Androgen Deprivation Therapy Use in the Setting of High-dose Radiation Therapy and the Risk of Prostate Cancer–Specific Mortality Stratified by the Extent of Competing Mortality

Androgen Deprivation Therapy Use in the Setting of High-dose Radiation Therapy and the Risk of Prostate Cancer–Specific Mortality Stratified by the Extent of Competing Mortality

Accepted Manuscript Androgen Deprivation Therapy Use in the Setting of High-Dose Radiation Therapy and the Risk of Prostate Cancer-Specific Mortality ...

2MB Sizes 0 Downloads 1 Views

Accepted Manuscript Androgen Deprivation Therapy Use in the Setting of High-Dose Radiation Therapy and the Risk of Prostate Cancer-Specific Mortality Stratified by the Extent of Competing Mortality Brent S. Rose, M.D., Ming-Hui Chen, Ph.D., Jing Wu, B.S., Michelle H. Braccioforte, M.P.H, Brian J. Moran, M.D., Daniel E. Doseretz, M.D., Michael J. Katin, M.D., Rudolf H. Ross, B.A., Sharon A. Salenius, M.A., Anthony V. D’Amico, M.D. Ph.D. PII:

S0360-3016(16)33099-1

DOI:

10.1016/j.ijrobp.2016.08.014

Reference:

ROB 23761

To appear in:

International Journal of Radiation Oncology • Biology • Physics

Received Date: 22 May 2016 Revised Date:

27 July 2016

Accepted Date: 12 August 2016

Please cite this article as: Rose BS, Chen M-H, Wu J, Braccioforte MH, Moran BJ, Doseretz DE, Katin MJ, Ross RH, Salenius SA, D’Amico AV, Androgen Deprivation Therapy Use in the Setting of HighDose Radiation Therapy and the Risk of Prostate Cancer-Specific Mortality Stratified by the Extent of Competing Mortality, International Journal of Radiation Oncology • Biology • Physics (2016), doi: 10.1016/j.ijrobp.2016.08.014. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

ANDROGEN DEPRIVATION THERAPY USE IN THE SETTING OF HIGH-DOSE RADIATION THERAPY AND THE RISK OF PROSTATE CANCER-SPECIFIC

RI PT

MORTALITY STRATIFIED BY THE EXTENT OF COMPETING MORTALITY

Brent S. Rose, M.D.1, Ming-Hui Chen, Ph.D.2, Jing Wu, B.S.2, Michelle H. Braccioforte, M.P.H3, Brian J. Moran, M.D.3, Daniel E. Doseretz M.D.4, Michael J. Katin, M.D.4, Rudolf

Harvard Radiation Oncology Program, Brigham and Women's Hospital/Dana-Farber

Cancer Institute, Boston, MA, USA

M AN U

1

SC

H. Ross, B.A.4, Sharon A. Salenius, M.A.4, and Anthony V. D’Amico, M.D. Ph.D.5

Department of Statistics, University of Connecticut, Storrs, CT, USA

3

Prostate Cancer Foundation of Chicago, Westmont, Illinois, USA

4

21st Century Oncology, Inc., Fort Myers, Florida, USA

5

Department of Radiation Oncology, Brigham and Women's Hospital/Dana-Farber

TE D

2

EP

Cancer Institute, Boston, MA, USA

Abstract word count: 249

AC C

Text word count: 3,199 Text pages: 18 Tables: 2

Figures: 1a and 1b

Corresponding Author: Brent S. Rose, M.D. 1

ACCEPTED MANUSCRIPT

Department of Radiation Oncology Brigham and Women's Hospital/Dana-Farber Cancer Institute 75 Francis St. Boston, MA, USA

RI PT

Phone: 617-732-8821, Fax: 617-264-5242

SC

Running Title: ADT Use and PCSM risk

Keywords: Prostate cancer, androgen deprivation therapy, radiation therapy,

M AN U

competing mortality

Conflict of Interest and Funding Declaration: The authors declare that there are no

AC C

EP

TE D

conflicts of interest. This study was not supported by any specific grant funding.

2

ACCEPTED MANUSCRIPT

Summary: Adding androgen deprivation therapy (ADT) to high-dose radiation therapy (RT) appears to be associated with decreased prostate cancer-specific mortality (PCSM) risk in men with low but not high competing mortality (CM). These data should

RI PT

serve to heighten awareness about the importance of considering competing risks when determining whether or not to add ADT to RT to treat older men with intermediate or

AC C

EP

TE D

M AN U

SC

high-risk PC.

