A and B, Pairwise P values for PCSM are as follows: P = .002 for MaxRT vs RP, P < .001 for MaxRT vs RP plus adjuvant RT, P = .41 for MaxRT vs RP plus adjuvant EBRT, and P = .29 for MaxRT vs MaxRP. Pairwise P values for ACM are as follows: P = .09 for MaxRT vs RP, P = .003 for MaxRT vs RP plus adjuvant RT, P = .52 for MaxRT vs RP plus adjuvant EBRT, and P = .95 for MaxRT vs MaxRP. ACM indicates all-cause mortality; ADT, androgen deprivation therapy; EBRT, external beam radiotherapy; MaxRP, RP and both adjuvant RT and ADT; MaxRT, EBRT, brachytherapy, and ADT; RP, radical prostatectomy; and RT, radiotherapy.
eMethods. Supplemental Methods
eFigure. A, Graphical Illustration of How the Plausibility Index Value Is Calculated for the Comparison of Treatment With MaxRP vs MaxRT for the End Point of the Risk of PCSM. B, Graphical Illustration of How the Plausibility Index Value Is Calculated for the Comparison of Treatment With MaxRP vs MaxRT for the End Point of the Risk of ACM.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Tilki D, Chen M, Wu J, et al. Surgery vs Radiotherapy in the Management of Biopsy Gleason Score 9-10 Prostate Cancer and the Risk of Mortality. JAMA Oncol. 2019;5(2):213–220. doi:10.1001/jamaoncol.2018.4836
Can treatment with radical prostatectomy, adjuvant external beam radiotherapy, and androgen deprivation therapy (termed MaxRP) or external beam radiotherapy, brachytherapy, and androgen deprivation therapy (termed MaxRT) in men with Gleason score 9-10 prostate cancer provide similar survival outcomes?
Among 639 men, this cohort study found no significant difference after MaxRP or MaxRT in prostate cancer–specific and all-cause death risk, with plausibility indexes for equivalence of 76.75% for prostate cancer–specific mortality risk and 77.97% for all-cause mortality risk.
It is plausible that treatment with MaxRP or MaxRT for men with Gleason score 9-10 prostate cancer provides equivalent survival outcomes.
It is unknown how treatment with radical prostatectomy (RP) and adjuvant external beam radiotherapy (EBRT), androgen deprivation therapy (ADT), or both (termed MaxRP) compares with treatment with EBRT, brachytherapy, and ADT (termed MaxRT).
To investigate whether treatment of Gleason score 9-10 prostate cancer with MaxRP vs MaxRT was associated with prostate cancer–specific mortality (PCSM) and all-cause mortality (ACM) risk.
Design, Setting, and Participants
The study cohort comprised 639 men with clinical T1-4,N0M0 biopsy Gleason score 9-10 prostate cancer. Between February 6, 1992, and April 26, 2013, a total of 80 men were consecutively treated with MaxRT at the Chicago Prostate Cancer Center, and 559 men were consecutively treated with RP and pelvic lymph node dissection at the Martini-Klinik Prostate Cancer Center. Follow-up started on the day of prostate EBRT or RP and concluded on October 27, 2017.
Of the 559 men managed with RP and pelvic lymph node dissection, 88 (15.7%) received adjuvant EBRT, 49 (8.8%) received ADT, and 50 (8.9%) received both.
Main Outcomes and Measures
Treatment propensity score–adjusted risk of PCSM and ACM and the likelihood of equivalence of these risks between treatments using a plausibility index.
The cohort included 639 men, with a mean (SD) age of 65.83 (6.52) years. After median follow-ups of 5.51 years (interquartile range, 2.19-6.95 years) among 80 men treated with MaxRT and 4.78 years (interquartile range, 4.01-6.05 years) among 559 men treated with RP-containing treatments, 161 men had died, 106 (65.8%) from prostate cancer. There was no significant difference in the risk of PCSM (adjusted hazard ratio, 1.33; 95% CI, 0.49-3.64; P = .58) and ACM (adjusted hazard ratio, 0.80; 95% CI, 0.36-1.81; P = .60) when comparing men who underwent MaxRP vs MaxRT, with plausibility indexes for equivalence of 76.75% for the end point of the risk of PCSM and 77.97% for the end point of the risk of ACM. Plausibility indexes for all other treatment comparisons were less than 63%.
