12

0.04

0.4 41 (13) 105 (34) 145 (46) 22 (7.0)

b

0.7

0.02 (7.4) (8.0) (9.3) (8.6) (13) (9.6) (9.0) (9.9) (12) (14)

p value

0.8

IQR = interquartile range; AJCC = American Joint Committee on Cancer; NLT = no local treatment; LT = local treatment; NA = not applicable. Comparing NLT versus LT (unmatched). b Comparing NLT versus LT (propensity score-adjusted cohorts). a

To test the potential effect of baseline PSA values, we repeated the analyses in individuals with available PSA data. Propensity score matching (1:1) for clinical parameters and PSA values resulted in 348 assessable mPCa patients (174 NLT and 174 LT). Here, MVA CRR showed that LT yields lower CSM (51%) versus NLT (SHR 0.49, 95% CI 0.28–0.86; p = 0.012), as in the main analysis. To address the effect of potential nonadherence to ADT use in mPCa patients, we performed interaction analyses where age, categorised as younger versus older (70 vs > 70 yr age) interacted with treatment type. We posited that in younger mPCa patients, nonadherence to ADT should not exist. According to the results of the interaction term analysis, where age 70 yr was set as a reference, the effect of LT relative to NLT was virtually the same: RP and RT

yielded lower CSM (68% and 52%, respectively) compared with NLT (SHR 0.32, 95% CI 0.23–0.45 and SHR 0.48, 95% CI 0.32–0.72; both p < 0.001). In landmark analyses performed at 6, 12, 18, and 24 mo after LT, the decrease in CSM recorded after RP, as well as after RT, remained unchanged relative to ‘‘naive’’ analyses, where the potentially favourable survival bias towards individuals who benefited from LT was unaccounted for. We also tested for the effect of an unmeasured confounder that could obliterate the effect of LT on CSM. The analyses revealed that an unmeasured confounder with an SHR of 2 would render the effect of LT statistically insignificant, if it affected 70% of NLT patients and  10% of LT patients. We tested a second scenario, where an unmeasured confounder with an SHR of 5 would render

Please cite this article in press as: Leyh-Bannurah S-R, et al. Local Therapy Improves Survival in Metastatic Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.03.020

EURURO-7302; No. of Pages 7 5

EUROPEAN UROLOGY XXX (2017) XXX–XXX

Table 3 – Multivariable competing risks regression of propensity score-adjusted patients diagnosed with metastatic prostate cancer between 2004 and 2013 from the Surveillance Epidemiology and End Results database, stratified according to treatment type Variables

Local treatment versus no local treatment SHR (95% CI)

Table 4 – Multivariable competing risks regression of propensity score-adjusted patients diagnosed with metastatic prostate cancer and locally treated between 2004 and 2013 from the Surveillance Epidemiology and End Results database, stratified according to treatment type Variables

Radical prostatectomy versus radiotherapy

p value SHR (95% CI)

Type of treatment No local therapy Radiotherapy Radical prostatectomy Biopsy Gleason score 7 8 Unknown Clinical T stage T1/T2 T3 T4 Clinical N stage N0/Nx N1 AJCC M stage M1a M1b M1c Age (yr) Race Caucasian African American Other/unknown Marital status Married Divorced/widowed Single Unknown

Ref. 0.48 (0.35–0.66) 0.35 (0.26–0.46)

<0.001 <0.001

Ref. 1.84 (1.59–2.13) 1.72 (1.28–2.31)

<0.001 <0.001

Ref. 1.10 (0.87–1.40) 1.85 (1.39–2.46)

0.4 <0.001

Ref. 1.18 (0.91–1.52)

0.20

Ref. 1.65 (1.31–2.08) 1.98 (1.52–2.58) 1.00 (0.99–1.01)

<0.001 <0.001 0.4

Ref. 0.91 (0.76–1.09) 0.77 (0.58–1.02)

0.3 0.07

Ref. 1.25 (1.03–1.51) 1.23 (0.97–1.57) 0.75 (0.54–1.03)

0.024 0.092 0.074

interval; SHR = subhazard CI = confidence AJCC = American Joint Committee on Cancer.

ratio;

Ref. = reference;

the effect of LT statistically insignificant, if it affected 30% of NLT patients and  10% of LT patients. In propensity score–matched MVA CRR testing RP versus RT CSM differences, RP yielded lower CSM (41%) relative to RT (SHR 0.59, 95% CI 0.35–0.99; p = 0.048; Table 4). Additionally, CSM was also lower in presence of Gleason 7 (vs GS 8; SHR 3.67, 95% CI 2.03–6.66; p < 0.001), cT3 (vs T4; SHR 5.48, 95% CI 2.64–11.4; p < 0.001), and M1a substage (vs M1c; SHR 4.7, 95% CI 1.88–11.7; p < 0.001). 4.

