Treatment and survival outcomes in young men diagnosed with prostate cancer

A Population-based Cohort Study

Abstract

BACKGROUND:

Outcomes of treatment for young men compared with older men with prostate cancer are poorly defined outside of limited institutional series. In this study, the authors examined the association between age at diagnosis and grade, stage, treatment, and survival outcomes in men who were diagnosed during the era of prostate-specific antigen testing.

METHODS:

The National Cancer Institute's Surveillance, Epidemiology, and End Results database was used to identify men who were diagnosed with prostate cancer between 1988 and 2003. Men ages 35 years to 74 years were stratified by age at diagnosis to examine differences in tumor characteristics, treatment, and survival within each age group.

RESULTS:

In total, 318,774 men ages 35 years to 74 years were identified who had been diagnosed with adenocarcinoma of the prostate between 1988 and 2003. The proportion of men aged ≤55 years at diagnosis increased over the study period from 2.3% between the years 1988 and 1991 to 9% between the years 2000 and 2003, and the median age at diagnosis decreased from 72 years in 1988 to 68 years in 2003. Younger men were diagnosed less frequently with organ-confined tumors (P < .001) but were less likely to be diagnosed with high-grade cancer (P < .001). Older men were more likely to receive no local therapy or external beam radiation than young men (P < .001 for trend). Among men who had tumors with a Gleason score between 5 and 7, overall survival was worse with advancing age. However, among all age groups with high grade and stage, the youngest men (ages 35-44 years) were at the highest risk of all-cause and cancer-specific death.

CONCLUSIONS:

Age at diagnosis among men with prostate cancer continued to decline. Younger men were more likely to undergo prostatectomy, have lower grade cancer, and, as a group, to have better overall and equivalent cancer-specific survival at 10 years compared with older men. Among men with high grade and locally advanced prostate cancer, the youngest men had a particularly poor prognosis compared with older men. Cancer 2009. Published 2009 by the American Cancer Society.

Age at diagnosis of cancer is a well recognized prognostic factor for patients with malignancy, and younger patients have better outcomes independent of comorbidity or performance status.1 A definition of “younger” men with prostate cancer has been poorly characterized and ranges from “<50” to “age <60 years” in previous reports.2–5 Although these divisions of younger and older men with prostate cancer have varied, previous reports typically have indicated that “younger” men have better biochemical progression-free survival after prostatectomy6, 7 and less advanced disease at prostatectomy.3 In contrast, Rosser et al reported that men aged <60 years who were treated with external-beam radiotherapy suffered higher rates of biochemical progression.4

To our knowledge, no studies have assessed the more significant endpoints of overall survival (OS) or disease-specific survival (DSS) in young men with local or regional prostate cancer, primarily because of limited follow-up and small numbers of patients in single-institution series. In clinical practice, patients usually are considered for definitive local therapy if their life expectancy is >10 years, a threshold that is crossed when men reach age 75 years.8 The objectives of the current study were to describe trends in age at diagnosis and to characterize the relation between age at diagnosis and survival outcomes in men diagnosed with prostate cancer who were considered candidates for definitive therapy based on their age. Specifically, we examined the association between age at diagnosis, tumor characteristics, and survival among men with prostate cancer in the US.

MATERIALS AND METHODS

Data Source

The Surveillance, Epidemiology, and End Results (SEER) Program database was used to identify the cohort of patients for this study. SEER collects cancer incidence and survival data from 17 population-based cancer registries that account for approximately 26% of the US population. Data from 1988 through 2003 from 17 SEER registries were used (San Francisco-Oakland standard metropolitan statistical area, Connecticut, Metropolitan Detroit, Hawaii, Iowa, New Mexico, Seattle [Puget Sound], rural Georgia, Utah, Metropolitan Atlanta, Alaska, San Jose-Monterey, Los Angeles, Kentucky, Louisiana, New Jersey, and Greater California). Cases are available only for the years 1992 through 2003 from Alaska, San Jose-Monterey, Los Angeles, and rural Georgia and for the years 2000 through 2003 from Kentucky, Louisiana, New Jersey, and Greater California.

