URO ONCO

Welcome, this website is intended for all international healthcare professionals in uro-oncology. By clicking the link below you are declaring and confirming that you are a healthcare professional.

You are here

Impact of the Site of Metastases on Survival in Patients with Metastatic Prostate Cancer

European Urology

Comment from Henk van der Poel:
In a large SEER registry population covering more than one-quarter of the US population, 3857 men over 65 years of age were identified with metastasized prostate cancer. More than 90% had bone metastases. Excluding men with prior local treatment, it became apparent that the presence of visceral metastases resulted in a drop in median overall survival of 10 months. Interestingly, men with visceral metastases were less likely to receive docetaxel chemotherapy. When correcting for this imbalance of chemotherapy use, men with visceral metastases did still worse compared to men with only bone or lymph node metastases.
The fact that men with both bone and visceral metastases did worse than men with visceral metastases-only with respect to survival clearly shows that the presence of bone metastases does not confer a more favorable course of disease. Rather metastases volume may be needed to be estimated in order to clearly show that location of metastases impacts upon survival and the worse outcome is not an effect of a larger tumor burden. Median overall survival of men with nodal metastases was almost double that of men with bone metastases.  Nodal metastasized disease is generally of low volume supporting the notion that metastases volume rather than location impacts upon prognosis.

Abstract

Background

Limited data exist on the impact of the site of metastases on survival in patients with stage IV prostate cancer (PCa).

Objective

To investigate the role of metastatic phenotype at presentation on mortality in stage IV PCa.

Design, setting, and participants

Overall, 3857 patients presenting with metastatic PCa between 1991 and 2009, included in the Surveillance Epidemiology and End Results–Medicare database were evaluated.

Outcome measurements and statistic analyses

Overall and cancer-specific survival rates were estimated in the overall population and after stratifying patients according to the metastatic site (lymph node [LN] alone, bone, visceral, or bone plus visceral). Multivariable Cox regression analyses tested the relationship between the site of metastases and survival. All analyses were repeated in a subcohort of patients with a single metastatic site involved.

Results and limitations

Respectively, 2.8%, 80.2%, 6.1%, and 10.9% of patients presented with LN, bone, visceral, and bone plus visceral metastases at diagnosis. Respective median overall survival and cancer-specific survival were 43 mo and 61 mo for LN metastases, 24 mo and 32 mo for bone metastases, 16 mo and 26 mo for visceral metastases, and 14 mo and 19 mo for bone plus visceral metastases (p < 0.001). In multivariable analyses, patients with visceral metastases had a significantly higher risk of overall and cancer-specific mortality versus those with exclusively LN metastases (p < 0.001). The unfavorable impact of visceral metastases persisted in the oligometastatic subgroup. Our study is limited by its retrospective design.

Conclusions

Visceral involvement represents a negative prognostic factor and should be considered as a proxy of more aggressive disease in patients presenting with metastatic PCa. This parameter might indicate the need for additional systemic therapies in these individuals.

Patient summary

Patients with visceral metastases should be considered as affected by more aggressive disease and might benefit from the inclusion in clinical trials evaluating novel molecules.

Take Home Message

The site of metastasis has a significant impact on survival in patients presenting with stage IV prostate cancer. Patients with visceral metastases should be considered as affected by more aggressive disease. These individuals might benefit from the inclusion in clinical trials evaluating the efficacy of novel molecules.

Keywords: Metastatic prostate cancer, Overall survival, Prostate cancer, Androgen deprivation therapy, Site of metastases, Number of metastases.

1. Introduction

Prostate cancer (PCa) represents the most common noncutaneous malignancy occurring in men [1] . Although studies have shown that the proportion of patients with metastatic disease has decreased over the last decades, up to 6.7 per 100 000 individuals still present with distant metastases [2] . Lymph nodes (LNs) and the bone represent the most common locations of metastatic disease [3] . However, numerous patients present with atypical metastases at diagnosis [3] and [4].

The spread of PCa cells to atypical sites of metastases might be the expression of different biologic characteristics (ie, a more aggressive disease that acquires the ability to localize to sites other than LNs and bone) [5] . In addition, visceral involvement might be associated with more severe clinical manifestations. However, limited data exist on the impact of the metastatic site on survival in patients with stage IV PCa [4], [6], [7], [8], [9], [10], and [11]. Indeed, several drawbacks limit the generalizability of previous findings. For example, these observations were obtained evaluating men enrolled in prospective randomized trials, which might represent highly selected patients. In addition, historical cohorts were evaluated. Thus, results obtained in this context might not be applicable to contemporary patients with metastatic PCa. To address this void, we sought to evaluate the role of the metastatic site on survival in a large contemporary cohort of US patients with PCa. Our hypothesis stated that the occurrence of visceral metastases might represent an adverse prognostic factor in these patients.

2. Materials and methods

2.1. Population source

The current study relied on the Surveillance, Epidemiology, and End Results (SEER)–Medicare insurance program-linked database. This database is 98% complete for case ascertainment. The SEER registries covered approximately 28% of the US population with Medicare administrative data. The Medicare insurance program encompasses approximately 97% of Americans aged ≥65 yr.

