Putting free PSA into play—why and how

Seeking answers from complexed PSA

April 2000
Cover Story

William Check, PhD

By consensus of clinicians and laboratory professionals, prostate-specific antigen is the best marker for screening for prostate cancer and is now accepted as an integral part of diagnosing and managing this malignancy.

"Total PSA is the best serum screening marker for any cancer," states Kevin Slawin, MD, assistant professor of urology and director, The Baylor Prostate Center, Baylor College of Medicine.

"PSA has revolutionized the diagnosis of prostate cancer," adds Michael Brawer, MD, director, Northwest Prostate Institute, Northwest Hospital, Seattle. "Using PSA screening, we find lots of cancers and at an earlier stage. It has changed the demography of prostate cancer today."

But as good as it is, total PSA measurement lacks specificity: It does not adequately differentiate between prostate cancer and benign prostatic hypertrophy in the so-called gray zone—total PSA values between 4 and 10 ng/mL. Values in this range are considered abnormal, requiring biopsy, yet 75 percent of men with such values and a negative digital rectal examination will not have cancer on initial biopsy.

It became possible about five years ago to measure the different forms in which PSA is found in the blood, either free (fPSA) or bound to serum proteinase inhibitors (complexed PSA; cPSA). It was observed that these molecular forms occur in different proportions in benign prostatic hypertrophy and prostate cancer. Patients with prostate cancer show a lower proportion of fPSA, while those with BPH have a higher percentage. Assays were developed to measure fPSA, and researchers set up prospective clinical trials. Even before results were in, many predicted that determination of percent fPSA (%fPSA) would add value to PSA screening.

Assays to measure cPSA directly were more difficult to devise but have now entered clinical testing. (See "Seeking answers from complexed PSA.")

Summing up results of the clinical studies with %fPSA, Milenko Tanasijevic, MD, associate director of Clinical Laboratories, Brigham and Women’s Hospital, Boston, says, "We believe that there is sufficient evidence showing the clinical usefulness of percent free PSA in clinical practice, especially for patients in whom total PSA is between 4 and 10 and who have a negative DRE."

The positive results of the clinical trials raise two questions: How can %fPSA best be used in practice, and how widely is it being adopted? Jerome Richie, MD, Elliott Cutler professor of surgery, Brigham and Women’s Hospital, and chairman of the Harvard Program in Urology, says, "Those are both controversial questions. Anytime a new test is introduced, studies are performed and physicians from academic centers publish and present data at national meetings. Over time, findings from these studies are gradually integrated into practice. I think that is what is happening now with free PSA." Several ways to use %fPSA values have been proposed, and practitioners are using different approaches.

Whether clinical outcomes improve with earlier detection of prostate cancer is a crucial question. One of the pioneers in PSA testing, William Catalona, MD, believes they do. "I think there is increasing evidence for the efficacy of radical prostatectomy and increasing evidence that PSA testing is reducing prostate cancer mortality rates," says Dr. Catalona, professor of urology, Washington University School of Medicine, St. Louis.

Clinical utility of the %fPSA assay was established by a prospective multicenter study conducted on 773 men with total PSA of 4 to 10 ng/mL and a negative DRE, with histopathological diagnosis available for all subjects (JAMA. 1998;279:1542-1547). As %fPSA rose, probability of cancer decreased. If a cutoff of 25 percent were used-that is, not performing biopsies in men with fPSA greater than 25 percent—80 (20 percent) of 394 unnecessary biopsies would have been avoided.

How does one assess the clinical value of this 20 percent increase in specificity from using %fPSA? Is it cost-effective to perform an additional assay on 773 men to avoid 80 biopsies? "It is difficult to do formal studies in this area," says Dr. Tanasijevic. He points to an analysis by Littrup showing that small changes in the specificity of a PSA assay could produce large reductions in net cost per patient screened (Cancer. 1995;75: 1987-1993). But these savings were derived from a model, not from actual study data or clinical practice.

Dr. Catalona cites ballpark costs for biopsy, including ultrasound at $1,000 to $2,000 and free PSA assay at $30 to $60. Using median figures ($1,500 for biopsy and $45 for free PSA), 800 free PSA tests would cost $36,000 and 80 biopsies averted would save $120,000. "I think that is cost-effective," he states.

But these are maximum savings. "In practice," says Daniel Chan, PhD, "there is a very tough question: Do you really want to tell a patient they do not need a biopsy when PSA is clearly elevated? And do you have any liability issues?" asks Dr. Chan, professor of pathology, urology, oncology, and radiology, Johns Hopkins University, and director of the Clinical Chemistry Division, Johns Hopkins Hospital.

These issues would not apply if free PSA were expanded to persons with a total PSA below 4 ng/mL, which some physicians have suggested, Dr. Chan notes. "Twenty to 30 percent of patients with total PSA under 4 could have cancer," he says. So a possible next step might be to lower the threshold for invoking %fPSA to a total PSA of 2.5 or 3 (Urology. 1999;54:220-224). "Patients with these PSA values will not be expecting biopsy, so they will not complain if they don’t get one," he says. "And they will be grateful if you pick up prostate cancer." [However, one study found that free PSA "is only useful in eliminating the need for prostate biopsies in men with total PSA of at least 4.0 ng/mL" (Urology. 1998;52:450-454).]