Conclusions and Relevance
Results of this study suggest that it is plausible that treatment with MaxRP or MaxRT for men with biopsy Gleason score 9-10 prostate cancer can lead to equivalent risk of PCSM and ACM.
While a large randomized trial1 comparing radical prostatectomy (RP) with external beam radiotherapy (EBRT) for the treatment of men with favorable-risk prostate cancer has been performed and reported, a large randomized clinical trial is lacking in men with high-risk prostate cancer.2 Evidence available to address this treatment decision comes from a multi-institutional retrospective study,3 which reported an association between a reduced risk of prostate cancer–specific mortality (PCSM) among men with high-risk prostate cancer based on biopsy Gleason score 9-10 prostate cancer and treatment with the combination of EBRT, brachytherapy, and androgen deprivation therapy (ADT) (termed MaxRT) for a median of 12 months compared with RP. However, a study comparing MaxRT with RP and adjuvant EBRT, ADT, or both (termed MaxRP) has not been performed, to our knowledge.
These comparisons are important because 3 randomized clinical trials4-6 have shown a significantly prolonged disease-free survival when adjuvant EBRT is administered compared with observation in men found to have extraprostatic extension, seminal vesicle invasion, and/or a positive surgical margin (R1) in the RP specimen. Moreover, another randomized study7 reported a survival benefit when adjuvant ADT was used compared with observation in the postoperative setting of pelvic lymph node–positive (N1) prostate cancer. Therefore, we investigated whether treatment of Gleason score 9-10 prostate cancer with RP and adjuvant EBRT, ADT, or MaxRP vs MaxRT was associated with PCSM and all-cause mortality (ACM) risk.
The study cohort comprised 639 men (mean [SD] age of 65.83 [6.52] years) with clinical T1-4,N0M0 (as per pelvic imaging and bone scan) biopsy Gleason score 9-10 prostate cancer. Between February 6, 1992, and April 26, 2013, a total of 80 men were consecutively treated with MaxRT at the Chicago Prostate Cancer Center (Westmont, Illinois), and 559 men were consecutively treated with RP and pelvic lymph node dissection at the Martini-Klinik Prostate Cancer Center (Hamburg, Germany). Men included in the study had a minimum follow-up of 6 months and a maximum follow-up of 8 years and initiated treatment within 4 to 6 weeks of diagnosis. Of the 559 men managed with RP and pelvic lymph node dissection, 372 (66.5%) received RP, 88 (15.7%) received RP plus adjuvant radiotherapy (RT) (ie, for prostate-specific antigen [PSA] level <0.1 ng/mL), 50 (8.9%) received MaxRT, and 49 (8.8%) received RP plus adjuvant ADT (to convert PSA level to micrograms per liter, multiply by 1.0).
Androgen deprivation therapy consisted of a luteinizing hormone–releasing hormone agonist or an antagonist with or without an antiandrogen and was delivered for a median duration of 6.00 months (interquartile range [IQR], 4.00-12.00 months) in men undergoing MaxRT. After RP, men found to have negative pelvic lymph nodes had a median ADT duration of 8.60 months (IQR, 4.70-19.80 months), and men found to have positive pelvic lymph nodes had a median ADT duration of 14.45 months (IQR, 7.00-23.90 months). External beam radiotherapy was delivered using photons in 25 fractions of 1.8 Gy to the prostate and seminal vesicles with a computed tomography–based simulation and intensity-modulated RT technique. Pelvic lymph node RT was delivered based on tumor characteristics at the discretion of the treating physician (B.J.M.). Brachytherapy was performed via a peripheral loading technique using preloaded iodine 125 (108 Gy), palladium 103 (90 Gy), or cesium 131 (100 Gy) sources and preplanned dosimetry.
