Long-Term Survival Rates of Patients With Prostate Cancer in the Prostate-Specific Antigen Screening Era: Population-Based Estimates for the
http://www.100md.com
《临床肿瘤学》
the German Centre for Research on Ageing, Heidelberg, Germany
ABSTRACT
METHODS: Five- and 10-year absolute and relative survival rates for the year 2000 were derived from the 1973 to 2000 database of the Surveillance, Epidemiology and End Results Program using the recently introduced period analysis methodology.
RESULTS: Overall, 5- and 10-year relative survival rates were approximately 99% and 95%; that is, excess mortality compared with the general population was as low as 1% and 5% within 5 and 10 years following diagnosis, respectively. Two-thirds of patients were diagnosed with well or moderately differentiated localized/regional prostate cancer, and among these patients, 5- and 10-year relative survival rates were above 100% (indicating the lack of any excess mortality) at all ages.
CONCLUSION: While the value of PSA screening for lowering mortality due to prostate cancer remains to be shown by randomized clinical trials, the majority of patients diagnosed with prostate cancer in the PSA screening era do not have excess mortality compared to the general population under current patterns of medical care. This information may be important for both clinical management of, and for patients' coping with, the disease.
INTRODUCTION
METHODS
The SEER Program is the most authoritative source of information on cancer incidence and survival in the United States, and it is considered as the standard for quality among cancer registries around the world. Quality control has been an integral part of SEER since its inception. Every year, studies are conducted in the SEER areas to evaluate the quality and completeness of the data being reported (SEER's standard for case ascertainment is 98%).
Data included in the 1973 to 2000 SEER database are from population-based cancer registries in Connecticut, New Mexico, Utah, Iowa, Hawaii, Atlanta (Georgia), Detroit (Michigan), Seattle-Puget Sound (Washington), and San Francisco-Oakland (California), which together cover a population of about 30 million people. Geographic areas were selected for inclusion in the SEER Program based on their ability to operate and maintain a high-quality, population-based cancer reporting system and for their epidemiologically significant population subgroups. The SEER population is comparable to the general United States population with regard to measures of poverty and education, even though it tends to be somewhat more urban and has a higher proportion of foreign-born persons than the latter.
This database included 183,484 patients with a first diagnosis of prostate cancer between 1990 and 2000 who have been followed for vital status until the end of the year 2000. After exclusion of 415 patients (0.2%) who were reported by autopsy only, 1,134 patients (0.6%) who were reported by death certificate only, and another 1,330 patients (0.7%) with unknown month of diagnosis, there remained 180,605 patients (98.4%) for the survival analysis.
In order to provide the most up-to-date possible survival estimates, only survival experience in the year 2000 was used by application of the recently introduced period analysis methodology3 (ie, all observations were left truncated at the beginning of 2000 in addition to being censored at the end of that year).
The principle of period analysis compared to traditional cohort-wise analysis of survival, which is explained in detail elsewhere,4 illustrated in Figure 1. The most up-to-date traditional cohort estimate of 5-year survival (10-year survival) that could be derived from the 1973 to 2000 SEER database would pertain to the survival experience in 1995 to 2000 (1990 to 2000) of patients diagnosed in 1995 (1990), as indicated by the solid frames in Figure 1. By contrast, an up-to-date estimate of both 5-year and 10-year survival could be obtained by period analysis for the year 2000, which is based exclusively on survival experience of patients in the year 2000 (dashed frame in Fig 1). With this approach, various parts of the survival function are obtained from cohorts of patients diagnosed in different years. This approach ensures that the most up-to date data are used to estimate conditional survival for each year following diagnosis: survival during the first year after diagnosis is obtained from patients diagnosed in 1999 and 2000, conditional survival in the second year after diagnosis is obtained from patients diagnosed in 1998 and 1999, and so on, until conditional survival in the 10th year after diagnosis, which is obtained from patients diagnosed in 1990 and 1991. These conditional survival estimates are then multiplied in the usual manner to derive 5-year or 10-year cumulative survival estimates.
It has been shown by extensive empirical evaluation that period analysis provides more up-to-date estimates of long-term survival rates than traditional methods of survival analysis,5-6 particularly in the case of strong improvement of survival rates over time. In particular, it has been shown that the period estimates of survival for a particular calendar year quite closely predict the survival rates later observed for patients diagnosed in that calendar year.7-8
Both absolute and relative survival rates are presented. Relative survival rates reflect survival rates of cancer patients compared with those of the general population. As described in detail elsewhere, they are calculated as the ratio of absolute survival rates of cancer patients divided by the expected survival rates of a group of individuals of the corresponding sex (and starting from the corresponding age) in the general population.9,10 In calculations of period estimates of relative survival, the period methodology is applied to the expected survival rates in the same way as to the observed survival rates. Typically, relative survival estimates are well below 100%, reflecting excess mortality among cancer patients compared to the general population. In contrast to absolute survival estimates, relative survival estimates may also equal or even exceed 100% if cancer patients have the same or even higher survival rates than the general population. This pattern may occur if cancer patients can generally be cured, or if patients diagnosed with cancer are otherwise more privileged (eg, in terms of socioeconomic factors or access to medical care) than the general population.