Further analysis of data from the multicenter study found that higher %fPSA was associated with more favorable histopathological findings—organ-confined tumors, Gleason score of less than 7, and 10 percent or less involvement of the prostate. Radical prostatectomy was performed on 268 patients. Among the 96 for whom complete data were available, 30 of 40 (75 percent) with fPSA greater than 15 percent had all three favorable findings on the pro-sta-tectomy specimen, compared with 19 of 56 (34 percent) who had fPSA of 15 percent or less. Percent free PSA "may assist both patient and physician in making more informed treatment decisions," investigators concluded (J Urology. 1999;162:1346-1351).

In the multicenter trial, %fPSA was measured using the Hybritech assay. Recently, Wayne Markus, MD, a pathologist at Physicians Laboratory, P.C., Omaha, performed beta-testing of the Hybritech assay on an automated Beckman Coulter Access instrument. "Subjectively, free PSA with the manual Hybritech RIA method looks very comparable to the nonradioactive assay done on the instrument," he says. On March 8, this combination became the first automated free PSA assay approved as an aid in detecting prostate cancer.

Other assays for free PSA are being tested or reviewed, though none are yet approved. Says Dr. Chan, "I have been involved with Abbott’s free PSA assay in clinical trials, and I know that it works." If approved, Abbott’s free PSA assay would most likely be performed on the AxSym instrument.

Different assays give different results and have different cutoffs for clinical response. New material sent by the CAP Surveys program minimizes differences between methods, Dr. Chan says, but variation still exists.

Most important, the same system needs to be used to determine total and free PSA. "This is a real problem, and I see it in my hospital," Dr. Catalona says. "A patient, one of our distinguished professors, goes to an internist for an evaluation, and the internist finds total PSA is 3.8, slightly higher than before. So I get a call from the internist: ’Does the patient need a biopsy?’ I say, ’Let’s get a free PSA on him.’ So they send it to a commercial laboratory, which reports a total PSA of 4.5 and a free PSA of 0.4." Now there are two slightly different total PSA values: Which one should be used to calculate %fPSA? "That is the scenario that I find frequently," Dr. Catalona says. "Total PSA at our hospital laboratory and from a reference laboratory are different. From my standpoint, the ideal situation would be to use one assay that has been tested in clinical trials and approved by FDA to be safe and effective for that procedure."

Physicians offer different estimates of the degree of adoption of %fPSA. Says Dr. Slawin, "A lot of physicians use it, but many don’t. I couldn’t practice without it. But some community urologists say to me, ’You use it? You believe that it works?’"

In Dr. Tanasijevic’s institution, "It has been adopted as much as we thought it would," he says. He reports that a recent increase in volume at his hospital has warranted setting up the test in-house.

But not all estimates are so clear-cut. Says Dr. Chan, "My expectation for free PSA usage was not as high as most other people’s." Some physicians predicted that %fPSA would be done on 50 percent of men having a total PSA assay. But right now that figure is between 10 and 20 percent. "I believe that the number of requests will go up as more clinicians have experience working with free PSA and understand how to use it," Dr. Chan says.

Is utilization lower in the community? "That may not be the case," Dr. Chan says. "Patients with prostate cancer read up and know things. I was told by some community urologists that patients are asking for it and are even willing to pay for it themselves."

In Dr. Markus’ experience, "Initially there was very little interest [in %fPSA]. There has not been any significant interest until the last several months. It is still not a high-volume test." One reason may be that a "small but significant" percentage of total PSA values are within the 4 to 10 range, where %fPSA is most valuable, he says. "The great majority of men [more than 85 percent] have a normal total PSA," Dr. Markus points out.

Even more interesting than how widely %fPSA is being used is how it is being used. In principle, the value of %fPSA is to reduce the number of false-positive biopsies in men with total PSA between 4 and 10 ng/mL. Above 10, the likelihood of prostate cancer, even with a negative DRE, is greater than 50 percent, high enough to mandate biopsy. With a total PSA of 4 to 10, about 25 percent of men would have prostate cancer. Within this group, %fPSA can segregate likelihoods of cancer from less than 10 percent to more than 50 percent. "So that is where free PSA has its greatest applicability," says Dr. Richie, who took part in the multicenter trial. "It’s all probabilities, but that is what we deal with in medicine all the time."

When total PSA is 4 to 10, %fPSA can be used as an initial screening test to help decide whether to perform a biopsy. "That is how I think it will be used once it is more accepted," Dr. Richie says. But that is not the case currently. "Right now, most urologists are using it in persons with one negative biopsy to decide whether they should do further biopsies," he explains. For free PSA higher than 20 to 25 percent, clinicians are more likely to follow patients; otherwise they recommend additional biopsies.