Among 559 men, RP and pelvic lymph node dissection were performed using an open technique (519 [92.8%]) or a robotic technique (40 [7.2%]) by 14 urologic oncologists (D.T., H.H., M.G., and other nonauthors) at the Martini-Klinik Prostate Cancer Center. The median number of lymph nodes obtained was 14 (IQR, 8-22). Adjuvant RT was delivered using EBRT at a median of 3.32 months (IQR, 2.60-4.14 months) after RP to the prostatic bed to a median dose of 68.4 Gy; when involved, the pelvic lymph nodes were initially treated to a total dose of 45.0 Gy. Adjuvant ADT was given a median of 0.95 months (IQR, 0.43-2.43 months) after RP and was recommended when more than 1 lymph node was involved.7 Salvage RT to the prostate and seminal vesicles bed was administered after post-RP PSA failure (PSA level >0.1 ng/mL) unless the patient previously had adjuvant RT, in which case salvage ADT was delivered. After MaxRT, salvage ADT was administered for PSA failure.8 Salvage RT and/or ADT was initiated within 3 months after documented PSA recurrence and always before symptomatic or radiographic progression.
Prostate needle biopsy specimens and prostatectomy specimens underwent review by a pathologist with expertise in genitourinary pathology at each center. This study was approved by the Ethik-Kommission der Ärztekamme institutional review board in Hamburg, Germany, and by the IntegReview institutional review board in Austin, Texas. In accord with federal and institutional guidelines, all men before study entry signed an institutional review board–approved, protocol-specific informed consent form permitting collection of deidentified patient data at baseline and follow-up, which were entered into a secure, password-protected database for subsequent outcome analysis.
Follow-up started on the day of prostate EBRT or RP and concluded on October 27, 2017, or the date of death, whichever came first; no patient was lost to follow-up. During follow-up, patients had a PSA test and rectal examination and were seen every 3 months for 1 year, every 6 months for an additional 4 years, and then annually thereafter. To record a death from prostate cancer, castrate-resistant metastatic prostate cancer based on a rising PSA level in the setting of a testosterone level less than 20 ng/dL before death needed to be confirmed (to convert testosterone level to nanomoles per liter, multiply by 0.0347).
The baseline group for the comparisons of the distribution of clinical characteristics and prostatectomy characteristics were MaxRT and RP alone, respectively. A Mantel-Haenszel χ2 metric9 was used to compare the distributions of categorical covariates; in the case of a small sample size, a Fisher exact test10 was used. For continuous covariates, such as serum PSA level, age, year of treatment, and percentage of positive lymph nodes, medians and their distributions were compared using a Wilcoxon 2-sample test.11 The distribution of the follow-up times was calculated using the method by Kaplan and Meier,12 and comparisons of these distributions across men treated using RP-based vs MaxRT therapy were made using the generalized Wilcoxon test for comparing 2 censored samples.13,14
Univariable and multivariable regression using methods by Cox15 and by Fine and Gray16 were used to evaluate whether PCSM risk and ACM risk, respectively, were significantly associated with treatment, adjusting for year of treatment, treatment propensity score (PS),17 and salvage therapy. Time zero was defined as the date of RP or the first day of MaxRT. The continuous covariate year of treatment was included in the model to adjust for any potential changes in patterns of care over time. The treatment PS was included in the model to ensure that the distributions of the baseline clinical characteristics between all RP-containing vs MaxRT treatments were balanced. The outcome for the PS was the treatment received (RP containing vs MaxRT), and the covariates included in the PS were prostate cancer prognostic factors, including the baseline PSA level (continuous), biopsy Gleason score (4 + 5 vs 5 + 4 vs 10), and American Joint Commission on Cancer (AJCC) tumor category (T3,4 vs T2 vs T1c), and patient factors, including age (>70 vs ≤70 years) and cardiometabolic comorbidity (minimal or moderate vs none) using the Adult Comorbidity Evaluation 27 (ACE-27) metric.18 Each patient had a detailed history and physical examination at the time of prostate cancer diagnosis, on which the ACE-27 score was based and assigned retrospectively at the time of the study analysis. Cardiometabolic comorbidity was categorized as moderate if the patient had a history of congestive heart failure and/or an acute myocardial infarction more than 6 months before the diagnosis of prostate cancer. No patient was reported to have had a history of congestive heart failure or an acute myocardial infarction within 6 months of the diagnosis of prostate cancer. Of the 639 men in the study cohort, 533 (83.4%) had no comorbidity, 101 (15.8%) had minimal comorbidity, and 5 (0.8%) had moderate comorbidity. Time-dependent treatment covariates15 for treatment with MaxRP were used to assess PCSM and ACM risk after these treatments, as well as RP alone compared with treatment with MaxRT. An adjustment for the use of salvage RT and/or ADT using time-dependent covariates was also included in the models. For the models by Fine and Gray16 and by Cox,15 an event was defined as prostate cancer and any death, respectively. The assumptions of both the model by Cox15 and the model by Fine and Gray16 were tested, and no evidence was found that these assumptions were violated. Unadjusted and adjusted hazard ratios (AHRs) for PCSM and ACM, with associated 95% CIs and P values, were calculated for each covariate.