Study Population

Cases initially were identified using International Classification of Diseases for Oncology (third edition) site codes for the prostate (C619) and histology classification codes for adenocarcinoma (8140) and acinar carcinoma (8550). Men aged <35 years were excluded. For the survival analyses, we only included men who were ages 35 years to 74 years at the time of diagnosis, because these men are most likely to be candidates for local therapy with curative intent, in that their actuarial life expectancy exceeds 10 years.

Data Collection and Coding

Demographic data included patient age, race, and year of treatment. Age was categorized into 10-year categories. Race was categorized as Caucasian, African American, or other based on SEER coding. Initial treatment type (prostatectomy, external beam radiotherapy, brachytherapy, external beam and brachytherapy, or no local therapy) also was determined. Pathologic data included SEER classification of tumor grade (low, Gleason score 2-4; intermediate, Gleason score 5-7; high, Gleason score 8-10) and SEER-modified American Joint Committee on Cancer stage (stages I and II, T1/T2N0M0; stage III, T3N0M0; and stage IV, T4 or N+ or M+). SEER reports pathologic stage when available, whereas patients who did not undergo prostatectomy have only clinical stage data recorded. Chemotherapy and comorbidity data are not available in SEER. Patients without complete stage and grade data were excluded from the survival analyses. Survival was calculated from the date of diagnosis to the date of death. If death was not observed, then patients were censored at the date of last follow-up. Prostate cancer-specific cause of death was determined using SEER site-specific death codes.

Statistical Analysis

Demographic and pathologic data are reported for the entire cohort. Multivariate Cox regression analysis was performed to evaluate differences in the risk of death based on age stratified by grade and stage at diagnosis. All multivariate analyses were adjusted for race, year of diagnosis, and initial treatment type. To minimize possible bias from inaccurate clinical staging data, a subanalysis of patients who underwent prostatectomy with associated pathologic staging data also was performed. The proportional hazards assumption for the Cox regression model was evaluated with Schoenfeld residuals. Variables that did not meet proportional hazards assumptions were stratified in the Cox model. Hazards ratios (HRs) are presented along with their 95% confidence intervals (95% CIs). All statistical analyses were conducted using Stata software (version 9; Stata, Inc., College Station, Tex).

RESULTS

We identified 453,195 men aged ≥35 years who were diagnosed with adenocarcinoma of the prostate between 1988 and 2003 in the SEER tumor registry. Of these, 318,774 men were between ages 35 years and 74 years, the age range during which active treatment typically is considered. The proportion of men who were aged ≤55 years at diagnosis increased over the study period from 2.3% between 1988 and 1991 to 9% between 2000 and 2003 (P < .001), whereas the median age at diagnosis decreased from 72 years in 1988 to 68 years in 2003. Table 1 provides demographic and tumor characteristics stratified by age for men who were ages 35 years to 74 years at the time of diagnosis. The younger men more commonly were African American, were less likely to have high-grade tumors (P < .001), were less likely to be diagnosed with organ-confined disease (P < .001), and were more likely to undergo radical prostatectomy (P < .001). Conversely, the use of radiotherapy or no local therapy was more common in older men.

Table 1. Baseline Characteristics of All Men Ages 35 to 74 Years With Prostate Cancer

	Age Group, Years
Patient Characteristic	35-44 (n=1673)	45-54 (n=28,665)	55-64 (n=107,337)	65-74 (n=181,099)
AJCC indicates American Joint Committee on Cancer; RP, radical prostatectomy; EBRT, external beam radiotherapy.
Grade, %
Low	5	5.7	7.4	10.2
Intermediate	74.4	73.5	70	64.4
High	17.1	17.8	19.1	21.1
Unknown	3.5	3	3.5	4.3
P (chi-square)				<.001
AJCC stage, %
I or II	32.1	35.9	38.4	42.5
III	13.5	15.3	15	11.6
IV	10.9	8.8	8.4	8
Unknown	43.5	40	38.2	37.9
P (chi-square)				<.001
Race, %
Caucasian	70.1	76.6	79.9	82
African American	24.7	18.3	14.5	11
Other	3.4	3.1	3.7	5.1
Unknown	1.9	2	1.9	1.9
P (chi-square)				<.001
Local therapy, %
None	16.1	15.1	18.4	28.3
RP	52.7	50.9	43	25.0
EBRT	10.8	12.7	17.6	28.9
Brachytherapy	3	5.3	6.8	6.7
Brachytherapy and EBRT	3	4.1	4.8	5.2
Unknown	14.4	12	9.4	5.9
P (chi-square)				<.001
Year of diagnosis, %
1988-1991	4	4.1	8.3	13.1
1992-1995	10.9	13.8	18.9	24.6
1996-1999	22.5	23.2	21.7	21
2000-2003	62.6	58.9	51.1	41.3
P (chi-square)				<.001

