A Method for Using Life Tables to Estimate Lifetime Risk for Prostate Cancer Death

View More View Less
  • a From the Division of Urology, Cedars-Sinai Medical Center, Los Angeles, California; Conseco Services, LLC, Carmel, Indiana; Department of Urologic Oncology, Roswell Park Cancer Institute, and Department of Urology, University of Buffalo, Buffalo, New York; and UNC Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, North Carolina.

Prostate cancer can have a long and indolent course, and management without curative therapy should be considered in select patients. When counseling patients, a useful way to convey the risk for death from competing causes is to estimate their lifetime risk for dying from prostate cancer. Double-decrement life tables were constructed to calculate age-specific death rates using the death probabilities from the Social Security Administration life tables and Gleason score–specific mortality rates reported from pre-PSA cohort study. The lifetime risk for prostate cancer death was calculated. Life tables provided life expectancy and risk for prostate cancer death based on age at diagnosis. For example, 60-year-old patient with a Gleason score 6, 7, or 8 tumor had an overall life expectancy of 14.4, 10.2, or 6.6 years, respectively. The risk for prostate cancer death during the expected years of life was 33%, 49%, or 57%, respectively. If a 10-year lead-time bias was assumed for PSA detection, the risks for death from prostate cancer decreased to 16%, 26%, or 37%, respectively. If the patient was in the bottom quartile for overall health and disease was detected by prostate examination, the risk for death from prostate cancer was 21%, 32%, or 40%, respectively. A Web-based tool for performing these calculations is available at http://www.roswellpark.org/Patient_Care/Specialized_Services/Prostate_Cancer_Estimator.html. Life tables can be created to estimate overall life expectancy and risk for prostate cancer death, and to assist with decision-making when considering management without curative therapy.

Patients with clinically localized prostate cancer face a wide array of management options. Treatment may include definitive local therapy. Although prostate cancer is the second leading cause of cancer mortality, with approximately 3% of all men dying of the disease,1 its natural history is heterogeneous and most prostate cancers are clinically insignificant. In one autopsy study, as many as 1 in 3 men older than 50 years had histologic evidence of prostate cancer.2 Estimates show that 8 men will be diagnosed with prostate cancer for every 1 who dies of the disease.3

Any active treatment leads to a decrease in quality of life.4 Ideally, treatment would be reserved for patients most likely to die of their disease. Currently, no accurate test exists to identify these patients. Various prognostic models have been described to stratify prostate cancers. Commonly used prognostic models include the Partin tables,5 the Kattan nomograms,6 and the CAPRA (Cancer of the Prostate Risk Assessment) score.7 Population-based cohort studies have defined the natural history of prostate cancer diagnosed in the pre–prostate-specific antigen (PSA) era and managed without curative treatment. Albertsen et al.8 defined the risk for prostate cancer mortality stratified by age and Gleason score. Johansson et al.9 reported 15-year survival rates stratified by stage and grade.9 Recently, Kattan et al.10 developed a nomogram incorporating PSA to predict 10-year disease-specific survival in men treated without curative therapy.

Elderly patients diagnosed with prostate cancer are at risk for non-prostate cancer mortality. Therefore, an estimate of life expectancy is essential for informed decision-making. Life expectancy is commonly estimated using the Social Security Administration (SSA) life tables, which provides life expectancies in years of life. However, most research studies report survival rates associated with prostate cancer. Therefore, during a patient counseling session, the risk posed by the prostate cancer may be difficult to convey when life expectancy is presented in years and prostate cancer survival is presented as a survival rate or chance at an arbitrary interval.

A better way to convey the risk for prostate cancer is to tailor the survival rate for patients by estimating their life-time risk for dying from prostate cancer, which can be done using standard actuarial methods. The authors used previously published mortality rates for prostate cancer to create life tables that estimate a patient's life expectancy and lifetime risk for dying from prostate cancer managed without curative therapy. The authors also illustrate how the calculations can be modified to account for lead-time bias from PSA screening and clinician assessment of overall health.