1

ACCEPTED MANUSCRIPT

ANDROGEN DEPRIVATION THERAPY USE IN THE SETTING OF HIGH-DOSE RADIATION THERAPY AND THE RISK OF PROSTATE CANCER-SPECIFIC

RI PT

MORTALITY STRATIFIED BY THE EXTENT OF COMPETING MORTALITY

SC

Running Title: ADT Use and PCSM risk

Keywords: Prostate cancer, androgen deprivation therapy, radiation therapy,

M AN U

competing mortality

Conflict of Interest and Funding Declaration: The authors declare that there are no

AC C

EP

TE D

conflicts of interest. This study was not supported by any specific grant funding.

1

ACCEPTED MANUSCRIPT

ABSTRACT PURPOSE: The addition of androgen deprivation therapy (ADT) to radiation therapy (RT) is standard of care for men with intermediate and high-risk prostate cancer (PC).

RI PT

However, whether competing mortality impacts the ability of ADT to improve survival remains unanswered.

METHODS AND MATERIALS: We calculated a competing mortality (CM) risk score

SC

using a Fine-Gray semi-parametric model including age and cardiometabolic

comorbidities from a cohort of 17,669 men treated with high-dose RT with or without

M AN U

supplemental ADT for non-metastatic PC. Fine and Gray competing risk regression was used to assess whether ADT reduced the risk of prostate cancer-specific mortality (PCSM) for men with low vs. high risk of CM amongst the 4,550 patients within the intermediate and high-risk cohort adjusted for established PC prognostic factors, year of

TE D

treatment, site, and ADT treatment propensity score.

RESULTS: After a median follow-up of 8.4 years 1,065 men died; 89 (8.36%) from PC. Among men with a low CM score, ADT use was associated with a significant reduction

EP

in the risk of PCSM (adjusted HR (AHR): 0.35, 95% CI: 0.14 to 0.87, p = 0.02) but not in men with a high CM score (AHR: 1.33, 95% CI: 0.77 to 2.30, p = 0.30).

AC C

CONCLUSIONS: Adding ADT to high-dose RT appears to be associated with decreased PCSM-risk in men with low but not high CM. These data should serve to heighten awareness about the importance of considering competing risks when determining whether or not to add ADT to RT to treat older men with intermediate or high-risk PC.

2

ACCEPTED MANUSCRIPT

INTRODUCTION

The addition of androgen deprivation therapy (ADT) to radiation therapy (RT) is a

RI PT

standard of care for men with intermediate and high-risk prostate cancer (PC) not

undergoing prostatectomy based on the results of multiple randomized trials (1-6).

However, recently updated information from one of the randomized trials suggested that

SC

ADT use was only associated with reduced risk of all-cause mortality (ACM) for men at low risk of competing mortality (CM) (1). For men with high risk of CM due to moderate

M AN U

to severe comorbidity assessed using the Adult Comorbidity Evaluation (ACE)-27 metric, ADT was associated with an increased risk of ACM.

The optimal method to stratify men according to their risk of CM remains to be defined. Multiple studies have shown that age and comorbidities are the primary

TE D

determinants of CM (7-9). To that end a multivariable model that combines age and comorbidity may improve the stratification of CM risk in men with PC and facilitate investigation of treatment effect heterogeneity in men with low and high risk of CM (10)

EP

which can enable a personalized approach to treatment selection. Therefore, we used a large prospectively assembled data base containing men

AC C

treated definitively with high dose radiation using brachytherapy with or without ADT for PC to determine the optimal combination of clinical factors that enable a calculation of CM risk. We then investigated the impact of ADT use on the risk of prostate cancerspecific mortality (PCSM) for men with intermediate or high-risk PC and low or high levels of CM risk adjusting for established PC prognostic factors, year of treatment and site as well as an ADT treatment propensity score.

3

ACCEPTED MANUSCRIPT

METHODS

RI PT

Patient Population and Treatment

We studied a prospectively assembled cohort of 17,669 men with localized or locally advanced PC who were consecutively treated with brachytherapy with or without

SC

ADT between May 8, 1991, and June 1, 2013 (Table 1). The study was performed with the approval of independent institutional review boards. Each patient signed an

M AN U

informed consent at the time of initial consultation allowing their de-identified clinical and prostate cancer-related information to be collected and entered into a secure, password-protected database for subsequent outcomes analysis. Technical details of the brachytherapy have been previously described (11).

TE D

Baseline and outcomes data were prospectively collected, and participants were categorized into risk groups according to the NCCN guidelines (12). Patients who received any external beam radiotherapy (EBRT) as a component of their treatment

EP

were excluded.