We examined both OS and DSS stratified by age, grade, and stage. The median follow-up for the entire cohort was 45 months. The Kaplan-Meier survival curves for the entire cohort stratified by age are shown in Figure 1A (OS) and Figure 1B (DSS). OS was better at 5 years and at 10 years for men aged <55 years, as expected. DSS for the entire group did not differ significantly for men aged <55 years or aged >55 years. To explore this issue further, we analyzed univariate 10-year OS and DSS Kaplan-Meier curves for patients who had tumors with a Gleason score between 5 and 7 stratified by age at diagnosis, and these survival cures are shown in Figure 2A and Figure 2B, respectively. Among all men who had tumors with a Gleason score between 5 and 7, the 10-year OS and DSS rates were 69% and 92%, respectively. When the analysis was stratified by age, as expected, advancing age was associated with decreased OS compared with the youngest men, although DSS was similar between age groups. Figure 3A presents OS and Figure 3B presents DSS for men who had tumors with a Gleason score between 8 and 10 stratified by age at diagnosis. Among all men who had tumors with a Gleason score between 8 and 10, the 10-year OS and DSS rates were 48% and 70%, respectively. However, in contrast to intermediate-grade tumors, young men with high-grade tumors had significantly decreased OS and DSS compared with older men.

Figure 1. Survival outcomes among men with prostate cancer are illustrated by age at diagnosis for (A) overall survival after diagnosis and (B) prostate cancer-specific survival after diagnosis.

Figure 2. Survival outcomes among men who had prostate cancer with a Gleason score between 5 and 7 are illustrated by age at diagnosis for (A) overall survival after diagnosis and (B) prostate cancer-specific survival after diagnosis.

Figure 3. Survival outcomes among men who had prostate cancer with a Gleason score between 8 and 10 are illustrated by age at diagnosis for (A) overall survival after diagnosis and (B) prostate cancer-specific survival after diagnosis.

Table 2 provides the results of the multivariate analysis stratified by age, grade, and stage adjusted for race, year of diagnosis, and initial treatment type. When men who had tumors with a Gleason score between 5 and 7 were stratified by grade and stage, the risk of all-cause death increased predictably with each advancing age group in men with organ-confined disease, and the oldest men had the highest risk of death (HR, 18.25; 95%CI, 4.56-73.02), although older men with stage IV disease had improved DSS. However, among men with high-grade and stage III and IV disease, the youngest men were approximately 3 times more likely to die of prostate cancer than all other age groups with locally advanced disease.

Table 2. Overall and Prostate Cancer-specific Survival of Patients With Prostate Cancer Stratified by Age, Grade, and Stage