Discussion

In mPCa, LT showed a survival benefit compared with NLT in six retrospective studies [8–13]. However, methodological limitations [9,10,13] and/or lack of adjustment for othercause mortality [8,10,12,13] limits their validity. Sample size limitations (eg, only 47 RP patients were included) undermine the generalisability of the single study that combined the necessary methodological tools: CRR and propensity score matching [11]. Last but not least, no study directly compared RP with nonsurgical LT, that is, RT. Based on those limitations, we tested CSM rates using the most bias-free approach, namely, with the combined use of CRR

Type of treatment Radiotherapy Radical prostatectomy Biopsy Gleason score 7 8 Unknown Clinical T stage T1/T2 T3 T4 Clinical N stage N0/Nx N1 AJCC M stage M1a M1b M1c Age (yr)

p value

Ref. 0.59 (0.35–0.99)

0.048

Ref. 3.67 (2.03–6.66) 0.80 (0.14–4.72)

<0.001 0.8

Ref. 1.01 (0.39–2.61) 5.48 (2.64–11.4)

>0.9 <0.001

Ref. 1.01 (0.34–2.99)

>0.9

Ref. 3.48 (1.51–8.04) 4.70 (1.88–11.7) 1.02 (0.98–1.05)

0.01 <0.001 0.3

interval; SHR = subhazard CI = confidence AJCC = American Joint Committee on Cancer.

ratio;

Ref. = reference;

accounting for other-cause mortality and propensity score matching. Several noteworthy findings emerged. First, we identified important and highly clinically meaningful differences between NLT and LT cohorts, as well as between RP and RT cohorts. Such differences validate the necessity of propensity score matching, which we applied. Moreover, such differences may not rest only on measured variables that are available for analyses. Instead, unmeasured confounders that are not accounted for by recorded variables also require attention. We addressed that need and showed that it is unlikely that an unmeasured confounder could negate the benefit of LT versus NLT. Second, our MVA demonstrated important and highly statistically significant survival benefit, when LT was compared with NLT. The CSM benefit withstood the rigours of increasingly complex hypothesis testing: sensitivity and landmark analyses. Both failed to corroborate the CSM decrease associated with LT, in patients with 2 risk criteria. Hence, there is a need for careful risk stratification. Interestingly, landmark analyses performed at 6, 12, 18, and 24 mo after LT failed to show that patients selected for LT, either RP or RT, have benefited from a favourable survival bias. Third, our MVA also demonstrated important and statistically significant survival benefit, expressed in CSM reduction, when RP was compared with RT. In both analyses, LT versus NLT and RP versus RT, the lowest CSM rates were recorded in patients with Gleason 7, cT3, or M1a substage. In consequence, important selection considerations are required, and patients with favourable grade, stage, and low metastatic burden should only be

Please cite this article in press as: Leyh-Bannurah S-R, et al. Local Therapy Improves Survival in Metastatic Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.03.020