2.2. Study population

Patients aged ≥65 yr (International Classification of Disease [ICD] for Oncology site code 61.9, histologic code 8140) diagnosed with metastatic PCa between 1991 and 2009 were identified (n = 14 103). Patients receiving definitive treatment aimed at primary PCa, such as radical prostatectomy (n = 418), radiotherapy (n = 2536), and/or brachytherapy (n = 632) were excluded. Because of the lack of radiotherapy site-specific codes, we were not able to differentiate between patients receiving radiotherapy with curative intent or in a palliative setting [12] . We limited this potential bias by excluding from our analyses patients receiving radiotherapy within 6 mo of diagnosis and thus more likely to undergo prostate-directed radiotherapy. Additional exclusions criteria consisted of unknown biopsy Gleason score (n = 1959), unknown site of metastases (n = 4576), unknown survival status (n = 41), and patients diagnosed at autopsy (n = 84). This resulted in a final population of 3857 assessable patients.

2.3. Sites of metastases

The metastatic sites were identified relying on secondary ICD, 9th revision (ICD-9) diagnostic codes: bone and bone marrow (198.5), LNs (196.x), liver (197.7), thorax (including lung [197.0], pleura [197.2], mediastinum [197.1], and other respiratory organs [197.3]), adrenal gland and kidney (198.7 and 198.0), brain and spinal cord (198.3), retroperitoneum and peritoneum (197.6), and digestive system (including large intestine and rectum [197.5], small intestine and duodenum [197.4], and other digestive organs and spleen [197.8]) [3] . For the purpose of analyses, we categorized patients according to the presence of exclusive LN metastases, bone metastases without visceral metastases, visceral metastases, and concomitant bone and visceral metastases. In addition, patients were stratified according to the number of metastatic sites involved (one vs two or more).

2.4. Outcomes

The end point of the study consisted of overall mortality and cancer-specific mortality (CSM), defined using the SEER code for cause of death. The duration of survival was defined as the time interval from PCa diagnosis to the date of death.

2.5. Covariates

For each patient, age, year of diagnosis, race, population density, marital status, education, income, region, and biopsy Gleason score were assigned. The Charlson Comorbidity Index (CCI) was derived from the Medicare claims 1 yr prior to PCa diagnosis using a previously validated algorithm [13] . Data on prostate-specific antigen (PSA) level at diagnosis were available for men diagnosed after the year 2003 (n = 1242). In addition, the use of androgen deprivation therapy (ADT) and chemotherapy was recorded [14] . Data on the type of ADT (gonadotropin-releasing hormone agonist administration or bilateral orchiectomy) were also abstracted [15] .

2.6. Statistical analyses

Means, medians, and interquartile ranges were reported for continuous variables. Frequencies and proportions were reported for categorical variables. The independent t test and chi-square test were used to compare means and proportions, respectively.

Our statistical analyses consisted of four steps. First, Kaplan-Meier curves were used to examine time to overall survival and CSM-free survival in the entire population, and according to sites of metastases and the number of metastatic sites involved. The log-rank test was used to compare mortality rates by patient categories. Second, multivariable Cox regression models were fitted to test the effect of the site of metastases and the number of metastatic sites on the risk of overall mortality and CSM after accounting for confounders (ie, biopsy Gleason score, the administration of ADT and chemotherapy, age, year of diagnosis, race, and CCI). Third, to get the most unbiased estimate of the effect of site of metastases on survival, we repeated our analyses in a cohort of patients who had only one metastatic site involved. Finally, our analyses were repeated after excluding patients with LN metastases only.

All statistical tests were performed using the R statistical package v.3.0.2 (R Project for Statistical Computing, http://www.r-project.org/ ). All tests were two-sided with a significance level set at p < 0.05.

3. Results

3.1. Baseline characteristics

Overall, 3857 patients with metastatic PCa were identified ( Table 1 ). Average age at diagnosis was 77.3 yr (median: 77 yr). The majority of patients had bone metastases (91.1%) ( Fig. 1 ). Other common sites of metastases were LNs (8.7%), thorax (5.7%), and liver (4.5%). Overall, 611 patients (15.8%) had two or more metastatic sites involved at diagnosis. In addition, 108 patients (2.8%) were recorded as having LN metastases, 3093 (80.2%) had bone metastases, 234 (6.1%) had visceral metastases, and 422 (10.9%) had bone plus visceral metastases. When patients were stratified according to the metastatic site, statistically significant differences were observed with regard to age, year of diagnosis, race, marital status, CCI, PSA level, use of ADT and chemotherapy, and region (all p ≤ 0.03) ( Table 1 ).

Table 1 Descriptive statistics of 3857 patients diagnosed with metastatic prostate cancer between 1991 and 2009 within the Surveillance Epidemiology and End Results database stratified according to the site of metastases