	HR (95% CI)
	Overall Survival		Prostate Cancer-specific Survival
Age Group, Years	Gleason Score 5-7	Gleason Score 8-10	Gleason Score 5-7	Gleason Score 8-10
HR indicates hazards ratio; 95% CI, 95% confidence interval. * Cox proportional hazards regression model was adjusted for stage, race, year of diagnosis, and initial treatment type. † P < .05. ‡ Cox proportional hazards regression model was adjusted for race, year of diagnosis, and initial treatment type. § This HR was incalculable because there were no events in 1 or more age strata.
All stages*
35-44	Referent	Referent	Referent	Referent
45-54	1.45 (0.95-2.22)	0.66 (0.50-0.86)†	0.80 (0.50-1.27)	0.64 (0.48-0.86)†
55-64	2.10 (1.38-3.19)†	0.71 (0.54-0.92)†	0.79 (0.50-1.24)	0.58 (0.44-0.77)†
65-74	3.53 (2.32-5.37)†	0.89 (0.69-1.16)	0.92 (0.59-1.45)	0.60 (0.46-0.80)†
Stages I and II‡
35-44	Referent	Referent	Referent	Referent
45-54	6.28 (1.56-25.22)†	1.03 (0.38-2.81)	2.07 (0.29-14.95)	1.36 (0.33-5.60)
55-64	9.96 (2.49-39.84)†	1.21 (0.45-3.25)	2.29 (0.32-16.34)	0.98 (0.24-3.95)
65-74	18.25 (4.56-73.02)†	1.70 (0.64-4.55)	3.27 (0.46-23.29)	1.06 (0.26-4.25)
Stage III‡
35-44	Referent	Referent	Referent	Referent
45-54	—§	0.38 (0.18-0.78)†	—§	0.30 (0.14-0.65)†
55-64	—§	0.49 (0.24-0.99)†	—§	0.30 (0.14-0.63)†
65-74	—§	0.73 (0.37-1.47)	—§	0.33 (0.16-0.70)†
Stage IV‡
35-44	Referent	Referent	Referent	Referent
45-54	0.75 (0.47-1.18)	0.71 (0.52-0.97)†	0.65 (0.40-1.06)†	0.67 (0.49-0.93)†
55-64	0.79 (0.50-1.23)	0.73 (0.54-0.98)†	0.60 (0.37-0.96)†	0.61 (0.45-0.84)†
65-74	0.96 (0.63-1.52)	0.83 (0.61-1.10)	0.60 (0.38-0.96)†	0.62 (0.46-0.85)†

In an attempt to minimize stage and grade bias, we performed a second multivariate analysis on the subgroup of men ages 35 years to 74 years who had undergone radical prostatectomy, because these men had tumor characteristics (pathologic stage and grade) determined from the radical prostatectomy specimen (Table 3). Trends similar to but more pronounced than the trends observed in the main analysis were observed in this subanalysis. Specifically, the youngest men were at least 5 times more likely to die of prostate cancer than any of their older counterparts with high-grade, stage III disease.

Table 3. Overall and Prostate Cancer-specific Survival of Patients Undergoing Radical Prostatectomy Stratified by Age

	HR (95% CI)
	Overall Survival		Prostate Cancer-specific Survival
Age Group, Years	Gleason Score 5-7	Gleason Score 8-10	Gleason Score 5-7	Gleason Score 8-10
HR indicates hazards ratio; 95% CI, 95% confidence interval. * Cox proportional hazards regression model adjusted for stage, race, year of diagnosis, and initial treatment type. † P < .05. ‡ Cox proportional hazards regression model adjusted for race, year of diagnosis, and initial treatment type. § This HR was incalculable because there were no events in 1 or more age strata.
All stages*
35-44	Referent	Referent	Referent	Referent
45-54	6.41 (1.59-25.8)†	0.47 (0.27-0.84)†	3.15 (0.44-23.7)	0.35 (0.19-0.65)†
55-64	10.1 (2.53-40.5)†	0.55 (0.32-0.96)†	2.79 (0.39-19.9)	0.29 (0.16-0.52)†
65-74	20.5 (5.12-81.9)†	0.86 (0.49-1.50)	4.19 (0.56-29.9)	0.35 (0.19-0.63)†
Stages I and II‡
35-44	Referent	Referent	Referent	Referent
45-54	5.95 (0.83-42.6)	0.45 (0.14-1.47)	—§	0.31 (0.07-1.33)
55-64	9.07 (1.28-64.5)†	0.41 (0.13-1.30)	—§	0.16 (0.04-0.68)†
65-74	20.0 (2.81-142.1)†	0.72 (0.23-2.27)	—§	0.18 (0.04-0.76)†
Stage T3‡
35-44	Referent	Referent	Referent	Referent
45-54	—§	0.28 (0.12-0.64)†	—§	0.17 (0.07-0.39)†
55-64	—§	0.36 (0.16-0.82)†	—§	0.16 (0.07-0.36)†
65-74	—§	0.58 (0.26-1.30)	—§	0.20 (0.09-0.44)†
Stage ≥T4‡
35-44	Referent	Referent	Referent	Referent
45-54	1.25 (0.17-9.21)	0.86 (0.30-2.36)	0.57 (0.07-4.31)	0.69 (0.24-1.98)
55-64	1.79 (0.25-12.8)	0.98 (0.36-2.67)	0.55 (0.07-4.06)	0.58 (0.21-1.60)
65-74	2.87 (0.40-20.6)	1.36 (0.50-3.68)	0.63 (0.09-4.57)	0.68 (0.25-1.87)

