- Email: [email protected]
PSA screening for prostate cancer: The current controversy – a viewpoint
Annals of Oncology 9: 1279-1282, 1998. © 1998 Kluwer Academic Publishers. Printed in the Netherlands.
Debate PSA screening for prostate cancer: The current controversy - a viewpoint M. J. Barry for the Patient Outcomes Research Team for Prostatic Diseases General Medicine Unit, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA
Summary
Introduction
Prostate cancer is an important health problem in the United States and in other developed Western countries. In 1998, the American Cancer Society estimates there will be 184,500 new cases of prostate cancer among Americans, with 39,200 deaths [1]. Historically, many prostate cancers had already spread systemically by the time of diagnosis, and were incurable. The advent of the prostate specific antigen (PSA) test has seen a remarkable increase, followed by a decrease in prostate cancer incidence in the US, due largely to use of this test for screening [2] (Figure 1). However, most authorities agree there is as yet no convincing evidence that efforts at early detection of prostate cancer with PSA (or other methods, for that matter), do more good than harm. At the same time, there is as yet no convincing evidence that these early detection efforts do more harm than good, either. Randomized trials of screening are underway in North America and Europe. Pending the results, should screening proceed until the evidence is negative, then be stopped; or should screening not be performed until the evidence is positive, then be started? Who should decide?
Key words: mass screening, prostatic neoplasia, prostatespecific antigen
For that matter, there is no evidence from clinical trials that aggressive treatment (with radical prostatectomy or radiotherapy) for men with prostate cancer that appears to be localized to the prostate gland improves outcomes, either. However, the negative evidence amounts to a single underpowered trial comparing radical prostatectomy and expectant management [5, 6]. In the absence of clinical trials, advocates of PSA screening cite data regarding an impressive stage shift
300
Rate per 100,000 (log scale)
100
Does PSA screening of asymptomatic men reduce mortality from prostate cancer? What is the strongest evidence available? No randomized trials have yet proven (or disproven) that early detection efforts with any modality improve outcomes (either morbidity and mortality) for any group of men. Two case-control studies have reached conflicting results regarding whether screening digital rectal exams prevent advanced or fatal prostate cancer [3, 4].
1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995
Year of Diagnosis/Death Figure 1. US prostate cancer mortality and incidence over time (ageadjusted to 1970 standard), by race. From SEER Cancer Statistics Review, 1973-1995 [2].
Downloaded from http://annonc.oxfordjournals.org/ by guest on March 15, 2016
Prostate cancer is an important health problem. Randomized trials have not yet proven whether or not screening with prostate-specific antigen measurements reduces morbidity or mortality. The potential for overtreatment of prostate cancers not destined to cause future mortality, the uncertainty about the benefits of aggressive treatment of screen-detected cancers,
and the relatively high costs of prostate cancer screening programs are all areas of concern. A shared approach to decisions about screening individual patients pending better evidence is one strategy for clinicians to consider in dealing with this controversial problem.
1280 Table 1. Ten-year cancer-specific survival (with 95% confidence intervals) for men with prostate cancer reported to the US SEER program, by degree of differentiation of tumor and treatment administered (intention to treat analysis)." Ten-year cancer-tumor differentiation Well Prostatectomy Radiation Conservative Moderate Prostatectomy Radiation Conservative Poor Prostatectomy Radiation Conservative
Specific survival 94% (91 %-95%) 90% (87%-92%) 93% (91 %-94%) 87% (85%-89%) 76% (72%-79%) 77% (74%-80%) 67% (62%-71%) 53% (47%-58%) 45% (40%-51 %)
a
Source: Lu-Yao GL, Yao SL. Population-based study of long-term survival in patients with clinically localised prostate cancer Lancet 1997; 349: 906-10 (used with permission), ©The Lancet Ltd.
Should PSA screening of the general population be routine practice, or is it still a research topic?