Methods

Double-decrement tables were constructed to calculate age-specific death rates using the death probabilities from the SSA life tables and Gleason score–specific mortality rates reported by Albertsen et al.8 The width of each age interval was 1 year. The probability of death [q(τ)] during the interval was calculated as the probability of death from prostate cancer [q′(1)] plus the probability of death from all causes [q′(2)] minus the intersection of the 2 probabilities. The intersection represents the clinically impossible scenario in which patients die twice from 2 different causes.

FD1

In a population in which all individuals have prostate cancer, the risk for dying from it is adjusted for risk of dying from all causes:

FD2

Similarly, the risk for dying from all causes (a′) is adjusted for risk of dying from prostate cancer:

FD3
Life Table

The mortality schedules represented by q(1) and q(2) were applied to a hypothetical population of 100,000 patients with prostate cancer starting at 50 years of age to calculate the numbers of deaths from prostate cancer [d(1)] and deaths from all causes [d(2)]. To calculate person-years lived during each yearly interval (L), all deaths were assumed to have occurred halfway through the interval.

FD4

At each interval, l was the starting population, and reflected deaths occurring during the previous interval. Table 1 provides an example for calculating person-years for patients with Gleason score 7 prostate cancer. At the age intervals shown, q(2) and q′(2) are identical at the number of decimal places shown. Total years remaining at each age (T) was calculated as the sum of all person-years starting at each age and going down to the bottom of column L. Life expectancy (e) was calculated:

FD5

The risk for prostate cancer death during a patient's life [q(3)] was calculated as the sum of all prostate cancer deaths occurring during a patient's expected length of life divided by the starting population:

FD6

For example, d(1)1 is the number of prostate cancer deaths during the interval corresponding to the age at diagnosis. Table 2 shows an example for calculating life expectancy and risk for death from prostate cancer in patients with Gleason score 7 disease. To illustrate calculations for total years remaining (T), ages 95 to 105 years are shown.

Table 1

Calculating Person-Years for Patients With Gleason 7 Prostate Cancer

Table 1
Table 2

Life Table for Patients With Gleason 7 Prostate Cancer

Table 2
Lead Time

Lead-time bias in years (b) can be used to adjust the risk for prostate cancer death at each age. The life table was adjusted so that no deaths occur from prostate cancer during the lead time [q′(1)1... q′(1)b is set to zero, where q′(1)1 is q′(1) at age of entry]. The risk for prostate cancer death during lifetime q(3) is calculated as:

FD7

Table 3 illustrates the inclusion of 5-year lead-time bias for a 50-year-old patient with Gleason score 7 prostate cancer. Note that q′(1) is set to zero for 5 years. During these 5 years, q′(2) and q(2) are identical. Starting with the sixth age interval, deaths occur from both prostate cancer [d(1)] and all causes [d(2)]. T, L, and e are calculated as previously described. q(3) is the sum of all prostate cancer deaths, d(1), during the patient's life expectancy of 15.7 years (rounded to 16 years) divided by l. These steps are repeated for every age interval.

Table 3

Life Table for 50-Year-Old Patient With Gleason 7 Prostate Cancer and 5-Year Lead-Time Bias

Table 3
Adjusting Life Expectancy for Overall Health

The NCCN guidelines recommend adjusting patient's life expectancy based on the clinician's assessment of patient's overall health.11 For patients in the best quartile of health, life expectancy is increased by 50%. For patients in the worse quartile of health, life expectancy is decreased by 50%. Therefore, the calculated life expectancy (e) can be adjusted before calculating the risk for prostate cancer death.

