Department of Urology, University of California at Davis, Davis Medical Center, Sacramento, California
Corresponding author: Marc Dall'Era, MD, Department of Urology, University of California at Davis, Davis Medical Center, 4860 Y St, Suite 3500, Sacramento, CA 95817; Fax: (916) 734-8094; Marc.firstname.lastname@example.org
Prostate cancer mortality in the United States has declined by nearly 40% over the last 25 years. However, to the authors' knowledge, the contribution of prostate-specific antigen (PSA) screening for the early detection of prostate cancer remains unclear and controversial. In the current study, the authors attempted to determine whether improvements in survival over time among patients with metastatic prostate cancer have contributed to the decline in mortality.
Men aged ≥ 45 years who presented with de novo metastatic prostate cancer from 1988 to 2009 were identified within the California Cancer Registry. Overall survival and disease-specific survival were estimated using the Kaplan-Meier method. A multivariate analysis with Cox proportional hazards modeling was performed to adjust for different distributions of variables between groups.
A total of 19,336 men presented with de novo metastatic prostate cancer during the study period. On multivariate analysis, overall survival was found to be better for men diagnosed from 1988 through 1992 and 1993 through 1998 than for men diagnosed in the most recent era (hazards ratio, 0.78; 95% confidence interval, 0.72-0.85 [P < .001] and HR, 0.79; 95% confidence interval, 0.74-0.86 [P < .001]). There was no improvement in disease-specific survival observed when comparing the most contemporary men (those diagnosed between 2004 and 2009) with those diagnosed between 1988 and 1997.
Prostate cancer mortality in the United States has declined by nearly 40% over the last 25 years. The reasons for this are likely multifactorial and to the best of our knowledge the contribution from prostate-specific antigen (PSA) screening for the early detection of prostate cancer remains unclear and controversial. Large randomized PSA screening trials have demonstrated conflicting results, with a suggestion of improved disease-specific survival (DSS) with screening noted primarily among younger, healthier men.[2, 3] However, any benefit comes at the substantial cost of overdiagnosis and overtreatment of indolent disease, leading the US Preventative Services Task Force in 2012 to recommend against the routine use of PSA screening for the early detection of prostate cancer. If PSA screening at best only has a modest effect on reducing deaths from prostate cancer, it remains unclear which factors account for the observed changes in prostate cancer mortality over the past 25 years.
Recent modeling studies have suggested that improvements in the treatment of localized disease may contribute to some of the observed improvements in prostate cancer mortality over time independent of PSA screening. This finding is predicted by improvements in techniques and quality of care for patients with localized prostate cancer coincident with the profound stage migration triggered by PSA screening. These studies have estimated that changes in treatment are responsible for 22% to 33% of the observed decline in mortality.
We hypothesized that survival for men with metastatic prostate cancer has improved over time and thus may account for part of the observed improvement in prostate cancer mortality. To study survival improvements for men independent of PSA screening, we elected to analyze men presenting with de novo metastatic disease across the pre-PSA and post-PSA eras.
MATERIALS AND METHODS
We identified all men aged ≥ 45 years who presented with de novo metastatic prostate cancer within the California Cancer Registry (CCR) from 1988 through 2009. The CCR is a comprehensive population-based database that has captured all cancer cases diagnosed in California since 1988, aside from basal and squamous cell carcinoma of the skin and carcinoma in situ of the cervix. The CCR is a part of the National Cancer Institute's Surveillance, Epidemiology, and End Results program. For descriptive purposes, we stratified patients into quintiles by year of diagnosis. The primary outcome measure was overall survival (OS). Prostate cancer survival was measured as a secondary endpoint along with survival stratified by race. Disease incidence rates were calculated per 100,000 men at risk and were age adjusted. OS and DSS were estimated using the Kaplan-Meier method. A multivariate analysis with Cox proportional hazards modeling was performed to adjust for different distributions of variables between groups.