Perhaps the answer is both. In many countries, such as the US, PSA is already being widely used for early detection in primary care and urologic practices alike. In the absence of data confirming or refuting the benefits of PSA screening, however, PSA tests should not be routinely administered by physicians without discussion. One reasonable approach is that patients considering PSA screening should be made aware of the pros and cons of testing and come to a shared decision with their physicians about whether to proceed. In the USA, several trials have demonstrated that providing basic information about the pros and cons of PSA testing, in written or video form, reduces patient interest in PSA testing [13] and leads to fewer tests being performed [14]. However, lack of time in many medical practices can make this approach challenging; many clinicians may reasonably feel that limited time should be better spent discussing screening maneuvers with better evidence of effectiveness. Who pays for PSA screening is another important issue. Even if patients must pay for their own PSA tests, given the lack of evidence of benefit of PSA screening, most of the costs of such a screening program are generated by the subsequent biopsies, courses of radiotherapy, and radical prostatectomies [15], costs that will likely be shared with other individuals through government health care systems or private health insurance plans. One may argue that such widespread financial support of a potentially ineffective screening program is a poor use of societal resources. Finally, it is important to highlight a particular prob-
Downloaded from http://annonc.oxfordjournals.org/ by guest on March 15, 2016
with PSA screening, with the proportion of clinically and even pathologically localized cancers being much higher among screen-detected cancers than among historical or concurrent controls [7]. On the other hand, stage shift data are necessary but not sufficient to prove a benefit from screening, as they may simply result from lead-time bias [8]. Critics of screening also worry that an enormous potential for overdetection exists with PSA testing, given the large reservoir of undetected prostate cancers, even larger ones, documented by autopsy studies [9]. While advocates argue that the great majority of screen-detected cancers have histologic features that suggest they could be dangerous [10], the epidemiologic facts contradict that assertion. Since the cumulative incidence of a prostate cancer death in the US is only 3.6% (and relatively stable back into the era of less aggressive treatment), compared with an estimated cumulative incidence of a prostate cancer diagnosis that has now risen to 19%, the great majority of prostate cancers being diagnosed in the 'PSA era' cannot, in fact, be destined to cause mortality [1]. The relatively long doubling time of early prostate cancer, measured on average at 3-4 years or more [11], means a relatively good prognosis for many men with this cancer, even without early detection and treatment. Therefore, the marginal benefit of aggressive treatment for many screen-detected cancers is limited. In the absence of controlled trials, observational studies have examined the outcomes of men with clinically localized prostate cancer treated with different modalities. The largest of these nonexperimental studies involved an examination of the outcomes of about 60,000 men reported to the US Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute between 1983 and 1992 [12]. The degree of differentiation of the tumor was the best predictor of outcome, regardless of treatment (Table 1). For men with well differentiated cancers, outcomes were essentially the
same across the treatments. For men with moderately differentiated cancers, 10-year cancer specific survival was about 10% better, in absolute terms, with radical prostatectomy. For men with poorly differentiated cancers (a minority of cancers detected in PSA screening programs), the absolute difference in cancer specific survival was about 20% with radical prostatectomy compared to expectant management, with radiation therapy having an intermediate position. However, these outcome data as defined in observational studies of men with cancers detected largely in the 'pre-PSA era' are likely flawed for several reasons. First, with early detection, cancer-specific survival will improve regardless of treatment, because of the effect of lead time. Second, SEER grade and stage data are based on clinical data in the absence of surgery and pathologic data following surgery, which results in relative undergrading and understaging of cancers treated nonsurgically. These biases and others tend to favor surgery; whether when the results of currently ongoing trials of radical prostatectomy versus expectant management in the US and Scandinavia will demonstrate any advantage of surgery remains to be seen.
1281 lem with the strategy of screening pending the evidence. Widespread screening as part of 'usual care' may in fact 'contaminate' the control groups of clinical trials, making it impossible to prove that screening is beneficial. Major concerns about contamination have been raised regarding US prostate cancer screening trials; a valid answer about the effectiveness of screening may well need to come from trials in Europe, where the background rate of PSA screening appears to be much lower.
