Abstract
Purpose
To provide patients and clinicians more accurate estimates of comorbidity-specific survival stratified by patient age, tumor stage, and tumor grade.
Patients and Methods
We conducted a 10-year competing risk analysis of 19,639 men 66 years of age and older identified by the Surveillance, Epidemiology, and End Results (SEER) program linked to Medicare program files. All men were diagnosed with localized prostate cancer and received no surgery or radiation within 180 days of diagnosis. The analysis was stratified by tumor grade and stage and by age and comorbidity at diagnosis classified using the Charlson comorbidity index. Underlying causes of death were obtained from SEER.
Results
During the first 10 years after diagnosis, men with moderately and poorly differentiated prostate cancer were more likely to die from causes other than their disease. Depending on patient age, Gleason score, and number of comorbidities present at diagnosis, 5-year overall mortality rates for men with stage T1c disease ranged from 11.7% (95% CI, 10.2% to 13.1%) to 65.7% (95% CI, 55.9% to 70.1%), and prostate cancer–specific mortality rates ranged from 1.1% (95% CI, 0.0% to 2.7%) to 16.3% (95% CI, 13.8% to 19.4%). Ten-year overall mortality rates ranged from 28.8% (95% CI, 25.3% to 32.6%) to 94.3% (95% CI, 87.4% to 100%), and prostate cancer–specific mortality rates ranged from 2.0% (95% CI, 0.0% to 5.3%) to 27.5% (95% CI, 21.5% to 36.5%).
Conclusion
Patients and clinicians should consider using comorbidity-specific data to estimate the threat posed by newly diagnosed localized prostate cancer and the threat posed by competing medical hazards.
INTRODUCTION
Because prostate cancer frequently has a long natural history, many professional organizations recommend that screening and treatment should be limited to men with a life expectancy greater than 10 years.1 The National Comprehensive Cancer Network prostate cancer practice guidelines suggest that radical surgery and/or radiation therapy should only be offered to men who are likely to survive more than 10 years, whereas the United States Preventive Services task force recently reaffirmed that men 75 years of age and older are unlikely to benefit from prostate cancer screening because relatively few of these men are likely to survive more than a decade.2,3
These recommendations have been validated by two widely publicized randomized trials evaluating the benefit of prostate-specific antigen (PSA) testing for prostate cancer. The Prostate, Lung, Colorectal, and Ovarian trial sponsored by the National Cancer Institute concluded that after 7 to 10 years of follow-up, prostate cancer mortality was very infrequent and did not differ significantly among men who were screened compared with the controls.4 The European Randomized Study of Screening for Prostate Cancer trial concluded that after 8 years of follow-up, PSA-based screening could reduce the death rate from prostate cancer by 20%, but at the risk of identifying many men who would not benefit from diagnosis and treatment.5
In contemporary clinical practice, treatment decisions are predominantly influenced by patients' life expectancy and tumor histology. Unfortunately, physicians often have difficulty providing accurate estimates of patients' life expectancy after initial cancer diagnosis.6 Patients have an even more distorted perception of the risks posed by their disease and the benefits of intervention. In a recently administered mail survey, Mohan et al7 demonstrated that patients with localized prostate cancer grossly underestimate their life expectancy without treatment and grossly overestimate the survival benefit associated with treatment.
We have previously published 10-year outcomes after conservative management of localized prostate cancer.8 These estimates did not highlight the impact of patient comorbidities. To provide patients and clinicians more accurate estimates of comorbidity-specific survival stratified by patient age, tumor stage, and tumor grade, we assembled a large population-based cohort of 19,639 men with localized stage T1c or T2 prostate cancer who were managed conservatively after diagnosis. The resulting competing risk analyses should assist patients and their physicians to evaluate the risks posed by localized prostate cancer and their existing medical conditions.
PATIENTS AND METHODS
Data Sources
Data were obtained from Medicare insurance program files linked to the population-based Surveillance, Epidemiology, and End Results (SEER) cancer registries. The SEER regions sampled approximately 14% of the US population before 2000 and 25% thereafter. The Medicare database includes information on approximately 97% of men age 65 years and older; 93% of patients are linked to the SEER database.
Cancer stage and grade for each case were abstracted from SEER data files. The SEER grading convention before 2003 characterized Gleason 2 to 4 tumors as well differentiated, Gleason 5 to 7 tumors as moderately differentiated, and Gleason 8 to 10 as poorly differentiated. Treatment information was obtained from both SEER and Medicare files. A Charlson comorbidity score was derived from Medicare claims during the year before prostate cancer diagnosis using a validated algorithm.9
Study Participants
The study cohort consisted of men 66 years of age and older who were diagnosed with localized prostate cancers between January 1, 1992, and December 31, 2005 (N = 157,381). The following men were excluded from the analysis: 95,256 men who received attempted curative therapy such as prostatectomy or radiation within 180 days of diagnosis, 6,603 men who died as a result of any cause within 180 days of diagnosis, 10,277 men who had other cancer diagnoses, 5,913 men with stage T1a or T1b disease, 258 men with well-differentiated disease, and 173 men who received androgen deprivation therapy before diagnosis. In addition, we excluded 16,333 men who did not have both Medicare Part A and Part B as their primary health insurance coverage during the study period, 2,537 men with missing data, and 392 men with an unknown cancer grade. The final population consisted of 14,130 men with Charlson score of 0, 3,535 men with Charlson score of 1, and 1,974 men with Charlson score ≥ 2. A summary of the Charlson scoring system is provided in Table 1 . Cancers detected by PSA testing were classified as T1c, whereas the remaining cancers were classified as T1a, T1b, or T2. In the SEER-Medicare database, the vital status of patients is derived from Medicare Enrollment Database, which is updated every night from the Social Security Administration. Overall mortality rates were derived from survival data available through December 31, 2007. Underlying causes of death were obtained from the SEER program, which derives this information from state death certificates. Usually this information becomes available within 2 years of the date of death. As a consequence, prostate cancer–specific mortality rates were derived from data available through December 31, 2005. Previous studies have confirmed that prostate cancer deaths can be identified reliably from death certificates if results are classified simply as death as a result of prostate cancer or death resulting from another cause.10,11
Table 1.