Given the AHRs and associated 95% CIs for treatment with MaxRP compared with MaxRT for the end points of PCSM and ACM, the plausibility index defines the likelihood that the true AHR equals 1.00, reflecting equivalence. The plausibility index was calculated for each treatment compared with MaxRT. The plausibility index can have a value between 0 and 1, corresponding to the least and the most likely chance, respectively, that the risk of PCSM and/or ACM between 2 treatments is equivalent (eMethods and eFigure in the Supplement).
Estimates of PCSM (cumulative incidence) and ACM (1 minus Kaplan-Meier estimates of overall survival) after the day of MaxRT initiation or RP stratified by time-dependent treatment groups were calculated using the extended Kaplan-Meier method with time-dependent covariates19 and adjusting for the year of treatment and the treatment PS.20 A 2-sided P < .05 was considered statistically significant, and a Bonferroni adjustment21 was applied for multiple comparisons (P < .05 divided by 4, or P < .01). R (version 3.4.4; R Foundation for Statistical Computing) was used for all calculations pertaining to cumulative incidence functions and Kaplan-Meier estimates with time-dependent treatment covariates. Version 3.4.1 of R was used to calculate the plausibility index. SAS (version 9.4; SAS Institute Inc) was used for all other calculations.
As summarized in Table 1, men who underwent RP plus adjuvant RT had a more favorable baseline biopsy Gleason score and AJCC tumor category compared with men who underwent MaxRT. They also had a significantly shorter median follow-up after treatment (3.87 years [IQR, 3.04-4.88 years] vs 5.51 years [IQR, 2.19-6.95 years], P = .03). However, as summarized in Table 2, after adjustment using the treatment PS, there were no significant differences in the distribution of the baseline clinical characteristics among men undergoing RP- vs RT-based treatments. Men who underwent RP were significantly more likely to have a more favorable prostatectomy AJCC tumor category and percentage of positive lymph nodes compared with men who underwent RP with adjuvant ADT or MaxRP.
After median follow-ups of 5.51 years (IQR, 2.19-6.95 years) among the 80 men treated with MaxRT and 4.78 years (IQR, 4.01-6.05 years) among the 559 men treated with RP-containing treatments, 161 men had died, 106 (65.8%) from prostate cancer. As summarized in Table 3, the risk of PCSM and ACM was increased in men undergoing RP compared with MaxRT (2.80; 95% CI, 1.26-6.22; P = .01 for PCSM and 1.65; 95% CI, 0.94-2.91; P = .08 for ACM). However, there was no significant difference in these risks when comparing men who underwent RP plus adjuvant RT (0.52; 95% CI, 0.14-1.98; P = .34 for PCSM and 0.70; 95% CI, 0.31-1.57; P = .39 for ACM) or MaxRP (1.33; 95% CI, 0.49-3.64; P = .58 for PCSM and 0.80; 95% CI, 0.36-1.81; P = .60 for ACM) vs MaxRT. Nevertheless, men who underwent RP plus adjuvant ADT had a significantly increased risk of PCSM and ACM compared with men who underwent MaxRT (3.15; 95% CI, 1.32-7.55; P = .01 for PCSM and 2.33; 95% CI, 1.23-4.42; P = .01 for ACM).