DISCUSSION

This population-based analysis demonstrates first that the diagnosis of prostate cancer in young men constitutes an increasing proportion of the number of men with newly diagnosed prostate cancer and that the primary treatment choice among young men and their physicians is radical prostatectomy over other modalities. Second, our analysis provides critical OS and DSS data beyond the prior institutional reports that assessed biochemical progression alone. The data support the intuitive concept that young men with fewer comorbidities will have better OS as a group. These data indicate that DSS at 10 years for the group as a whole is not significantly better for younger men. Other institutional series of outcomes in very young men have suggested that biochemical recurrence rates are superior for the youngest men compared with older men.7 The current study focuses on survival outcomes rather than biochemical recurrence and surveys a substantially larger and more diverse range of patients, making direct comparison difficult. The advantages of this analysis are that it reflects practice patterns and outcomes from a broad range of providers and includes a significantly larger cohort of men aged <50 years than any other reported series to our knowledge. This analysis also unexpectedly revealed that young men with high-grade tumors have significantly worse DSS outcomes. Specifically, these young men with high-grade tumors had worse OS when they were diagnosed with stage III and IV disease and worse DSS in all stages, depending on primary treatment. Conversely, in general, older men who have tumors with a Gleason score between 5 and 7 have worse outcomes, presumably related to other comorbidities. Specifically, these older men with low/intermediate-grade tumors have decreased OS in stage I/II disease and decreased DSS in stage IV disease.

Younger age at diagnosis has a positive impact on prognosis for most adult men and women who are diagnosed with malignancy. For all patients who are diagnosed with cancer, subsets of histologies are associated with worse outcomes in young adult patients, including breast cancer, sarcoma, colon cancer, lymphoma, and leukemia.1 In breast cancer, which is another hormone-dependent malignancy, young age is a risk factor independent of stage, grade, hormone receptor expression, and mode of primary therapy.9, 10 Better tolerance for aggressive therapy, earlier recognition of the disease, and attendant lower grade and stage of disease all have been invoked as reasons why young men with prostate cancer have better recurrence-free survival after primary therapy in other smaller institutional series.11

There are several potential explanations for our current findings with regard to the more significant endpoints of OS and DSS. In our study, younger men were more likely to have extraprostatic disease, possibly reflecting the greater use of prostatectomy and subsequent pathologic upstaging in these men. Despite the trend toward better outcomes with younger age for the group as a whole, the youngest men diagnosed with high-grade disease paradoxically are at much higher risk for death from prostate cancer regardless of the form of therapy. Why men ages 35 years to 44 years with high-grade disease have dramatically higher cancer-specific mortality compared with older men is unclear, although the relative lack of competing comorbidities in younger men likely plays a role in this finding. In support of our study, an analysis of the Cancer of the Prostate Strategic Urologic Research Endeavor or CaPSURE dataset suggested that younger men who present with metastatic prostate cancer were at higher risk of early death, with an HR of 0.47 (95%CI, 0.28-0.78) for men aged >65 years.12 Family history of prostate cancer also may play a role in screening men earlier in life and potentially may enrich the population of young patients for high-risk disease. However, in the prostate-specific antigen era, a family history of prostate cancer reportedly imparts no significant impact on biochemical recurrence rates, suggesting that this factor is unlikely to play a significant role in our analysis.13, 14 Genetic factors, such as BRCA2 mutations, are associated with a higher risk of locally advanced disease and death from prostate cancer, although only a small minority of patients carries the mutation at diagnosis.15, 16 In addition, the youngest men may be diagnosed as a result of symptomatic disease, as opposed to a screening diagnosis, and, thus, may harbor more aggressive and higher volume disease. However, the proportion of men aged <45 years did not change significantly over the period of analysis despite a substantial increase in prostate-specific antigen screening, making it unlikely that symptomatic disease is more common in this group of patients. Finally, young men simply may have biologically more aggressive disease, because these men are more likely to be African American and to present with metastatic disease, suggesting a shorter latency and a different biology. Despite the small number of very young men diagnosed with high-grade disease, more completely defining a significant clinical phenotype in this set of very young men and its associated molecular signature may provide additional insights into the biology of aggressive prostate cancer.17

