Fax: (313) 874-6656
Trends in the utilization of androgen-deprivation therapy for patients with prostate carcinoma suggest an effect on mortality
Article first published online: 23 OCT 2001
Copyright © 2001 American Cancer Society
Volume 92, Issue 9, pages 2309–2317, 1 November 2001
How to Cite
Demers, R. Y., Tiwari, A., Wei, J., Weiss, L. K., Severson, R. K. and Montie, J. (2001), Trends in the utilization of androgen-deprivation therapy for patients with prostate carcinoma suggest an effect on mortality. Cancer, 92: 2309–2317. doi: 10.1002/1097-0142(20011101)92:9<2309::AID-CNCR1577>3.0.CO;2-8
- Issue published online: 23 OCT 2001
- Article first published online: 23 OCT 2001
- Manuscript Accepted: 23 JUL 2001
- Manuscript Revised: 13 JUN 2001
- Manuscript Received: 19 MAR 2001
- prostate carcinoma;
- and End Results (SEER) Program
After a surge in the incidence of prostate carcinoma in the early 1990s, diminishing rates of mortality became apparent in 1993. This decrease in mortality is unlikely to be explained entirely by treatment with curative intent alone following screen-detected cases, because the time frame between detection and mortality remains relatively brief.
This study used incidence and initial treatment data from the Detroit area SEER registry between 1973 and 1998 in addition to mortality data covering the Metropolitan Detroit area obtained from the Michigan Department of Community Health. Data for Caucasian and African-American men were analyzed. The use of androgen-deprivation therapy, which evolved during the study period, was evaluated in conjunction with mortality and incidence trend data for consideration of etiologic contributions.
The incidence of prostate carcinoma, as noted previously in national data, increased sharply in 1988, peaking in 1992 in Southeast Michigan, whereas mortality rates began to decrease in approximately 1993, with a sustained decrease to the latest recorded data in 1998. These trends were identical in Caucasians and African Americans. A sharp increase in the use of androgen-deprivation therapy began in 1990. This use of androgen-deprivation therapy is high and sustained for patients with early-stage disease, increases for several years, and then diminishes for patients with regional disease. The use also diminished through the 1990s for patients with late-stage disease, paralleling the decrease in the incidence rate for late-stage disease.
The pattern of androgen-deprivation therapy usage was consistent with that for hormonal monotherapy and adjuvant and neoadjuvant therapy. These findings suggest that androgen-deprivation therapy may contribute, along with advances in diagnostic techniques and curative therapy with radiation or surgery, toward decreasing prostate carcinoma mortality rates in Southeast Michigan. Cancer 2001;92:2309–17. © 2001 American Cancer Society.
The decreasing trend in prostate carcinoma mortality rates in the United States in recent years has stimulated numerous hypothetical considerations, many of which attempt to link the trend to a prostate specific antigen (PSA) screening effect. However, controversy over the effectiveness of PSA screening remains. Leading national organizations continue to debate possible PSA screening benefits and express opinions in various directions, contributing to confusion on this topic for both medical providers and men considering PSA screening. The United States Preventative Services Task Force, the Canadian Task Force on the Periodic Health Examination, and the Canadian Urological Association emphatically do not recommend PSA screening.1–3 The National Cancer Institute does not actively recommend it but leaves the judgment with clinicians.4 The American Cancer Society and the American Urological Association recommend that clinicians discuss PSA screening with all men age ≥ 50 years as part of an annual prostate examination and that PSA screening should begin at age 40 years in men who are at high risk.5, 6
This debate may be diminished once data from ongoing randomized clinical trials become available. The largest such trial in the United States is the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO) funded through the National Cancer Institute. The PLCO trial will include 74,000 men ages 55–74 years and has a design power of 90% to detect a 20% reduction in prostate carcinoma mortality using a baseline screening examination and four subsequent annual screening examinations, including a PSA blood test and digital rectal examination. This randomized trial began in 1993, but final results are still years away.7 Other trials, including the European Randomized Study of Screening for Prostate Cancer, also are underway.8 This trial involves the study of over 20,000 volunteers, ages 55–76 years, who were enrolled between 1993 and 1997.