  Overall (n = 3857) Lymph node metastases (n = 108; 2.8%) Bone metastases (n = 3093; 80.2%) Visceral metastases (n = 234; 6.1%) Bone and visceral metastases (n = 422; 10.9%) p value
Age at diagnosis, yr
 Mean (median) 77.3 (77) 75.3 (75) 77.2 (77) 78.7 (78) 78.1 (78) 0.01
 IQR 69–81 71–82 72–85 73–85 72–84  
Year of diagnosis
 Mean (median) 2000 (2001) 2001 (2003) 2000 (2001) 2000 (2001) 2000 (2000) 0.03
 IQR 1996–2007 1994–2005 1994–2005 1994–2005 1994–2006  
Race, %
 White 2869 (74.4) 84 (77.8) 2348 (75.9) 167 (71.4) 270 (64.0) <0.001
 Other * 988 (25.6) 24 (22.2) 745 (24.1) 67 (28.6) 152 (36.0)  
Income, $, no. (%)
 <31 101 715 (25.0) 22 (25.9) 554 (24.3) 49 (28.3) 90 (28.2) 0.5
 31 101–41 575 714 (25.0) 18 (21.2) 576 (25.3) 41 (23.7) 79 (24.8)  
 41 575–56 094 715 (25.0) 18 (21.2) 573 (25.1) 40 (23.1) 84 (26.3)  
 >56 094 712 (25.0) 27 (31.8) 576 (25.3) 43 (24.9) 66 (20.7)  
Education, yr, no. (%)            
 <10.6 718 (25.1) 18 (21.2) 561 (24.6) 41 (23.7) 98 (30.7) 0.3
 10.6–18.0 716 (25.1) 17 (20.0) 579 (25.4) 44 (25.4) 76 (23.8)  
 18.0–32.3 718 (24.8) 29 (34.1) 564 (24.7) 43 (24.9) 72 (22.6)  
 >32.3 716 (25.1) 21 (24.7) 577 (25.3) 45 (26.0) 73 (22.9)  
Marital status, no. (%)
 Married 2324 (60.3) 75 (69.4) 1880 (60.8) 140 (59.8) 229 (54.3) 0.01
 Unmarried ** 1533 (39.7) 33 (30.6) 1213 (39.2) 94 (40.2) 193 (45.7)  
Population density, no. (%)
 Nonmetropolitan 490 (12.7) 13 (12.0) 399 (12.9) 29 (12.4) 49 (11.6) 0.9
 Metropolitan 3367 (87.3) 95 (88.0) 2694 (87.1) 205 (87.6) 373 (88.4)  
CCI, no. (%)
 0 1522 (39.5) 38 (35.2) 1255 (40.6) 91 (38.9) 138 (32.7) 0.02
 1 579 (15.0) 22 (20.4) 464 (15.0) 33 (14.1) 60 (14.2)  
 ≥2 1756 (45.5) 48 (44.4) 1374 (44.4) 110 (47.0) 224 (53.1)  
Gleason score, no. (%)
 ≤7 1311 (34.0) 43 (39.8) 1053 (34.0) 86 (36.8) 130 (30.8) 0.2
 8–10 2546 (66.0) 65 (60.2) 2041 (66.0) 148 (63.2) 292 (69.2)  
PSA at diagnosis, ng/ml, no. (%) ***
 <20 262 (21.1) 13 (29.5) 200 (20.2) 26 (41.3) 23 (15.6) <0.001
 20.0–39.9 151 (12.2) NA 118 (11.9) NA 22 (15.0)  
 40.0–59.9 115 (9.3) NA 94 (9.5) NA NA  
 60.0–79.9 77 (6.2) NA 61 (6.2) NA NA  
 80.0–98.0 49 (3.9) NA 40 (4.0) NA NA  
 >98.0 588 (47.3) 16 (36.4) 475 (48.1) 21 (33.3) 76 (51.7)  
Androgen deprivation therapy, no. (%) 3323 (86.2) 95 (88.0) 2725 (88.1) 176 (75.2) 427 (77.5) <0.001
 GnRH agonists 2267 (58.8) 73 (67.6) 1849 (59.8) 137 (58.5) 208 (49.3) <0.001
 Bilateral orchiectomy 1232 (31.9) 25 (23.1) 1025 (33.1) 48 (20.5) 134 (31.8) <0.001
Chemotherapy, no. (%) 1030 (26.7) 32 (29.6) 856 (27.7) 51 (21.8) 91 (21.6) 0.02
Region, no. (%)
 East 1199 (31.1) 43 (39.8) 928 (30.0) 88 (37.6) 140 (33.2) 0.02
 Northern plain 1042 (27.0) NA 836 (27.0) 65 (27.8) 121 (28.7)  
 Pacific coast 1381 (35.8) 39 (36.1) 1132 (36.6) 66 (28.2) 144 (34.1)  
 Southwest 235 (6.1) NA 197 (6.4) 15 (6.4) 17 (4.0)  

* Including black, Asian, Hispanic, North American Native, and unknown.

** Including single, separated, divorced, widowed, and unknown.

*** Available only for patients diagnosed after the year 2003 (n = 1242).

Due to the small number of patients stratification is not possible, per National Cancer Institute indications.

CCI = Charlson comorbidity index; GnRH = gonadotropin-releasing hormone; IQR = interquartile range; NA = not available; PSA = prostate-specific antigen.

gr1

Fig. 1 Distribution of metastatic sites in 2607 patients diagnosed with stage IV prostate cancer between 1991 and 2009 within the Surveillance Epidemiology and End Results–Medicare database.

3.2. Unadjusted survival analyses

Mean and median follow-up were 115 mo (95% confidence interval [CI]: 109.8–121.8) and 114 mo (95% CI, 98.7–129.3), respectively. The median survival time was 22 mo (95% CI, 21.0–22.9) ( Fig. 2 a). The median CSM-free survival time was 30 mo (95% CI, 28.3–31.7) ( Fig. 2 b).

gr2

Fig. 2 Kaplan-Meier curves depicting (a) time to overall mortality and (b) cancer-specific mortality in the overall population. CI = confidence interval; CSM = cancer-specific mortality.