Charlson Comorbidity Index
| Condition | Assigned Weight |
|---|---|
| Myocardial infarction | 1 |
| Congestive heart failure | 1 |
| Peripheral vascular disease | 1 |
| Cerebrovascular disease | 1 |
| Dementia | 1 |
| Chronic pulmonary disease | 1 |
| Connective tissue disease | 1 |
| Ulcer disease | 1 |
| Liver disease, mild | 1 |
| Diabetes | 1 |
| Hemiplegia | 2 |
| Renal disease, moderate or severe | 2 |
| Diabetes with end organ damage | 2 |
| Any malignancy | 2 |
| Leukemia | 2 |
| Malignant lymphoma | 2 |
| Liver disease, moderate or severe | 3 |
| Metastatic solid malignancy | 6 |
Statistical Analyses
The primary study end points were time to death resulting from prostate cancer and time to death resulting from other causes, stratified by patient age, cancer grade, cancer stage, and Charlson comorbidity score at the time of diagnosis. Estimates of competing risks were computed using cumulative incidence functions with competing risk adjustments for death resulting from prostate cancer and from other causes as described in Putter, Fiocco, and Geskus.12,13 To provide more stable estimates of the survival curves, we used a nearest-neighbor hazard smoother with an Epanechnikov Kernel as implemented in the R statistical system (R Foundation for Statistical Computing, Vienna, Austria).14 Covariate adjustment for competing risks was conducted as described in Putter et al.13 Bootstrap confidence intervals for the 10-year survival rates were obtained using 1,000 bootstrap replications.15 Median follow-up time was computed following the method described in Clark et al16 and Schemper and Smith.17
RESULTS
Most patients (72%) in our study cohort had no significant comorbidities (Charlson score = 0), 18% had one significant comorbidity, and 10% had two or more comorbidities. About two thirds (66%) of the patients were classified as having moderately differentiated disease (Gleason score 5 to 7), and about one third (34%) had poorly differentiated disease (Gleason score 8 to 10). Approximately 45% of the patients had screen-detected disease (T1c), whereas 55% had palpable disease (T2). The median age at diagnosis was 77 years, and the median follow-up for the study population was 6 years.
At the end of the study period, 60%, 51%, and 41% of men with zero, one, or two or more comorbidities were still alive, respectively. Only 5% and 11% of men with moderately or poorly differentiated disease, respectively, died as a result of prostate cancer. Baseline characteristics and long-term outcomes of the study population classified by comorbidity are summarized in Table 2 . Sample sizes by age, comorbidity, and vital status as of December 31, 2005, are summarized in Table 3.
Table 2.
Characteristics of Men Receiving Active Surveillance/Conservative Management for Clinically Localized Prostate Cancer by Comorbidity (T1c/ T2)
| Characteristic | Comorbidity = 0 (n = 14,130) |
Comorbidity = 1 (n = 3,535) |
Comorbidity = 2+ (n = 1,974) |
|||
|---|---|---|---|---|---|---|
| No. | % | No. | % | No. | % | |
| Age at diagnosis, years | ||||||
| Median | 77 | 78 | 78 | |||
| IQR | 72-82 | 74-82 | 74-82 | |||
| Black race | 1,752 | 12.4 | 519 | 14.7 | 380 | 19.3 |
| Married at diagnosis | 8,746 | 61.9 | 2,074 | 58.7 | 1,100 | 55.7 |
| Urban residence | 11,579 | 81.9 | 2,946 | 83.3 | 1,641 | 83.1 |
| Zip code-level income, US$ | ||||||
| Median | 44,025 | 43,003 | 41,573 | |||
| IQR | 33,757-59,397 | 32,580-57,554 | 31,396-56,041 | |||
| SEER regions | ||||||
| Northeast | 2,632 | 18.6 | 745 | 21.1 | 406 | 20.6 |
| North-central | 2,715 | 19.2 | 657 | 18.6 | 427 | 21.6 |
| West | 6,633 | 46.9 | 1554 | 43.9 | 785 | 39.8 |
| South | 2,150 | 15.2 | 579 | 16.4 | 356 | 18.0 |
| Cancer grade, Gleason score | ||||||
| Moderately differentiated, 5-7 | 9,499 | 67.2 | 2,309 | 65.3 | 1,199 | 60.7 |
| Poorly differentiated, 8-10 | 4,631 | 32.8 | 1,226 | 34.7 | 775 | 39.3 |
| Clinical stage | ||||||
| T1c | 6,389 | 45.2 | 1,629 | 46.1 | 900 | 45.6 |
| T2 | 7,741 | 54.8 | 1,906 | 53.9 | 1,074 | 54.4 |
| Year of diagnosis | ||||||
| 1992-1996 | 2,757 | 19.5 | 540 | 15.3 | 296 | 14.9 |
| 1997-2003 | 11,373 | 80.5 | 2,995 | 84.7 | 1,678 | 85.0 |
| Use of PADT | 6,280 | 44.4 | 1,893 | 53.6 | 1,093 | 55.4 |
| Vital status at last follow-up: alive as of December 31, 2007 | 8,512 | 60.2 | 1,808 | 51.2 | 805 | 40.8 |
Abbreviations: IQR, interquartile range; SEER, Surveillance, Epidemiology, and End Results; PADT, primary androgen deprivation therapy.
