PSA Testing: Title and subTitle BreakPublic Policy or Private Penchant?

Physicians order prostate-specific antigen (PSA) tests for many reasons: to confirm the presence of suspected cancer, to monitor progression of prostate cancer or the effect of treatment, or to predict the likelihood that prostate cancer will occur in the future (ie, screening). In this issue of JAMA, Walter et al¹ document that many clinicians in the Veterans Affairs medical system order PSA tests for elderly male patients. In 2003, 56% of men older than 70 years who had no previous history of prostate cancer, elevated PSA level, or prostate cancer symptoms had a PSA test performed. Among men older than 85 years, 34% of those in good health and 36% in poor health had a PSA test performed. Most guidelines do not recommend PSA testing in elderly men, so why would physicians perform these screening tests? Why does practice not comply with policy?

Five key questions should drive the decision to perform a screening test.² Is the disease a significant, serious disease? Is the test accurate? Will the test improve the outcome of the disease? Will the test result cause the patient any harm? Is screening likely to do more harm than good? A review of these 5 key questions should help in understanding why PSA testing has been performed so frequently.

Does a PSA test screen for a significant, serious disease? After lung cancer, prostate cancer is the second leading cause of cancer death in men in the United States. While the lifetime risk of a prostate cancer diagnosis is about 16%, the lifetime risk of prostate cancer death is only 3.4%.³ Of the 29 000 men destined to die from prostate cancer this year, the vast majority were diagnosed before age 75 years. Prostate cancer, even poorly differentiated disease, takes several years to progress when first identified by PSA testing. Even under favorable assumptions, routine PSA screening is unlikely to benefit most men who have a low probability of surviving 10 years.

These statistics, however, are often meaningless to individual patients and their physicians. Most men are overly optimistic about their own longevity. Men who are healthy at age 80 years generally believe they will be alive at age 90 years. Even if the chance is remote, many men fear that prostate cancer may strike them and therefore request a PSA test. Health care policy is driven by research studies that use population estimates of longevity and mortality. Patients make decisions on the basis of their own personal fears. Being diagnosed with cancer is one of them.

Is PSA testing accurate? As men age, they often develop benign prostate hypertrophy, which confounds PSA testing. A large prostate gland often produces increased quantities of PSA, which can generate false-positive screening results. To increase both the sensitivity and specificity of PSA testing, alternative metrics such as PSA density, the ratio of free to bound PSA, and age-specific reference ranges have been proposed.^{4 - 6} Unfortunately, not all clinicians use these adjustments, in part because many are concerned about potentially missing a cancer and the potential for associated professional liability risks.⁷ In the US medical culture, it is much easier to recommend medical intervention than to suggest withholding care.

Will testing for PSA improve patient outcomes? To determine whether PSA testing can improve patient outcomes, data that document the natural history of screen-detected disease are of crucial importance. Unfortunately, studies that have assessed the natural history of prostate cancer began patient accrual before the advent of PSA testing. The best data available were reported by Johansson et al,⁸ who assessed the long-term outcome of 648 Swedish men diagnosed with prostate cancer between 1977 and 1984, and our study that documented the long-term outcome of 767 Connecticut men who were diagnosed with clinically localized prostate cancer between 1971 and 1984.⁹ Both studies showed that older men with low-grade tumors are much more likely to die from other causes than prostate cancer. Conversely, younger men, especially men with high-grade cancers, have a very high probability of dying from their disease. Patients with screen-detected disease are likely to have outcomes that are considerably better than those described in these studies because screening advances the date of diagnosis and increases the likelihood of identifying less aggressive disease.

A more significant issue involves the efficacy of interventions to alter the outcomes of men with screen-detected prostate cancers. In one clinical trial with 10 years of follow-up, prostate cancer mortality was 14% among patients randomly assigned to watchful waiting compared with 8% mortality in men having radical prostatectomy.¹⁰ Overall survival between the 2 groups favored men having radical prostatectomy, but the difference was relatively small (83 deaths in the surgery group vs 106 deaths in the watchful-waiting group). Comparable data supporting the efficacy of radiation therapy or surgery in men with prostate cancer detected by PSA screening are not available. Results from large case series¹¹ suggest the outcomes may be similar, but this remains to be determined.