There is historic precedence for using epidemiologic data to reach consensus on the effectiveness of cancer screening. With cervical carcinoma, randomized studies were an almost impossible means of assessing whether the Papanicolaou (Pap) test was efficacious in lowering mortality. The medical standard of care for the use of this test reached widespread acceptance rapidly in the 1940s, precluding studies that may have included women who had not been screened at the time the study began and who would remain unscreened through the course of a randomized clinical trial. Longitudinal data showed decreasing mortality rates for cervical carcinoma in addition to increasing survival rates that seemed to coincide with the occurrence of Pap smear testing. These trends continued further with the intensity of testing. Eventually, this indirect evidence led to the universal endorsement of Pap test screening.9
Evidence from epidemiologic studies suggests a positive impact of early detection for patients with prostate carcinoma.10 A trend toward decreasing mortality rates for patients with prostate carcinoma led to some early enthusiasm, speculating that PSA screening was responsible for a screening effect resulting in curative treatment that was lowering mortality rates.11 However, conclusive evidence does not exist, and the Cancer Surveillance Series published in the Journal of the National Cancer Institute in 1999 suggested that a sense of caution should be applied to early optimism over the efficacy of PSA screening.12–14 Although there is considerable enthusiasm over decreasing prostate carcinoma mortality rates and a desire to link the decrease causally to PSA screening, clinicians, patients, and policy makers alike remain at a loss for a correct decision regarding advice for or against screening for prostate carcinoma.
In this study, the most current data are presented showing a sustained decrease in prostate carcinoma mortality rates in Southeast Michigan, both for African-American men and Caucasian men. We propose a possible contributing explanation for these findings that is based on the practice of androgen-deprivation therapy (ADT), a therapy that, by itself, does not lead to cure. In addition, we place the current trends in the context of other recent historic occurrences: advances in both diagnosis and treatment that, together with PSA-driven early detection and ADT, may contribute to a theory of a multifactorial explanation for decreasing mortality rates.
MATERIALS AND METHODS
Ascertainment of Prostate Carcinoma Cases and Mortality Data
Prostate carcinoma incidence data for this analysis were obtained from the Metropolitan Detroit Cancer Surveillance System, one of the founding registries of the Surveillance, Epidemiology, and End Results (SEER) Program of the National Cancer Institute. These population-based data included Wayne, Oakland, and Macomb Counties in Southeastern Michigan. The 1990 U.S. Census recorded that these counties had a total population of 3.9 million, of whom 24% were African American. Cases were restricted to those men who 1) were newly diagnosed with a neoplasm of the prostate (ICD-0-2 code of C61.9); 2) were diagnosed during the period from January 1973 through December 1998; 3) had a malignant neoplasm (fifth digit of ICD-0-2 morphology code, 3); and 4) were residents of Wayne, Oakland, or Macomb Counties at the time of diagnosis.15 Within the SEER Program, the stage of prostate carcinoma is recorded as local if the disease is confined to the prostate without penetration of the capsule. Regional prostate carcinoma indicates that the disease has extended beyond the capsule into periprostatic tissues and/or regional lymph nodes. Distant prostate carcinoma is coded if the disease has metastasized to remote organ sites. Mortality data were obtained from the Michigan Department of Community Health. Deaths in this study were restricted to men who 1) expired due to a neoplasm of the prostate; 2) were residents of Wayne, Oakland, or Macomb Counties at the time of death; and 3) died between January 1973 and December 1998.
Ascertainment of Treatment Data
Treatment that occurs within the first 4 months after the diagnosis of prostate carcinoma is recorded by SEER as a first course of treatment. If the medical records indicate (within this 4-month period) that a treatment is planned, then that treatment also is recorded as the first course of treatment if it occurs within 1 year of diagnosis. The attempted curative treatments (e.g., radical prostatectomy or radiation therapy) and palliative treatments (e.g., ADT) are identified within medical and pathology records then reported in SEER abstracts.