We identified a total of 19,336 men presenting with de novo metastatic prostate cancer during the study period. Exclusions were made for 325 men for whom there were insufficient data regarding race and 14 men for whom no survival time was recorded. Patient characteristics of the study cohort are shown in Table 1. The age-adjusted incidence rates of newly diagnosed metastatic prostate cancer by era of diagnosis are shown in Figure 1. The median age at the time of diagnosis decreased from 73 years to 71 years over the study period (P < .001). Of the entire cohort, 11,299 men (58%) died of prostate cancer. The unadjusted Kaplan-Meier OS and DSS estimates for patients with newly diagnosed metastatic prostate cancer stratified by era of diagnosis are shown in Figure 2 and Figure 3. Although these estimates demonstrated some significant improvements in survival for men diagnosed between 1993 and 2003 compared with men diagnosed between 1988 and 1992, no survival improvement was observed in the most contemporary cohort.
Table 1. Baseline Characteristics of the Study Cohort
Time Period of Diagnosis
Abbreviations: NOS, not otherwise specified; PI, Pacific Islander; WHO, World Health Organization.
Total no. of patients
Median age (range), y
Chemotherapy (alone or +)
Hormonal therapy only
Surgery plus radiation
Surgery plus hormonal therapy
Hormones plus radiation
Other treatment combinations/unknown
Table 2 displays the calculated hazards ratios (HRs) after adjusting for differences in baseline characteristics between the treatment groups. On multivariate analysis, OS was better for patients diagnosed between 1988 and 1992 and 1993 and 1998 than for men diagnosed in the most recent era (HR, 0.78; 95% confidence interval [95% CI], 0.72-0.85 [P < .0001] and HR, 0.79; 95% CI, 0.74-0.86 [P < .0001], respectively). There was no improvement in DSS noted when comparing the most contemporary men (those diagnosed between 2004 and 2009) with men diagnosed between 1988 and 1997. There was a slightly higher risk of prostate cancer death for men diagnosed between 1998 and 2003 compared with those diagnosed between 2004 and 2009 (HR, 1.09; 95% CI, 1.02-1.17 [P = .01]).
Table 2. Cox Proportional Hazards Models for OS and DSS for De Novo Metastatic Prostate Cancer Stratified by Year of Diagnosis
Race was associated with OS and DSS for the entire cohort, with Asian men demonstrating the best OS and DSS (Fig. 4). When stratified by era of diagnosis, the disparity between white and African American men was significant only for the pre-PSA era (1988-1992) (P = .0042 for OS and P = .0015 for DSS). There was no significant difference in survival noted between African American and white men presenting with metastatic prostate cancer in the post-PSA era.
The age-adjusted incidence rates of de novo metastatic prostate cancer were reported to decrease by 65% between 1988 and 2009. This well-demonstrated stage migration in prostate cancer is a direct result of widespread PSA screening over this time period. However, the impact of PSA screening on mortality remains unclear and controversial. Our hypothesis was that advances in managing metastatic prostate cancer and general end-of-life care would translate into improved survival for this group of men, contributing to the observed improvement in prostate cancer mortality since the introduction of PSA screening. In this analysis of men presenting with de novo metastatic prostate cancer, we were unable to demonstrate a consistent improvement in OS or DSS over time. Despite a dramatic decline in incidence rates, after adjusting for baseline covariates, there was a suggestion of poorer OS for contemporary men presenting with de novo metastatic disease. These data suggest that improvements in survival for patients with advanced disease have not contributed substantially to the observed decrease in prostate cancer mortality over this time period and that stage migration secondary to PSA screening plays a more prominent role.
A recent analysis was able to demonstrate some improvements in the median OS during the PSA era for men with advanced prostate cancer who were enrolled in Southwest Oncology Group (SWOG) clinical trials S8494, S8894, and S9346. This analysis demonstrated separation of the survival curves within 2 years, with clear advantages for patients from the SWOG S9346 trial. Our unadjusted curves indicated some separation after 40 months, which was not found to be statistically significant on multivariate analysis. These discordant results likely reflect differing patient populations between our studies (clinical trial vs general) and the differing time periods analyzed. The majority of men within the SWOG phase 3 clinical trials received aggressive systemic therapies with regular monitoring, and clinical trial enrollment has been shown to independently predict lower mortality for patients with certain solid malignancies. However, these findings provide further support that patients enrolled in clinical trials may have better survival outcomes compared with patients in general. Similar to our findings, data from the SWOG trials were also able to demonstrate comparable survival for African American and white men in the post-PSA era. These findings are likely also explained by the profound stage migration afforded by PSA screening, with men diagnosed during the contemporary period (between 2004 and 2009) with metastatic disease having a uniformly poor prognosis regardless of race. In a separate analysis, we were able to demonstrate that socioeconomic status plays a more important role in predicting DSS than race or even health insurance status (data not shown).