As summarized in Table 4, the plausibility index for equivalence was highest for the treatment comparison of MaxRP with MaxRT, being 76.75% for the end point of the risk of PCSM and 77.97% for the end point of the risk of ACM. For all other treatment comparisons, these respective values ranged from 4.75% to 58.24% and from 4.62% to 62.32%.
As shown in the Figure, estimates of PCSM and ACM were lower for men treated with MaxRT compared with RP or RP plus adjuvant ADT. These estimates were not significantly different in men undergoing MaxRT, RP plus adjuvant RT, or MaxRP. Five-year PCSM estimates were 2.22% (95% CI, 0.91%-5.37%) for men undergoing MaxRT, 21.89% (95% CI, 17.07%-27.82%) for men undergoing RP, 3.93% (95% CI, 1.35%-11.19%) for men undergoing RP plus adjuvant RT, 9.83% (95% CI, 3.82%-24.02%) for men undergoing MaxRP, and 27.04% (95% CI, 20.39-35.32) for men undergoing RP plus adjuvant ADT. Likewise, these estimates for ACM were 6.79% (95% CI, 4.40%-10.40%) for men undergoing MaxRT, 26.55% (95% CI, 22.02%-34.43%) for men undergoing RP, 12.26% (95% CI, 6.58%-22.20%) for men undergoing RP plus adjuvant RT, 15.85% (95% CI, 8.27%-29.19%) for men undergoing MaxRP, and 36.88% (95% CI, 28.53%-44.76%) for men undergoing RP plus adjuvant ADT.
In this study, we first validated the prior finding3 of a significantly reduced PCSM risk in men with biopsy Gleason score 9-10 prostate cancer when treated using MaxRT compared with RP but with a median ADT duration of 6 months as opposed to 12 months. Specifically, the point estimate of the AHR for the risk of PCSM comparing MaxRT with RP in the present study was 0.36 (1 divided by 2.80), which is similar to the previous report3 of 0.38. While prospective validation is needed in a randomized clinical trial, this result provides evidence to support that, in the setting of EBRT and brachytherapy, adding 6 months compared with 12 months of ADT may provide a similar PCSM risk reduction compared with RP.
Second, when assessing the likelihood of equivalence in the risk of PCSM and ACM after treatment using the plausibility index, men undergoing MaxRP had the highest likelihood of achieving equivalent risk of PCSM and ACM compared with men undergoing MaxRT. Lending further support to this result, we observed that, despite having more adverse postoperative prostate cancer prognostic factors compared with men undergoing RP, who fared worse than men undergoing MaxRT, men undergoing MaxRP did not have an increased risk of PCSM or ACM when compared with men undergoing MaxRT, whereas men who underwent RP plus adjuvant ADT without EBRT did. The clinical significance of these observations is that they provide evidence to support the importance of adding both adjuvant EBRT and ADT after RP in men with biopsy Gleason score 9-10 prostate cancer to reduce the risk of PCSM and ACM so that their outcomes may become comparable to those of men undergoing MaxRT.
Some points require further discussion. First, while not significant, men undergoing RP plus adjuvant RT had an AHR for PCSM and ACM risk less than 1.0, which can be explained by the more favorable baseline distribution of prostate cancer prognostic factors in men undergoing RP plus adjuvant RT vs MaxRT, in addition to the significantly shorter median follow-up of 3.87 vs 5.51 years (P = .03), providing less time to observe deaths. Moreover, when treatment with RP plus adjuvant RT was evaluated for possible equivalence with MaxRT using the plausibility index, the values were 58.24% and 62.32% for the risk of PCSM and ACM, respectively, which were lower compared with the respective values of 76.75% and 77.97% for men treated with MaxRP.