After stratifying patients according to the metastatic site, respective median overall survival and CSM-free survival were 43 mo and 61 mo for men with LN metastases, 24 mo and 32 mo for men with bone metastases, 16 mo and 26 mo for men with visceral metastases, and 14 mo and 19 mo for men with visceral plus bone metastases (p < 0.001) ( Fig. 3 a and 3b).

gr3

Fig. 3 Kaplan-Meier curves depicting time to (a) overall mortality and (b) cancer-specific mortality, after stratifying patients according to the site of metastases. CI = confidence interval; CSM = cancer-specific mortality.

After stratifying patients according to the number of metastatic sites involved, men with one site involved had significantly longer overall and CSM-free survival than those with two or more metastatic sites involved (24 and 33 vs 15 and 21, respectively; all log-rank comparisons p < 0.001) ( Fig. 4 a and 4b). When focusing exclusively on patients with one site involved, respective median overall survival and CSM-free survival were 43 mo and 61 mo for patients with LN metastasis, 24 mo and 32 mo for patients with bone metastases, and 17 mo and 26 mo for patients with visceral metastases (all log-rank comparisons, p < 0.001) (Supplementary Fig. 1).

gr4

Fig. 4 Kaplan-Meier curves depicting time to (a) overall mortality and (b) cancer-specific mortality, after stratifying patients according to the number of metastatic sites. CI = confidence interval; CSM = cancer-specific mortality.

3.3. Multivariable Cox regression models

The site of metastases represented an independent predictor of overall mortality, after adjusting for confounders (model 1 in Table 2 ). Particularly, patients with bone metastases had 1.5-fold higher risk of dying compared with those with nodal involvement (p < 0.001). This held true also in patients with visceral and bone plus visceral metastases, where the risk of mortality was, respectively, 1.7-fold and 2-fold higher compared with their counterparts with nodal involvement (all p < 0.001). Similarly, increasing number of metastatic sites involved was confirmed as a predictor of mortality, after adjusting for confounders (p = 0.02).

Table 2 Multivariable Cox regression analysis predicting overall mortality in 3857 patients diagnosed with metastatic prostate cancer within the Surveillance Epidemiology and End Results–Medicare database between 1991 and 2009

  Multivariate analyses in the overall population (model 1) * Multivariate analyses in patients with one site involved (model 2) ** Multivariate analyses including only patients with bone and/or visceral metastases (model 2) *
  HR (95% CI) p value HR (95% CI) p value HR (95% CI) p value
Site of metastases
 Lymph nodes 1 (Ref)   1 (Ref)  
 Bone 1.52 (1.20–1.93) <0.001 1.50 (1.18–1.91) 0.001 1 (Ref)  
 Visceral 1.76 (1.34–2.32) <0.001 1.71 (1.29–2.27) <0.001 1.15 (1.03–1.33) 0.03
 Bone plus visceral 1.99 (1.48–2.68) <0.001 1.30 (1.09–1.57) <0.001
Metastatic sites involved (continuously coded) 1.13 (1.02–1.26) 0.02 1.13 (1.02–1.26) 0.02
Biopsy Gleason score
 ≤7 1 (Ref)   1 (Ref)   1 (Ref)  
 8–10 1.28 (1.18–1.38) <0.001 1.34 (1.23–1.45) <0.001 1.28 (1.18–1.37) <0.001
Androgen deprivation therapy 0.36 (0.32–0.40) <0.001 0.37 (0.33–0.42) <0.001 0.36 (0.32–0.40) <0.001
Chemotherapy 0.87 (0.79–0.94) 0.001 0.87 (0.79–0.95) 0.01 0.86 (0.79–0.94) 0.001
Age at diagnosis 1.03 (1.00–1.18) <0.001 1.03 (1.02–1.03) <0.001 1.03 (1.02–1.03) <0.001
Year of diagnosis 1.01 (1.00–1.01) 0.01 1.00 (1.00–1.01) 0.04 1.00 (1.00–1.01) 0.01
Race
 White 1 (Ref)   1 (Ref)   1 (Ref)  
 Other 1.09 (1.00–1.18) 0.03 1.11 (1.02–1.22) 0.02 1.08 (1.00–1.18) 0.04
CCI
 0 1 (Ref)   1 (Ref)   1 (Ref)  
 1 1.12 (1.01–1.25) 0.03 1.11 (0.99–1.25) 0.1 1.13 (1.02–1.25) 0.02
 ≥2 1.42 (1.32–1.54) <0.001 1.40 (1.29–1.53) <0.001 1.41 (1.30–1.52) <0.001

* Model adjusted for the site of metastases, number of sites involved, biopsy Gleason score, androgen deprivation therapy, chemotherapy, age, year of diagnosis, race, and CCI.

** Model adjusted for the site of metastases, biopsy Gleason score, androgen deprivation therapy, chemotherapy, age, year of diagnosis, race, and CCI.

CCI = Charlson comorbidity index; CI = confidence interval; HR = hazard ratio; Ref = reference.

When focusing exclusively on patients with only one site involved, the location of metastases was confirmed as a predictor of mortality (model 2 in Table 2 ). Specifically, patients with bone metastases had 1.5-fold higher probability of dying compared with those with nodal involvement (p = 0.02). In addition, for patients with visceral metastases, the risk of mortality was 1.7-fold higher compared with their counterparts with LN involvement (p < 0.001).

Finally, when considering only patients with bone and/or visceral metastases, the metastatic site was confirmed as a predictor of overall mortality (model 3 in Table 2 ). Particularly, patients with visceral metastases had 1.15-fold higher probability of dying compared with their counterparts with bone involvement (p = 0.03). Similarly, men with bone plus visceral metastases had 1.3-fold higher probability of dying compared with men with bone metastases (p < 0.001).