How should the effectiveness of PSA screening in the general population be measured?
Who should be screened? Should 'high risk' groups have different screening protocols? What are the 'high-risk' groups? Age is the dominant risk factor for prostate cancer in Western countries. While screening has not been proven effective for any age group, decision models have suggested that, assuming a benefit for aggressive treatment similar in magnitude to the nonexperimental studies cited above, that men age 50-69 are more likely to benefit at a reasonable cost [15]. Older men have fewer life-years to harvest the benefits of early detection, bear greater risks with treatment, and are often destined to die with, rather than of, any asymptomatic prostate cancer they may be harboring. Even advocates of PSA
What are the side effects of screening? The risks of a prostate cancer screening program are engendered through the psychological consequences of suspicious screening results or an actual prostate cancer diagnosis, as well as the morbidity and mortality of the cascade from diagnosis through staging to treatment. The psychological consequences of a suspicious screening test result or of a prostate cancer diagnosis have been poorly studied, but in early reports appear to be appreciable [18]. Particularly now that the false reassurance rate with biopsy in the setting of an elevated PSA has been demonstrated to be as high as 10%-15% [19], many men may remain anxious about the implications of their PSA results even after one or more sets of negative biopsies. The risks of biopsy, principally infection and bleeding, are small and usually self-limited. The real risks come with aggressive treatment, which has a small but finite mortality, and considerable morbidity. The major ongoing morbidities of treatment include incontinence and erectile dysfunction, both more common with surgery, and bowel dysfunction, which appears more common with radiation. When side effects are measured by independent surveys of random samples of men age 65 and older who have undergone surgery or radiatiotherapy, rates of side effects tend to be higher than reported from centers of excellence which directly define their own complication rates [20].
Conclusion In the absence of solid evidence of benefit, one reasonable approach to screening at the individual level is to
Downloaded from http://annonc.oxfordjournals.org/ by guest on March 15, 2016
Randomized clinical trials are the best way to determine whether prostate cancer screening is effective at reducing morbidity and mortality. Case-control studies in this area have methodologic limitations; in particular, it can be difficult to differentiate a screening from a diagnostic test for prostate cancer, and this imprecision in classification can have a major impact on the results of such studies [16]. Advocates have noted a small recent decrease in rates of population-based prostate cancer mortality in the US, and hope to attribute this finding to the impact of screening. However, not only is the downturn in mortality small, but also it follows a recent upswing in mortality coincident with the rise in incidence attributable to screening (Figure 1). Furthermore, these changes are being seen for a number of tumors other than prostate cancer [17], and may be attributable to other factors, such as population trends in dietary fat intake and other cancer risk factors. At the same time, a continued, more dramatic decrease in population-based prostate cancer mortality in the US would eventually legitimately raise the question of whether such data would ever be sufficient in itself to conclude prostate cancer screening is effective. Unfortunately, there is no consensus about this threshold; without clinical trials, the benefits of screening will likely continue to be a subject of controversy, regardless of population-based trends.
screening caution against screening men with a life expectancy of less 10 years; in the US, men with average comorbidity reach this point at about age 75. The other main high risk groups for prostate cancer in the United States are African-American men and men with first-degree relatives with prostate cancer. Screening high-risk groups is a strategy that is usually employed to improve the positive predictive value of the screening test by employing it in a population with an enriched 'prior probability' of disease. However, the prior probability of prostate cancer is usually not the issue, given its high prevalence. The problem with prostate cancer screening is whether it is effective at improving outcomes; if it is, it will likely be effective for men even at average risk. Men with risk factors should be made aware of them; however, targeted screening based on risk factors does not make sense for this malignancy. An exception might be that, if screening is proven effective, the lower age threshold for screening might reasonably be dropped for men with risk factors, as they will likely have a similar prior probability of disease as somewhat older men without risk factors.