Results

Table 4 provides the estimated life expectancy for patients based on Gleason score and age. The mortality rates used for the calculations were from a pre-PSA population-based cohort study. Therefore, life expectancy was estimated after assuming 5- and 10-year lead-time bias. Table 5 provides the life-time risk for dying from prostate cancer based on Gleason score, age, and assumed lead-time bias. Table 6 provides the life-time risk for dying from prostate cancer based on Gleason score, age, and quartile of health. A Web-based tool for performing these calculations is available at http://www.roswellpark.org/Patient_Care/Specialized_Services/Prostate_Cancer_Estimator.html. This tool allows users to input age and Gleason score based on biopsy to obtain estimates of life expectancy and lifetime risk for death from prostate cancer. The patient's quartile of health can be provided or an estimated lead-time bias can be included. The tool requires users to assume that patients are in the middle quartiles of health when lead-time bias is provided, and conversely, lead-time bias cannot be included in the calculation when patients are not in the middle quartiles of health.

Discussion

The mortality rate for prostate cancer detected by PSA screening and managed without definitive local therapy is not known. Population-based cohort studies with 20 years of follow-up have been reported for patients diagnosed in the pre-PSA era.8,9 A recent study characterized a cohort of patients with prostate cancer with baseline PSAs from 6 English cancer registries managed without definitive local therapy. The study described a nomogram that incorporated PSA for predicting cancer-specific survival.10 However, the study was based on patients diagnosed in England between 1990 and 1996, when PSA screening was infrequently performed. Nearly half of the patients were diagnosed with prostate cancer after transurethral resection of the prostate. Therefore, it is not surprising that survivals predicted by the nomogram are similar to survival rates reported in the pre-PSA era cohort studies.8,12

Table 4

Life Expectancy Based on Assumed Lead-Time Bias

Table 4
Table 5

Risk for Prostate Cancer Death Based on Assumed Lead-Time Bias

Table 5

Most contemporary prostate cancers are detected based on PSA screening, before the tumor is palpable. PSA screening has nearly doubled the number of men diagnosed with prostate cancer since the 1980s.1 This increase in incidence has been associated with a corresponding stage migration. In the Cancer of the Prostate Strategic Urologic Research Endeavor (CaPSURE) registry, percent of locally advanced (T3 and T4) tumors decreased from 11.8% in the early 1990s to 3.5% in 2000.13 Therefore, previously reported mortality and survival rates based on pre-PSA screening cohort studies may not mirror contemporary American men diagnosed with prostate cancer, and results of randomized trials designed to define the natural history of untreated prostate cancer detected by PSA screening are years away.14,15

Table 6

Risk for Prostate Cancer Death Based on Overall Health

Table 6

Despite this important limitation, prostate cancer mortality can be estimated for modern patients when considering lead-time bias. Estimates of lead-time bias resulting from prostate cancer screening range from 5 to 13.3 years.8,9,16,17 In the European Randomized Study of Screening for Prostate Cancer, lead-time bias resulting from PSA screening was related to age of diagnosis.18 A 55-year-old man was estimated to have a lead time of approximately 12.3 years, whereas a man aged 75 years had a lead time of approximately 6 years. Therefore, when using the method described in this study, assuming a 10-year lead time may be appropriate when estimating risk for prostate cancer death for patients in their 50s. Assuming a 5-year lead time may be appropriate when estimating the same risk for patients in their 70s.

For patients diagnosed with prostate cancer after presenting with symptoms or an abnormal digital rectal examination, life expectancy and risk for cancer mortality should be obtained without including a lead time.

The authors' model makes several assumptions. When constructing the life times, the mortality rate was specific for each Gleason score; however, for a given Gleason score, the mortality rate was assumed to be constant for all age groups. Also assumed was that deaths are distributed evenly throughout each yearly interval and that all deaths can be considered to occur at the halfway point of the interval. The death probabilities from the SSA include men who died of all causes, including prostate cancer. However, because the proportion of men dying of prostate cancer is expected to be small, these men were assumed to have died of causes not related to prostate cancer. Finally, when modeling the lead time, no deaths were assumed to have occurred from prostate cancer during the lead time. However, at the end of the lead time, prostate cancer mortality resumed at the rate observed in the pre-PSA era.

The authors have disclosed that they have no financial interests, arrangements, or affiliations with the manufacturers of any products discussed in the article or their competitors.

References

  • 1.

    Jemal A, Siegel R, Ward E. Cancer statistics, 2008. CA Cancer J Clin 2008;58:7196.