Estimates of expected survival were derived according to Hakulinen's method11 using the 2000 US sex, age, and race-specific life tables.12 Standard errors were derived according to Greenwood's method.13 All analyses were performed with the SAS (SAS Institute, Cary, NC) software package using an adapted version of a publicly available macro.14
RESULTS
Follow-up with respect to vital status was complete (ie, patients were followed until death or until the end of the year 2000) for 96.5% of patients. Among patients lost to follow-up, survival time was censored at the date they were last known to be alive. The proportion of patients lost to follow-up was below 5% for all subgroups of patients shown in Table 1, except for patients below age 65 years among whom 8.9% had incomplete follow-up. The period analysis of 10-year survival for the year 2000 is based on a total number of 124,803 person-years and 6,902 deaths. Prostate cancer was coded as the underlying cause for 1,723 (25.0%) of deaths.
Overall, absolute survival 5 and 10 years after diagnosis was estimated to be 79.3% and 54.8%, respectively. Five- and 10-year period estimates of relative survival for the year 2000 were 98.9% and 94.8%, which indicates that excess mortality due to prostate cancer was as low as 1% and 5% within 5 and 10 years following diagnosis, respectively. A more comprehensive picture of relative survival within 10 years following diagnosis is given in Figure 3. Besides the 10-year relative survival curve for the year 2000 obtained by period analysis, we also provide the most up-to date 5- and 10-year survival curves (pertaining to the cohorts of patients diagnosed in 1995 and 1990, respectively) that would have been obtained by traditional cohort-wise survival analysis. The 5- and 10-year relative survival estimates for the 1990 cohort would be as low as 90.3% and 82.2%, respectively. Patients diagnosed in 1995 already had a substantially higher 5-year relative survival (96.3%), but their 10-year relative survival will only be known after another 5 years of follow-up. The period survival curve for the year 2000 indicates that there has been further major improvement in relative survival in recent years that has so far remained undisclosed by traditional cohort analysis.
Table 1 additionally provides period estimates for the year 2000 of 5- and 10-year survival by race, age, tumor stage, and grade. Survival rates were considerably lower among black patients than among white patients. To assess recent time trends in relative survival by race, we additionally carried out period analyses of 5- and 10-year relative survival rates over the 1990 to 2000 period. To reduce random variation, smoothed trend curves were derived based on period estimates of relative survival for successive overlapping 3-year periods (1990 to 1992, 1991 to 1993... 1998 to 2000). There was a steady increase in 10-year relative survival rates during the 1990s for both black and white patients (see Fig 4). Similar trends were seen (albeit at higher levels) for 5-year relative survival rates (data not shown). Although the increase was stronger among black patients than among white patients, a substantial gap between both groups persisted until the end of the century.
Obviously, absolute survival rates were highest in the youngest age group, with a first diagnosis of prostate cancer below age 65 years, but relative survival rates were highest in 65- to 74-year-old patients who had essentially the same mortality as the general population (see Table 1). As expected, the strongest prognostic factors were tumor stage and grade. Whereas prognosis remained very poor among the minority of patients with distant tumor spread (6.5% of all patients), moderate excess mortality was observed among patients with poorly differentiated or undifferentiated localized/regional prostate cancer. Patients with well- or moderately differentiated localized/regional cancer did not have any excess mortality compared with the general population, as reflected by relative 5- and 10-year survival rates above 100%.
In more detailed, age-specific analyses (Table 2), the absence of excess mortality among patients with well- or moderately differentiated localized/regional prostate cancer could be confirmed for all age groups.
DISCUSSION
The survival estimates obtained in this analysis are the highest ever reported for patients with prostate cancer. The fact that relative survival rates above 100% were estimated for some groups of patients appears counterintuitive on first view. It is unlikely that occurrence of prostate cancer would increase chances of survival compared to the general population. A more plausible explanation is that this pattern reflects a selection effect of PSA screening, as screening tests tend to be less often used by socially disadvantaged population groups, who, in general, also have higher mortality.1
The period estimates of survival obtained in this article are considerably higher than those derived by traditional cohort-based methods reported in previous work (including the clinical trials), as they more efficiently capture the major increase in survival, which started with the introduction of widespread PSA screening in the 1980s and was ongoing in the 1990s.16 The period estimates may still turn out to be too pessimistic regarding the long-term survival expectations of newly diagnosed patients in the case of ongoing major improvement in survival, albeit less so than the traditional survival estimates, as extensive empirical evaluation has shown.4,7,8 On the other hand, the exclusion of patients reported to the registries by death certificate or autopsy only may have led to some overestimation of survival rates. However, given that these patients made up less than 1% of all patients, potential bias from this source can only be small.