1282 involve the patient in decisions about whether to perform a PSA test. Launching of mass screening programs for the early detection of prostate cancer is premature. Acknowledgement Supported by grant No. HS08397 from the US Agency for Health Care Policy and Research. References
Received 28 October 1998; accepted 28 October 1998.
Correspondence to. Michael J. Barry, MD Chief, General Medicine Unit Massachusetts General Hospital 50 Staniford Street - 9th Floor Boston, MA 02114, USA E-mail: barry [email protected]
Downloaded from http://annonc.oxfordjournals.org/ by guest on March 15, 2016
1. Landis SH, Murray T, Bolden S et al. Cancer statistics, 1998. CA Cancer J Clin 1998; 48: 6-29. 2. Ries L, Kosary CL, Hankey BF et al. SEER Cancer Statistics Review, 1973-1995. Bethesda, MD: National Cancer Institute 1998. 3. Friedman GD, Hiatt RA, Quesenberry CP et al. Case-control study of screening for prostatic cancer by digital rectal examinations. Lancet 1991; 337: 1526-9. 4. Jacobsen SJ, Berkstralh EJ, Katusic SK. et al. Screening digital rectal examination and prostate cancer mortality. A populationbased case-control study. Urology 1998; 52: 173-9. 5. Byar D, Corle D. VACURG randomized trial of radical prostatectomy for stages 1 and II prostate cancer. Urology (Suppl) 1981; 17:7-11. 6. Graversen PH, Nielsen KT, Gasser TC et al. Radical prostatectomy versus expectant primary treatment in Stages I and II prostatic cancer: A 15-year follow-up. Urol 1990; 36: 493-8. 7. Catalona WJ, Richie JP, Ahmann FR et al. Comparison of digital rectal examination and serum prostate specific antigen in the early detecction of prostate cancer: Results of a multicenter clinical trial of 6,630 men. J Urol 1994; 151: 1283-90. 8. Sackett DL, Holland WW. Controversy in the detection of disease. Lancet 1975; August 23: 357-9. 9. Coley CM, Barry MJ, Fleming C et al. Early detection of prostate cancer. Part I: Prior probability and effectiveness of tests. Ann Intern Med 1997; 126(5): 394-406. 10. Walsh PC. Prostate cancer kills: Strategy to reduce deaths. Urology 1994; 44(4): 463-6.
11. Schmid HP, McNeal JE, Stamey TA. Observations on the doubling time of prostate cancer: The use of serial prostate-specific antigen in patients with untreated disease as a measure of increasing cancer volume. Cancer 1993; 71: 2031—40. 12. Lu-Yao GL, Yao SL. Population-based study of long-term survival in patients with clinically localised prostate cancer. Lancet 1997; 349: 906-10. 13. Wolf AMD, Nasser JF, Wolf AM et al. The impact of informed consent on patient interest in prostate-specific antigen screening. Arch Int Med 1996; 156: 1333-6. 14. Flood AB, Wennberg JE, Nease RF et al. The importance of patient preference in the decision to screen for prostate cancer. J Gen Int Med 1996; 11:342-9. 15. Coley CM, Barry MJ, Fleming C et al. Early detection of prostate cancer. Part II: Estimating the risks, benefits, and costs. Ann Intern Med 1997; 126 (6): 468-79. 16. Concato J. What is a screening test? Misclassification bias in observational studies of screening for cancer. J Gen Int Med 1997; 12- 607-12. 17. Cole P, Rodu B. Declining cancer mortality in the United States. Cancer 1996; 78: 2045-8. 18. Katz DA, McHorney CA, Schapira MM. Health-related quality of life in men with localized prostate cancer. J Gen Int Med 1998; 13 (Suppl): 16 19. Levine MA, Ittman M, Melamed J et al. Two consecutive sets of transrectal ultrasound guided sextant biopsies of the prostate for the detection of prostate cancer. J Urol 1998; 159: 471-6. 20. Fowler FJ, Barry MJ, Lu-Yao et al. Outcomes of external-beam radiation therapy for prostate cacner: A study of Medicare Beneficiaries in three Surveillance, Epidemiology, and End results areas. J Clin Oncol 1996, 14: 2258-65.
