In May 2012, the United States Preventive Services Task Force recommended against PSA testing in all men based upon a flawed analysis that overestimated the harms and underestimated the benefits of testing.
In May 2012, the United States Preventive Services Task Force (USPSTF) recommended against PSA testing in all men based upon a flawed analysis that overestimated the harms and underestimated the benefits of testing (Ann Intern Med 2012; 157:137-8).
This highly controversial recommendation “outraged” the AUA, which characterized it as a great disservice to men. The American Society of Clinical Oncology recommended that for men with a life expectancy of 10 years, physicians should discuss whether PSA testing is appropriate for them. Similarly, the American Cancer Society recommended a shared decision-making process.
Harms of PSA testing
The most important potential harm from screening is overdiagnosis and overtreatment of cancers that would never cause symptoms. While there may be anxiety associated with having a prostate biopsy, the side effects are usually mild and short lasting. A randomized trial reported that only 6% of patients had high anxiety and 45% reported no anxiety (Eur J Cancer 2011; 47:545-53). Overdiagnosis is inevitable with all cancer screening programs, but the true magnitude is unknown and cannot be determined for individual patients. Moreover, not all men who meet the criteria for biopsy undergo biopsy (many do not), and not all overdiagnosed patients are treated. Furthermore, the harms of treatment vary with the treating physician, the hospital, and the patient’s comorbidities.
Benefits of PSA testing
In assessing the benefits, the recommendation of the USPSTF was strongly influenced by the Prostate, Lung, Colorectal, and Ovarian (PLCO) cancer screening trial (J Natl Cancer Inst 2012; 104:125-32), European Randomized Study of Screening for Prostate Cancer (ERSPC) (N Engl J Med 2012; 366:981-90), and the Radical Prostatectomy versus Observation Trial (PIVOT) (N Engl J Med 2012; 367:203-13) (figure 1). The USPSTF was wrong to give equal weight to ERSPC on one side and PLCO and PIVOT on the other, considering the extensive flaws of PLCO and PIVOT.
PLCO has been portrayed as a negative prostate cancer screening trial. However, because of many flaws (eg, prescreening 40% of the participants before enrollment, >50% “contamination” rate by screening controls, and 70% of men with abnormal screening tests not having prompt biopsy (although most had a markedly delayed biopsy procedure), PLCO does not qualify as a true screening trial. Prostate cancer epidemiologist Ruth Etzioni, PhD, declared that PLCO can neither be interpreted as a negative study nor as providing evidence that screening has no benefit (Evid Based Med 2012; July 7 PMID 22773765). In fact, considering only the healthy men, there was a significant 27% cancer-specific mortality benefit in PLCO.
In ERSPC, having 15% to 20% contamination and prompt biopsy performed in 86% of men with positive tests, the screening arm had 40% fewer advanced cases, and prostate cancer mortality was 21% lower-29% lower after adjustment for non-compliance and 38% lower for those with 10 to 11 years of follow-up.
The USPSTF gave little weight to the independent, longer, and better-designed Göteborg population-based screening trial (The Lancet Oncol 2010; 11:725-32) having 3% contamination, 93% complying with biopsy recommendation, and 77% having 14-year follow-up. There was a 41% decrease in advanced disease (66% in men actually screened) and a 44% decrease in prostate cancer mortality (56% in men actually screened) despite one-third being managed with active surveillance (figure 2). The number needed to treat to save one life was 12, comparing favorably with breast cancer screening programs. Assuming overdiagnosis and screening efficacy consistent with U.S. incidence, it has been projected that, after 25 years of follow-up, 186 to 220 men need to be screened, and only two to five additional men need to have screen-detected prostate cancer to prevent one prostate cancer death (J Clin Epidemiol 2011; 64:1412-7).
PSA testing as performed in these trials was not the way it is practiced today, and many of the screened men who died of prostate cancer were diagnosed with advanced disease at their first screening visit. With serial PSA testing, the mortality benefit would be even greater.
The PIVOT trial comparing radical prostatectomy with observation concluded that surgery did not significantly reduce mortality among men with PSA ≤10.0 ng/mL or those with low-risk tumors, while in men with a higher PSA there was a reduction in prostate cancer and all-cause mortality. This result contrasts with a Scandinavian trial that showed fewer metastases, a 38% lower prostate cancer mortality, and 25% lower all-cause mortality rate with surgery (N Engl J Med 2011; 364:1708-17). Although, unlike the Scandinavian trial, PIVOT was conducted in the PSA screening era, the Scandinavian trial reported that a low-risk group (Gleason grade ≤6 and PSA <10.0 ng/mL) treated with surgery had a lower prostate cancer and overall mortality.
PIVOT was statistically underpowered. The goal was to enroll 2,000 men, but the investigators could only enroll 731 (N Engl J Med 2012; 367:270-1). Patients were largely older Veterans Administration hospital patients, known to have more comorbidities (J Urol 2003; 170:905-8). An enrollment criterion was a 10-year life expectancy, but by 10 years, half the patients had died. Thus, instead of recruiting healthy candidates for either surgery or observation, PIVOT investigators recruited men with a limited life expectancy who were only candidates for observation and randomized half of them to surgery. Only 77% assigned to surgery actually underwent surgery, while 20% assigned to observation were treated with surgery or radiotherapy. Despite these limitations, in the entire study cohort nearly all outcomes favored surgery (60% fewer metastases and 37% lower prostate cancer mortality rate in the surgery arm, similar to the Scandinavian study).