In an era of widespread screening, increasing survival rates do not by themselves ensure progress against the cancer of interest. In theory, increasing survival rates might simply reflect increased lead time, that is, mere prolongation of the "patient career" by advancement in diagnosis, or even over-diagnosis of patients whose cancer would possibly never become clinically manifest in the absence of screening.17
For example, the shift from late-stage disease diagnoses to early-stage diagnoses (see Fig 2) will ultimately only reduce mortality if it increases cure rates of patients. Otherwise, the much higher 10-year relative survival rates of patients with localized/regional disease compared to patients with distant disease spread may just reflect postponement of cancer-related deaths beyond 10 years following diagnosis. In other words, patients might just have earlier knowledge of their disease, which by itself may have adverse psychological consequences. According to recent estimates, lead time and overdiagnosis may, in fact, be substantial and strongly dependent on age at and time intervals between PSA screenings.18 Another possibility that could contribute to an apparent increase in survival rates of patients with early-stage cancer would be that screening selectively advances early detection of slow growing, prognostically favorable tumors.
If and to what extent PSA screening lowers prostate cancer mortality will have to be answered by large scale randomized trials that are ongoing.19,20 In the European Randomized Screening for Prostate Cancer (ERSPC) Trial, more than 160,000 men aged 55 to 69 years at entry were recruited in seven European countries between 1992 and 2001, and more than 75,000 men aged 55 to 74 years were recruited in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial at 10 centers across the United States in 1993 to 2001. Long-term follow-up of participants is on its way. It has been estimated that the ERSPC trial will reach a power of 0.86 to detect a 25% reduction in prostate cancer mortality by the year 2008. Pooling data from both trials might further increase the power or advance availability of conclusive results.21 Additional end points to be addressed by the randomized trials will also include health-related quality of life, potential harms of unnecessary treatment, and cost-effectiveness of PSA testing.
In the meantime, the benefits of PSA screening are subject to ongoing debate. While some over-diagnosis appears to be out of question,18 there is also increasing indirect evidence that PSA screening may, in the long run, lead to some reduction in prostate cancer mortality. Multiple recent studies have reported a reduction in prostate cancer mortality.15,22-26 The decrease in incidence of distant-stage disease shown in Figure 2 is in agreement with these findings; however, any pertinent evidence will remain inconclusive before the long-term results of the randomized studies are available.27,28
The aim of our study was not to contribute to the ongoing discussion regarding the usefulness of PSA screening, and its results should not be mistaken as providing support for any particular screening strategy. Like other methods of survival analysis, period analysis by itself does not disclose to what extent the strongly increased survival rates are as a result of the above mentioned artifacts, true prevention of prostate cancer mortality by PSA screening, or other reasons, such as recent advances in therapy.29-31 Our study does make a major contribution, however, in providing the most up-to-date estimates of survival of patients diagnosed with prostate cancer today (ie, under the current patterns of screening and medical care). Our analyses were stratified by age, race, tumor stage, and grade, which are known strong prognostic factors. Other prognostic factors, including pretreatment PSA level, which are not included in the SEER database, would further improve prediction of cancer-specific survival and should be considered in pertinent analyses in the clinical setting as well as in individual patient care.32
Regardless of the relative contributions of each of the aforementioned factors to the improvement of survival rates in recent years, clinicians, as well as their patients diagnosed with prostate cancer, should have access to the most up-to-date survival statistics possible. In this context, the finding that, under the current patterns of screening and medical care, two thirds of these patients do not have excess mortality compared with the general population, may be important for both clinical management of, and patients' coping with, the disease.