Debate PSA screening for prostate cancer: The current controversy - a viewpoint M. J. Barry for the Patient Outcomes Research Team for Prostatic Diseases General Medicine Unit, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA
Summary
Introduction
Prostate cancer is an important health problem in the United States and in other developed Western countries. In 1998, the American Cancer Society estimates there will be 184,500 new cases of prostate cancer among Americans, with 39,200 deaths [1]. Historically, many prostate cancers had already spread systemically by the time of diagnosis, and were incurable. The advent of the prostate specific antigen (PSA) test has seen a remarkable increase, followed by a decrease in prostate cancer incidence in the US, due largely to use of this test for screening [2] (Figure 1). However, most authorities agree there is as yet no convincing evidence that efforts at early detection of prostate cancer with PSA (or other methods, for that matter), do more good than harm. At the same time, there is as yet no convincing evidence that these early detection efforts do more harm than good, either. Randomized trials of screening are underway in North America and Europe. Pending the results, should screening proceed until the evidence is negative, then be stopped; or should screening not be performed until the evidence is positive, then be started? Who should decide?
Key words: mass screening, prostatic neoplasia, prostatespecific antigen
For that matter, there is no evidence from clinical trials that aggressive treatment (with radical prostatectomy or radiotherapy) for men with prostate cancer that appears to be localized to the prostate gland improves outcomes, either. However, the negative evidence amounts to a single underpowered trial comparing radical prostatectomy and expectant management [5, 6]. In the absence of clinical trials, advocates of PSA screening cite data regarding an impressive stage shift
300
Rate per 100,000 (log scale)
100
Does PSA screening of asymptomatic men reduce mortality from prostate cancer? What is the strongest evidence available? No randomized trials have yet proven (or disproven) that early detection efforts with any modality improve outcomes (either morbidity and mortality) for any group of men. Two case-control studies have reached conflicting results regarding whether screening digital rectal exams prevent advanced or fatal prostate cancer [3, 4].
1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995
Year of Diagnosis/Death Figure 1. US prostate cancer mortality and incidence over time (ageadjusted to 1970 standard), by race. From SEER Cancer Statistics Review, 1973-1995 [2].
Downloaded from http://annonc.oxfordjournals.org/ by guest on March 15, 2016
Prostate cancer is an important health problem. Randomized trials have not yet proven whether or not screening with prostate-specific antigen measurements reduces morbidity or mortality. The potential for overtreatment of prostate cancers not destined to cause future mortality, the uncertainty about the benefits of aggressive treatment of screen-detected cancers,
and the relatively high costs of prostate cancer screening programs are all areas of concern. A shared approach to decisions about screening individual patients pending better evidence is one strategy for clinicians to consider in dealing with this controversial problem.
1280 Table 1. Ten-year cancer-specific survival (with 95% confidence intervals) for men with prostate cancer reported to the US SEER program, by degree of differentiation of tumor and treatment administered (intention to treat analysis)." Ten-year cancer-tumor differentiation Well Prostatectomy Radiation Conservative Moderate Prostatectomy Radiation Conservative Poor Prostatectomy Radiation Conservative
Specific survival 94% (91 %-95%) 90% (87%-92%) 93% (91 %-94%) 87% (85%-89%) 76% (72%-79%) 77% (74%-80%) 67% (62%-71%) 53% (47%-58%) 45% (40%-51 %)
a
Source: Lu-Yao GL, Yao SL. Population-based study of long-term survival in patients with clinically localised prostate cancer Lancet 1997; 349: 906-10 (used with permission), ©The Lancet Ltd.
Should PSA screening of the general population be routine practice, or is it still a research topic?