The metastatic site represented an independent predictor of CSM after accounting for confounders ( Table 3 ). Particularly, the risk of dying from PCa was 1.5-, 1.6-, and 2.1-fold higher in patients with bone, visceral, and bone plus visceral involvement compared with those with LN metastases (all p < 0.001).

Table 3 Multivariable Cox regression analysis predicting cancer-specific mortality in 3857 patients diagnosed with metastatic prostate cancer within the Surveillance Epidemiology and End Results–Medicare database between 1991 and 2009

  Multivariate analyses in the overall population (model 1) * Multivariate analyses in patients with one site involved (model 2) ** Multivariate analyses including only patients with bone and/or visceral metastases (model 2) *
  HR (95% CI) p value HR (95% CI) p value HR (95% CI) p value
Site of metastases
 Lymph nodes 1 (Ref)   1 (Ref)
 Bone 1.51 (1.13–2.00) 0.001 1.50 (1.13–1.99) 0.005 1 (Ref)  
 Visceral 1.58 (1.13–2.20) 0.001 1.56 (1.10–2.20) 0.01 1.10 (1.00 –1.31) 0.05
 Bone plus visceral 2.15 (1.51–3.06) <0.001 1.42 (1.15–1.77) <0.001
Number of metastatic sites involved (continuously coded) 1.10 (1.00–1.25) 0.04 1.09 (0.96–1.25) 0.1
Biopsy Gleason score
 ≤7 1 (Ref)   1 (Ref)   1 (Ref)  
 8–10 1.39 (1.27–1.53) <0.001 1.45 (1.31–1.61) <0.001 1.38 (1.26–1.52) <0.001
Androgen deprivation therapy 0.37 (0.33–0.42) <0.001 0.38 (0.33–0.44) <0.001 0.37 (0.33–0.42) <0.001
Chemotherapy 0.97 (0.88–1.07) 0.5 0.97 (0.87–1.08) 0.6 0.96 (0.87–1.06) 0.4
Age at diagnosis 1.02 (1.01–1.02) <0.001 1.01 (1.00–1.02) <0.001 1.02 (1.01–1.03) <0.001
Year of diagnosis 1.01 (1.00–1.02) 0.01 1.01 (1.00–1.02) 0.04 1.01 (1.00–1.02) 0.01
Race
 White 1 (Ref)   1 (Ref)   1 (Ref)  
 Other 1.14 (1.04–1.26) 0.01 1.19 (1.07–1.33) 0.001 1.14 (1.03–1.26) 0.01
CCI
 0 1 (Ref)   1 (Ref)   1 (Ref)  
 1 1.08 (0.95–1.22) 0.2 1.08 (0.95–1.24) 0.2 1.09 (0.96–1.25) 0.1
 ≥2 1.20 (1.09–1.31) <0.001 1.19 (1.08–1.32) <0.001 1.19 (1.09–1.31) <0.001

* Model adjusted for the site of metastases, number of sites involved, biopsy Gleason score, androgen deprivation therapy, chemotherapy, age, year of diagnosis, race, and CCI.

** Model adjusted for the site of metastases, biopsy Gleason score, androgen deprivation therapy, chemotherapy, age, year of diagnosis, race, and CCI.

CCI = Charlson comorbidity index; CI = confidence interval; HR = hazard ratio; Ref = reference.

4. Discussion

Although PCa commonly spreads to the regional nodes and to the skeleton, the risk of atypical metastases is not negligible [3] . In this context, up to 15% of patients diagnosed with metastatic PCa present with visceral involvement [3] . However, the role of the metastatic site on survival has not been comprehensively addressed so far, and the largest available observations come from clinical trials or small institutional studies [4], [6], [7], [8], [9], [10], [11], and [16]. Results obtained in these settings might not be generalizable to the US population. Given such a paucity of data, we sought to evaluate the impact of the site of metastases on mortality in a large contemporary cohort of metastatic patients.

We made several findings. First, although previous studies described longer survival in patients developing metastases after primary treatment [12] and [17], our investigation reports the overall- and CSM-free survival rates in a large contemporary cohort of patients with metastatic disease at diagnosis. Particularly, we demonstrated that the survival rates varied according to site and number of metastases. Visceral metastases alone or with concomitant bone involvement conferred worst survival compared with bone metastases. Previous smaller studies investigated this issue in patients with castration-resistant PCa [6], [8], and [11]. For example, Halabi et al. [6] and [11] included the presence of visceral metastases in their prognostic models. In addition, Armstrong et al. [8] showed that liver metastases were associated with shorter overall survival. However, their model did not include a distinction between visceral or nonvisceral metastases. Our findings confirmed that visceral metastases confer the worst survival and that multiple metastatic sites involved also result in shorter survival than presence of oligometastases. These observations postulate that PCa may have several phenotypes that predispose to different natural histories when survival represents the end point. This is in line with what Pond et al. [4] recently reported when evaluating a highly selected population of approximately 1000 patients in a clinical trial.

Second, substantial differences exist with regard to baseline characteristics at diagnosis according to the metastatic site. For example, patients with visceral metastases were older and had a higher CCI and a higher PSA level compared with those with LN or bone involvement. In addition, the proportion of white and married patients was higher among individuals with LN and bone involvement. These observations might reflect the impact of patient characteristics on the predisposition to experience more aggressive disease [18] . On the other hand, they might also impact the probability of receiving early detection [19] , where individuals less likely to be screened present with more aggressive diseases at diagnosis [20] . In addition, the proportion of patients treated with ADT and chemotherapy during follow-up was lower among those with visceral metastases compared with their counterparts with LN or bone involvement. In this light, we should emphasize that the association between the metastatic site and survival held true even after adjusting for the administration of cancer therapies during follow-up.