Authors' Disclosures of Potential Conflicts of Interest
NOTES
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Swan J, Breen N, Coates RJ, et al: Progress in cancer screening practices in the United States. Results from the 2000 National Health Interview Survey. Cancer 97:1528-1540, 2003
2. Surveillance, Epidemiology, and End Results (SEER) Program Public-Use Data (1973–2000), Bethesda, MD, National Cancer Institute, DCCPS, Cancer Surveillance Research Program, Cancer Statistics Branch, released April 2003
3. Brenner H, Gefeller O: An alternative approach to monitoring cancer patient survival. Cancer 78:2004-2010, 1996
4. Brenner H, Gefeller O, Hakulinen T: Period analysis for up-to-date cancer survival data: Theory, empirical evaluation, computational realization and applications. Eur J Cancer 40:326-335, 2004
5. Cutler SJ, Ederer F: Maximum utilization of the life table method in analyzing survival. J Chronic Dis 8:699-712, 1958
6. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 58:457-481, 1958
7. Brenner H, Hakulinen T: Up-to-date long-term survival estimates of patients with cancer by period analysis. J Clin Oncol 20:826-832, 2002
8. Brenner H, Hakulinen T: Advanced detection of time trends in long-term cancer patient survival: Experience from 50 years of cancer registration in Finland. Am J Epidemiol 156:566-577, 2002
9. Ederer F, Axtell LM, Cutler SJ: The relative survival rate: A statistical methodology. Natl Cancer Inst Monogr 6:101-121, 1961
10. Henson DE, Ries LA: The relative survival rate. Cancer 76:1687-1688, 1995
11. Hakulinen T: Cancer survival corrected for heterogeneity in patient withdrawal. Biometrics 38:933-942, 1982
12. Arias E: United States life tables, 2000. Natl Vital Stat Rep 51:1-38, 2002
13. Greenwood M: A report on the natural duration of cancer. London, Ministry of Health, His Majesty's Stationery Office, 1926
14. Brenner H, Hakulinen T, Gefeller O: Computational realization of period analysis for monitoring cancer patient survival. Epidemiology 13:611-612, 2002
15. Potosky AL, Feuer EJ, Levin DL: Impact of screening on incidence and mortality of prostate cancer in the United States. Epidemiol Rev 23:181-186, 2001
16. Brenner H: Long-term survival rates of cancer patients achieved by the end of the 20th century: A period analysis. Lancet 360:1131-1135, 2002
17. Welch HG, Schwartz LM, Woloshin S: Are increasing 5-year survival rates evidence of success against cancer JAMA 283:2975-2978, 2000
18. 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 95:868-878, 2003
19. Schrder FH, Bangma CH: The European Randomized Study of Screening for Prostate Cancer (ERSPC). Br J Urol 79:68-71, 1997
20. Gohagan JK, Prorok PC, Hayes RB, et al: The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial of the National Cancer Institute: History, organization, and status. Control Clin Trials 21:251S-272S, 2000 (suppl 6)
21. de Koning HJ, Liem MK, Baan CA, et al: ERSPC: Prostate cancer mortality reduction by screening: Power and time frame with complete enrollment in the European Randomised Screening for Prostate Cancer (ERSPC) trial. Int J Cancer 98:268-273, 2002
22. Hankey BF, Feuer EJ, Clegg LX, et al: Cancer surveillance series: Interpreting trends in prostate cancer—part I: Evidence of the effects of screening in recent prostate cancer incidence, mortality and survival rates. J Natl Cancer Inst 91:1017-1024, 1999
23. Etzioni R, Legler JM, Feuer EJ, et al: Cancer surveillance series: Interpreting trends in prostate cancer—part III: Quantifying the link between population prostate-specific antigen testing and recent declines in prostate cancer mortality. J Natl Cancer Inst 91:1033-1039, 1999
24. Sarma AV, Schottenfeld D: Prostate cancer incidence, mortality, and survival trends in the United States: 1981–2001. Semin Urol Oncol 20:3-9, 2002
25. Stephenson RA: Prostate cancer trends in the era of prostate-specific antigen. An update of incidence, mortality, and clinical factors from the SEER database. Urol Clin North Am 29:173-181, 2002
26. Chu KC, Tarone RE, Freeman HP: Trends in prostate cancer mortality among black men and white men in the United States. Cancer 97:1507-1516, 2003
27. Harris R, Lohr KN: Screening for prostate cancer: An update of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med 137:917-929, 2002
28. Frankel S, Smith GD, Donovan J, et al: Screening for prostate cancer. Lancet 361:1122-1128, 2003
29. Bolla M, Gonzalez D, Warde P, et al: Improved survival in patients with locally advanced prostate cancer treated with radiotherapy and goserelin. N Engl J Med 337:295-300, 1997
30. Roach M 3rd: Neoadjuvant therapy prior to radiotherapy for clinically localized prostate cancer. Eur Urol 32:48-54, 1997 (suppl 3)
31. Messing EM, Manola J, Sarosdy M, et al: Immediate hormonal therapy compared with observation after radical prostatectomy and pelvic lymphadenectomy in men with node-positive prostate cancer. N Engl J Med 341:1781-1788, 1999
32. D'Amico AV, Moul J, Carroll PR, et al: Cancer-specific mortality after surgery or radiation for patients with clinically localized prostate cancer managed during the prostate-specific antigen era. J Clin Oncol 21:2163-2172, 2003(Hermann Brenner, Volker A)
ABSTRACT
METHODS: Five- and 10-year absolute and relative survival rates for the year 2000 were derived from the 1973 to 2000 database of the Surveillance, Epidemiology and End Results Program using the recently introduced period analysis methodology.