The total cohort was first used to develop the CM risk score. We then restricted

AC C

our attention to the cohort of 4,550 men with intermediate or high-risk PC with a minimum of two years of follow-up to investigate whether ADT was beneficial in men with low vs. high CM risk. Androgen deprivation therapy included a gonadotropinreleasing hormone agonist with or without an antiandrogen for a median of 4 months (IQR: 3 months, 4 months) and 4 months (IQR: 3 months, 6 months) in men with intermediate and high risk PC respectively.

4

ACCEPTED MANUSCRIPT

Follow-up and Determination of Cause of Death The primary endpoint was PCSM, defined as time to death from prostate cancer.

RI PT

CM was defined as death from any other cause and was considered a competing event. Follow-up began on the date of prostate brachytherapy and continued to the date of death. Follow-up included serial PSA measurements followed by digital rectal

SC

examination every 3 months for 2 years, every 6 months for an additional 3 years, then annually thereafter. To be considered to have died of PC, men must have had

M AN U

radiographically confirmed metastases and a rising PSA despite salvage ADT and/or cytotoxic chemotherapy. The data for all surviving patients was administratively censored at the last data update, June 1, 2013. Median follow-up for the total cohort was 6.8 years (Inter-quartile range (IQR): 4.0 years, 10.0 years). Median follow-up for

Statistical Methods

TE D

the intermediate and high-risk cohort was 8.4 years (IQR: 6.2 years, 11.0 years).

EP

Calculating the Competing Mortality Score To estimate the effects of age and comorbidity on the risk of CM, we used the

AC C

Fine-Gray semi-parametric model for subdistribution hazards (13) for the total cohort of 17,669 patients. The CM risk score was constructed by multiplying the model parameter estimates and corresponding data vectors. Age was analyzed as a continuous variable. Comorbidity was analyzed as categorical variable according to the presence of 0, 1, or 2 or more comorbid disease. Comorbidities that were recorded included diabetes,

5

ACCEPTED MANUSCRIPT

coronary artery disease, prior myocardial infarction or congestive heart failure. The CM risk score was calculated according to the following equation:

RI PT

CM score = 0.090 x Age + 0.174 x only one comorbidity + 0.374 x two or more comorbidities

SC

We then calculated the CM score for all 4,550 patients within the intermediate and highrisk cohort. The minimum, median, and maximum CM scores were 3.703, 6.557, and

M AN U

8.285, respectively. Patients with CM score above the median CM score were categorized as being at high-risk of CM. Patients > 72.8 years old without comorbidities, > 70.9 years old with 1 comorbidity, and > 68.6 years old with 2 or more

TE D

comorbidities were included in the high CM cohort.

Comparison of Distribution of Patient Factors Stratified by Competing Mortality Group

EP

Using descriptive statistics, patient and tumor characteristics were enumerated and displayed stratified by the CM group (Table 2) for the 4,550 patients within the

AC C

intermediate and high-risk cohort. Categorical variables age (≤60, 61-70, 71-80, >80), number of comorbidities (0, 1, ≥ 2), Gleason category (≤6, 3+4, 4+3, 8-10), T-stage, prostate specific antigen (PSA) category (≤ 4, > 4 to 10, > 10 to 20, > 20) NCCN risk group, treatment setting and ADT use were compared by the Mantel Haenszel statistic. Continuous variables including age, PSA, and year of treatment were compared by the Wilcoxon rank sum test.

6

ACCEPTED MANUSCRIPT

Competing Risk Regression for Prostate Cancer-Specific Mortality Fine and Gray competing risk regression was used to assess whether ADT

RI PT

reduced the risk of PCSM for men with low vs. high risk of CM amongst the 4,550

patients within the intermediate and high-risk cohort. An interaction term with ADT use and high vs. low CM group was used to determine if the effect of ADT varied according

SC

to CM group. The model was adjusted for Gleason score, T-stage, PSA, year of

treatment, propensity score for ADT use which was obtained using the covariates of

M AN U

age, PSA, Gleason score, and T stage, and treatment setting (reference: Institution 1). Gleason score was categorized as ≤ 6, 7, or ≥ 8. T-stage was categorized as T1, T2, or T3. PSA (log transformed) and propensity score were analyzed as continuous variables. A propensity score for ADT use was included in the model to account for

TE D

potential biases in treatment effect arising from non-random selection of patients for ADT use. The methodology for the propensity analysis has been previously described (xx). Proportional hazards model assumptions were tested and assessed via various

EP

residuals. No evidence was found to suggest these assumptions were violated. Univariable and multivariable hazard ratios and associated 95% confidence intervals

AC C

(CI) and p-values were calculated for each covariate. Adjusted hazard ratios for the treatment effect in the low and high CM groups were derived from a single, interaction model with the non-treated group as the reference group within each CM cohort. All statistical tests were two-sided, and p < 0.05 was considered significant.