EURURO-7302; No. of Pages 7 6

EUROPEAN UROLOGY XXX (2017) XXX–XXX

considered, as is recommended in ongoing prospective trials (NCT02454543, NCT02458716, and NCT01751438). Fourth, our results emphasised the importance of risk stratification within the LT versus NLT cohort, as well as within the RP versus RT cohort. For example, interaction term analyses showed that RT and RP are less effective in patients with 2 risk criteria compared with those with 1 risk criterion. Similarly, in the comparison between RP and RT patients, the CSM benefit was not observed in patients with 2 risk criteria. Fifth, our results corroborate the findings based on the same and other data repositories, namely the SEER [8–10], SEER-Medicare [11], NCDB [12], and Munich Cancer registry [13]. However, none of those four comparisons directly examined RP against RT, as was done in the current study. Survival advantage related to LT delivery in select mPCa patients was also recently reported in a German case– control study with a small sample size (n = 23) [15]. RP was limited to mPCa patients treated with ADT, with minimal extent of bone metastases, with limited extent of lymph node metastases, and without visceral involvement. RP patients experienced similar complication rates to patients treated with ADT alone. Our study is not devoid of limitations. First, similar to all previous studies, this study is retrospective in nature. Currently, three prospective clinical trials addressing RP in the metastatic setting are open (NCT02454543, NCT02458716, and NCT01751438). Second, the SEER database does not provide information on comorbidities and/or performance status, which may be used for patient selection. Third, post-treatment PSA values are available only in some SEER database patients. They represent important prognostic markers, and favourable post-ADT PSA responses might have represented an argument for stronger consideration of LT. Similarly, baseline PSA represents an independent risk factor for CSM. Indeed, the protective association of RP with CSM in mPCa patients was previously not recorded in patients with baseline PSA levels above 20 ng/ml [11]. Fourth, site-specific EBRT codes are unavailable in the SEER database and only brachytherapy codes were analyzed. Fifth, the SEER database does not provide complete information regarding the exact extent, for example, number and/or site(s) of metastatic disease, except for M substages that were included in all analyses [21]. Finally, data regarding systemic therapies, such as ADT [22], chemotherapy, or agents targeting the androgen receptor and/or autocrine/paracrine androgen synthesis, were also unavailable. In the United States, during the study span and until today, ADT represents the gold standard therapy for men with mPCa, as per National Comprehensive Cancer Guidelines.

from RP, when LT types are compared (RP vs RT). It is important to consider study limitations until ongoing clinical trials confirm the proposed benefits. Author contributions: Sami-Ramzi Leyh-Bannurah had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Leyh-Bannurah, Karakiewicz, Graefen, Buda¨us, Huland. Acquisition of data: Leyh-Bannurah, Gazdovich, Zaffuto. Analysis and interpretation of data: Leyh-Bannurah, Karakiewicz, Zaffuto, Buda¨us, Chun, Briganti, Montorsi, Graefen, Steuber, Huland. Drafting of the manuscript: Leyh-Bannurah, Karakiewicz, Gazdovich, Buda¨us. Critical revision of the manuscript for important intellectual content: Huland, Graefen, Buda¨us, Steuber, Fisch, Chun, Montorsi, Briganti, Abdollah, Menon, Shariat, Schiffmann. Statistical analysis: Leyh-Bannurah, Karakiewicz, Gazdovich, Chun, Zaffuto. Obtaining funding: None. Administrative, technical, or material support: Huland, Graefen, Fisch, Buda¨us, Steuber, Chun, Montorsi, Briganti, Abdollah. Supervision: Karakiewicz, Huland, Graefen, Fisch, Briganti, Montorsi, Menon, Shariat. Other: None. Financial disclosures: Sami-Ramzi Leyh-Bannurah certifies that all conflicts of interest, including specific financial interests and relationships and affiliations relevant to the subject matter or materials discussed in the manuscript (eg, employment/affiliation, grants or funding, consultancies, honoraria, stock ownership or options, expert testimony, royalties, or patents filed, received, or pending), are the following: None. Funding/Support and role of the sponsor: None.

Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. eururo.2017.03.020.

References [1] Bristow RE, Tomacruz RS, Armstrong DK, Trimble EL, Montz FJ. Survival effect of maximal cytoreductive surgery for advanced ovarian carcinoma during the platinum era: a meta-analysis. J Clin Oncol 2002;20:1248–59. [2] Heng DY, Wells JC, Rini BI, et al. Cytoreductive nephrectomy in patients with synchronous metastases from renal cell carcinoma: results from the International Metastatic Renal Cell Carcinoma Database Consortium. Eur Urol 2014;66:704–10. [3] Choueiri TK, Xie W, Kollmannsberger C, et al. The impact of cytor-

5.

Conclusions

Our study provides evidence that LT yields lower mortality relative to NLT in mPCa. Within LT, our study also suggests that RP decreased mortality more than RT. Furthermore, the results of our study show that patients with most favourable grade, local stage, and metastatic substage derive most benefit from LT. They also derive most benefit

eductive nephrectomy on survival of patients with metastatic renal cell carcinoma receiving vascular endothelial growth factor targeted therapy. J Urol 2011;185:60–6. [4] Mickisch GHJ, Garin A, van Poppel H, de Prijck L, Sylvester R. Radical nephrectomy plus interferon-alfa-based immunotherapy compared with interferon alfa alone in metastatic renal-cell carcinoma: a randomised trial. Lancet 2001;358:966–70. [5] Roviello F, Caruso S, Marrelli D, et al. Treatment of peritoneal carcinomatosis with cytoreductive surgery and hyperthermic