Table 3.
Sample Sizes by Age, Comorbidity, and Vital Status As of December 31, 2005
| Characteristic | Age at Diagnosis |
|||
|---|---|---|---|---|
| 66-74 Years |
75+ Years |
|||
| No. | % | No. | % | |
| Moderately differentiated cancer | ||||
| Gleason 5-7, T1c, comorbidity = 0 | ||||
| Overall sample size | 1,949 | 100 | 2,689 | 100 |
| Died of prostate cancer | 33 | 2 | 126 | 5 |
| Died of other causes | 203 | 10 | 603 | 22 |
| Alive | 1,713 | 88 | 1,960 | 73 |
| Gleason 5-7, T1c, comorbidity = 1 | ||||
| Overall sample size | NA | 771 | 100 | |
| Died of prostate cancer | NA | 32 | 4 | |
| Died of other causes | 72 | 19 | 245 | 32 |
| Alive | 295 | 80 | 494 | 64 |
| Gleason 5-7, T1c, comorbidity ≥ 2 | ||||
| Overall sample size | NA | 394 | 100 | |
| Died of prostate cancer | NA | 12 | 3 | |
| Died of other causes | 60 | 30 | 150 | 38 |
| Alive | 131 | 66 | 232 | 59 |
| Gleason 5-7, T2, comorbidity = 0 | ||||
| Overall sample size | 1,811 | 100 | 3,050 | 100 |
| Died of prostate cancer | 98 | 5 | 231 | 8 |
| Died of other causes | 289 | 16 | 922 | 30 |
| Alive | 1,424 | 79 | 1,897 | 62 |
| Gleason 5-7, T2, comorbidity = 1 | ||||
| Overall sample size | 355 | 100 | 813 | 100 |
| Died of prostate cancer | 20 | 6 | 47 | 6 |
| Died of other causes | 89 | 25 | 327 | 40 |
| Alive | 246 | 69 | 439 | 54 |
| Gleason 5-7, T2, comorbidity ≥ 2 | ||||
| Overall sample size | NA | 433 | 100 | |
| Died of prostate cancer | NA | 26 | 6 | |
| Died of Other Causes | 80 | 46 | 217 | 50 |
| Alive | 92 | 53 | 190 | 44 |
| Poorly differentiated | ||||
| Gleason 8-10, T1c, comorbidity = 0 | ||||
| Overall sample size | 481 | 100 | 1,270 | 100 |
| Died of prostate cancer | 42 | 9 | 138 | 11 |
| Died of other causes | 37 | 8 | 214 | 17 |
| Alive | 402 | 84 | 918 | 72 |
| Gleason 8-10, T1c, comorbidity = 1 | ||||
| Overall sample size | NA | 363 | 100 | |
| Died of prostate cancer | NA | 26 | 7 | |
| Died of other causes | 17 | 14 | 89 | 25 |
| Alive | 102 | 82 | 248 | 68 |
| Gleason 8-10, T1c, comorbidity ≥ 2 | ||||
| Overall sample size | NA | 222 | 100 | |
| Died of prostate cancer | NA | 17 | 8 | |
| Died of other causes | 29 | 34 | 61 | 27 |
| Alive | 51 | 60 | 144 | 65 |
| Gleason 8-10, T2, comorbidity = 0 | ||||
| Overall sample size | 813 | 100 | 2,067 | 100 |
| Died of prostate cancer | 111 | 14 | 288 | 14 |
| Died of other causes | 114 | 14 | 536 | 26 |
| Alive | 588 | 72 | 1,243 | 60 |
| Gleason 8-10, T2, comorbidity = 1 | ||||
| Overall sample size | NA | 563 | 100 | |
| Died of prostate cancer | NA | 54 | 10 | |
| Died of other causes | 44 | 25 | 165 | 29 |
| Alive | 125 | 71 | 344 | 61 |
| Gleason 8-10, T2, comorbidity ≥ 2 | ||||
| Overall sample size | NA | 365 | 100 | |
| Died of prostate cancer | NA | 27 | 7 | |
| Died of other causes | 29 | 28 | 117 | 32 |
| Alive | 66 | 64 | 221 | 61 |
Abbreviation: NA, not available because Surveillance, Epidemiology, and End Results–Medicare privacy rules prohibit disclosure of numbers of less than 11 in any specific cell.