RESULTS: Overall, 5- and 10-year relative survival rates were approximately 99% and 95%; that is, excess mortality compared with the general population was as low as 1% and 5% within 5 and 10 years following diagnosis, respectively. Two-thirds of patients were diagnosed with well or moderately differentiated localized/regional prostate cancer, and among these patients, 5- and 10-year relative survival rates were above 100% (indicating the lack of any excess mortality) at all ages.
CONCLUSION: While the value of PSA screening for lowering mortality due to prostate cancer remains to be shown by randomized clinical trials, the majority of patients diagnosed with prostate cancer in the PSA screening era do not have excess mortality compared to the general population under current patterns of medical care. This information may be important for both clinical management of, and for patients' coping with, the disease.
INTRODUCTION
METHODS
The SEER Program is the most authoritative source of information on cancer incidence and survival in the United States, and it is considered as the standard for quality among cancer registries around the world. Quality control has been an integral part of SEER since its inception. Every year, studies are conducted in the SEER areas to evaluate the quality and completeness of the data being reported (SEER's standard for case ascertainment is 98%).
Data included in the 1973 to 2000 SEER database are from population-based cancer registries in Connecticut, New Mexico, Utah, Iowa, Hawaii, Atlanta (Georgia), Detroit (Michigan), Seattle-Puget Sound (Washington), and San Francisco-Oakland (California), which together cover a population of about 30 million people. Geographic areas were selected for inclusion in the SEER Program based on their ability to operate and maintain a high-quality, population-based cancer reporting system and for their epidemiologically significant population subgroups. The SEER population is comparable to the general United States population with regard to measures of poverty and education, even though it tends to be somewhat more urban and has a higher proportion of foreign-born persons than the latter.
This database included 183,484 patients with a first diagnosis of prostate cancer between 1990 and 2000 who have been followed for vital status until the end of the year 2000. After exclusion of 415 patients (0.2%) who were reported by autopsy only, 1,134 patients (0.6%) who were reported by death certificate only, and another 1,330 patients (0.7%) with unknown month of diagnosis, there remained 180,605 patients (98.4%) for the survival analysis.
In order to provide the most up-to-date possible survival estimates, only survival experience in the year 2000 was used by application of the recently introduced period analysis methodology3 (ie, all observations were left truncated at the beginning of 2000 in addition to being censored at the end of that year).
The principle of period analysis compared to traditional cohort-wise analysis of survival, which is explained in detail elsewhere,4 illustrated in Figure 1. The most up-to-date traditional cohort estimate of 5-year survival (10-year survival) that could be derived from the 1973 to 2000 SEER database would pertain to the survival experience in 1995 to 2000 (1990 to 2000) of patients diagnosed in 1995 (1990), as indicated by the solid frames in Figure 1. By contrast, an up-to-date estimate of both 5-year and 10-year survival could be obtained by period analysis for the year 2000, which is based exclusively on survival experience of patients in the year 2000 (dashed frame in Fig 1). With this approach, various parts of the survival function are obtained from cohorts of patients diagnosed in different years. This approach ensures that the most up-to date data are used to estimate conditional survival for each year following diagnosis: survival during the first year after diagnosis is obtained from patients diagnosed in 1999 and 2000, conditional survival in the second year after diagnosis is obtained from patients diagnosed in 1998 and 1999, and so on, until conditional survival in the 10th year after diagnosis, which is obtained from patients diagnosed in 1990 and 1991. These conditional survival estimates are then multiplied in the usual manner to derive 5-year or 10-year cumulative survival estimates.
It has been shown by extensive empirical evaluation that period analysis provides more up-to-date estimates of long-term survival rates than traditional methods of survival analysis,5-6 particularly in the case of strong improvement of survival rates over time. In particular, it has been shown that the period estimates of survival for a particular calendar year quite closely predict the survival rates later observed for patients diagnosed in that calendar year.7-8
Both absolute and relative survival rates are presented. Relative survival rates reflect survival rates of cancer patients compared with those of the general population. As described in detail elsewhere, they are calculated as the ratio of absolute survival rates of cancer patients divided by the expected survival rates of a group of individuals of the corresponding sex (and starting from the corresponding age) in the general population.9,10 In calculations of period estimates of relative survival, the period methodology is applied to the expected survival rates in the same way as to the observed survival rates. Typically, relative survival estimates are well below 100%, reflecting excess mortality among cancer patients compared to the general population. In contrast to absolute survival estimates, relative survival estimates may also equal or even exceed 100% if cancer patients have the same or even higher survival rates than the general population. This pattern may occur if cancer patients can generally be cured, or if patients diagnosed with cancer are otherwise more privileged (eg, in terms of socioeconomic factors or access to medical care) than the general population.