  • 2.

    Yatani R, Chigusa I, Akazaki K. Geographic pathology of latent prostatic carcinoma. Int J Cancer 1982;29:611616.

  • 3.

    Parkin DM, Bray FI, Devesa SS. Cancer burden in the year 2000. The global picture. Eur J Cancer 2001;37(Suppl 8):S466.

  • 4.

    Sanda MG, Dunn RL, Michalski J. Quality of life and satisfaction with outcome among prostate-cancer survivors. N Engl J Med 2008;358:12501261.

    • Search Google Scholar
    • Export Citation
  • 5.

    Partin AW, Mangold LA, Lamm DM. Contemporary update of prostate cancer staging nomograms (Partin Tables) for the new millennium. Urology 2001;58:843848.

    • Search Google Scholar
    • Export Citation
  • 6.

    Kattan MW, Eastham JA, Stapleton AM. A preoperative nomogram for disease recurrence following radical prostatectomy for prostate cancer. J Natl Cancer Inst 1998;90:766771.

    • Search Google Scholar
    • Export Citation
  • 7.

    Cooperberg MR, Pasta DJ, Elkin EP. The University of California, San Francisco Cancer of the Prostate Risk Assessment score: a straightforward and reliable preoperative predictor of disease recurrence after radical prostatectomy. J Urol 2005;173:19381942.

    • Search Google Scholar
    • Export Citation
  • 8.

    Albertsen PC, Hanley JA, Fine J. 20-year outcomes following conservative management of clinically localized prostate cancer. JAMA 2005;293:20952101.

    • Search Google Scholar
    • Export Citation
  • 9.

    Johansson JE, Andren O, Andersson SO. Natural history of early, localized prostate cancer. JAMA 2004;291:27132719.

  • 10.

    Kattan MW, Cuzick J, Fisher G. Nomogram incorporating PSA level to predict cancer-specific survival for men with clinically localized prostate cancer managed without curative intent. Cancer 2008;112:6974.

    • Search Google Scholar
    • Export Citation
  • 11.

    Mohler J, Amling CL, Bahnson RR. NCCN clinical practice guidelines in oncology: prostate cancer. Version 2, 2009. Available at: http://www.nccn.org/professionals/physician_gls/PDF/prostate.pdf. Accessed December 29, 2009.

    • Search Google Scholar
    • Export Citation
  • 12.

    Albertsen P. Predicting survival for men with clinically localized prostate cancer: what do we need in contemporary practice? Cancer 2008;112:13.

    • Search Google Scholar
    • Export Citation
  • 13.

    Cooperberg MR, Lubeck DP, Meng MV. The changing face of low-risk prostate cancer: trends in clinical presentation and primary management. J Clin Oncol 2004;22:21412149.

    • Search Google Scholar
    • Export Citation
  • 14.

    Schroder FH, Denis LJ, Roobol M. The story of the European Randomized Study of Screening for Prostate Cancer. BJU Int 2003;92(Suppl 2):113.

  • 15.

    Andriole GL, Reding D, Hayes RB. The prostate, lung, colon, and ovarian (PLCO) cancer screening trial: status and promise. Urol Oncol 2004;22:358361.

    • Search Google Scholar
    • Export Citation
  • 16.

    Gann PH, Hennekens CH, Stampfer MJ. A prospective evaluation of plasma prostate-specific antigen for detection of prostatic cancer. JAMA 1995;273:289294.

    • Search Google Scholar
    • Export Citation
  • 17.

    Pearson JD, Carter HB. Natural history of changes in prostate specific antigen in early stage prostate cancer. J Urol 1994;152:17431748.

  • 18.

    Draisma G, Boer R, Otto SJ. 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;95:868878.

    • Search Google Scholar
    • Export Citation

If the inline PDF is not rendering correctly, you can download the PDF file here.

Correspondence: Hyung L. Kim, MD, Division of Urology, Cedars-Sinai Medical Center, 8635 W. Third St, Suite 1070, Los Angeles, CA 90048. E-mail: kimhl@cshs.org
  • 1.