With the appreciation that increasing proportions of prostate carcinoma diagnoses and treatments occur in the ambulatory setting, the Detroit SEER Program initiated attempts to acquire these data beginning in the 1970s. The system for acquiring these data has remained the same through 1998. If initially available data are limited to evidence of a pathologic diagnosis of prostate carcinoma, then a letter is sent to the physician of record, usually a urologist. The physician returns this letter indicating further data on the diagnosis and treatment, including the use of various hormone-related treatment modalities. To ascertain an estimate of the proportion of patients receiving ADT in the ambulatory setting, we reviewed physician responses over the last 4 recorded years. From 1994 through 1998, physicians responded at a rate of 33%, with 29% of respondents indicating that their patients were receiving hormone therapy as a component of their treatment.
The prime intent of tracking data on the first course of treatment in this study was to decipher patterns of ADT either alone or in conjunction with radiation or surgical therapy. Age-adjusted ADT usage rates were calculated in a manner similar to that used to calculated incidence rates. Treatment trends were analyzed by disease stage at diagnosis and whether ADT was the sole treatment or was used in combination with other treatment approaches, most commonly radical prostatectomy or radiation therapy. ADT is a general category of treatment recorded by the SEER data base; however, due to the inconsistency of capture through the medical records, differentiation between neoadjuvant versus adjuvant use was not attempted.
Annual age-adjusted, race specific incidence and mortality rates were computed for all men combined and separately for African-American men and Caucasian men. Denominator data were derived from interpolations based on the 1970, 1980, and 1990 U.S. Census and were computed by the SEER Program. Rates were adjusted to the 1970 U.S. standard population using the direct method. Unless otherwise noted, all rates are expressed per 100,000 population per year. Time trends were plotted using 3-year moving averages. Statistical tests for trends were computed using linear regression models.16
Complementing SEER data are historic events that evolved between 1973 and 1998 and that may have had an effect on prostate carcinoma incidence, mortality, or both. These events include both diagnostic and screening breakthroughs, such as transrectal ultrasound (TRUS), biopsy, and PSA, and treatment advances, such as nerve-sparing radical prostatectomy and ADT.17–23 These diagnostic and treatment advances were adopted into clinical practice throughout the United States, including Michigan. These events are included in Figure 1, which displays time trends for incidence and mortality rates along with the publication date of the earliest cited references; thus, Figure 1 only approximates the chronology of changing practice patterns that may have changed either before or after the publication date.
New cases of prostate carcinoma totaled 56,425 between 1973 and 1998 among the residents of Wayne, Oakland, and Macomb Counties of metropolitan Detroit. Table 1 summarizes the age, stage at diagnosis, and race of the incident cases. Older men were more likely to be diagnosed with prostate carcinoma (> 75% are diagnosed at age ≥ 65 years), and African-American men were more likely to present with distant disease compared with Caucasian men (19.5% vs. 12.2%, respectively). Table 2 depicts mortality rates for both African-American men and Caucasian men from 1988 through 1998, showing a persistent, two-fold greater mortality rate among African-American men. A gradual rise in mortality rates is noted among Caucasian men until 1991 and among African-American men until 1992, at which time the mortality rates for both races show a steady decrease through 1998. Mortality rate trends for black men and white men are in parallel.
|Characteristic||White males||Black males||Other malesc||All males|
|Disease stage at diagnosis|
|Year||White males||Black males|
Figure 1 summarizes prostate carcinoma incidence and mortality rates for African-American and Caucasian men from 1973 through 1998. The sharp rise in prostate carcinoma incidence for both African-American and Caucasian men began in 1988 and peaked between 1993 and 1994. Prostate carcinoma mortality rates, as noted above, began decreasing in approximately 1992, adjusted for the 3-year moving average. The decreasing trend in mortality rates reached statistical significance (P = 0.0004) over the most recently recorded 6 years.16 Figure 2 shows age-adjusted prostate carcinoma incidence rates by disease stage at the time of diagnosis from 1973 through 1998. The figure shows the harvesting effect of prevalent cases subsequent to the onset of PSA screening in 1988. Not only were patients with early-stage disease identified, but (probably asymptomatic) patients with regional disease also were also diagnosed, as would be expected with rapid start-up screening activity. Those who were diagnosed in subsequent years (approximately 1993 and beyond) appear to have been diagnosed with a greater preponderance of early-stage disease. The proportion of patients with prostate carcinoma who were diagnosed with distant disease decreased from 1993 onward relative to baseline proportions from 1973 through 1989.