7

ACCEPTED MANUSCRIPT

Estimates of Prostate Cancer-Specific Mortality for Men Treated with and without Androgen Deprivation Therapy Stratified by Competing Mortality Group Eight-year point estimates of the cumulative incidence of PCSM with associated

RI PT

95% CI were calculated for men treated with and without ADT stratified by CM group. For the purpose of illustration, the cumulative incidence of PCSM plots are covariateadjusted to reflect the results in the multivariable model for men treated with and without

SC

ADT and are displayed graphically in the low CM (Figure 1a) and high CM groups

(Figure 1b). A Bonferroni correction was applied to adjust for multiple comparisons.

M AN U

Given that the effect of ADT was compared in two CM groups, there were a total of two comparisons. Therefore, a Bonferroni-corrected significance level of 0.05/2 = 0.025 was considered significant. SAS version 9.4 was used for all analyses except for eight-year point estimates of the cumulative incidence of PCSM with associated 95% CI which

EP

RESULTS

TE D

used R version 3.0.1.

Comparison of Distribution of Patient Factors Stratified by Competing Mortality

AC C

Group

As shown in table 1, men with a high as compared to low CM score were older

(median age: 76.4 years vs. 66.9 years, p < 0.001), with 18.3% versus 0% age 80 or older (p < 0.001), more likely to have 2 or more cardiometabolic comorbidities (6.5% vs. 2.4%, p < 0.001), be treated at Institution 2 (27.0% vs. 15.6%, p < 0.001), and have more aggressive prostate cancer. Specifically, Gleason score 8-10 (10.4% vs. 5.0%, p <

8

ACCEPTED MANUSCRIPT

0.001), PSA > 20 ng/ml (9.1% vs. 6.3%, p <0.001) and T2 (44.2% vs. 31.2%, p < 0.001) or T3 (1.5% vs. 0.4%) PC, and therefore high-risk PC (19.1% vs. 11.2%, p < 0.001) was more prevalent amongst men with a high as compared to low CM score. As a result of

RI PT

the more advanced PC indices these men were also more likely to receive ADT (42.9% vs. 29.8%, p < 0.001).

SC

Competing Risk Regression for Prostate Cancer-Specific Mortality

After a median follow up of 8.4 years (interquartile range: 6.2 years to 11.0

M AN U

years), 1,065 men died; 317 (29.8%) and 748 (70.2%) of these deaths were in men with low versus high CM scores, respectively. Eighty-nine deaths (8.36% of total) were from PC; 33 (10.4%) and 56 (7.5%) of men with low and high CM scores respectively. After adjustment for established PC prognostic factors, year of treatment and site as well as

TE D

an ADT treatment propensity score, among men with a low CM score there was a significant reduction in the risk of PCSM if ADT was administered (adjusted HR (AHR): 0.35, 95% CI: 0.14 to 0.87, p = 0.02) as shown in table 2. However, ADT use was not

EP

associated with a significant reduction in PCSM in men with a high CM score (AHR: 1.33, 95% CI: 0.77 to 2.30, p = 0.30). There was a statistically significant interaction for

AC C

the association between ADT use and PCSM amongst men with high versus low CM score (p = 0.01) indicating evidence of heterogeneity of ADT treatment effect according to the risk of CM.

Estimates of Prostate Cancer-Specific Mortality for Men Treated with and without Androgen Deprivation Therapy Stratified by Competing Mortality Group

9

ACCEPTED MANUSCRIPT

As shown in figure 1a, cumulative incidence estimates of PCSM were significantly lower in men with a low CM score if they received ADT as compared to if they did not (p = 0.02). In contrast, these estimates were not significantly different when

RI PT

stratified by ADT use (p = 0.34) in men with a high CM score (figure 1b). Specifically, 8year point estimates were 0.79% (95% CI: 0.30% to 1.80%) versus 1.29% (95% CI: 0.80% to 1.99%) and 2.44% (95% CI: 1.24% to 4.32%) versus 2.49% (95% CI: 1.65%

SC

to 3.62%) for men who received ADT versus not amongst those with a low versus high

M AN U

CM score, respectively.

DISCUSSION

We observed that the risk of PCSM significantly decreased with the addition of

TE D

ADT in the setting of high-dose RT for men with intermediate or high-risk PC who had a low but not a high CM score. The clinical significance of this finding is that adding ADT to high-dose RT may only reduce the risk of death from PC in the setting of low

EP

competing risks, raising the question of whether competing risks should be taken into account when considering the optimal management of intermediate or high risk PC in

AC C

older men or those with significant comorbidities. In our study we found 70.2% (748 / 1,065 deaths) of all deaths occurred in men

with a high CM score and that 92.5% (692 / 748 deaths) of these deaths were from causes other than prostate cancer. Therefore, we hypothesize that younger patients and those with few comorbidities may benefit from ADT because they have a lower risk of mortality from competing causes of death. In contrast, elderly men and those with

10

ACCEPTED MANUSCRIPT

more comorbidities may not benefit from ADT because of the high competing risk of deaths from non-PC causes. These competing causes of death are not necessarily related to ADT use, but more likely related to advanced age or cardiometabolic

RI PT

comorbidity.