Perhaps the answer is both. In many countries, such as the US, PSA is already being widely used for early detection in primary care and urologic practices alike. In the absence of data confirming or refuting the benefits of PSA screening, however, PSA tests should not be routinely administered by physicians without discussion. One reasonable approach is that patients considering PSA screening should be made aware of the pros and cons of testing and come to a shared decision with their physicians about whether to proceed. In the USA, several trials have demonstrated that providing basic information about the pros and cons of PSA testing, in written or video form, reduces patient interest in PSA testing [13] and leads to fewer tests being performed [14]. However, lack of time in many medical practices can make this approach challenging; many clinicians may reasonably feel that limited time should be better spent discussing screening maneuvers with better evidence of effectiveness. Who pays for PSA screening is another important issue. Even if patients must pay for their own PSA tests, given the lack of evidence of benefit of PSA screening, most of the costs of such a screening program are generated by the subsequent biopsies, courses of radiotherapy, and radical prostatectomies [15], costs that will likely be shared with other individuals through government health care systems or private health insurance plans. One may argue that such widespread financial support of a potentially ineffective screening program is a poor use of societal resources. Finally, it is important to highlight a particular prob-
Downloaded from http://annonc.oxfordjournals.org/ by guest on March 15, 2016
with PSA screening, with the proportion of clinically and even pathologically localized cancers being much higher among screen-detected cancers than among historical or concurrent controls [7]. On the other hand, stage shift data are necessary but not sufficient to prove a benefit from screening, as they may simply result from lead-time bias [8]. Critics of screening also worry that an enormous potential for overdetection exists with PSA testing, given the large reservoir of undetected prostate cancers, even larger ones, documented by autopsy studies [9]. While advocates argue that the great majority of screen-detected cancers have histologic features that suggest they could be dangerous [10], the epidemiologic facts contradict that assertion. Since the cumulative incidence of a prostate cancer death in the US is only 3.6% (and relatively stable back into the era of less aggressive treatment), compared with an estimated cumulative incidence of a prostate cancer diagnosis that has now risen to 19%, the great majority of prostate cancers being diagnosed in the 'PSA era' cannot, in fact, be destined to cause mortality [1]. The relatively long doubling time of early prostate cancer, measured on average at 3-4 years or more [11], means a relatively good prognosis for many men with this cancer, even without early detection and treatment. Therefore, the marginal benefit of aggressive treatment for many screen-detected cancers is limited. In the absence of controlled trials, observational studies have examined the outcomes of men with clinically localized prostate cancer treated with different modalities. The largest of these nonexperimental studies involved an examination of the outcomes of about 60,000 men reported to the US Surveillance, Epidemiology, and End Results (SEER) program of the National Cancer Institute between 1983 and 1992 [12]. The degree of differentiation of the tumor was the best predictor of outcome, regardless of treatment (Table 1). For men with well differentiated cancers, outcomes were essentially the
same across the treatments. For men with moderately differentiated cancers, 10-year cancer specific survival was about 10% better, in absolute terms, with radical prostatectomy. For men with poorly differentiated cancers (a minority of cancers detected in PSA screening programs), the absolute difference in cancer specific survival was about 20% with radical prostatectomy compared to expectant management, with radiation therapy having an intermediate position. However, these outcome data as defined in observational studies of men with cancers detected largely in the 'pre-PSA era' are likely flawed for several reasons. First, with early detection, cancer-specific survival will improve regardless of treatment, because of the effect of lead time. Second, SEER grade and stage data are based on clinical data in the absence of surgery and pathologic data following surgery, which results in relative undergrading and understaging of cancers treated nonsurgically. These biases and others tend to favor surgery; whether when the results of currently ongoing trials of radical prostatectomy versus expectant management in the US and Scandinavia will demonstrate any advantage of surgery remains to be seen.
1281 lem with the strategy of screening pending the evidence. Widespread screening as part of 'usual care' may in fact 'contaminate' the control groups of clinical trials, making it impossible to prove that screening is beneficial. Major concerns about contamination have been raised regarding US prostate cancer screening trials; a valid answer about the effectiveness of screening may well need to come from trials in Europe, where the background rate of PSA screening appears to be much lower.
How should the effectiveness of PSA screening in the general population be measured?