Figures 3–5 display age-adjusted rates for the first course of treatment with ADT either alone or in combination with other treatments, such as surgery or radiation. Figure 3 depicts a sharp and sustained rise in the use of ADT in combination with other forms of treatment for patients with local prostate carcinoma from the early 1990s through 1998. This trend probably reflects the increasing application of neoadjuvant and adjuvant therapies—likely gonadotropin-releasing hormone (leutenizing hormone-releasing hormone [LHRH] agonists; see Fig. 1)—beginning in the same period. Figures 4 and 5 show ADT treatment trends for patients with regional and distant disease. Rates of ADT use for those with regional disease increased in the early 1990s, then began decreasing in approximately 1993, paralleling the rise and fall of regional disease incidence. For ADT as a sole treatment, probably for palliation in patients with late-stage disease, the rates begin to decrease in approximately 1990, corresponding to the decreased incidence of distant disease shown in Figure 2. The SEER data show that approximately 60% of Detroit area men with late-stage prostate carcinoma received ADT as initial treatment since 1988 compared with 15% of men with local or regional disease.
Decreasing mortality rates over the past 7–8 years in Southeast Michigan may be associated with the use of LHRH agonists administered as adjuvant or neoadjuvant therapy for patients with local disease and, to a lesser extent, regional disease. It is plausible that increasing survival experienced from the use of ADT, combined with other forms of therapy for local and regional disease, explain at least in part the lowering of mortality rates.
We have presented data suggesting that the introduction of LHRH agonist depot formulations in early 1989 resulted in a sharp increase in the use of ADT. Furthermore, we have found that this use often is combined with other therapies, including radical prostatectomy and radiation therapy (Fig. 3). The data from this study do not distinguish clearly between those men who have received neoadjuvant ADT and those who have received adjuvant ADT; therefore, both applications may be important. It is also worth noting that neoadjuvant ADT in the setting of prostatectomy and radiation therapy is a recent development and precedes the fall in mortality by a relatively brief time, approximately 2–3 years.24, 25 The findings from this study, taken together with the growing body of evidence that support an effect of protracted ADT on survival, suggest that the drop in prostate carcinoma mortality reported in Southeast Michigan is due in part to the increased use of adjuvant ADT or early therapy based on rising PSA levels only.
Caution about the use of treatment data obtained from the SEER data base is warranted, as described above (see Materials and Methods). In this study, the key analytic variable, hormone-based treatment, must be scrutinized even more closely, because much of this treatment is performed in the outpatient setting. Collection of such data requires interface with physicians' offices, in our study, through a mailed survey. The survey methodology remained unchanged through the entire study period, removing a change in methodology as an explanation for increasing hormone treatment rates in the 1990s. Although true hormone treatment rates are likely to be underestimated, the trend in change of usage probably is a valid depiction.
Animal studies have shown that hormonal manipulation, if applied early in the course of tumor growth, results in a considerable increase in survival among study animals.26, 27 This increased survival in animal studies is supported further by clinical trials. Messing et al.28 recently suggested that early ADT actually may increase survival in subsets of patients with advanced disease after radical prostatectomy. In that report, 98 men with clinically localized prostate carcinoma were randomized to receive immediate or delayed ADT after the detection of positive lymph node disease by surgery. The difference in overall survival between the two groups was remarkable, with 77% of the early treatment group alive at 36 months compared with only 18% in the observation group. Furthermore, prostate carcinoma specific death was statistically significantly greater in the observation group compared with the early treatment group. Similar evidence also exists for radiation therapy for patients with clinically localized prostate carcinoma. Bolla and colleagues29 demonstrated improved survival in patients who were treated with adjuvant ADT that was extended for 3 years after radiation therapy. This randomized trial demonstrated that the proportion of surviving men who received the adjuvant therapy were more likely to be free of disease (85%) compared with only 48% of men in the radiation alone group.
A theoretical explanation for decreasing mortality rates evolves from these animal and human studies. Increasing ADT use beginning in the early 1990s for patients with early-stage and regional disease may have resulted in extended survival, particularly for patients with aggressive local disease and asymptomatic regional disease, in addition to those with clinically staged local disease, which is actually regional disease that is undocumented by surgical exploration. Men with prostate carcinoma in these categories may have died sooner if ADT had not been used as a component of their treatment. Taken together, these data support the hypothesis that ADT given early together with other local therapy in the course of the disease can favorably impact mortality.