Several hypotheses can be proposed to explain the survival differences and/or the predilection for metastatic location and extent. For example, the interaction between metastatic cells and the bone microenvironment (ie, the seeds and soil hypothesis), together with the ability of the bone to attract cancer cells through the release of chemotactic factors (ie, the homing theory), might explain the high prevalence of bone and skeleton metastases [21] and [22]. In contrast, undifferentiated cells with a more aggressive phenotype might acquire the ability to invade distant visceral organs, leading to faster progression and shorter survival in these patients. In this light, molecular data supporting this hypothesis are scarce [4], [9], and [23], and further well-designed studies investigating the biologic mechanisms of metastatic spread are urgently needed [4] . The reason of such unfavorable impact of visceral metastases on overall survival might also merely reflect the role of disease burden. In this context, our findings show that any additional metastatic site involved increased by approximately 20% the subsequent risk of mortality. Nevertheless, we should underline that the negative impact of visceral metastases on mortality held true even when focusing solely on patients with one metastatic site involved. Consequently, we can speculate that although skeletal-related events significantly increase the risk of mortality [24] , clinical manifestations related to liver, lung, or brain involvement might be associated with worse outcomes. In addition, patients with LN metastases represented individuals with a better prognosis compared with their counterparts with bone or visceral involvement. Although several studies reported better oncologic outcomes in node-positive patients [25] , they included highly selected individuals treated with radical prostatectomy and pelvic LN dissection. Conversely, our observations originate from a cohort of stage IV PCa patients at diagnosis. Thus, those individuals did not receive any definitive therapy aimed at the primary tumor. This fact removes a potentially important favorable bias that may have undermined studies where definitive therapies were used. In addition, due to the low sensitivity of imaging modalities such as computed tomography and magnetic resonance imaging in the detection of micrometastatic nodal involvement [26] , we can hypothesize that patients with LN metastases included in our study might have high nodal burden. In this regard, it is noteworthy that they had better oncologic outcomes compared with men with bone and/or visceral involvement. The reason for these observations might reside in the tumor biology. Evidence coming from preclinical studies suggests that tumor cells metastasizing only to LNs might have specific epigenetic modifications that predispose to the invasion of LNs instead of visceral organs [25] and [27]. These cells might acquire the ability to spread into vascular circulation and to invade specific metastatic sites only after subsequent neoplastic transformations [28] . Consequently, PCa cells invading the LNs might harbor a less aggressive phenotype compared with those invading other sites, resulting in a better prognosis [16] .

The current study has several clinical implications. First, the identification of the site of metastases and the number of metastatic sites as prognostic factors might be critical to properly counsel patients and their families about the oncologic outcomes. Second, our observations might help physicians better understand the biology and natural history of the disease, thus allowing the development of novel therapeutic approaches [29] . Indeed, individuals with visceral involvement represent patients with poorer prognosis. Thus, they are less likely to respond to conventional treatment modalities and they might warrant more aggressive approaches. Third, our findings allow improving the risk stratification of stage IV patients. Although recent randomized phase 3 trials showed the efficacy of novel molecules in patients with castration-resistant PCa [30], [31], and [32], the optimal candidates to receive these cytotoxic agents and their appropriate sequencing still represent matters of debate [11] . It is also noteworthy that several clinical trials evaluating novel molecules did not include patients with visceral metastases [30] and [31]. Thus, our findings might be used in the design and analysis of prospective studies evaluating the effectiveness of novel treatments in metastatic patients. On the opposite side, our observations show that relatively good cancer control might be achieved in patients with LN involvement. Since previous studies showed that the survival is not invariably poor in patients with node-positive disease [16] , even after primary treatment, it might be speculated that some of these patients might benefit from local therapies combined with primary tumor control [33] and [34].

The current study reflects the largest population-based assessment of sites of metastases and cancer-control outcomes after stage IV PCa diagnosis. Despite its novelty and significance, a major limitation involves the identification of metastatic sites, which was based on ICD-9 diagnostic codes. Including these codes on a Medicare claim does not affect the amount of reimbursement to the provider. Thus, these codes might be inaccurate and incomplete in some cases. In addition, previous studies showed that the Medicare claims for bone metastases in PCa patients have a low accuracy [35] . Moreover, the use of these codes resulted in the exclusion of a non-negligible proportion of patients due to missing values and might have introduced a selection bias. That being said, Dolan et al. [36] recently reviewed the medical records of 300 men hospitalized at one center and evaluated the accuracy of ICD-9 codes for metastases, using the chart review as the gold standard. They demonstrated a high sensitivity, specificity, positive predictive value, and negative predictive value for the use of ICD-9 codes in identifying site of metastases [36] . Nonetheless, further larger studies are needed to validate the use of ICD-9 codes in the context of population-based studies. Consequently, caution should be used when interpreting the results of our study. Other limitations apply to our study. First, the retrospective nature of our database prevented us determining the exact location of bone metastases or their number. However, these covariates were not included in previously published tools predicting survival [6], [7], [8], and [11]. Second, the lack of information regarding the dose and duration of ADT and chemotherapy prevented us from adjusting our analyses for these potential confounders. Similarly, since novel molecules, such as sipuleucel-T, radium-223, and abiraterone acetate, were not available over the study period, we were not able to assess the impact of the metastatic site on survival, according to the administration of these therapies. Third, details regarding PSA values and the onset of castration-resistant disease during follow-up were not available. Fourth, laboratory parameters and performance status were not available in our large population-based cohort. We tried to circumvent this potential limitation by adjusting for baseline comorbidity status. Last, these observations were obtained in the context of men aged ≥65 yr. Thus, they might not be applicable to younger patients.