    Jemal A, Siegel R, Ward E. Cancer statistics, 2008. CA Cancer J Clin 2008;58:7196.

  • 2.

    Yatani R, Chigusa I, Akazaki K. Geographic pathology of latent prostatic carcinoma. Int J Cancer 1982;29:611616.

  • 3.

    Parkin DM, Bray FI, Devesa SS. Cancer burden in the year 2000. The global picture. Eur J Cancer 2001;37(Suppl 8):S466.

  • 4.

    Sanda MG, Dunn RL, Michalski J. Quality of life and satisfaction with outcome among prostate-cancer survivors. N Engl J Med 2008;358:12501261.

    • Search Google Scholar
    • Export Citation
  • 5.

    Partin AW, Mangold LA, Lamm DM. Contemporary update of prostate cancer staging nomograms (Partin Tables) for the new millennium. Urology 2001;58:843848.

    • Search Google Scholar
    • Export Citation
  • 6.

    Kattan MW, Eastham JA, Stapleton AM. A preoperative nomogram for disease recurrence following radical prostatectomy for prostate cancer. J Natl Cancer Inst 1998;90:766771.

    • Search Google Scholar
    • Export Citation
  • 7.

    Cooperberg MR, Pasta DJ, Elkin EP. The University of California, San Francisco Cancer of the Prostate Risk Assessment score: a straightforward and reliable preoperative predictor of disease recurrence after radical prostatectomy. J Urol 2005;173:19381942.

    • Search Google Scholar
    • Export Citation
  • 8.

    Albertsen PC, Hanley JA, Fine J. 20-year outcomes following conservative management of clinically localized prostate cancer. JAMA 2005;293:20952101.

    • Search Google Scholar
    • Export Citation
  • 9.

    Johansson JE, Andren O, Andersson SO. Natural history of early, localized prostate cancer. JAMA 2004;291:27132719.

  • 10.

    Kattan MW, Cuzick J, Fisher G. Nomogram incorporating PSA level to predict cancer-specific survival for men with clinically localized prostate cancer managed without curative intent. Cancer 2008;112:6974.

    • Search Google Scholar
    • Export Citation
  • 11.

    Mohler J, Amling CL, Bahnson RR. NCCN clinical practice guidelines in oncology: prostate cancer. Version 2, 2009. Available at: http://www.nccn.org/professionals/physician_gls/PDF/prostate.pdf. Accessed December 29, 2009.

    • Search Google Scholar
    • Export Citation
  • 12.

    Albertsen P. Predicting survival for men with clinically localized prostate cancer: what do we need in contemporary practice? Cancer 2008;112:13.

    • Search Google Scholar
    • Export Citation
  • 13.

    Cooperberg MR, Lubeck DP, Meng MV. The changing face of low-risk prostate cancer: trends in clinical presentation and primary management. J Clin Oncol 2004;22:21412149.

    • Search Google Scholar
    • Export Citation
  • 14.

    Schroder FH, Denis LJ, Roobol M. The story of the European Randomized Study of Screening for Prostate Cancer. BJU Int 2003;92(Suppl 2):113.

  • 15.

    Andriole GL, Reding D, Hayes RB. The prostate, lung, colon, and ovarian (PLCO) cancer screening trial: status and promise. Urol Oncol 2004;22:358361.

    • Search Google Scholar
    • Export Citation
  • 16.

    Gann PH, Hennekens CH, Stampfer MJ. A prospective evaluation of plasma prostate-specific antigen for detection of prostatic cancer. JAMA 1995;273:289294.

    • Search Google Scholar
    • Export Citation
  • 17.

    Pearson JD, Carter HB. Natural history of changes in prostate specific antigen in early stage prostate cancer. J Urol 1994;152:17431748.

  • 18.

    Draisma G, Boer R, Otto SJ. 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;95:868878.

    • Search Google Scholar
    • Export Citation
All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 684 129 21
PDF Downloads 103 28 3
EPUB Downloads 0 0 0