Our study has potential limitations that merit review. First, there is the potential for misclassification of grade and stage, because only a subset of men in our study underwent radical prostatectomy, which is the gold standard for obtaining accurate stage and grade. It is apparent in Table 3 that we attempted to control for this potential shortcoming by analyzing only patients who had postprostatectomy pathologic data available, and we observed that the results were very similar to the results from the analysis that included all patients, suggesting that misclassification bias minimally affected the results. Second, the SEER database does not report comorbidity measures, and the effects of comorbidity on OS were unmeasured. However, the most significant results of this analysis are the opposite of what may be expected with increasing comorbidity associated with age, because younger men had equivalent or worse outcomes compared with older men. If we had been able to adjust for comorbidity, then it most likely would have increased the magnitude of the hazard ratios associated with the youngest age group. Second, the group of patients who had ≥T4 disease was heterogeneous, and that heterogeneity may lead to imbalances between the different age groups. In a secondary analysis, the dataset demonstrated that the proportions of men who had lymph node-positive disease by age group were 8.1% for ages 35 years to 44 years, 6.1% for ages 45 years to 54 years, 6.9% for ages 55 years to 64 years, and 8.4% for ages 65 years to 74 years. There does not appear to be a bias for a greater proportion of lymph node-positive patients in the youngest men. The grouping of T4 patients with those who had lymph node-positive disease increases the heterogeneity of the group; however, because the large majority of men with T4 disease most likely harbor lymph node metastases, this appeared to be a logical grouping. Another consideration is that the SEER database lacks stage and grade information for approximately 40% of patients. We elected to analyze only those patients who had complete data available, because it appears unlikely that patient age would change the availability of clinical staging information. The proportion of missing data was 43%, 40%, 38%, 38% for the men ages 35 years to 44 years, 45 years to 54 years, 55 years to 64 years, and 65 years to 74 years, suggesting that there was no clear bias for the selective lack of information based on age. The analyses of the group as a whole and of the patients who underwent prostatectomy, as discussed above, demonstrate very similar results, suggesting that the missing data does not compromise the conclusions. Finally, the majority of young men diagnosed with prostate cancer are at low risk for cancer-specific mortality within 10 years of diagnosis, and the majority of men diagnosed with prostate cancer are aged ≥60 years; thus, the clinical applicability of these findings is limited to a defined small, yet clinically relevant population of men.

The proportion of young men with prostate cancer will continue to rise with the current and evolving practices of screening and treatment.18 The number of men with high-grade cancer also may continue to increase, and finding ways to prevent cancer-related morbidity and mortality in young men, who will remain at risk for the greatest period of time, is critical. In this SEER-based analysis, which reflects “real-world” practice, young men did not have uniformly better cancer-specific survival than older men, contrary to most institutional series. Our finding that high-grade cancer was associated with >25% cancer-related mortality at 10 years in men aged <55 emphasizes the significant incidence of disease-related morbidity in this group of men. These data provide a strong argument for the need to consider multimodality therapy for young men with high-risk disease and to support the ongoing neoadjuvant and adjuvant studies of systemic therapies, which could improve the efficacy of local therapy and the chance of achieving a cure for these men. The paradoxical effect of very young age and high-grade disease suggests that the biology of prostate cancer in young men may be inherently different and may provide new insights into the development and behavior of the disease. More detailed studies of the clinical phenotype and molecular changes in primary tumors that develop in very young men may provide additional tools to improve screening in young men and answer the question of whether there are important biologic differences that could lead to different systemic and local treatment strategies in these men.

Conflict of Interest Disclosures

Supported by the Veterans Affairs Cooperative Studies Program (CSP 553) and the Northwest Specialized Program of Research Excellence (P50 CA097186-06).