Center News

Improving early detection of prostate cancer

Ruth Etzioni's study is first to formally evaluate the combined PSA test's ability to distinguish between true cancers and benign conditions

Nov. 4, 2004
Dr. Ruth Etzioni shows Seth Falcon (behind, left) and Dr. Charles Kooperberg

In a mathematical visualization, Dr. Ruth Etzioni shows Seth Falcon (behind, left) and Dr. Charles Kooperberg (right) how a PSA test variation could improve the test's accuracy for men with borderline-normal total PSA levels.

Photo by Todd McNaught

Measurement of a man's prostate-specific antigen, or PSA, level is currently the most commonly used test for early detection of prostate cancer. Yet researchers question the test's adequacy as a cancer-screening tool since studies show that it may detect tumors that would never require treatment while, at the same time, fails to find some aggressive cancers.

A new study led by Dr. Ruth Etzioni now provides conclusive evidence that a variation of the test that relies on two types of PSA measurements could improve the test's accuracy for men with borderline-normal total PSA levels. Although the benefit of this approach was modest, the researchers concluded that its use could translate into a significant reduction in medical costs and complications for this group of men by eliminating unnecessary biopsies to confirm the PSA test results.

The study appears in the October issue of Cancer Epidemiology, Biomarkers and Prevention. Co-authors include Seth Falcon and Dr. Charles Kooperberg of the Public Health Sciences Division and colleagues at the University of Washington and at the Harvard School of Public Health.

The researchers conducted the study in order to resolve a controversy over whether analyzing a man's free PSA level — the amount of PSA protein that is not bound to other proteins in the blood serum — in conjunction with his total PSA level provides any benefit for accurate early cancer detection. Although this combination test has become popular in recent years, Etzioni said that its effectiveness compared to other types of PSA testing, which gauge levels of either total PSA alone or PSA that is in a form that is bound to other serum proteins (complexed PSA), has never been rigorously tested.

"Since the beginning of the PSA era, there's been a concern about what should be done for men who have moderately-elevated total PSA levels to determine whether they in fact have cancer," said Etzioni, a PHS investigator who has led multiple studies on the test's utility as a screening tool. "This is particularly an issue for older men and men who have benign prostate conditions — two groups in which total PSA levels may be elevated."

Researchers consider levels of total PSA below a threshold of 4 ng/ml to be in the normal range. But there is debate about how to treat men whose total PSA levels are moderately elevated but not clearly indicative of cancer. To address this, some men whose total PSA levels are moderately elevated — between 4 and 10 ng/ml — also undergo testing of their free PSA levels, which tend to be low in those with cancer.

Earlier studies have suggested that the ratio of free PSA to total PSA results in fewer false positive outcomes than total PSA testing, with only small losses in the number of cancers that can be detected. Etzioni's study is the first to formally evaluate the combined test's ability to discriminate between true cancers and benign conditions. The study used a statistical method called logic regression, developed by Kooperberg and his former student, Ingo Ruczinski, and Dr. Michael LeBlanc of PHS, to sort through and compare a multitude of potentially useful combination tests.

Serum samples studied

Etzioni and colleagues analyzed free and total PSA levels in stored serum samples taken prior to diagnosis in 429 men with prostate cancer and 1,640 healthy men and compared the levels to tests based on total and complexed PSA. The samples were from a subset of participants in the Physicians' Health Study, a large study of more than 22,000 U.S. male physicians ages 40 to 84 years of age, some of whom developed prostate cancer.

The researchers found the free/total combination tests gave the fewest false positive outcomes — an attribute known as specificity — particularly in cases where total PSA levels were below or slightly above 4ng/ml. When compared with the standard PSA-based test, which declares a man to be positive if his total PSA exceeds 4 ng/ml, one combination test reduced false positive rates by 30 percent while simultaneously increasing true-positive rates by more than 10 percent. The authors concluded that the ability to accurately diagnose cancer in men whose only symptom is moderately elevated total PSA level may be enhanced by use of the combination test.

Etzioni described the improvement as a small clinical benefit.

"Still, even a small benefit could translate into an absolute number that would provide a significant savings in cost and unnecessary side effects due to biopsy or other treatment," she said, adding that "the use of combination tests could be particularly important if the total PSA threshold is lowered, as has been recommended by several recent publications."

Fred Hutchinson Cancer Research Center is a world leader in research to prevent, detect and treat cancer and other life-threatening diseases.