Appendix Figures A1 and A2 (online only) illustrate the competing risk of death according to age at diagnosis, cancer stage, cancer grade, and comorbidity score. Our results demonstrate that comorbidities negatively impact overall survival in men with localized prostate cancer. Men with stage T1c Gleason score 5 to 7 prostate cancer and no comorbidities have a 2% to 4% chance of dying as a result of prostate cancer within 5 years of diagnosis and a 5% to 14% chance within 10 years. During this same period they have a 12% to 48% chance of dying as a result of any cause within 5 years of diagnosis and a 29% to 83% chance within 10 years, depending on their age at diagnosis and comorbidity. Men with two or more comorbidities have a 43% to 48% chance of dying as a result of any cause within 5 years and a 74% to 83% chance within 10 years, depending on their age at diagnosis. Prostate cancer mortality rates for men with stage T2 Gleason score 5 to 7 prostate cancer are only slightly higher than the rates for men with stage T1c disease.
Men with stage T1c Gleason score 8 to 10 localized prostate cancer also have a substantial risk of dying as a result of competing hazards. Men with no comorbidities have an equal chance of dying as a result of prostate cancer or a competing hazard as compared with men with two or more comorbidities, who have a substantially higher risk of dying as a result of a competing medical hazard. Men with stage T2 Gleason score 8 to 10 prostate cancer have a similar risk of dying as a result of prostate cancer or a competing hazard.
Table 4 summarizes the 5- and 10-year prostate cancer–specific and overall mortality rates for men with screen-detected (stage T1c) moderately and poorly differentiated disease. For men age 66 to 74 years, the overall 5-year mortality rate for men with Gleason score 5 to 7 disease rises from 11.7 (95% CI, 10.2 to 13.1) to 42.5 (95% CI, 36.1 to 48.5), and the overall 10-year mortality rate rises from 28.8 (95% CI, 25.3 to 32.6) to 83.1 (95% CI, 67.4 to 97.2) per 100 as the number of comorbidities present at diagnosis increases from zero to two or more. In contrast, these men face 5- and 10-year prostate cancer–specific mortality rates of 1.6 (95% CI, 1.1 to 2.4) to 4.3 (95% CI, 1.6 to 8.3) and 4.8 (95% CI, 2.8 to 8.4) to 5.3 (95% CI, 2.5 to 10.0) per 100, respectively. For men with Gleason score 8 to 10 disease, the overall 5- and 10-year mortality rates ranged from 26.4 (95% CI, 22.2 to 30.8) to 52.0 (95% CI, 42.1 to 64.5) and 55.0 (95% CI, 43.9 to 65.9) to 64.3 (95% CI, 52.0 to 84.9) per 100, respectively. As expected, the 5- and 10-year prostate cancer–specific mortality rates were somewhat higher in this high-risk group, ranging from 13.6 (95% CI, 9.6 to 17.8) to 9.6 (95% CI, 2.4 to 19.3) and 25.7 (95% CI, 15.9 to 40.6) to 13.7 (95% CI, 2.7 t o 33.4) per 100 for men with no comorbidities to two or more comorbidities, respectively.
Table 4.
Overall and Prostate Caner–Specific Mortality Rates for Men by Years With Localized Prostate Cancer (T1c)
| Characteristic | Age at Diagnosis |
|||||||
|---|---|---|---|---|---|---|---|---|
| 66-74 Years |
75+ Years |
|||||||
| 5-Year Mortality |
10-Year Mortality |
5-Year Mortality |
10-Year Mortality |
|||||
| Rate per 100* | 95% CI† | Rate per 100* | 95% CI† | Rate per 100* | 95% CI† | Rate per 100* | 95% CI† | |
| T1c, Gleason 5-7, comorbidity = 0 | ||||||||
| Overall mortality | 11.7 | 10.2 to 13.1 | 28.8 | 25.3 to 32.6 | 26.3 | 24.8 to 28.0 | 67.1 | 63.3 to 72.4 |
| Prostate cancer specific mortality | 1.6 | 1.1 to 2.4 | 4.8 | 2.8 to 8.4 | 4.4 | 3.4 to 5.1 | 14.0 | 10.6 to 20.9 |
| T1c, Gleason 5-7, comorbidity = 1 | ||||||||
| Overall mortality | 25.3 | 20.7 to 29.5 | 50.5 | 41.5 to 59.2 | 39.4 | 35.9 to 42.5 | 76.8 | 70.5 to 82.9 |
| Prostate cancer–specific mortality | 1.1 | 0.0 to 2.7 | 2.0 | 0.0 to 5.3 | 5.1 | 3.3 to 7.2 | 9.1 | 5.5 to 14.4 |
| T1c, Gleason 5-7, comorbidity ≥ 2 | ||||||||
| Overall mortality | 42.5 | 36.1 to 48.5 | 83.1 | 67.4 to 97.2 | 48.1 | 42.7 to 52.7 | 74.4 | 63.7 to 84.7 |
| Prostate cancer–specific mortality | 4.3 | 1.6 to 8.3 | 5.3 | 2.5 to 10.0 | 4.0 | 1.7 to 6.5 | 5.0 | 2.5 to 8.7 |
| T1c, Gleason 8-10, comorbidity = 0 | ||||||||
| Overall mortality | 26.4 | 22.2 to 30.8 | 55.0 | 43.9 to 65.9 | 41.4 | 38.3 to 44.0 | 77.0 | 71.5 to 82.5 |
| Prostate cancer specific mortality | 13.6 | 9.6 to 17.8 | 25.7 | 15.9 to 40.6 | 16.3 | 13.8 to 19.4 | 27.5 | 21.5 to 36.5 |
| T1c, Gleason 8-10, comorbidity = 1 | ||||||||
| Overall mortality | 30.7 | 23.7 to 41.0 | 52.0 | 38.0 to 77.0 | 47.2 | 41.8 to 52.5 | 92.4 | 79.3 to 99.7 |
| Prostate cancer–specific mortality | 11.6 | 3.0 to 23.4 | 20.2 | 4.1 to 46.6 | 11.2 | 7.3 to 16.1 | 23.7 | 9.5 to 44.7 |
| T1c, Gleason 8-10, comorbidity ≥ 2 | ||||||||
| Overall mortality | 52.0 | 42.1 to 64.5 | 64.3 | 52.0 to 84.9 | 65.7 | 55.9 to 70.1 | 94.3 | 87.4 to 100.0 |
| Prostate cancer–specific mortality | 9.6 | 2.4 to 19.3 | 13.7 | 2.7 to 33.4 | 12.8 | 7.3 to 18.9 | 18.8 | 9.3 to 36.8 |
The mortality rates were derived by using smoothed cumulative incidence curves as described in the text.