Who should be screened? Should 'high risk' groups have different screening protocols? What are the 'high-risk' groups? Age is the dominant risk factor for prostate cancer in Western countries. While screening has not been proven effective for any age group, decision models have suggested that, assuming a benefit for aggressive treatment similar in magnitude to the nonexperimental studies cited above, that men age 50-69 are more likely to benefit at a reasonable cost [15]. Older men have fewer life-years to harvest the benefits of early detection, bear greater risks with treatment, and are often destined to die with, rather than of, any asymptomatic prostate cancer they may be harboring. Even advocates of PSA
What are the side effects of screening? The risks of a prostate cancer screening program are engendered through the psychological consequences of suspicious screening results or an actual prostate cancer diagnosis, as well as the morbidity and mortality of the cascade from diagnosis through staging to treatment. The psychological consequences of a suspicious screening test result or of a prostate cancer diagnosis have been poorly studied, but in early reports appear to be appreciable [18]. Particularly now that the false reassurance rate with biopsy in the setting of an elevated PSA has been demonstrated to be as high as 10%-15% [19], many men may remain anxious about the implications of their PSA results even after one or more sets of negative biopsies. The risks of biopsy, principally infection and bleeding, are small and usually self-limited. The real risks come with aggressive treatment, which has a small but finite mortality, and considerable morbidity. The major ongoing morbidities of treatment include incontinence and erectile dysfunction, both more common with surgery, and bowel dysfunction, which appears more common with radiation. When side effects are measured by independent surveys of random samples of men age 65 and older who have undergone surgery or radiatiotherapy, rates of side effects tend to be higher than reported from centers of excellence which directly define their own complication rates [20].
Conclusion In the absence of solid evidence of benefit, one reasonable approach to screening at the individual level is to
Downloaded from http://annonc.oxfordjournals.org/ by guest on March 15, 2016
Randomized clinical trials are the best way to determine whether prostate cancer screening is effective at reducing morbidity and mortality. Case-control studies in this area have methodologic limitations; in particular, it can be difficult to differentiate a screening from a diagnostic test for prostate cancer, and this imprecision in classification can have a major impact on the results of such studies [16]. Advocates have noted a small recent decrease in rates of population-based prostate cancer mortality in the US, and hope to attribute this finding to the impact of screening. However, not only is the downturn in mortality small, but also it follows a recent upswing in mortality coincident with the rise in incidence attributable to screening (Figure 1). Furthermore, these changes are being seen for a number of tumors other than prostate cancer [17], and may be attributable to other factors, such as population trends in dietary fat intake and other cancer risk factors. At the same time, a continued, more dramatic decrease in population-based prostate cancer mortality in the US would eventually legitimately raise the question of whether such data would ever be sufficient in itself to conclude prostate cancer screening is effective. Unfortunately, there is no consensus about this threshold; without clinical trials, the benefits of screening will likely continue to be a subject of controversy, regardless of population-based trends.
screening caution against screening men with a life expectancy of less 10 years; in the US, men with average comorbidity reach this point at about age 75. The other main high risk groups for prostate cancer in the United States are African-American men and men with first-degree relatives with prostate cancer. Screening high-risk groups is a strategy that is usually employed to improve the positive predictive value of the screening test by employing it in a population with an enriched 'prior probability' of disease. However, the prior probability of prostate cancer is usually not the issue, given its high prevalence. The problem with prostate cancer screening is whether it is effective at improving outcomes; if it is, it will likely be effective for men even at average risk. Men with risk factors should be made aware of them; however, targeted screening based on risk factors does not make sense for this malignancy. An exception might be that, if screening is proven effective, the lower age threshold for screening might reasonably be dropped for men with risk factors, as they will likely have a similar prior probability of disease as somewhat older men without risk factors.
1282 involve the patient in decisions about whether to perform a PSA test. Launching of mass screening programs for the early detection of prostate cancer is premature. Acknowledgement Supported by grant No. HS08397 from the US Agency for Health Care Policy and Research. References
Received 28 October 1998; accepted 28 October 1998.