Estimates of expected survival were derived according to Hakulinen's method11 using the 2000 US sex, age, and race-specific life tables.12 Standard errors were derived according to Greenwood's method.13 All analyses were performed with the SAS (SAS Institute, Cary, NC) software package using an adapted version of a publicly available macro.14
RESULTS
Follow-up with respect to vital status was complete (ie, patients were followed until death or until the end of the year 2000) for 96.5% of patients. Among patients lost to follow-up, survival time was censored at the date they were last known to be alive. The proportion of patients lost to follow-up was below 5% for all subgroups of patients shown in Table 1, except for patients below age 65 years among whom 8.9% had incomplete follow-up. The period analysis of 10-year survival for the year 2000 is based on a total number of 124,803 person-years and 6,902 deaths. Prostate cancer was coded as the underlying cause for 1,723 (25.0%) of deaths.
Overall, absolute survival 5 and 10 years after diagnosis was estimated to be 79.3% and 54.8%, respectively. Five- and 10-year period estimates of relative survival for the year 2000 were 98.9% and 94.8%, which indicates that excess mortality due to prostate cancer was as low as 1% and 5% within 5 and 10 years following diagnosis, respectively. A more comprehensive picture of relative survival within 10 years following diagnosis is given in Figure 3. Besides the 10-year relative survival curve for the year 2000 obtained by period analysis, we also provide the most up-to date 5- and 10-year survival curves (pertaining to the cohorts of patients diagnosed in 1995 and 1990, respectively) that would have been obtained by traditional cohort-wise survival analysis. The 5- and 10-year relative survival estimates for the 1990 cohort would be as low as 90.3% and 82.2%, respectively. Patients diagnosed in 1995 already had a substantially higher 5-year relative survival (96.3%), but their 10-year relative survival will only be known after another 5 years of follow-up. The period survival curve for the year 2000 indicates that there has been further major improvement in relative survival in recent years that has so far remained undisclosed by traditional cohort analysis.
Table 1 additionally provides period estimates for the year 2000 of 5- and 10-year survival by race, age, tumor stage, and grade. Survival rates were considerably lower among black patients than among white patients. To assess recent time trends in relative survival by race, we additionally carried out period analyses of 5- and 10-year relative survival rates over the 1990 to 2000 period. To reduce random variation, smoothed trend curves were derived based on period estimates of relative survival for successive overlapping 3-year periods (1990 to 1992, 1991 to 1993... 1998 to 2000). There was a steady increase in 10-year relative survival rates during the 1990s for both black and white patients (see Fig 4). Similar trends were seen (albeit at higher levels) for 5-year relative survival rates (data not shown). Although the increase was stronger among black patients than among white patients, a substantial gap between both groups persisted until the end of the century.
Obviously, absolute survival rates were highest in the youngest age group, with a first diagnosis of prostate cancer below age 65 years, but relative survival rates were highest in 65- to 74-year-old patients who had essentially the same mortality as the general population (see Table 1). As expected, the strongest prognostic factors were tumor stage and grade. Whereas prognosis remained very poor among the minority of patients with distant tumor spread (6.5% of all patients), moderate excess mortality was observed among patients with poorly differentiated or undifferentiated localized/regional prostate cancer. Patients with well- or moderately differentiated localized/regional cancer did not have any excess mortality compared with the general population, as reflected by relative 5- and 10-year survival rates above 100%.
In more detailed, age-specific analyses (Table 2), the absence of excess mortality among patients with well- or moderately differentiated localized/regional prostate cancer could be confirmed for all age groups.
DISCUSSION
The survival estimates obtained in this analysis are the highest ever reported for patients with prostate cancer. The fact that relative survival rates above 100% were estimated for some groups of patients appears counterintuitive on first view. It is unlikely that occurrence of prostate cancer would increase chances of survival compared to the general population. A more plausible explanation is that this pattern reflects a selection effect of PSA screening, as screening tests tend to be less often used by socially disadvantaged population groups, who, in general, also have higher mortality.1
The period estimates of survival obtained in this article are considerably higher than those derived by traditional cohort-based methods reported in previous work (including the clinical trials), as they more efficiently capture the major increase in survival, which started with the introduction of widespread PSA screening in the 1980s and was ongoing in the 1990s.16 The period estimates may still turn out to be too pessimistic regarding the long-term survival expectations of newly diagnosed patients in the case of ongoing major improvement in survival, albeit less so than the traditional survival estimates, as extensive empirical evaluation has shown.4,7,8 On the other hand, the exclusion of patients reported to the registries by death certificate or autopsy only may have led to some overestimation of survival rates. However, given that these patients made up less than 1% of all patients, potential bias from this source can only be small.