Because of the low statistical power of PIVOT, these favorable results were not statistically significant. The only thing that PIVOT really shows is that surgery is not a good option for men with low-risk disease whose life expectancy is less than 10 years-something we have known for decades-and is not informative for healthy men in their 40s to 60s trying to figure out what they should do.
Decreasing locally advanced/metastatic PCa
Lowering the rate of advanced disease at diagnosis reduces suffering and death from prostate cancer. Moreover, patients with advanced disease have more treatments and related complications. The USPSTF analysis paid little attention to reducing advanced disease. Unlike mortality rates that are affected both by early detection and better treatments, the frequency of metastases at diagnosis is determined only by early detection. In all of the screening trials, there was a lower rate of advanced disease in the screening arm. Scosyrev et al reported that if the pre-PSA era metastases rates are applied to today’s U.S. population, there would be three times as many metastatic cases (17,000 more) diagnosed each year (Cancer 2012; 118:5768-76).
The USPSTF dismissed epidemiologic data showing a 44% decrease in prostate cancer mortality and a 75% decrease in the percentage of metastatic cases at diagnosis in the U.S., according to data from the Surveillance, Epidemiology, and End Results (SEER) program (accessed July 22, 2012) (figure 3). Statistical teams have reported that 45% to 70% of this mortality decrease is due directly to PSA testing, while improved treatment explained only 17% to 33% (Cancer 2012; 118:5955-63). Similar results have been reported in Tyrol, Austria (Int J Public Health 2012; 57:45-7) and other countries where widespread PSA testing has been practiced. It is likely that if the USPSTF recommendations against PSA testing are widely implemented, the incidence of metastatic disease will return to pre-PSA era levels, the 44% prostate cancer mortality gains achieved during the PSA era could be erased, and we would witness a significant increase in prostate cancer deaths.
The USPSTF recommendation also ignores high-risk populations, such as those with a strong family history or men of African descent who have a 50% higher incidence rate and a 200% to 300% higher mortality rate, as shown by SEER data (accessed July 22, 2012).
Men over age 75
The USPSTF strongly recommends against testing men older than 75 years of age. However, recent studies show that while this age group represents 26% of prostate cancer cases, it represents 48% of metastatic cases and 53% of prostate cancer deaths (Cancer 2012; 118:3062-70). Older men with aggressive disease may benefit from early detection and treatment.
Attempts at ‘quality-adjusting’ benefits
Decision modeling has been used in an attempt to quantify how benefits of screening are offset by harms. In a study evaluating 5-year follow-up data, quality of life scores for men with prostate cancer were actually above those of the normal population (Int J Cancer 2005; 116:291-6). Thus, patients may re-evaluate the value of life itself when after curative treatment they are alive and in a reasonably good functional state. And although sexual, urinary, and bowel dysfunction are relatively common after treatment and important, their impact on the health status of prostate cancer patients may be overstated.
Using ERSPC data, Heijnsdijk et al reported that annual screening of men aged 55 to 69 years would result in a 28% decrease in prostate cancer deaths, 35% decrease in palliative treatments, and a total of 73 life-years gained per 1,000 men screened (average, 8.4 years per death avoided) (N Engl J Med 2012; 367:595-605). However, they had to discount the gain in years of life by 23% because of offsetting treatment complications; ie, the net gain was 56 quality-adjusted life years.
In decision-modeling studies such as these, researchers attempt to quantify the trade-offs by expressing both benefits and harms in terms of life-years, like comparing apples and oranges (N Engl J Med 2012; 367:669-71). If the assumptions made or the “utilities” of various health states (eg, urinary, sexual, or bowel dysfunction, suffering from metastases, dying of prostate cancer, etc.) used in the models are invalid, the models fail to predict the truth. For instance, based on European data, Heijnsdijk et al assumed 43% of screen-detected cancers are overdiagnosed (N Engl J Med 2012; 367:595-605). This is higher than estimates from U.S. models (23%-28%) or surgical data (7%-20%) (J Urol 2007; 178:88-92).
Moreover, while the years gained by avoiding prostate cancer death is objective, converting how a man feels about the number of years of life he would trade to avoid a certain risk for future problems (utilities) that he has never experienced remains subjective. For instance, how accurately could a man who has never been incontinent or died of prostate cancer answer a question such as whether he would rather live for 6 years with normal urinary continence and then die of prostate cancer or survive for 14 years with a 10% chance of having to wear a pad and not die of prostate cancer? These calculations are profoundly flawed (N Engl J Med 2012; 367:987-9). In any case, impotence and incontinence should not be put on the same level as having terrible bone pain or dying of prostate cancer.
Harold Sox, MD, declared that until we have accurate information about how most men value one health state versus another, guidelines should not recommend for or against PSA screening (N Engl J Med 2012; 367:669-71). Even so, not all men would value any given health state the same. Therefore, a shared decision process that includes assessing the man’s feelings about what he could go through after screening is preferable. Thus, the USPSTF was wrong to downgrade the PSA screening recommendation.
I favor the National Comprehensive Cancer Network guidelines designed to help implement screening for men who have made an affirmative decision to be tested (J Natl Compr Canc Netw 2010; 8:240-62). They recommend early evaluation of men in their 40s to assess risk through consideration of race, family history, PSA, and digital rectal examination and then tailor subsequent screening and biopsy recommendations based on the results.