The increased use of prostate-specific antigen (PSA) has altered the diagnostic and treatment landscape of prostate cancer. The true impact of PSA on prostate cancer mortality is frequently debated and is influenced by multiple variables. One crucial factor that should be considered when evaluating the future of PSA testing in prostate cancer is the role of “sticky diagnosis bias.”1 This term refers to the type of bias seen in screening trials when the screening group has more subjects diagnosed with the disease, which can lead to some deaths from other causes falsely attributed to the disease.1 Recently, Welch and Albertsen published a review in the New England Journal of Medicine about how this type of bias can influence the future of PSA testing.2
An example of sticky diagnosis bias can be observed during the 1970s and 1980s when urologists performed transurethral resections of the prostate (TURP) to treat benign prostate enlargement in older men.2 As more TURP procedures were performed and specimens were sent to pathology, cancer was frequently incidentally found, leading to increased incidence figures. By 1986, Merrill and colleagues reported that approximately half of all prostate cancers were detected on TURP.3
Eventually, there was a 50% decrease in TURP-detected prostate cancer incidence in favor of more medical therapy, however, the overall prostate cancer incidence doubled from 1986 to 1992, which can be linked to the widespread use of PSA test.2,3 Many older men were eventually diagnosed with prostate cancer based on PSA screening, at which point their diagnosis “stuck” with them at the time of their death.2 This can continue to be an issue even in men who may not be considered “older,” especially as more than half of men who have died at an age older than 60 years from another cause have pathologic evidence of prostate cancer on autopsy.4
The increase in prostate cancer mortality during these periods most likely was because the cause of death for these men was attributed to prostate cancer, rather than being linked to sequelae from the disease. A study conducted by Thompson and colleagues found that many men with PSA-detected prostate cancer were found to survive up to 15 to 20 years following their diagnosis — typically with minimal treatment — and eventually died from another disease process.5
Another confounding factor is that patients with poorly differentiated and more aggressive prostate cancers can have normal PSA values.6 Therefore, the authors argued that most PSA-detected cancers are well-differentiated cancers and are not more aggressive forms of disease.
The authors noted that if peak prostate cancer mortality from the early 1990s is used as a goal post, there has been a 51% decrease in mortality from this point until the year 2016. In comparison, if the baseline levels from the 1950s to 1970s (prior to routine TURP utilization) are used, there is still a 37% decrease in prostate cancer mortality.