Breakdown of patient assignments into high-risk groups.
Ryan P. Smith, S. Bruce Malkowicz, Richard Whittington, Keith VanArsdalen, Zelig Tochner, Alan J. Wein. Identification of Clinically Significant Prostate Cancer by Prostate-Specific Antigen Screening. Arch Intern Med. 2004;164(11):1227–1230. doi:10.1001/archinte.164.11.1227
The importance of screening for early-stage prostate cancer has been debated in the literature. However, there are well-established prognostic factors (Gleason score [GS], pretreatment prostate-specific antigen [PSA], and percent positive biopsy findings [%+Bx]) that predict biologically aggressive disease. These factors, together with a patient's age and general state of health, will permit physicians to project the effect of a prostate tumor over the patient's expected lifetime. This study was performed to determine the proportion of clinically significant prostate cancers diagnosed in a screened population.
From 1991 through 2002, 977 patients with nonpalpable (T1c) prostate cancer were seen for evaluation and comprise the study group. Patients were classified according to pretreatment PSA level, GS, %+Bx, and age.
Based on tumor characteristics alone, 130 patients were noted to be at high risk (GS = 8-10 or PSA level >20 ng/mL; or GS = 7 or PSA level >10-≤20 ng/mL and >50%+Bx), with a historical 4-year PSA control of 10% to 30% after definitive therapy. An additional 45 patients were at intermediate risk (GS
= 7, PSA level >10-≤20 ng/mL, and 34%-50%+Bx), with a historical 4-year PSA control of 50% to 60% after definitive therapy. Additional patients were identified who had a cumulative anticipated prostate cancer mortality greater than 30% to 50% based on age and GS (GS = 7, age ≤70 years [n = 89]; GS
= 6, age ≤65 years [n = 337]). The total at risk for clinically significant tumors was 601 (61.5%) of 977 patients.
A significant proportion of patients with nonpalpable disease diagnosed as having prostate cancer by PSA screening have clinically significant cancers. This supports the continued use of PSA screening.
Adenocarcinoma of the prostate is the most common malignancy diagnosed in men. It is estimated that 189 000 new cases will be diagnosed in 2002, with 30 200 deaths.1 This disparity between incidence and mortality reflects the wide range of biological potential of these tumors. There has been a long debate over the treatment of prostate cancer. This debate is amplified when speaking about early-stage prostate cancer (ie, T1c prostate cancer) and whether men should even be tested for prostate cancer by prostate-specific antigen (PSA) screening.
Prostate-specific antigen screening has resulted in earlier detection of many prostatic tumors.2After PSA screening became available for widespread use, the incidence of prostate cancer rose from approximately 37 000 cases in 1985-1986 to over 200 000 in 1995.2- 4 Because PSA screening is now widely used, an increasing number of prostate cancers are being diagnosed and treated when there is no evidence of disease on clinical (digital rectal) examination. The question becomes how many of these tumors pose no threat to the patients and might be followed without immediate treatment and which tumors pose significant risk to the patient's life or well-being.
There have been several studies that argue the lack of benefit in treating early-stage prostate cancer5- 8 and several studies that favor the treatment of early-stage prostate cancer.9,10 Because of these studies, there has been a debate as to whether PSA screening affects mortality and quality of life in a positive or negative manner. It may increase the potential to cure the tumor, but the predictive value of a positive test result is not as high as desired, since testing may detect cancers that will not cause clinically significant symptoms within the patients' lifetimes. This study attempts to identify patients with clinically significant cancers, as defined by disease with a 50% probability of cause-specific mortality based on historical data, in a prostate cancer population diagnosed by PSA screening. This will serve to determine the risk to these patients of the nonpalpable tumors that are diagnosed based on screening.
The study group consists of 977 consecutive patients with nonpalpable (T1c) prostate cancer seen for evaluation in the Division of Urology and the Department of Radiation Oncology of the University of Pennsylvania, Philadelphia, between 1991 and 2002. All were diagnosed by ultrasound-guided needle biopsy of the prostate after presenting with an elevated screening PSA or PSA velocity. All had a negative workup for metastatic disease, including magnetic resonance imaging or computed tomography of the abdomen and pelvis and bone scan. None had a palpable prostatic lesion during 2 digital rectal examinations (DREs) performed by both a urologist and a radiation oncologist.