Fortunately, RTOG 0815 (NCT00936390) is a randomized trial enrolling men with intermediate risk PC and randomizing them to high dose RT (79.2 Gy or 45 Gy and

SC

brachytherapy boost) with or without 6 months of ADT and includes a pre-randomization stratification using a validated metric of comorbidity (ACE-27) (14). Therefore, that

M AN U

study will be able to validate or refute the hypothesis that ADT use is associated with reduction in PCSM in men at low but not high risk of CM. If validated, reconsidering ADT use in the management of intermediate-risk PC in older men and those who have significant comorbidity would be warranted.

TE D

Given that our study utilized brachytherapy which is not one of the standard arms of RTOG 0815 the following considerations are important to note. Specifically, one of the randomized treatment arms on RTOG 0815 permits 79.2 Gy of EBRT with or

EP

without 6 months of ADT. Given that PSA nadir following prostate brachytherapy is typically lower than that following EBRT (15) likely related to the higher biologically

AC C

equivalent dose (BED) (16) and that both decreasing PSA nadir and increasing BED have been shown to be associated with a decreased risk of post RT PSA failure (17,18) suggests that brachytherapy monotherapy as performed in our study yields intraprostatic doses that are no less and possibly greater than the RT doses delivered using 79.2 Gy of EBRT. However, it is important to note that there is no assurance that the dosimetric margin around the prostate using brachytherapy would be comparable to that

11

ACCEPTED MANUSCRIPT

using EBRT which is important in men with prostate cancer in our study that have a significant chance of having occult extracapsular extension (ECE) despite having clinical category T1c, 2 prostate cancer by digital rectal examination. Therefore, it is

RI PT

possible that the addition of ADT to brachytherapy may be associated with a reduction in the risk of death from prostate cancer due to sterilization of subclinical ECE by ADT that is not being addressed with brachytherapy and that may be addressed with the

SC

margins provided using EBRT. Therefore, it is possible that RTOG 0815 may find no reduction in death from PC for men randomized to EBRT, brachytherapy boost and 6

M AN U

months of ADT as compared to EBRT and brachytherapy boost given this consideration. However, the lack of association between the risk of death from PC and ADT use in men with high CM scores cannot be explained by a potentially narrower dosimetric margin when using brachytherapy as compared to EBRT. It is more likely

TE D

that these men do not live long enough given the high CM score to realize any benefit in terms of PC death risk reduction when adding ADT to brachytherapy. Several points require further discussion. First, while the CM score calculated in

EP

this study used age and specific cardiometabolic comorbidities including diabetes, coronary artery disease, prior myocardial infarction or congestive heart failure, it did not

AC C

include all possible comorbidities. However, it has been established that cardiometabolic comorbidities are the major non-cancer competing risk in men of PCbearing age and, therefore, our scores should closely approximate the CM score that would have been obtained if all comorbidities were used (19). Moreover it has been recently shown that multiple cardiometabolic comorbidities similar to those defined in our study, substantially lower life expectancy in patients with multimorbidity (20). This

12

ACCEPTED MANUSCRIPT

study found that the presence of two or three cardiometabolic comorbidities reduced the life expectancy for an average 60 year old by 12 and 15 years, respectively. Second, the median duration of ADT used in this study was 4 months. Evidence

RI PT

from the Grupo de Investigacion Clinica en Oncologia Radioterapica (GICOR) study (21) showed that a longer course of 28 months as compared to 4 to months of ADT was superior in terms of overall survival in men with high-risk PC in the setting of a median

SC

dose of 78 Gy of external beam RT. Therefore, it is possible that a significant reduction in PCSM would have been observed if a longer duration of ADT had been used in men

M AN U

with a high CM score. However, given the absolute magnitude of the estimates of PCSM in these men approximating only 2.5% after 8 years, it seems more likely that competing risks will overshadow and negate any reduction in PCSM afforded by longer term ADT.