Patients often overestimate both the risk posed by prostate cancer and the efficacy of treatment. With even modest potential benefit, many patients are likely to request screening and treatment. From a public health perspective, how large an improvement in overall mortality is needed to justify PSA screening?

Does PSA testing cause any harm? How can a simple blood test cause any harm? It rarely does. Most of the morbidity associated with PSA testing is related to the procedures that follow a prostate cancer diagnosis. The Prostate Cancer Outcomes Study was designed to obtain information concerning the effect of treatment on quality of life. This large, prospective, population-based study enrolled men with localized prostate cancer diagnosed in 1994 and 1995 and collected follow-up data through 2001. Of the 1291 men undergoing radical prostatectomy, 59.9% were impotent and 8.4% were incontinent within 18 months after surgery.¹² Forty-one percent of the study participants undergoing surgery reported that sexual performance was a moderate to large problem after treatment. Of the 497 patients who received external beam radiation, 43% of the previously potent men were impotent within 2 years and 5.4% had significant bowel dysfunction.¹³ Similar findings have been noted in a large randomized trial in Sweden that compared radical prostatectomy with conservative management.¹⁴ Considering these outcomes, it is important to counsel patients who request a PSA test regarding the risks of intervention and the uncertainty of benefit.

Not all elevated PSA levels need to prompt a transrectal ultrasound and biopsy. Benign prostate hypertrophy is associated with an increase in PSA values of approximately 0.75 ng/y.¹⁵ A more rapidly increasing PSA level raises concerns about clinically significant disease.¹⁶ If a physician and patient are willing to risk a delay in diagnosis, a physician can simply monitor the rate of change in PSA over time. Periodic monitoring of PSA levels may be best suited to older patients in whom the likelihood of significant morbidity and mortality from prostate cancer are low.

Will PSA testing cause more harm than good? Widespread, repeated testing for PSA has increased the incidence of prostate biopsy and has raised the probability of detecting indolent rather than life-threatening disease. Draisma et al¹⁷ estimate that at age 55 years, the lead time associated with PSA testing is approximately 12.3 years and the likelihood of detecting clinically insignificant disease is 27%. By age 75 years, the lead time associated with PSA testing decreases to 6.0 years, while the likelihood of detecting clinically insignificant disease rises to 56%.

Changes in the application of Gleason scoring have also contributed to the prostate cancer screening conundrum.¹⁸ Most researchers and clinicians agree that Gleason tumors with scores of 2 to 5 pose little threat. Unfortunately, pathologists are reluctant to report these scores on biopsy in contemporary practice.¹⁹ Accordingly, the risk posed by small, contemporary Gleason score 6 tumors is exaggerated.

Two large randomized trials currently in progress were designed to test whether PSA screening reduces prostate cancer mortality.^{20 - 21} Both studies have been ongoing for approximately 12 years and have yet to demonstrate a clinically significant survival advantage for the men assigned to the screening group. Clinicians and patients should understand that the efficacy of PSA testing for all men remains in doubt.²² Natural history studies suggest that younger men between the ages of 50 and 70 years who have a life expectancy greater than 10 years are those most likely to benefit from PSA screening. When the test is applied to older men, the benefits are not likely to exceed the potential harms.

Public health policy does not support PSA screening among elderly men with limited life expectancy because the potential gains are limited and the potential losses are great. Physicians order the tests anyway because of patients' exaggerated fear of prostate cancer mortality and their overestimation of treatment efficacy. Physicians also order PSA tests because the reward for treatment can be significant and the penalty for failing to diagnose can be severe.⁷ This dilemma is quite common in the current health care system and certainly requires urgent attention in the near future.