CIs were estimated using a bootstrap with 1,000 replications.
For men age 75 years and older at diagnosis, 5- and 10-year overall mortality rates ranged from 26.3 (95% CI, 24.8 to 28.0) to 39.4 (95% CI, 35.9 to 42.5) and 67.1 (95% CI, 63.3 to 72.4) to 76.8 (95% CI, 70.5 to 82.9) per 100 for those men with Gleason score 5 to 7 disease, respectively. These rates are somewhat higher than those noted for younger men. Five- and 10-year prostate cancer–specific rates were also somewhat higher, ranging from 4.4 (95% CI, 3.4 to 5.1) to 4.0 (95% CI, 1.7 to 6.5) and 14.0 (95% CI, 10.6 to 20.9) to 5.0 (95% CI, 2.5 to 8.7) per 100 for men with no comorbidities to two or more comorbidities, respectively. For men age 75 years and older at diagnosis with Gleason score 8 to 10 disease, the 5- and 10-year overall mortality rates ranged from 41.4 (95% CI, 38.3 to 44.0) to 65.7 (95% CI, 55.9 to 70.1) and 77.0 (95% CI, 71.5 to 82.5) to 94.3 (95% CI, 87.4 to 100) per 100, respectively. The 5- and 10-year prostate cancer–specific mortality rates were comparable to those of the younger men, ranging from 16.3 (95% CI, 13.8 to 19.4) to 12.8 (95% CI, 7.3 to 18.9) and 27.5 (95% CI, 21.5 to 36.5) to 18.8 (95% CI, 9.3 to 36.8), respectively, per 100.
Table 5 summarizes the 5- and 10-year prostate cancer–specific and overall mortality rates for men with stage T2 moderately and poorly differentiated disease. Overall, the 5- and 10-year rates and trends were comparable to those of the men presenting with T1c disease. Both the overall mortality rate and the prostate cancer–specific mortality rates for men with stage T2 disease were slightly higher when compared with those of men with stage T1c disease. The 5- and 10-year prostate cancer–specific mortality rates were also somewhat higher among men with Gleason score 8 to 10 disease when compared with men with Gleason score 5 to 7 disease for men with no comorbidities to two or more comorbidities.
Table 5.
Overall and Prostate Cancer–Specific Mortality Rates for Men by Years With Localized Prostate Cancer (T2)
| Characteristic | Age at Diagnosis |
|||||||
|---|---|---|---|---|---|---|---|---|
| 66-74 Years |
75+ Years |
|||||||
| 5-Year Mortality |
10-Year Mortality |
5-Year Mortality |
10-Year Mortality |
|||||
| Rate per 100* | 95% CI† | Rate per 100* | 95% CI† | Rate per 100* | 95% CI† | Rate per 100* | 95% CI† | |
| T2, Gleason 5-7, comorbidity = 0 | ||||||||
| Overall mortality | 14.0 | 12.7 to 15.5 | 35.5 | 32.3 to 39.0 | 31.2 | 29.9 to 32.9 | 67.4 | 64.9 to 75.3 |
| Prostate cancer–specific mortality | 3.7 | 2.9 to 4.6 | 9.5 | 7.5 to 11.7 | 6.6 | 5.5 to 7.4 | 14.2 | 12.0 to 16.8 |
| T2, Gleason 5-7, comorbidity = 1 | ||||||||
| Overall mortality | 26.4 | 22.0 to 31.0 | 65.5 | 56.4 to 79.2 | 43.3 | 39.7 to 46.4 | 77.3 | 72.5 to 82.8 |
| Prostate cancer–specific mortality | 5.7 | 3.4 to 8.5 | 11.9 | 6.9 to 23.9 | 5.5 | 4.1 to 7.2 | 13.3 | 7.9 to 21.3 |
| T2, Gleason 5-7, comorbidity ≥ 2 | ||||||||
| Overall mortality | 40.4 | 33.9 to 48.2 | 74.6 | 64.2 to 85.1 | 57.8 | 53.6 to 62.1 | 88.9 | 84.0 to 94.4 |
| Prostate cancer specific mortality | 0.6 | 0.0 to 2.0 | 1.0 | 0.0 to 3.7 | 6.0 | 3.8 to 8.7 | 9.9 | 5.9 to 18.2 |
| T2, Gleason 8-10, comorbidity = 0 | ||||||||
| Overall mortality | 31.5 | 28.0 to 34.3 | 61.5 | 55.0 to 67.1 | 50.0 | 47.7 to 52.4 | 83.4 | 80.2 to 86.7 |
| Prostate cancer–specific mortality | 18.2 | 14.6 to 21.5 | 23.9 | 23.9 to 35.6 | 18.9 | 16.5 to 20.9 | 27.9 | 24.3 to 31.7 |
| T2, Gleason 8-10, comorbidity = 1 | ||||||||
| Overall mortality | 42.7 | 35.6 to 51.1 | 77.2 | 63.7 to 93.6 | 56.5 | 51.6 to 60.5 | 87.6 | 81.4 to 93.8 |
| Prostate cancer–specific mortality | 5.5 | 1.7 to 10.2 | 8.7 | 2.1 to 18.9 | 14.8 | 10.8 to 19.1 | 19.6 | 14.0 to 25.8 |
| T2, Gleason 8-10, comorbidity ≥ 2 | ||||||||
| Overall mortality | 51.2 | 40.8 to 60.1 | 93.8 | 80.1 to 99.9 | 63.7 | 58.8 to 69.2 | 88.8 | 80.0 to 99.5 |
| Prostate cancer–specific mortality | 13.0 | 5.4 to 23.1 | 18.4 | 7.6 to 37.6 | 13.8 | 8.3 to 18.3 | 15.7 | 9.6 to 22.6 |
The mortality rates were derived by using smoothed cumulative incidence curves as described in the text.
CIs were estimated using a bootstrap with 1,000 replications.