TE D

Third, men with a high CM scores were significantly older than men with low CM scores as reflected in table 1 which can explain the more aggressive PC indices in these men as shown in table 1 given that a follow up study to the Prostate Cancer

EP

Prevention Trial reported that advancing age was associated with the presentation of high-grade PC (22). Specifically they found a 3% increase in the diagnosis of Gleason

AC C

score 7 or higher PC per advancing year. Therefore, it is even more noteworthy that despite the presence of higher-grade PC in men with a high as compared to low CM risk score and where the addition of ADT has been shown to prolong survival (1, 2, 4, 22) that ADT use was not associated with a reduction in the risk of PCSM, further highlighting the importance of competing risk when evaluating outcomes in men with unfavorable-risk and non-metastatic PC.

13

ACCEPTED MANUSCRIPT

Fourth, since all men received RT in our study we cannot comment on whether RT is beneficial in the high CM cohort of patients. It may be that active surveillance may be appropriate for men with very high risk of CM.

RI PT

Despite these considerations, it appears that adding ADT to high-dose RT is only associated with a decreased risk of PCSM in the setting of low but not high CM risk. While awaiting the results of RTOG 0815, these data should serve to heighten

SC

awareness about the importance of considering competing risks when determining

AC C

EP

TE D

those with significant comorbidities.

M AN U

whether or not to add ADT to RT to treat intermediate or high-risk PC in elderly men and

14

ACCEPTED MANUSCRIPT

REFERENCES D'Amico AV, Chen MH, Renshaw A, Loffredo M, Kantoff PW. Long-term Follow-

RI PT

1)

up of a Randomized Trial of Radiation with or without Androgen Deprivation Therapy for Localized Prostate Cancer. JAMA. 2015;314:1291-3.

Denham JW, Steigler A, Lamb DS, et al. Short-term neoadjuvant androgen

SC

2)

deprivation and radiotherapy for locally advanced prostate cancer: 10-year data

3)

M AN U

from the TROG 96.01 randomised trial. Lancet Oncol. 2011;12:451-9. Jones CU, Hunt D, McGowan DG, et al. Radiotherapy and short-term androgen deprivation for localized prostate cancer. N Engl J Med. 2011;365:107-18. 4)

Bolla M, Van Tienhoven G, Warde P, et al. External irradiation with or without

TE D

long-term androgen suppression for prostate cancer with high metastatic risk: 10year results of an EORTC randomised study. Lancet Oncol. 2010;11:1066-73. 5)

Bolla M, de Reijke TM, Van Tienhoven G, et al. Duration of androgen

27.

Horwitz EM, Bae K, Hanks GE, et al. Ten-year follow-up of radiation therapy

AC C

6)

EP

suppression in the treatment of prostate cancer. N Engl J Med. 2009;360:2516-

oncology group protocol 92-02: a phase III trial of the duration of elective

androgen deprivation in locally advanced prostate cancer. J Clin Oncol.

2008;26:2497-504.

15

ACCEPTED MANUSCRIPT

7)

Daskivich TJ, Fan KH, Koyama T, et al. Effect of age, tumor risk, and comorbidity on competing risks for survival in a U.S. population-based cohort of men with prostate cancer. Ann Intern Med. 2013;158:709-17. Briganti A, Spahn M, Joniau S, et al. Impact of age and comorbidities on long-

RI PT

8)

term survival of patients with high-risk prostate cancer treated with radical prostatectomy: a multi-institutional competing-risks analysis. Eur Urol.

9)

SC

2013;63:693-701.

Albertsen PC, Moore DF, Shih W, et al. Impact of comorbidity on survival among

10)

M AN U

men with localized prostate cancer. J Clin Oncol. 2011;29:1335-41 Hayward RA, Kent DM, Vijan S, Hofer TP. Multivariable risk prediction can greatly enhance the statistical power of clinical trial subgroup analysis. BMC Med Res Methodol. 2006;6:18.

Nanda A, Chen M-H, Braccioforte MH, Moran BJ, D’Amico AV. Hormonal therapy

TE D

11)

use for prostate cancer and mortality in men with coronary artery diseaseinduced congestive heart failure or myocardial infarction. JAMA. 2009;302:866-

12)

EP

873.

National Comprehensive Cancer Network. Prostate Cancer, Version 1.2016.

AC C

http://www.nccn.org/professionals/physician_gls/pdf/prostate.pdf. Accessed January 18, 2016.

13)

Fine JP, Gray RJ. A proportional hazards model for the subdistribution of a competing risk. J Am Stat Assoc. 1999;94:496-509.

16

ACCEPTED MANUSCRIPT

14)

Piccirillo JF, Tierney RM, Costas I, Grove L, Spitznagel EL Jr. Prognostic importance of comorbidity in a hospital-based cancer registry. JAMA. 2004;291:2441-7. https://www.auanet.org/education/guidelines/prostate-specific-antigen.cfm.