In the Metropolitan Detroit area, ADT has been employed consistently alone or in combination with primary therapies since 1973, although its use increased dramatically in 1990, especially for patients with local and regional disease (Figs. 3, 4). This is despite a declining proportion of patients with advanced disease, leading to the conclusion that ADT has become more common as a neoadjuvant or adjuvant therapy for patients with prostate carcinoma. This marked change in the use of ADT likely is related to three developments that occurred in the late 1980s through the early 1990s: increased survival attributed to maximal androgen blockade, introduction of depot injections for LHRH agonists, and increasing occurrence of rising PSA levels after previous radical prostatectomy or radiation therapy. PSA failure after primary treatment, with or without identifiable metastases, often has prompted the initiation of ADT, even without confirmatory evidence supporting this strategy. In 1989, a well-publicized report suggested that the combined use of leuprolide with flutamide may improve the median survival in patients with disseminated prostate carcinoma.30 In the next year, the safety and clinical efficacy of depot deuprorelin acetate formulation was reported. Although developments have encouraged the use of ADT for patients with advanced disease, reports in the early 1990s of potential benefit with neoadjuvant and adjuvant therapies have led to even more widespread use earlier in the natural history of the disease.31–33 Such use of ADT may result in greater disease specific survival, as suggested by Messing et al.28 and Bolla et al.
The correlation between PSA testing and evolving practice patterns changing incidence and mortality rates is complex and must not be oversimplified. The notable and sharp increase in prostate carcinoma incidence beginning in 1990 has been attributed to the widespread use of PSA screening for prostate carcinoma. Other contributing factors may include ultrasound guidance for prostate biopsies, application of sextant pattern biopsies, and an increase in public awareness of prostate carcinoma.20, 34, 35 Because these diagnostic interventions were introduced into clinical practice gradually at about the same time, it becomes impossible to assign the contribution of each separate cause to the increase in prostate carcinoma incidence. Recognizing the role of PSA screening while failing to acknowledge the multifactorial nature of the sharp increase in incidence may lead clinicians to an incorrect interpretation of the cause of promising trends in prostate carcinoma incidence and mortality.
Events introduced in the distant past, including the introduction of ADT, needle aspiration of the prostate, and bone imaging for patients with metastatic disease, are unlikely to be sentinel events, because they have been applied widely for more than 10 years without any significant effect on mortality trends.33 Indeed, from 1973 to 1990, mortality rates increased at a slow but constant rate in the Southeast Michigan region. Because the fall in prostate carcinoma mortality began in approximately 1993, we considered a number of sentinel events that occurred in the previous 10–15-year period. Important advances during that period included the development and widespread application of TRUS-guided needle biopsy of the prostate, radical retropubic prostatectomy, modern external beam radiation therapy, and depot formulations of LHRH agonists.35–38 Other therapeutic strategies, such as brachytherapy, are unlikely candidates, because they became common practices only after the decrease in mortality had begun.39 Again, like the incidence rates for prostate carcinoma, it may be difficult to discern clearly the contribution of individual modalities to decreasing mortality rates of prostate carcinoma, they were phased into clinical practice gradually over a period of 5–10 years.
Treatment with curative intent of patients with truly localized prostate carcinoma is unlikely to demonstrate a substantial impact on decreasing mortality for 5–10 years or more after therapy because of the protracted natural history of most localized prostate carcinoma. The availability of TRUS and core needle biopsies has simplified the diagnostic process; thus, the threshold for a prostate biopsy based on an abnormal digital rectal examination or TRUS dropped substantially in the mid-1980s. Nevertheless, the contribution of such early detection with only TRUS and core biopsy to overall reduction in mortality must be modest, because an increase in the detection of localized disease did not become evident until 1989.
More recent developments since 1980 have greater potential as contributing factors to the fall in mortality. Numerous articles have been published that have attempted to attribute the recent fall in mortality with the increase in prostate carcinoma screening. The early identification of prostate carcinoma should be combined with potentially curative strategies to affect a benefit for the population, which is true for any screening program. Although screening for prostate carcinoma may play a role, it is unlikely to be the sole factor that explains the epidemiologic findings. Curative strategies need not be in the absolute sense; rather, these strategies merely have to increase survival such that patients succumb to other comorbidities. Thus, it becomes incumbent upon the scientific community to identify other potential contributing factors as soon as possible so that patients can be educated appropriately about treatment specific outcomes.