5. Conclusions

Heterogeneity exists regarding oncologic outcomes in stage IV PCa patients. Visceral involvement represents a negative prognostic factor in these patients. Consequently, these patients should be considered as affected by more aggressive disease and might benefit from the inclusion in clinical trials evaluating the efficacy of novel molecules.


Author contributions: Giorgio Gandaglia 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: Gandaglia, Karakiewicz, Sun, Briganti, Montorsi.

Acquisition of data: Sun, Gandaglia, Trudeau, Schiffmann.

Analysis and interpretation of data: Gandaglia, Sun, Passoni, Karakiewicz.

Drafting of the manuscript: Gandaglia, Sun, Karakiewicz.

Critical revision of the manuscript for important intellectual content: Montorsi, Graefen, Briganti, Karakiewicz.

Statistical analysis: Gandaglia, Sun.

Obtaining funding: Montorsi, Graefen, Karakiewicz.

Administrative, technical, or material support: None.

Supervision: Karakiewicz, Montorsi, Graefen, Briganti.

Other (specify): None.

Financial disclosures: Giorgio Gandaglia 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

 

mmc1

Supplementary Fig. 1 – Kaplan-Meier curves depicting time to overall mortality (a) and cancer-specific mortality (CSM) (b) in patients with only one metastatic site involved after stratification according to the site of metastases.