The following groups of patients at high risk for clinically significant disease were identified:
Patients deemed to have clinically significant disease were those with a 50% probability of cause-specific mortality based on historical data. These patients were noted to be at high risk based on the following previously published data:
D'Amico high-risk group (PSA level >20 ng/mL or biopsy Gleason score [GS] ≥8; or PSA level >10-≤20 ng/mL or GS = 7 and percent positive prostate biopsy findings for cancer [%+Bx] >50%).11,12 Patients with high pretreatment PSA levels and high biopsy GSs have been noted to be at high risk for biochemical failure after radical prostatectomy in many studies.13- 16 More recently, %+Bx has been shown to be an independent predictor of posttreatment biochemical failure. Specifically, those previously classified as "intermediate group" (GS = 7 or PSA level >10-≤20 ng/mL) can be classified as high risk if %+Bx is greater than 50%.11,12 Patients in these groups have an approximate 3-year PSA failure-free survival of 33% after definitive treatment.17 Prostate-specific antigen failure has been shown to be directly correlated with cause-specific survival.18
D'Amico intermediate risk group (PSA level >10-≤20 ng/mL or GS = 7 and 34% to 50%+Bx [ie, more than 2 of 6]). Patients in this group have an approximate 3-year PSA failure-free survival of 50%.17
Patients 70 years or younger with a GS of 7 and less than 34%+Bx. American men, age 70 years, have a projected life expectancy of 12 to 16 years.19,20 Patients with a prostate cancer GS of 7 have an annual risk of death from cancer of approximately 7%.21 Therefore, over their lifetimes, younger patients (≤70 years) have at least a 66% chance of dying of prostate cancer. This is likely an underestimation of clinically significant prostate cancers, since it does not account for symptoms caused by prostate cancer.
Patients 65 years or younger with a GS of 6 with less than 34%+Bx. This group encompasses the remaining patients 65 years or younger with a GS of 6. American men, age 65 years, have a projected life expectancy of 12 to 20 years.19,20 Patients with a prostate cancer GS of 6 have an annual risk of death from cancer of approximately 4.5%.21 Therefore, over their lifetimes, younger patients (age ≤65 years) have at least a 50% chance of dying of prostate cancer. As above, this is likely an underestimation of the actual number of clinically significant cancers.
The clinical characteristics for the 977 study patients are listed in Table 1. In keeping with historical data, most patients had moderately increased PSA levels (4-10 ng/mL) and moderately differentiated carcinomas (GS = 5-6). Biopsy results were scored as the total number of cores obtained or involved with tumor, but if only the number of specimens submitted was reported, the number of specimens was used to determine the number of positive biopsy findings. The median number of biopsy specimens taken per patient was 6 (range, 1-22 [mean 7.48]), with a median of 2 positive biopsy findings (range, 1-11 [mean 2.0]). The median patient age was 63 years (range, 40-86 years).
A breakdown of clinically significant groups is listed below and in Figure 1.
Of the 977 patients studied, 130 were placed into the D'Amico high-risk group. This assignment was based on high GS (45/130), PSA level greater than 20 ng/mL (44/130) or a GS of 7 and greater than 50%+Bx (35/130), or PSA level between 10.1 ng/mL and 20 ng/mL and greater than 50%+Bx (24/130).
Forty-five patients were classified into the D'Amico intermediate risk group. Patients in this group had %+Bx between 34% and 50% and either a GS of 7 (29/45) or PSA level between 10.1 and 20 ng/mL (25/45).
Most of the patients with clinically significant prostate cancer were relatively young with relatively high GSs. There were 89 patients who were 70 years or younger with a cancer with a GS of 7. An additional 337 patients were 65 years or younger with a cancer with a GS of 6.
Overall, of the 977 patients with stage T1c prostate cancer who were diagnosed by PSA screening alone, 601 (61.5%) had clinically significant prostate carcinomas when judging strictly by mortality alone. The combined number of patients in these groups provides a good approximation of the proportion of patients with T1c prostate cancer who are diagnosed as having high-risk disease.
Prostate-specific antigen screening has become more widely used over the past 15 years. Therefore, more patients are being diagnosed as having prostate cancer by PSA screening.2 Most of these patients present at an earlier time than they would have without PSA screening. Those without a palpable nodule, diagnosed by PSA screening alone, have been staged with T1c tumors. Given the earlier diagnoses of these patients, there has been great debate over the proper management in these cases.
There have been several reports that argue against the universal treatment of early-stage prostate cancer.8,22- 25 Many of these studies are small with short follow-up. However, a large study of 223 patients with 15 years of follow-up has been reported by Johansson and colleagures.5- 7 In these studies, prostate cancer mortality was very low compared with mortality from other causes. In addition, they cited a lower progression rate in patients with early-stage prostate cancer. These data should be interpreted with caution. These 223 patients who received no initial treatment were selected from an initial group of 642 patients. The authors' selection criteria were such that their patient population comprised mainly older men with low-grade tumors, a significant proportion of which were T1a tumors, which have been shown to be unworthy of treatment.26,27 However, even in this highly selected population, there were groups of patients at high risk of death from prostate cancer and/or symptoms that substantially affected the quality of life.
Chodak et al,8 who reported data from a pooled analysis of 6 randomized studies (of 10 originally targeted for study), also showed a low mortality rate in 828 men with clinically localized prostate cancer at diagnosis. However, they also included patients with T1a disease, and over half of their patient population comprised men who had grade 1 tumors. Additionally, metastatic disease was high and disease-specific survival relatively low in those patients with higher grade tumors.
In contrast to these studies, there have also been reports on prostate cancer treated conservatively that have a much higher mortality and morbidity. Aus et al9 published a study that focused on 301 patients with nonmetastatic prostate cancer who underwent either deferred therapy or immediate hormonal treatment only. This population of patients had fewer T1a tumors, with a more representative distribution of tumor grade, since most patients in this study did not have grade 1 tumors. In this patient group, 50% died as a direct or indirect consequence of prostate cancer, with over 70% dying within 10 years of diagnosis. In addition, there was a significant incidence of complications requiring palliation that arose from their prostate cancer.10 Risk of death from prostate cancer was higher in younger patients, probably because their longer life expectancy allowed more time for the prostate cancer to become more clinically significant. It should be noted that there is also criticism of these studies. Aus et al9 used as their patient population men who died of prostate cancer in a certain calendar period. This method is statistically weakened because it potentially underestimates the survival.28,29
Others have attempted to divide early-stage prostate cancer patients into subsets to determine which cancers place the patients at high risk for prostate cancer death. Albertsen et al30,31 attempted to stratify patients into risk groups based on age and GS, finding that patients with higher GSs and younger ages had higher prostate cancer–specific mortalities. Others have also demonstrated that younger patients have a greater risk of dying from prostate cancer after diagnosis compared with older patients. Gronberg et al32 found that 45- to 54-year-old patients diagnosed as having prostate cancer lost 16 years of life expectancy to prostate cancer compared with 65- to 74-year-old patients, who lost a mean 5 years of life expectancy from prostate cancer. Aus et al33 reported that 75% of men younger than 65 years with nonmetastatic prostate cancer at time of diagnosis died of prostate cancer.
The issue of whether to treat early-stage prostate cancer is clouded by these conflicting results. As more patients are diagnosed by PSA screening, without other signs or symptoms of disease (T1c), this issue becomes more important. The present study attempted to determine what percentage of T1c prostate cancer patients have clinically significant disease, as defined as having a cause-specific mortality of greater than 50% or a greater than 50% risk of tumor recurrence after radical prostatectomy. This identifies a group of men who need more aggressive therapy including radiation with or without hormone therapy.
There are some prognostic factors, such as PSA and GS, which are undeniably significant. Patients with a PSA level greater than 20 ng/mL are at high risk of mortality from prostate cancer. Similarly, patients with higher prostate cancer GSs are at high risk as well. It is on these data, along with %+Bx, that D'Amico et al11,12 based the stratification of risk groups. As stated previously, patients who fall into D'Amico's high-risk group have only an approximate 3-year PSA failure-free survival of 33%. Patients in D'Amico's intermediate risk group have an approximate PSA failure-free survival of 50%. However, there are patients classified as low risk by D'Amico's criteria who are also at risk for clinically significant disease because of their age and anticipated survival. These younger patients' long-anticipated life span offsets the potential benefit of relatively biologically less aggressive disease, as judged by GS and PSA screening.
This study reports a high incidence of clinically significant disease in a population of patients with T1c prostate cancers. It is likely that this study actually underestimates the incidence of clinically significant disease because of the conservative definition of significant disease. Younger patients with low PSA level or prostate cancer GS of 2 to 5 are also likely to develop symptoms related to progression of their prostate cancer if they have a long enough survival. They also may have a high mortality from prostate cancer, even though these patients are not included in the definition of clinically significant tumors. Although cause-specific mortality may be lower in older patients since they are more likely to die from intercurrent disease, symptoms from prostate cancer are very likely to occur from either local progression or metastatic disease, which can substantially affect the quality of life. It is impossible to take into account those patients who will want to treat their prostate cancer simply because they do not want to take a passive course in the management of their cancer. The prevalence of clinically significant disease in the T1c tumor population supports the continued use of PSA screening in asymptomatic patients who are aware of the issues and desire screening.
Corresponding author and reprints: Ryan P. Smith, MD, 3400 Spruce St, 2 Donner, Philadelphia, PA 19104 (e-mail: firstname.lastname@example.org).
Accepted for publication August 1, 2003.