RI PT

15)

Accessed 3/23/2016. 16)

Nath R, Bice WS, Butler WM et al. AAPM recommendations on dose prescription

SC

and reporting methods for permanent interstitial brachytherapy for prostate cancer: report of Task Group 137. Med Phys. 2009;36:5310-22.

Stock, R.G., Stone, N.N., Cesaretti J.A., Rosentein, B.S. Biologically Effective

M AN U

17)

Dose Values for Prostate Brachytherapy: Effects on PSA Failure and Posttreatment Biopsy Results. Int J Radiat Oncol Biol Phys. 2006;64:527-33. 18)

Zelefsky, M., Kuban, D.A., Levy, L.B., et al. Multi-institutional analysis of long-

TE D

term outcome for stages T1-T2 prostate cancer treated with permanent seed implantation. Int J Radiat Oncol Biol Phys. 2007;67:327-33. Heron M. Deaths: Leading Causes for 2012. Natl Vital Stat Rep. 2015;64:1-93.

20)

Di Angelantonio, Kaptoge, Wormser, et al. Association of Cardiometabolic

EP

19)

Multimorbidity With Mortality. JAMA. 2015;314:52-60 Zapatero A, Guerrero A, Maldonado X, et al. High-dose radiotherapy with short-

AC C

21)

term or long-term androgen deprivation in localised prostate cancer (DART01/05 GICOR): a randomised, controlled, phase 3 trial. Lancet Oncol. 2015;16:320-7.

22)

Thompson IM, Ankerst DP, Chi C, et al. Assessing prostate cancer risk: results

from the Prostate Cancer Prevention Trial. J Natl Cancer Inst. 2006;98:529-34.

17

ACCEPTED MANUSCRIPT

FIGURE LEGENDS

Figure 1: Adjusted cumulative incidence estimates of prostate cancer-specific mortality

RI PT

stratified by treatment with high-dose radiation with or without androgen suppression therapy in men with a (a) low or (b) high competing mortality risk. The y-axis ranges

SC

from 0 to 10%.

1a: p-value = 0.02

AC C

EP

TE D

M AN U

1b: p-value = 0.34

18

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Table 1. Distribution of the clinical characteristics of 17,669 men in the study cohort and comparison of these distributions stratified by competing mortality score for the 4,550 men with intermediate and high-risk Prostate Cancer. Patient, Prostate 17, 669 men in the 4,550 men with intermediate and high-risk Prostate Cancer, and Treatment study cohort Cancer Factors Low CM High CM p-value Patients, no. (%) 17, 669 2,274 2,276 Median Follow-up 6.76 (3.96, 10.00) 8.21 (6.07, 10.72) 8.60 (6.31, 11.35) 0.003 (IQR), years Median age (IQR), 70.63 (65.15, 66.89 (62.45, 76.42 (74.30, <0.001 years 75.19) 69.85) 79.04) Age in years, no. (%) ≤ 60 2,021 (11.44%) 357 (15.70%) 0 (0%) <0.001 61-70 6,233 (35.28%) 1,374 (60.42%) 16 (0.70%) 71-80 8,342 (47.21%) 543 (23.88%) 1,842 (80.93%) > 80 1,073 (6.07%) 0 (0%) 418 (18.37%) Cardiometabolic Comorbidity, no. (%) 0 13,810 (78.16%) 1,855 (81.57%) 1,583 (69.55%) <0.001 1 3,285 (18.59%) 365 (16.05%) 546 (23.99%) ≥2 574 (3.25%) 54 (2.37%) 147 (6.46%) Highest Gleason Score, no. (%) ≤6 11,543 (65.33%) 1,002 (44.06%) 944 (41.48%) <0.001 7 4,818 (27.27%) 1,159 (50.97%) 1,096 (48.15%) 8 to 10 1,308 (7.40%) 113 (4.97%) 236 (10.37%) AJCC T category, no. (%) 11,768 (66.60%) 1,555 (68.38%) 1,238 (54.39%) <0.001 T1 5,618 (31.80%) 710 (31.22%) 1,005 (44.16%) T2 283 (1.60%) 9 (0.40%) 33 (1.45%) T3 Median PSA, ng/mL 6.9 (5.1, 9.9) 10.1 (6.2, 12.60) 10.2 (6.4, 13.37) 0.006 (IQR) PSA in ng/mL, no. (%) ≤4 1,781 (10.08%) 132 (5.80%) 153 (6.72%) 0.12 > 4 to 10 11,595 (65.62%) 989 (43.49%) 936 (41.12%) > 10 to 20 3,381 (19.14%) 1,010 (44.42%) 979 (43.01%) > 20* 912 (5.16%) 143 (6.29%) 208 (9.14%) NCCN Risk Group, no. (%) 8,599 (48.67%) NA NA <0.001 Low 6,916 (39.14%) 2,019 (88.79%) 1,842 (80.93%) Intermediate 2,154 (12.19%) 255 (11.21%) 434 (19.07%) High Year of brachytherapy, 2003 (2000, 2004 (2001, 2002 (2000, 2005) <0.001 median (IQR) 2006) 2006) Treatment Setting