References

  • [1] R. Siegel, D. Naishadham, A. Jemal. Cancer statistics, 2013. CA Can J Clin. 2013;63:11-30
  • [2] K. Cetin, J.L. Beebe-Dimmer, J.P. Fryzek, R. Markus, M.A. Carducci. Recent time trends in the epidemiology of stage IV prostate cancer in the United States: analysis of data from the Surveillance, Epidemiology, and End Results Program. Urology. 2010;75:1396-1404
  • [3] G. Gandaglia, F. Abdollah, J. Schiffmann, et al. Distribution of metastatic sites in patients with prostate cancer: a population-based analysis. Prostate. 2014;74:210-216
  • [4] G.R. Pond, G. Sonpavde, R. de Wit, M.A. Eisenberger, I.F. Tannock, A.J. Armstrong. The prognostic importance of metastatic site in men with metastatic castration-resistant prostate cancer. Eur Urol. 2014;65:3-6
  • [5] A.H. Vinjamoori, J.P. Jagannathan, A.B. Shinagare, et al. Atypical metastases from prostate cancer: 10-year experience at a single institution. AJR Am J Roentgenol. 2012;199:367-372
  • [6] S. Halabi, E.J. Small, P.W. Kantoff, et al. Prognostic model for predicting survival in men with hormone-refractory metastatic prostate cancer. J Clin Oncol. 2003;21:1232-1237
  • [7] A. Omlin, C. Pezaro, D. Mukherji, et al. Improved survival in a cohort of trial participants with metastatic castration-resistant prostate cancer demonstrates the need for updated prognostic nomograms. Eur Urol. 2013;64:300-306
  • [8] A.J. Armstrong, E.S. Garrett-Mayer, Y.C. Yang, R. de Wit, I.F. Tannock, M. Eisenberger. A contemporary prognostic nomogram for men with hormone-refractory metastatic prostate cancer: a TAX327 study analysis. Clin Canc Res. 2007;13:6396-6403
  • [9] C.J. Pezaro, A. Omlin, D. Lorente, et al. Visceral disease in castration-resistant prostate cancer. Eur Urol. 2014;65:270-273
  • [10] P. Ost, K. Decaestecker, B. Lambert, et al. Prognostic factors influencing prostate cancer-specific survival in non-castrate patients with metastatic prostate cancer. Prostate. 2014;74:297-305
  • [11] S. Halabi, C.Y. Lin, W.K. Kelly, et al. Updated prognostic model for predicting overall survival in first-line chemotherapy for patients with metastatic castration-resistant prostate cancer. J Clin Oncol. 2014;32:671-677
  • [12] S.H. Culp, P.F. Schellhammer, M.B. Williams. Might men diagnosed with metastatic prostate cancer benefit from definitive treatment of the primary tumor? A SEER-based study. Eur Urol. 2014;65:1058-1066
  • [13] C.N. Klabunde, A.L. Potosky, J.M. Legler, J.L. Warren. Development of a comorbidity index using physician claims data. J Clin Epidemiol. 2000;53:1258-1267
  • [14] G.L. Lu-Yao, P.C. Albertsen, H. Li, et al. Does primary androgen-deprivation therapy delay the receipt of secondary cancer therapy for localized prostate cancer?. Eur Urol. 2012;62:966-972
  • [15] Gandaglia G, Sun M, Hu JC, et al. Gonadotropin-releasing hormone agonists and acute kidney injury in patients with prostate cancer. Eur Urol. In press. http://dx.doi.org/10.1016/j.eururo.2014.01.026
  • [16] Y. Furuya, K. Akakura, S. Akimoto, H. Ito. Prognosis of patients with prostate carcinoma presenting as nonregional lymph node metastases. Urol Intern. 1998;61:17-21
  • [17] C.R. Pound, A.W. Partin, M.A. Eisenberger, D.W. Chan, J.D. Pearson, P.C. Walsh. Natural history of progression after PSA elevation following radical prostatectomy. JAMA. 1999;281:1591-1597
  • [18] G. Chornokur, K. Dalton, M.E. Borysova, N.B. Kumar. Disparities at presentation, diagnosis, treatment, and survival in African American men, affected by prostate cancer. Prostate. 2011;71:985-997
  • [19] A.B. Mariotto, R. Etzioni, M. Krapcho, E.J. Feuer. Reconstructing PSA testing patterns between black and white men in the US from Medicare claims and the National Health Interview Survey. Cancer. 2007;109:1877-1886
  • [20] F.H. Schroder, J. Hugosson, M.J. Roobol, et al. Prostate-cancer mortality at 11 years of follow-up. N Engl J Med. 2012;366:981-990
  • [21] A.S. Tsingotjidou, G. Zotalis, K.R. Jackson, et al. Development of an animal model for prostate cancer cell metastasis to adult human bone. Anticanc Res. 2001;21:971-978
  • [22] K. Jacob, M. Webber, D. Benayahu, H.K. Kleinman. Osteonectin promotes prostate cancer cell migration and invasion: a possible mechanism for metastasis to bone. Cancer Res. 1999;59:4453-4457
  • [23] F. O’Kelly, S. Elamin, A. Cahill, et al. Characteristics of modern Gleason 9/10 prostate adenocarcinoma: a single tertiary centre experience within the Republic of Ireland. World J Urol. 2014;32:1067-1074
  • [24] N. Sathiakumar, E. Delzell, M.A. Morrisey, et al. Mortality following bone metastasis and skeletal-related events among men with prostate cancer: a population-based analysis of US Medicare beneficiaries, 1999–2006. Prostate Cancer Prostatic Dis. 2011;14:177-183
  • [25] A. Briganti, N.M. Passoni, F. Abdollah, A. Nini, F. Montorsi, R.J. Karnes. Treatment of lymph node-positive prostate cancer: teaching old dogmas new tricks. Eur Urol. 2014;65:26-28
  • [26] R.A. Heesakkers, A.M. Hovels, G.J. Jager, et al. MRI with a lymph-node-specific contrast agent as an alternative to CT scan and lymph-node dissection in patients with prostate cancer: a prospective multicohort study. Lancet Oncol. 2008;9:850-856
  • [27] Z. Barekati, R. Radpour, Q. Lu, et al. Methylation signature of lymph node metastases in breast cancer patients. BMC Cancer. 2012;12:244
  • [28] M. Arya, S.R. Bott, I.S. Shergill, H.U. Ahmed, M. Williamson, H.R. Patel. The metastatic cascade in prostate cancer. Surg Oncol. 2006;15:117-128
  • [29] S. Halabi, K. Owzar. The importance of identifying and validating prognostic factors in oncology. Semin Oncol. 2010;37:e9-e18
  • [30] C.J. Ryan, M.R. Smith, J.S. de Bono, et al. Abiraterone in metastatic prostate cancer without previous chemotherapy. N Engl J Med. 2013;368:138-148
  • [31] C. Parker, S. Nilsson, D. Heinrich, et al. Alpha emitter radium-223 and survival in metastatic prostate cancer. N Engl J Med. 2013;369:213-223
  • [32] J.S. De Bono, C.J. Logothetis, A. Molina, et al. Abiraterone and increased survival in metastatic prostate cancer. N Engl J Med. 2011;364:1995-2005
  • [33] H.G. Frohmuller, M. Theiss, A. Manseck, M.P. Wirth. Survival and quality of life of patients with stage D1 (T1-3 pN1-2 M0) prostate cancer. Radical prostatectomy plus androgen deprivation versus androgen deprivation alone. Eur Urol. 1995;27:202-206
  • [34] Suardi N, Gandaglia G, Gallina A, et al. Long-term outcomes of salvage lymph node dissection for clinically recurrent prostate cancer: results of a single-institution series with a minimum follow-up of 5 years. Eur Urol. In press. http://dx.doi.org/10.1016/j.eururo.2014.02.011
  • [35] E. Onukwugha, D. Mullins, C. Yong, D.L. McNally, B.S. Seal, A. Hussain. Bone metastasis (BM) based on SEER registry versus Medicare claims among metastatic prostate cancer (PCa) patients (pts) in SEER-Medicare [abstract e15148]. J Clin Oncol. 2012;30(suppl)
  • [36] Dolan MT, Kim S, Shao Y, Lu-Yao GL. Authentication of algorithm to detect metastases in men with prostate cancer using ICD-9 codes. Epid Res Intern. http://dx.doi.org/10.1155/2012/970406

Footnotes

a Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, Canada

b Division of Oncology/Unit of Urology, URI, IRCCS Ospedale San Raffaele, Milan, Italy

c Department of Urology, University of Montreal Health Center, Montreal, Canada

d Martini-Clinic, Prostate Cancer Center Hamburg-Eppendorf, Hamburg, Germany

lowast Corresponding author. University of Montreal Health Center, 1058, rue St-Denis, Montreal, QC, H2X 3J4, Canada. Tel. +1 514 890 8000 ext. 35335; Fax: +1 514 227 5103.

These authors contributed equally.