Correspondence to. Michael J. Barry, MD Chief, General Medicine Unit Massachusetts General Hospital 50 Staniford Street - 9th Floor Boston, MA 02114, USA E-mail: barry [email protected]
Downloaded from http://annonc.oxfordjournals.org/ by guest on March 15, 2016
1. Landis SH, Murray T, Bolden S et al. Cancer statistics, 1998. CA Cancer J Clin 1998; 48: 6-29. 2. Ries L, Kosary CL, Hankey BF et al. SEER Cancer Statistics Review, 1973-1995. Bethesda, MD: National Cancer Institute 1998. 3. Friedman GD, Hiatt RA, Quesenberry CP et al. Case-control study of screening for prostatic cancer by digital rectal examinations. Lancet 1991; 337: 1526-9. 4. Jacobsen SJ, Berkstralh EJ, Katusic SK. et al. Screening digital rectal examination and prostate cancer mortality. A populationbased case-control study. Urology 1998; 52: 173-9. 5. Byar D, Corle D. VACURG randomized trial of radical prostatectomy for stages 1 and II prostate cancer. Urology (Suppl) 1981; 17:7-11. 6. Graversen PH, Nielsen KT, Gasser TC et al. Radical prostatectomy versus expectant primary treatment in Stages I and II prostatic cancer: A 15-year follow-up. Urol 1990; 36: 493-8. 7. Catalona WJ, Richie JP, Ahmann FR et al. Comparison of digital rectal examination and serum prostate specific antigen in the early detecction of prostate cancer: Results of a multicenter clinical trial of 6,630 men. J Urol 1994; 151: 1283-90. 8. Sackett DL, Holland WW. Controversy in the detection of disease. Lancet 1975; August 23: 357-9. 9. Coley CM, Barry MJ, Fleming C et al. Early detection of prostate cancer. Part I: Prior probability and effectiveness of tests. Ann Intern Med 1997; 126(5): 394-406. 10. Walsh PC. Prostate cancer kills: Strategy to reduce deaths. Urology 1994; 44(4): 463-6.
11. Schmid HP, McNeal JE, Stamey TA. Observations on the doubling time of prostate cancer: The use of serial prostate-specific antigen in patients with untreated disease as a measure of increasing cancer volume. Cancer 1993; 71: 2031—40. 12. Lu-Yao GL, Yao SL. Population-based study of long-term survival in patients with clinically localised prostate cancer. Lancet 1997; 349: 906-10. 13. Wolf AMD, Nasser JF, Wolf AM et al. The impact of informed consent on patient interest in prostate-specific antigen screening. Arch Int Med 1996; 156: 1333-6. 14. Flood AB, Wennberg JE, Nease RF et al. The importance of patient preference in the decision to screen for prostate cancer. J Gen Int Med 1996; 11:342-9. 15. Coley CM, Barry MJ, Fleming C et al. Early detection of prostate cancer. Part II: Estimating the risks, benefits, and costs. Ann Intern Med 1997; 126 (6): 468-79. 16. Concato J. What is a screening test? Misclassification bias in observational studies of screening for cancer. J Gen Int Med 1997; 12- 607-12. 17. Cole P, Rodu B. Declining cancer mortality in the United States. Cancer 1996; 78: 2045-8. 18. Katz DA, McHorney CA, Schapira MM. Health-related quality of life in men with localized prostate cancer. J Gen Int Med 1998; 13 (Suppl): 16 19. Levine MA, Ittman M, Melamed J et al. Two consecutive sets of transrectal ultrasound guided sextant biopsies of the prostate for the detection of prostate cancer. J Urol 1998; 159: 471-6. 20. Fowler FJ, Barry MJ, Lu-Yao et al. Outcomes of external-beam radiation therapy for prostate cacner: A study of Medicare Beneficiaries in three Surveillance, Epidemiology, and End results areas. J Clin Oncol 1996, 14: 2258-65.