For men age 75 years and older at diagnosis, the 5- and 10-year overall mortality rates for men with Gleason score 5 to 7 disease were somewhat higher than those of younger men. Five- and 10-year prostate cancer–specific rates were also somewhat higher. For men age 75 years and older diagnosed with Gleason score 8 to 10 disease, the 5- and 10-year overall mortality rates were comparable to those of the younger men.
DISCUSSION
An accurate estimate of overall and disease-specific survival is critical for an informed discussion of treatment options among men with newly diagnosed prostate cancer. Long-term survival depends, in part, on risks posed by the tumor and risks posed by existing comorbidity. Our study, the largest population-based study in the United States, shows that a higher comorbidity score is, in general, associated with higher overall mortality and lower prostate cancer–specific mortality.
Over a decade ago, we explored these issues using data from a population-based cohort of men diagnosed with localized prostate cancer in Connecticut who did not receive either surgery or radiation therapy.18 We demonstrated that an assessment of patient comorbidities using standardized instruments could be used to stratify patients according to their risk of dying from competing medical hazards. This concept has been validated by several authors who have demonstrated that the measurement of patient comorbidities combined with patient age and tumor grade can provide estimates of long-term survival with moderate accuracy.19–21 The prediction rules developed have c-statistics in the range of 0.70 to 0.84, suggesting that survival estimates will be correct approximately three of every four times. Unfortunately, the estimates provided in our 1996 publication were based on a cohort of patients diagnosed before the advent of PSA testing. They do not incorporate the impact of the lead time associated with screening for PSA, and as a consequence, they underestimate long-term survival outcomes.
The tables presented in this article as well as the figures available in the Appendix (online only) use the Charlson comorbidity instrument to assess patient comorbidities. This instrument was originally developed to predict 1-year mortality in patients suffering from breast cancer, but has since been validated for use in estimating long-term survival among patients with multiple different cancers.22,23 It has the added advantage of being validated in a clinical setting, as well as in the analysis of large administrative data sets.24
The impact of PSA testing most likely explains the substantial improvement in survival that we observed in this report when compared with our earlier analysis conducted more than a decade ago. PSA testing advances the date of diagnosis by as much as 13 years, resulting in longer patient survival after diagnosis.25 Furthermore, systematic upgrading of Gleason scores has resulted in an apparent survival improvement for men classified by tumor grade.26
Our results have several limitations. Our analysis only involves men age 66 years and older. Men younger than age 66 are much less likely to succumb to a competing medical hazard, and therefore our results cannot be applied to younger patients. Our results could not be stratified by individual tumor grade. Before 2003, the SEER system grouped Gleason scores 5 to 7 together as moderately differentiated disease and Gleason scores 8 to 10 together as poorly differentiated disease. Our report most likely overestimates prostate cancer–specific survival for men with Gleason 7 disease and underestimates prostate cancer–specific survival for men with Gleason 5 disease. This is balanced by the fact that many of these men received antiandrogen therapy, which may negatively impact overall survival, thereby underestimating prostate cancer–specific survival and overestimating overall survival.
Our follow-up was limited to 10 years. Most men with Charlson score ≥ 2 succumbed to a competing medical hazard during that time period. Men with no or one comorbidity often survived for at least 10 years. Longer follow-up will be needed to determine the relative impact of prostate cancer and competing medical hazards on long-term survival. Finally, we do not have data concerning why the men identified in this study chose active surveillance or watchful waiting as their primary treatment option. There may be unmeasured patient or disease factors that impact prostate specific–survival or all-cause survival. Accordingly, our results may or may not be applicable to men who undergo surgery or radiation therapy.