Androgen deprivation therapy remains the standard initial therapy for men with metastatic prostate cancer. Inevitably, the disease will progress to castration-resistant disease, and many of the treatment advances for this stage of disease are relatively recent. Docetaxel was approved in 2004 as initial systemic chemotherapy for patients with castration-resistant disease, with randomized trials demonstrating a mean survival advantage of 3.4 months over other regimens. Although we were unable to analyze treatments received over time with this data set, several studies have demonstrated that a number of patients still do not receive systemic therapy.[9, 10] More recently, improved survival has been shown with dendritic cell immunotherapy and novel agents targeting the androgen biosynthesis axis[11-13] or the androgen receptor. However, any effect of these agents has not been captured by the current study data.
The findings of the current study can be explained by the profound migration in stage at the time of diagnosis of prostate cancer with PSA screening. The rates of de novo metastatic prostate cancer have declined dramatically and it is conceivable that the less biologically aggressive tumors with a longer lead time are selected out by PSA screening more frequently than the aggressive phenotypes. Although there has been ample evidence of grade migration using the Gleason system over time, it is interesting to note that approximately 81.7% of tumors from the most recent stratum of men were considered to be high grade (World Health Organization 3/4) as opposed to only 49% of those from men diagnosed between 1988 and 1992. World Health Organization grading may also be subject to grade migration over time, but these data also suggest that more aggressive tumors are less likely to be diagnosed at a localized stage with PSA screening. However, PSA screening has clearly shifted the diagnostic lead time of many tumors with high metastatic potential. This may also explain why we were unable to find a difference over time and perhaps improvements in patient care are masked by the findings that men diagnosed with newly diagnosed metastatic disease during the contemporary period (between 2004 and 2009) have biologically more aggressive tumors. Tumors with a longer preclinical phase are selectively diagnosed earlier with PSA screening and there is supportive evidence that the aggressive treatment of high-risk early-stage disease is beneficial. Two randomized trials have shown a survival benefit with radical prostatectomy over conservative management, especially in the high-risk disease setting. At 10 years, the Scandinavian Prostate Cancer Group Trial-4 demonstrated a 17.2% absolute prostate cancer risk reduction for men in the high-risk group. Subgroup analysis from the Prostate Cancer Intervention Versus Observation Trial (PIVOT) also has suggested an OS advantage with surgery for men at higher risk. These trials support the belief that the aggressive treatment of men with high-risk PSA-detected disease can lead to improved survival. The findings of the current study suggest that de novo metastatic prostate cancer diagnosed in the contemporary period (between 2004 and 2009) represents a worse disease than previously believed, translating into no observed changes in DSS for this group of men as a whole.
The current analysis has several limitations. Using the CCR, we were only able to study men presenting with metastatic disease and therefore did not include men who progressed to metastatic disease after initial therapy. We were also unable to analyze treatments received over time (because treatment beyond the first course is not recorded in the CCR database) and what effect this may have on survival. We were able to adjust for the initial treatment received and this was included in our model. Although we were able to adjust for important baseline characteristics such as age, race, and socioeconomic status, we were unable to make adjustments for differences in such important variables as comorbidities or family history. The majority of patients from this cohort died of prostate cancer (58%), whereas the mortality rate from other causes was only 27%. We suspect that differences in comorbidity are unlikely to affect the results of the current study, especially given that other-cause mortality rate was relatively low.
Survival for men presenting with de novo metastatic prostate cancer has not improved substantially over the last 25 years and likely has not contributed to the observed declines in prostate cancer mortality. The stage migration associated with PSA screening has resulted in a dramatically reduced incidence of men presenting with metastatic prostate cancer, a diagnosis with a persistently poor prognosis. This effect of PSA testing likely has had an impact on reducing mortality from prostate cancer over the last 30 years. It has been suggested that perhaps with longer follow-up from the European Randomized Study of Screening for Prostate Cancer and the Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial, we may realize a more substantial benefit with PSA screening.
Supported by a University of California at Davis Cancer Center Epidemiology Shared Resource.