Limitations of these data must be considered. The SEER Program is an incidence registry; therefore, with SEER data, one is unable to relate individual therapy to mortality. Conclusions are drawn on population samples separately for incidence and mortality in the same geographic area. Treatment classification of patients is based on the therapy captured during the 4 months after diagnosis. If the treatment decision is delayed beyond 4 months, then it may be missed. Stage misclassification also may occur. If a patient is treated with surgery, then his stage classification may change based on intraoperative pathologic findings. This information is not available for other types of primary therapy and results in a potential stage misclassification bias among patients who undergo surgery. Another limitation of the SEER registry is the lack of PSA utilization data. Without such data on individual patients, one cannot establish the precise time course of events from PSA screening, to detection, to mortality. However, the timing of introduction of PSA into clinical use is reliable, and the estimate that PSA screening did not become widespread for several more years also is credible.40 Finally, treatment data in the SEER registry are primarily hospital based, whereas most ADT is provided as outpatient therapy in the physician's office. Although, as described above (see Materials and Methods and Discussion), additional efforts are made to obtain further data from outpatient treatment through physicians' offices, a net underestimation of the utilization of ADT is likely to have occurred.
Taken together, these data support the hypothesis that decreasing mortality rates for prostate carcinoma in Southeast Michigan is multifactorial. Data have been presented that suggest an association between mortality, early detection, utilization of surgery or radiation therapy, and utilization of androgen-deprivation therapies, especially when they are applied early in the course of the disease. We conclude that there is a need to conduct a more focused case–control or cohort study to determine whether the use of adjuvant or neoadjuvant androgen-deprivation therapies indeed explains the decrease in prostate carcinoma mortality rates.
The authors acknowledge the in-depth support of Tsui-Ying Kau, M.P.H., who provided invaluable biostatistical support for our needed access to Detroit area SEER data. They also acknowledge the SEER data itself (contract N01-CN-65054) and appreciate the SEER Program as a vital resource to investigators nationwide.
- 1U.S. Preventive Services Task Force. Guide to clinical preventive services, 2nd ed. Baltimore: Williams & Wilkins, 1996.
- 2National Cancer Institute. Prostate cancer screening state-of-the-art statement. Physician data query [computer data base]. Bethesda: National Cancer Institute, 1996.
- 3Early detection of prostate cancer. Recommendations from the Canadian Urological Association. Can J Oncol 1997; 4(Suppl:8) 2–85..
- 6American Urological Association Executive Committee Report. Early detection of prostate cancer. Baltimore: American Urological Association, 1992.
- 15International classification of diseases for oncology, 2nd ed. Geneva: World Health Organization, 1990., , .
- 16Applied linear statistical models. Homewood: Richard D. Irwin, Inc., 1974., .
- 17Transrectal needle method for biopsy of the prostate: review of 90 case proceedings; frozen aspiration biopsy of prostate. Br J Urol 1975; 47: 227–61., , , .
- 26Concepts related to prostatic cancer as revealed by a study of the dunning R3327 prostatic adenocarcinoma models. In: WilkinsonDM, editor. Advances in medical oncology research and education, 2nd ed. Clinical cancer, principle sites. New York: Pergamon Press, 1979: 313–21., , , .
- 31Clinical efficacy and safety of a new leuprorelin acetate depot formulation in patients with advanced prostatic cancer. J Intern Med Res 1990; 18(1): 57–68., .
- 33Randomized prospective study comparing radical prostatectomy alone versus radical prostatectomy preceded by androgen blockade in clinical stage B2 (T2bNxM0) prostate cancer. The Lupron Depot Neoadjuvant Prostate Cancer Study Group. J Urol 1995; 154(2 Pt 1): 424–8., , , , , .
- 38Clinical efficacy and safety of a new leuprorelin acetate depot formulation in patients with advanced prostatic cancer. J Intern Med Res 1990; 18(Suppl 1): 57–68., .