Please cite this article in press as: Leyh-Bannurah S-R, et al. Local Therapy Improves Survival in Metastatic Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.03.020

EURURO-7302; No. of Pages 7 EUROPEAN UROLOGY XXX (2017) XXX–XXX

intraperitoneal chemotherapy: state of the art and future developments. Surg Oncol 2011;20:e38–54. [6] Glehen O, Mohamed F, Gilly FN. Peritoneal carcinomatosis from digestive tract cancer: new management by cytoreductive surgery and intraperitoneal chemohyperthermia. Lancet Oncol 2004;5: 219–28. [7] Heidenreich A, Bastian PJ, Bellmunt J, et al. EAU guidelines on prostate cancer. Part II: treatment of advanced, relapsing, and castration-resistant prostate cancer. Eur Urol 2014;65:467–79.

7

[14] Sooriakumaran P, Karnes J, Stief C, et al. A multi-institutional analysis of perioperative outcomes in 106 men who underwent radical prostatectomy for distant metastatic prostate cancer at presentation. Eur Urol 2016;69:788–94. [15] Heidenreich A, Pfister D, Porres D. Cytoreductive radical prostatectomy in patients with prostate cancer and low volume skeletal metastases: results of a feasibility and case–control study. J Urol 2015;193:832–8. [16] Gandaglia G, Fossati N, Stabile A, et al. Radical prostatectomy in

[8] Antwi S, Everson TM. Prognostic impact of definitive local therapy

men with oligometastatic prostate cancer: results of a single-

of the primary tumor in men with metastatic prostate cancer at

institution series with long-term follow-up. Eur Urol. In press.

diagnosis: a population-based, propensity score analysis. Cancer Epidemiol 2014;38:435–41. [9] Culp SH, Schellhammer PF, Williams MB. Might men diagnosed with metastatic prostate cancer benefit from definitive treatment of the primary tumor? A SEER-based study. Eur Urol 2014;65:1058–66. [10] Fossati N, Trinh Q-D, Sammon JD, et al. Identifying optimal candi-

http://dx.doi.org/10.1016/j.eururo.2016.08.040. [17] D’Agostino RB. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17:2265–81. [18] Anderson JR, Cain KC, Gelber RD. Analysis of survival by tumor response. J Clin Oncol 1983;1:710–9.

dates for local treatment of the primary tumor among patients

[19] Lin DY, Psaty BM, Kronmal RA. Assessing the sensitivity of regres-

diagnosed with metastatic prostate cancer: a SEER-based study. Eur

sion results to unmeasured confounders in observational studies.

Urol 2015;67:3–6.

Biometrics 1998;54:948–63.

[11] Satkunasivam R, Kim AE, Desai M, et al. Radical prostatectomy or

[20] Mitra N, Heitjan DF. Sensitivity of the hazard ratio to nonignorable

external beam radiation therapy vs no local therapy for survival

treatment assignment in an observational study. Stat Med 2007;26:

benefit in metastatic prostate cancer: a SEER-Medicare analysis. J Urol 2015;194:378–85. [12] Loppenberg B, Dalela D, Karabon P, et al. The impact of local

1398–414. [21] James ND, Sydes MR, Mason MD, et al. Celecoxib plus hormone therapy versus hormone therapy alone for hormone-sensitive

treatment on overall survival in patients with metastatic prostate

prostate cancer: first results from the STAMPEDE multiarm,

cancer on diagnosis: a national cancer data base analysis. Eur Urol.

multistage, randomised controlled trial. Lancet Oncol 2012;13:

In press. http://dx.doi.org/10.1016/j.eururo.2016.04.031. [13] Gratzke C, Engel J, Stief CG. Role of radical prostatectomy in

549–58. [22] Attard G, Sydes MR, Mason MD, et al. Combining enzalutamide with

metastatic prostate cancer: data from the Munich Cancer Registry.

abiraterone, prednisone, and androgen deprivation therapy in the

Eur Urol 2014;66:602–3.

STAMPEDE trial. Eur Urol 2014;66:799–802.

Please cite this article in press as: Leyh-Bannurah S-R, et al. Local Therapy Improves Survival in Metastatic Prostate Cancer. Eur Urol (2017), http://dx.doi.org/10.1016/j.eururo.2017.03.020