ACCEPTED MANUSCRIPT

Institution 1 Institution 2 ADT use, no. (%)

10,418 (58.96%) 7,251 (41.04%)

1,920 (84.43%) 354 (15.57%)

1,661 (72.98%) 615 (27.02%)

<0.001

AC C

EP

TE D

M AN U

SC

RI PT

No 10,322 (58.42%) 1,595 (70.14%) 1,300 (57.12%) <0.001 Yes 7,347 (41.58%) 679 (29.84%) 976 (42.88%) Abbreviations: No., number; IQR, Interquartile range; AJCC, American Joint Committee on Cancer; CM, competing mortality; PSA, prostate-specific antigen; NCCN, National Comprehensive Cancer Network; ADT, androgen deprivation therapy. *p-value comparing PSA > 20 ng/ml vs. 20 ng/ml or less: < 0.001

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Table 2. Prostate cancer-specific mortality hazard ratios by patient, prostate cancer, and treatment factors. Patient, Prostate Cancer, No. of No. of Univariable Analysis Multivariable Analysis and Treatment Factors Men Prostate HR (95% CI) p-value AHR (95% CI) p-value Cancer Deaths Interaction Terms Low CM ADT use 679 6 0.51 0.13 0.35 0.02 (0.21, 1.23) (0.14, 0.87) No ADT use 1,595 27 1.0 (Ref) 1.0 (Ref) High CM ADT use 976 29 1.51 0.13 1.33 0.30 (0.89, 2.55) (0.77, 2.30) No ADT use 1,300 27 1.0 (Ref) 1.0 (Ref) ADT x CM group 4,550 89 2.98 0.04 3.76 0.01 (1.07, 8.34) (1.31, 10.83) Prostate Cancer Prognostic Factors PSA, ng/ml 4,550 89 1.83 0.005 1.23 0.37 (1.20, 2.78) (0.78, 1.96) AJCC tumor category T1 2,793 38 1.0 (Ref) 1.0 (Ref) T2 1,715 47 1.90 0.004 1.40 0.16 (1.23, 2.92) (0.88, 2.24) T3 42 4 6.67 <0.001 0.55 0.41 (2.39, 18.54) (0.13, 2.33) Highest Gleason Score ≤6 1,946 23 1.0 (Ref) 1.0 (Ref) 7 2,255 40 1.63 0.06 2.19 0.004 (0.97, 2.72) (1.28, 3.74) 8-10 349 26 7.33 <0.001 4.32 <0.001 (4.19, 12.83) (1.96, 9.53) Treatment Setting Institution 1 3,581 70 1.0 (Ref) 1.0 (Ref) Institution 2 969 19 1.23 0.42 0.34 0.003 (0.74, 2.04) (0.17, 0.70) Year of brachytherapy 4,550 89 0.89 <0.001 0.83 <0.001 (0.84, 0.93) (0.76, 0.90) ADT propensity score (%) 4,550 89 1.05 <0.001 1.05 0.01 (1.04, 1.07) (1.01, 1.08) Abbreviations: CM, competing mortality; ADT, androgen deprivation therapy; PSA, prostate specific antigen; AJCC, American Joint Committee on Cancer; No., number; HR, hazard ratio; AHR, adjusted hazard ratio; CI, confidence interval.

10

no ADT use ADT use

9

SC

8 7

M AN U

6 5 4

1 0

0 1

2 3 4

EP

2

TE D

3

5 6

7 8 9 10 11 12 13 14 15

Time (Years) Following Brachytherapy

AC C

Percent Prostate Cancer Specific Mortality

RI PT

ACCEPTED MANUSCRIPT

1595 1595 1595 1465 1419 1332 1129 902 714 538 420 282 200 134 77 679 679 679 629 597 543 466 407 334 246 184 132 83 57 29

Number at Risk

27 11

10

no ADT use ADT use

9

SC

8 7

M AN U

6 5 4

1 0

0 1

2 3 4

EP

2

TE D

3

5 6

7 8 9 10 11 12 13 14 15

Time (Years) Following Brachytherapy

AC C

Percent Prostate Cancer Specific Mortality

RI PT

ACCEPTED MANUSCRIPT

1300 1300 1300 1198 1134 1045 868 691 531 412 317 223 157 85 976 976 976 881 798 715 597 488 368 273 190 129 72 36

Number at Risk

45 14

18 7