Our analysis has several strengths. The study was population-based, encompassed a wide geographic area, and had a large sample size, giving us sufficient statistical power to provide stable survival estimates. Our results also report outcomes specific for older men. Most case series reports do not include information concerning outcomes among men older than 70 years. Our analysis has information on outcomes of approximately 20,000 men who were managed conservatively after a diagnosis of localized prostate cancer.
In summary, our findings suggest that relatively few men diagnosed with localized prostate cancer older than 65 years will die as a result of prostate cancer within 10 years of diagnosis. Most men with either no or one comorbidity will survive at least 10 years, whereas men with two or more comorbidities have a substantial risk of dying as a result of a competing medical hazard within this time frame. Men with poorly differentiated disease face a more substantial risk of dying as a result of prostate cancer when compared with men with moderately differentiated disease and an equivalent risk of dying as a result of competing medical hazards. Patients and clinicians should use the tables presented in this article (and the figures presented in the Appendix) to understand the risks posed by prostate cancer and by competing medical hazards before selecting the therapeutic approach that is most appropriate for them.
Acknowledgment
We thank the Applied Research Branch, Division of Cancer Prevention and Population Science, National Cancer Institute, the Office of Information Services, and the Office of Strategic Planning, Center for Medicare and Medicaid Services; Information Management Services, and the Surveillance, Epidemiology, and End Results (SEER) Program tumor registries in the creation of the SEER–Medicare database; and Shunhua Shen, supported by the Cancer Institute of New Jersey, for her outstanding technical assistance.
Appendix
Fig A1.
Competing risk of mortality by age at diagnosis, cancer stage, grade, and comorbidity: cancer stage T1c. White area, probability of being alive; light blue area, non–prostate cancer mortality; dark blue area, prostate cancer mortality.
Fig A2.
Competing risk of mortality by age at diagnosis, cancer stage, grade, and comorbidity: cancer stage T2. White area, probability of being alive; light blue area, non–prostate cancer mortality; dark blue area, prostate cancer mortality.
Footnotes
Supported by the National Cancer Institute Grant No. R01 CA116399, and Cancer Institute of New Jersey biometric shared resources (National Cancer Institute Grant No. CA7272010).
Presented in part at the 2011 Genitourinary Cancers Symposium, February 17-19, 2011, Orlando, FL.
The National Cancer Institute did not play any role in the design and conduct of the study, analysis and interpretation of the data, and preparation of the manuscript but did collect and manage the data, as well as review and approve the manuscript.
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST
Although all authors completed the disclosure declaration, the following author(s) indicated a financial or other interest that is relevant to the subject matter under consideration in this article. Certain relationships marked with a “U” are those for which no compensation was received; those relationships marked with a “C” were compensated. For a detailed description of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
Employment or Leadership Position: None Consultant or Advisory Role: Peter C. Albertsen, GlaxoSmithKline (C), Ferring Pharmaceuticals (C), Blue Cross and Blue Shield (C), sanofi-aventis (U); Dirk F. Moore, InnoCentive (C) Stock Ownership: None Honoraria: None Research Funding: Grace L. Lu-Yao, OHL Foundation Expert Testimony: None Other Remuneration: None
AUTHOR CONTRIBUTIONS
Conception and design: Peter C. Albertsen, Dirk F. Moore, Grace L. Lu-Yao
Financial support: Grace L. Lu-Yao
Collection and assembly of data: Dirk F. Moore, Hui Li, Grace L. Lu-Yao
Data analysis and interpretation: Peter C. Albertsen, Dirk F. Moore, Weichung Shih, Yong Lin, Hui Li, Grace L. Lu-Yao
Manuscript writing: Peter C. Albertsen, Dirk F. Moore, Weichung Shih, Yong Lin, Hui Li, Grace L. Lu-Yao
Final approval of manuscript: Peter C. Albertsen, Dirk F. Moore, Weichung Shih, Yong Lin, Hui Li, Grace L. Lu-Yao
REFERENCES
- 1.Aus G, Abbou CC, Bolla M, et al. European Association of Urology: EAU guidelines on prostate cancer. Eur Urol. 2005;48:546–551. doi: 10.1016/j.eururo.2005.06.001. [DOI] [PubMed] [Google Scholar]
- 2.Scardino P. Update: NCCN prostate cancer Clinical Practice Guidelines. J Natl Compr Canc Netw. 2005;3(suppl 1):S29–S33. [PubMed] [Google Scholar]
- 3.US Preventive Services Task Force Screening for prostate cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2008;149:185–191. doi: 10.7326/0003-4819-149-3-200808050-00008. [DOI] [PubMed] [Google Scholar]