In an era of widespread screening, increasing survival rates do not by themselves ensure progress against the cancer of interest. In theory, increasing survival rates might simply reflect increased lead time, that is, mere prolongation of the "patient career" by advancement in diagnosis, or even over-diagnosis of patients whose cancer would possibly never become clinically manifest in the absence of screening.17
For example, the shift from late-stage disease diagnoses to early-stage diagnoses (see Fig 2) will ultimately only reduce mortality if it increases cure rates of patients. Otherwise, the much higher 10-year relative survival rates of patients with localized/regional disease compared to patients with distant disease spread may just reflect postponement of cancer-related deaths beyond 10 years following diagnosis. In other words, patients might just have earlier knowledge of their disease, which by itself may have adverse psychological consequences. According to recent estimates, lead time and overdiagnosis may, in fact, be substantial and strongly dependent on age at and time intervals between PSA screenings.18 Another possibility that could contribute to an apparent increase in survival rates of patients with early-stage cancer would be that screening selectively advances early detection of slow growing, prognostically favorable tumors.
If and to what extent PSA screening lowers prostate cancer mortality will have to be answered by large scale randomized trials that are ongoing.19,20 In the European Randomized Screening for Prostate Cancer (ERSPC) Trial, more than 160,000 men aged 55 to 69 years at entry were recruited in seven European countries between 1992 and 2001, and more than 75,000 men aged 55 to 74 years were recruited in the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial at 10 centers across the United States in 1993 to 2001. Long-term follow-up of participants is on its way. It has been estimated that the ERSPC trial will reach a power of 0.86 to detect a 25% reduction in prostate cancer mortality by the year 2008. Pooling data from both trials might further increase the power or advance availability of conclusive results.21 Additional end points to be addressed by the randomized trials will also include health-related quality of life, potential harms of unnecessary treatment, and cost-effectiveness of PSA testing.
In the meantime, the benefits of PSA screening are subject to ongoing debate. While some over-diagnosis appears to be out of question,18 there is also increasing indirect evidence that PSA screening may, in the long run, lead to some reduction in prostate cancer mortality. Multiple recent studies have reported a reduction in prostate cancer mortality.15,22-26 The decrease in incidence of distant-stage disease shown in Figure 2 is in agreement with these findings; however, any pertinent evidence will remain inconclusive before the long-term results of the randomized studies are available.27,28
The aim of our study was not to contribute to the ongoing discussion regarding the usefulness of PSA screening, and its results should not be mistaken as providing support for any particular screening strategy. Like other methods of survival analysis, period analysis by itself does not disclose to what extent the strongly increased survival rates are as a result of the above mentioned artifacts, true prevention of prostate cancer mortality by PSA screening, or other reasons, such as recent advances in therapy.29-31 Our study does make a major contribution, however, in providing the most up-to-date estimates of survival of patients diagnosed with prostate cancer today (ie, under the current patterns of screening and medical care). Our analyses were stratified by age, race, tumor stage, and grade, which are known strong prognostic factors. Other prognostic factors, including pretreatment PSA level, which are not included in the SEER database, would further improve prediction of cancer-specific survival and should be considered in pertinent analyses in the clinical setting as well as in individual patient care.32
Regardless of the relative contributions of each of the aforementioned factors to the improvement of survival rates in recent years, clinicians, as well as their patients diagnosed with prostate cancer, should have access to the most up-to-date survival statistics possible. In this context, the finding that, under the current patterns of screening and medical care, two thirds of these patients do not have excess mortality compared with the general population, may be important for both clinical management of, and patients' coping with, the disease.