Second, while a prior study22 has shown that at least 75% of men undergoing RP for biopsy Gleason score 9-10 prostate cancer will have at least 1 adverse pathologic factor at RP (ie, extraprostatic extension, seminal vesicle invasion, R1, or N1) and thus have an indication for adjuvant EBRT and/or ADT, only 33.5% (187 of 559) of men received adjuvant EBRT, ADT, or both in the present study. The reason for the low use of adjuvant EBRT in the post-RP setting arises from the concerns regarding overtreatment in the current era in which PSA monitoring after RP is routinely practiced and early salvage EBRT at the time of PSA failure can occur. Specifically, the 3 randomized adjuvant EBRT trials4-6 were conducted during the pre-PSA and early PSA era. As a result, the observation arm of those studies started salvage EBRT later then would be practiced today, appropriately calling into question whether the earlier use of salvage EBRT would be as effective as adjuvant EBRT. This observation led to the formation of 2 randomized clinical trials (Radiotherapy–Adjuvant Versus Early Salvage [RAVES; NCT00860652] and Radiation Therapy and Androgen Deprivation Therapy in Treating Patients Who Have Undergone Surgery for Prostate Cancer [RADICALS; NCT00541047]) that together will help to determine the influence of timing (ie, adjuvant or early salvage), type (ie, EBRT, ADT, or both), and duration of ADT (none, 6 months, or 2 years) on the long-term end points, including recurrence-free and prostate cancer–specific survival. Based on the Surveillance, Epidemiology, and End Results database, only 5.2% of men have prostatectomy Gleason score 9-10 prostate cancer after RP,23 and RAVES (NCT00860652) and RADICALS (NCT00541047) did not stratify men by prostatectomy Gleason score 9-10 vs 8 or less prostate cancer before randomization; as a result, any benefit in the prostatectomy Gleason score 9-10 prostate cancer subset to adjuvant vs early salvage may be diluted by the more common lower-risk subsets of prostatectomy Gleason score 7 and 8 prostate cancer and not observed. This fact emphasizes the importance of our results, which provide the only available evidence to date to support the contention that the use of both adjuvant EBRT and ADT in men with biopsy Gleason score 9-10 prostate cancer may reduce the risk of PCSM to that seen after MaxRT. Moreover, this potential reduction in PCSM risk is accompanied by minimal, if any, risk of overtreatment given the high rate of PSA failure, being at least 80% by 15 years after RP19 and which necessitates the subsequent use of salvage EBRT and ADT.24,25
The strength of our study is the adjustment for the use of adjuvant and salvage therapies using time dependent covariates enabling a comparison of Max RT with Max RP with respect to the endpoints of PCSM and ACM-risk. A limitation is the lack of randomization.
Given no ongoing or planned randomized equivalence trials, to our knowledge, comparing PCSM and ACM after MaxRP vs MaxRT for men with biopsy Gleason score 9-10 prostate cancer, our results herein are important. They provide the only available evidence to date to support that it is plausible that treatment with MaxRP or MaxRT can lead to equivalent risk of PCSM and ACM in men with biopsy Gleason score 9-10 prostate cancer.
Accepted for Publication: August 10, 2018.
Corresponding Author: Anthony V. D’Amico, MD, PhD, Department of Radiation Oncology, Dana-Farber Cancer Institute, Brigham and Women’s Hospital, 75 Francis St, Boston, MA 02115 (firstname.lastname@example.org).
Published Online: November 15, 2018. doi:10.1001/jamaoncol.2018.4836
Author Contributions: Dr D’Amico had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Tilki, Huland, D’Amico.
Acquisition, analysis, or interpretation of data: Tilki, Chen, Wu, Graefen, Braccioforte, Moran, D’Amico.
Drafting of the manuscript: Tilki, Chen, D’Amico.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Chen, Wu, Braccioforte, Moran.
Obtained funding: Braccioforte.
Administrative, technical, or material support: Braccioforte, D’Amico.
Supervision: Tilki, Huland, Graefen, Braccioforte, D’Amico.
Conflict of Interest Disclosures: None reported.
Create a personal account or sign in to: