Methods for the detection of early prostate cancer have certainly improved, but is this a good thing? It has been argued that aggressive cancer detection is expensive and meddlesome and leads to the overtreatment of tumors that may be indolent or at low risk for significant progression. In the background paper [1] that accompanies the American College of Physicians guidelines on screening for prostate cancer in this issue [2], Coley and colleagues point out that digital rectal examination, PSA measurement, transrectal ultrasonography, and ultrasonography-guided biopsy have not been proven to reduce mortality related to prostate cancer. We know that as a way to detect suspicious induration in the prostate, digital rectal examination tends to be inaccurate and its results are not very reproducible from one examiner to another. Levels of PSA can increase modestly with benign prostatic hyperplasia and probably with advancing age. There is much debate over which criteria should lead to prostate biopsy and when aggressive treatment, such as radical prostatectomy or radiation therapy, is appropriate. The grade and estimated volume of the tumor and the age and general health of the patient are all important factors in deciding which treatment method to choose. However, the outcome data that have been published are inadequate to allow the formation of a guideline for decision making. As a result, we are in the midst of a vigorous and uncontrolled nationwide effort to detect prostate cancer and eradicate localized prostate cancer.

The American Urological Association Guidelines Panel for the Management of Clinically Localized Prostate Cancer reviewed all published data on the outcomes of radical prostatectomy, external-beam radiation therapy, brachytherapy, and surveillance (expectant management or "watchful waiting") [3, 4]. This review makes it clear that the outcomes of these treatment methods could not be compared because of extreme variation in the patient populations that had been assigned to these therapeutic approaches. For example, the mean age of patients reported in the urologic studies who had had radical prostatectomy was 62.7 years. In contrast, the mean age of patients who had only received surveillance was 70 years-a difference of 7.3 years. In the reports of patients who had had radical prostatectomy, 23% of the tumors were well differentiated, 57% were moderately differentiated, and 20% were poorly differentiated. In contrast, 62% of the tumors from patients who had received surveillance were well differentiated, 35% were moderately differentiated, and only 3% were poorly differentiated. Thus, patients who had received surveillance were generally older and had a lower grade of tumor. Although many of these patients were appropriately assigned to surveillance, we cannot accurately compare the treatment outcomes of patients having surgery and those of patients receiving surveillance. Because the characteristics of patients receiving radiation therapy and their tumors vary, it is also impossible to compare the outcomes of such patients with those of patients who were treated with radical prostatectomy and those who were managed with surveillance.

Another disturbing problem with the literature on treatment outcomes in prostate cancer is that nearly all of the outcome data are projected from Kaplan-Meier curves. Of patients whose 5-year outcomes have been reported in the literature, only one in seven (13%) has actually been followed for 5 years; of patients who have had 10- and 15-year outcomes reported, an even smaller fraction has actually been followed to these milestones. Thus, nearly all outcomes are projected from results in patients who have only been followed for a few years. The validity of such projected outcomes is uncertain.

The American Urological Association Prostate Cancer Panel reviewed complications of the treatment of localized prostate cancer. The literature that has been published in this area is inadequate and frustrating. Many reports simply do not mention complications. The complications that have been reported vary greatly from one series to another, and specific complications are defined differently by different authors. For example, proctitis after external-beam radiation therapy is reported in 16% to 55% of cases. Although we do know that radiation proctitis of a mild-to-moderate degree is common and often resolves quite well, we also know that it can be severe enough to warrant a colostomy. However, the degree to which proctitis is a problem after treatment is difficult to determine from the published data.

Incontinence is common after radical prostatectomy. Reports of its prevalence vary; as many as 50% of patients have stress incontinence and 0% to 15% have severe incontinence, although this complication has been reported less frequently in recent years. Moreover, the interpretation of the effect of this outcome is clearly an emotional issue. I (and probably most urologists) regard mild and occasional stress incontinence after surgery as a mere trifle-we do not find it to be a significant issue in most patients who are otherwise doing well. Other observers (who are generally not urologists) consider any degree of postoperative incontinence to be an extremely dire outcome that greatly compromises the patient's quality of life.

Impotence is also often viewed as a demoralizing result of radical prostatectomy. Impotence is common after this surgical procedure and is almost inevitable unless a nerve-sparing surgical technique is used. When biopsy results show an apical or bilateral tumor, it is probably best to do a wide removal of the prostate without nerve sparing in an attempt at an improved outcome [5]. The patient should certainly understand and agree to the surgeon's intent. The informed patient is almost always willing to accept impotence if he understands the need for wide removal of the prostate. In this context, impotence is not really a complication but rather the anticipated result of an attempt to eradicate a tumor. Fortunately, fairly good methods are now available to restore erectile ability after radical prostatectomy.

Coley and colleagues have thus faced an exceptionally difficult task in weighing the benefits and risks of screening for prostate cancer, but when all is said and done, I believe that the model they have proposed to estimate the risk–benefit ratio of screening for prostate cancer and treatment of localized disease is not reliable or particularly useful. They selectively emphasize certain reports from the literature and exclude many others. The simple fact remains that studies of treatment outcomes are confusing because groups of patients that receive various treatments are different and follow-up methods are poor. We therefore cannot adequately measure outcomes and, as a consequence, cannot effectively measure risks because the data on complications of treatment vary so widely. It is also easy to exaggerate the significance of some complications and perhaps to downplay the significance of others.

As clinicians, we continue to struggle to decide which prostate tumors are likely to be indolent or relatively insignificant to the patient and which warrant aggressive treatment. If we embark on aggressive treatment in the process, are we really helping the patient? No good randomized studies provide the answer. Radical prostatectomy has appeal for a relatively healthy man in his 50s or 60s who seems to have a tumor of significant volume (at least moderate-grade adenocarcinoma) that has probably not spread beyond the prostate. But we will not know whether the patient has been properly selected for surgery until the lymph nodes have been removed and analyzed and we have carefully examined the specimen obtained through radical prostatectomy. However, even if the tumor is confined to the prostate, cancer may recur in the future. On the other hand, a tumor that extends through the prostate capsule to the margin or into the seminal vesicle makes recurrence more likely but certainly not inevitable. Radical prostatectomy provides the real possibility of eradicating all tumors and gives us a specimen to help us determine outcome and the accuracy of our preoperative assessment. In contrast, in patients who receive radiation or surveillance, we never know the true extent of the tumor, its volume, or the spectrum of tumor grades; therefore, such patients are difficult to compare with patients who have been treated surgically.

Well-controlled studies of treatment outcomes will be difficult to do and will take a long time. Accurate data collection and better and longer follow-up periods will help, but prostate cancer is a complicated disease. In the absence of good long-term outcome data, patients should certainly be fully informed about treatment options and should have a strong role in helping to identify the treatment method that they prefer. We are a long way from having a formula with which to select the appropriate treatment pathway for our patients.

In the meantime, I subscribe to the American Cancer Society recommendation that a man older than 50 years of age should have an annual digital rectal examination directed at assessment of the prostate. I also support measuring the PSA level at baseline. If this level is low and the prostate is unchanged over time, the PSA measurement can perhaps be repeated in 2 to 3 years.

Prostate cancer is important in our society, and metastatic prostate cancer is a devastating disease; it takes a substantial toll and accounts for substantial cost in our medical system. We do not know whether our efforts will ultimately reduce mortality related to prostate cancer, but we can be hopeful, particularly because of the decrease in the incidence of new cases of metastatic prostate cancer as shown in the Utah Cancer registry and other local registries that are part of the SEER (Surveillance, Epidemiology, and End Results Program) national registry [6]. Only an intense and unrelenting effort to improve our detection and treatment methods will bring us closer, over time, to the goal of controlling this most discouraging and confusing of scourges.