Authors' Disclosures of Potential Conflicts of Interest
NOTES
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
1. Swan J, Breen N, Coates RJ, et al: Progress in cancer screening practices in the United States. Results from the 2000 National Health Interview Survey. Cancer 97:1528-1540, 2003
2. Surveillance, Epidemiology, and End Results (SEER) Program Public-Use Data (1973–2000), Bethesda, MD, National Cancer Institute, DCCPS, Cancer Surveillance Research Program, Cancer Statistics Branch, released April 2003
3. Brenner H, Gefeller O: An alternative approach to monitoring cancer patient survival. Cancer 78:2004-2010, 1996
4. Brenner H, Gefeller O, Hakulinen T: Period analysis for up-to-date cancer survival data: Theory, empirical evaluation, computational realization and applications. Eur J Cancer 40:326-335, 2004
5. Cutler SJ, Ederer F: Maximum utilization of the life table method in analyzing survival. J Chronic Dis 8:699-712, 1958
6. Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 58:457-481, 1958
7. Brenner H, Hakulinen T: Up-to-date long-term survival estimates of patients with cancer by period analysis. J Clin Oncol 20:826-832, 2002
8. Brenner H, Hakulinen T: Advanced detection of time trends in long-term cancer patient survival: Experience from 50 years of cancer registration in Finland. Am J Epidemiol 156:566-577, 2002
9. Ederer F, Axtell LM, Cutler SJ: The relative survival rate: A statistical methodology. Natl Cancer Inst Monogr 6:101-121, 1961
10. Henson DE, Ries LA: The relative survival rate. Cancer 76:1687-1688, 1995
11. Hakulinen T: Cancer survival corrected for heterogeneity in patient withdrawal. Biometrics 38:933-942, 1982
12. Arias E: United States life tables, 2000. Natl Vital Stat Rep 51:1-38, 2002
13. Greenwood M: A report on the natural duration of cancer. London, Ministry of Health, His Majesty's Stationery Office, 1926
14. Brenner H, Hakulinen T, Gefeller O: Computational realization of period analysis for monitoring cancer patient survival. Epidemiology 13:611-612, 2002
15. Potosky AL, Feuer EJ, Levin DL: Impact of screening on incidence and mortality of prostate cancer in the United States. Epidemiol Rev 23:181-186, 2001
16. Brenner H: Long-term survival rates of cancer patients achieved by the end of the 20th century: A period analysis. Lancet 360:1131-1135, 2002
17. Welch HG, Schwartz LM, Woloshin S: Are increasing 5-year survival rates evidence of success against cancer JAMA 283:2975-2978, 2000
18. 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 95:868-878, 2003
19. Schrder FH, Bangma CH: The European Randomized Study of Screening for Prostate Cancer (ERSPC). Br J Urol 79:68-71, 1997
20. Gohagan JK, Prorok PC, Hayes RB, et al: The Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial of the National Cancer Institute: History, organization, and status. Control Clin Trials 21:251S-272S, 2000 (suppl 6)
21. de Koning HJ, Liem MK, Baan CA, et al: ERSPC: Prostate cancer mortality reduction by screening: Power and time frame with complete enrollment in the European Randomised Screening for Prostate Cancer (ERSPC) trial. Int J Cancer 98:268-273, 2002
22. Hankey BF, Feuer EJ, Clegg LX, et al: Cancer surveillance series: Interpreting trends in prostate cancer—part I: Evidence of the effects of screening in recent prostate cancer incidence, mortality and survival rates. J Natl Cancer Inst 91:1017-1024, 1999
23. Etzioni R, Legler JM, Feuer EJ, et al: Cancer surveillance series: Interpreting trends in prostate cancer—part III: Quantifying the link between population prostate-specific antigen testing and recent declines in prostate cancer mortality. J Natl Cancer Inst 91:1033-1039, 1999
24. Sarma AV, Schottenfeld D: Prostate cancer incidence, mortality, and survival trends in the United States: 1981–2001. Semin Urol Oncol 20:3-9, 2002
25. Stephenson RA: Prostate cancer trends in the era of prostate-specific antigen. An update of incidence, mortality, and clinical factors from the SEER database. Urol Clin North Am 29:173-181, 2002
26. Chu KC, Tarone RE, Freeman HP: Trends in prostate cancer mortality among black men and white men in the United States. Cancer 97:1507-1516, 2003
27. Harris R, Lohr KN: Screening for prostate cancer: An update of the evidence for the U.S. Preventive Services Task Force. Ann Intern Med 137:917-929, 2002
28. Frankel S, Smith GD, Donovan J, et al: Screening for prostate cancer. Lancet 361:1122-1128, 2003
29. Bolla M, Gonzalez D, Warde P, et al: Improved survival in patients with locally advanced prostate cancer treated with radiotherapy and goserelin. N Engl J Med 337:295-300, 1997
30. Roach M 3rd: Neoadjuvant therapy prior to radiotherapy for clinically localized prostate cancer. Eur Urol 32:48-54, 1997 (suppl 3)
31. Messing EM, Manola J, Sarosdy M, et al: Immediate hormonal therapy compared with observation after radical prostatectomy and pelvic lymphadenectomy in men with node-positive prostate cancer. N Engl J Med 341:1781-1788, 1999
32. D'Amico AV, Moul J, Carroll PR, et al: Cancer-specific mortality after surgery or radiation for patients with clinically localized prostate cancer managed during the prostate-specific antigen era. J Clin Oncol 21:2163-2172, 2003(Hermann Brenner, Volker A)