- 4.Andriole GL, Grubb RL, III, Buys SS, et al. Mortality results from a randomized prostate-cancer screening trial. N Engl J Med. 2009;360:1310–1319. doi: 10.1056/NEJMoa0810696. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Schröder FH, Hugosson J, Roobol MJ, et al. Screening and prostate-cancer mortality in a randomized European study. N Engl J Med. 2009;360:1320–1328. doi: 10.1056/NEJMoa0810084. [DOI] [PubMed] [Google Scholar]
- 6.Wilson JRM, Clarke MG, Ewings P, et al. The assessment of patient life expectancy: How accurate are urologists and oncologists? BJU Int. 2005;95:794–798. doi: 10.1111/j.1464-410X.2005.05403.x. [DOI] [PubMed] [Google Scholar]
- 7.Mohan R, Beydoun H, Barnes-Ely ML, et al. Patients' survival expectations before localized prostate cancer treatment by treatment status. J Am Board Fam Med. 2009;22:247–256. doi: 10.3122/jabfm.2009.03.080200. [DOI] [PubMed] [Google Scholar]
- 8.Lu-Yao GL, Albertsen PC, Moore DF, et al. Outcomes of localized prostate cancer following conservative management. JAMA. 2009;302:1202–1209. doi: 10.1001/jama.2009.1348. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Klabunde CN, Potosky AL, Legler JM, et al. Development of a comorbidity index using physician claims data. J Clin Epidemiol. 2000;53:1258–1267. doi: 10.1016/s0895-4356(00)00256-0. [DOI] [PubMed] [Google Scholar]
- 10.Albertsen PC, Walters S, Hanley JA. A comparison of cause of death determination in men previously diagnosed with prostate cancer who died in 1985 or 1995. J Urol. 2000;163:519–523. [PubMed] [Google Scholar]
- 11.Penson DF, Albertsen PC, Nelson PS, et al. Determining cause of death in prostate cancer: Are death certificates valid? J Natl Cancer Inst. 2001;93:1822–1823. doi: 10.1093/jnci/93.23.1822. [DOI] [PubMed] [Google Scholar]
- 12.Cox DR, Oakes D. New York, NY: Chapman & Hall; 1984. Analysis of Survival Data. [Google Scholar]
- 13.Putter H, Fiocco M, Geskus RB. Tutorial in biostatistics: Competing risks and multi-state models. Stat Med. 2007;26:2389–2430. doi: 10.1002/sim.2712. [DOI] [PubMed] [Google Scholar]
- 14.Müller H, Wang J. Hazard rates estimation under random censoring with varying kernels and bandwidths. Biometrics. 1994;50:61–76. [PubMed] [Google Scholar]
- 15.Efron B, Tibshirani R. New York, NY: Chapman & Hall; 1994. An Introduction to the Bootstrap. [Google Scholar]
- 16.Clark TG, Bradburn MJ, Love SB, et al. Survival analysis part I: Basic concepts and first analyses. Br J Cancer. 2003;89:232–238. doi: 10.1038/sj.bjc.6601118. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Schemper M, Smith TL. A note on quantifying follow-up in studies of failure time. Control Clin Trials. 1996;17:343–346. doi: 10.1016/0197-2456(96)00075-x. [DOI] [PubMed] [Google Scholar]
- 18.Albertsen PC, Fryback DG, Storer BE, et al. The impact of co-morbidity on life expectancy among men with localized prostate cancer. J Urol. 1996;156:127–132. [PubMed] [Google Scholar]
- 19.Tewari A, Johnson CC, Divine G, et al. Long-term survival probability in men with clinically localized prostate cancer: A case-control, propensity modeling study stratified by race, age, treatment and comorbidities. J Urol. 2004;171:1513–1519. doi: 10.1097/01.ju.0000117975.40782.95. [DOI] [PubMed] [Google Scholar]
- 20.Cowen ME, Halasyamani LK, Kattan MW. Predicting life expectancy in men with clinically localized prostate cancer. J Urol. 2006;175:99–103. doi: 10.1016/S0022-5347(05)00018-2. [DOI] [PubMed] [Google Scholar]
- 21.Walz J, Gallina A, Saad F, et al. A nomogram predicting 10-year life expectancy in candidates for radical prostatectomy or radiotherapy for prostate cancer. J Clin Oncol. 2007;25:3576–3581. doi: 10.1200/JCO.2006.10.3820. [DOI] [PubMed] [Google Scholar]
- 22.Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chron Dis. 1987;40:373–383. doi: 10.1016/0021-9681(87)90171-8. [DOI] [PubMed] [Google Scholar]
- 23.Klabunde CN, Potosky AL, Legler JM, et al. Development of a comorbidity index using physician claims data. J Clin Epidemiol. 2000;53:1258–1267. doi: 10.1016/s0895-4356(00)00256-0. [DOI] [PubMed] [Google Scholar]
- 24.Klabunde DN, Legler JM, Warren JL, et al. A refined comorbidity measurement algorithm for claims-based studies of breast, prostate, colorectal and lung cancer patients. Ann Epidemiol. 2007;17:584–590. doi: 10.1016/j.annepidem.2007.03.011. [DOI] [PubMed] [Google Scholar]
- 25.Draisma G, Boer R, Otto SJ, et al. Lead times and overdetection due to prostate-specific antigen screening: Estimates from the European Randomized Study of Screening for Prostate Cancer. J Natl Cancer Inst. 2003;95:868–878. doi: 10.1093/jnci/95.12.868. [DOI] [PubMed] [Google Scholar]
- 26.Albertsen PC, Hanley JA, Barrows GH, et al. Prostate cancer and the Will Rogers phenomenon. J Natl Cancer Inst. 2005;97:1248–1253. doi: 10.1093/jnci/dji248. [DOI] [PubMed] [Google Scholar]