Dr. Richie follows this course. "I use it to decide who should get additional biopsies," he says. "As it becomes more available and more accepted, I may use it as an initial screening test." Dr. Chan cites several ways %fPSA can be used in clinical practice. First, and most obvious, using a single cutoff—25 percent—yields 95 percent sensitivity and a potential 20 percent reduction in unnecessary biopsies.

Another approach is to estimate the likelihood of prostate cancer based on %fPSA. "Perhaps it is better not to use a single cutoff but to look at whether percent free PSA is high or low, then tell the patient his risk of prostate cancer," Dr. Chan says. The Hybritech and Abbott assays give a graded probability of prostate cancer, with a 60 to 80 percent probability at %fPSA values of less than 10 percent and a probability of less than 10 percent when %fPSA is greater than 25 percent.

"We actually encourage people to use a probability based on percent free PSA," Dr. Chan says. Such decisions are challenging and best made among laboratory professionals, clinicians, and patients. "I have gone into discussions about that with the urologist and patient," Dr. Chan says. "I would think that urologists who are knowledgeable can use those probabilities. It depends on the comfort level of the patient and physician to make that decision."

In the above techniques, %fPSA is used diagnostically. But it is also useful for advising patients who might need a repeat biopsy, as Dr. Richie described. Many patients with elevated total PSA have a biopsy that is negative. "Percent free PSA can help decide about a repeat biopsy," Dr. Chan says. "It is being used here by our clinicians [in] that way as well."

Biopsy is a gold standard, but it is not perfect. A standard sextant biopsy takes six segments of the prostate gland. "So the result depends on whether the needle hits where the cancer is," Dr. Chan notes. To reduce the problem of false-negative biopsies, he adds, "Some clinicians want to do nine- or 12-core biopsies, but that is harder on patients."

Dr. Chan also mentions the prognostic value of %fPSA. "Some urologists feel it is not necessary to detect small cancers that are not so significant and slow-growing, especially for an older man," he says. It can be helpful to use total PSA and %fPSA, along with the Gleason score and other clinical findings, to decide whether the patient has significant tumor and should undergo surgery. But, Dr. Chan notes, "The assay is not FDA-approved in the prognostic setting."

For monitoring, using %fPSA is still controversial. Dr. Chan notes that, in principle, %fPSA could be helpful in patients being treated with hormonal therapy, which affects production of PSA. With this therapy, changes in total PSA would not necessarily reflect the effect of the treatment on the cancer, while %fPSA might. "But in practice, this has not proven to be useful so far," Dr. Chan says. He offers a similar evaluation for using %fPSA to help detect recurrence after radical prostatectomy. "Perhaps it needs more studies," he suggests.

In Dr. Catalona’s experience, "At this stage [percent free PSA] is probably used more often in problem cases than routinely across the board." A problem case might be a patient who has a persistently elevated total PSA or perhaps a progressively rising total PSA and who has had one or more negative biopsies.

Does the physician need to worry about cancer missed on previous biopsies? "Our work and other reports have shown that the percent free PSA is probably one of the best ways to identify men who need repeat biopsies," Dr. Catalona says. In this context, based on referral calls from community urologists, Dr. Catalona explains, "Physicians basically dichotomize. If percent free PSA is above 25 percent, they feel comfortable. And they get scared if it is below 10 percent. That is the message I am hearing: They are using it more at its extremes than at its mid-zones."

Percent free PSA might be used routinely in the type of patient included in the multicenter study—a man with total PSA of 4 to 10 ng/mL and a negative DRE who has not yet had a biopsy. (This is equivalent to what Dr. Richie calls initial screening and Dr. Chan calls diagnostic testing.)

"The way I envisioned percent free PSA to be used, and the way we use it, is to screen a bunch of men using total PSA," says Dr. Catalona. If total PSA comes back less than 4 (less than 2.5 might also be used as a cutoff) or greater than 10, they would not need %fPSA. But if total PSA is in the 4 to 10 range, Dr. Catalona’s laboratory would automatically run free PSA. "Such reflex testing is not done in places where many patients get screened," Dr. Catalona notes. "That only happens in places where the percent free PSA assay is available and it has been set up that way." In practice, many clinicians have to order %fPSA specifically.

To illustrate how the %fPSA value can be integrated into clinical decision-making, Dr. Catalona describes a typical case. "Let’s say you have an elderly patient who has an enlarged prostate and a high total PSA. And you do a free PSA that comes back 35 percent. You can take some comfort in the fact that, based on the free PSA test result, the chance that a biopsy will be positive is only one of 12. So 11 times out of 12 that patient will go through biopsy unnecessarily. And even if he is the one in 12 who has cancer, it is probably going to be a slow-growing cancer. So it would be safe to follow him for a while."

In Dr. Markus’ laboratory, the most common scenario is to use total PSA as a screen and %fPSA as a followup test if the physician orders it. He does not do reflex %fPSA. "The people we are doing PSA tests for are urologists, and they have their own criteria for ordering percent free PSA," he says.

His informal conversation with a few urologists reveals that, generally, a patient with elevated total PSA undergoes ultrasound and digital rectal examination. During ultrasound examination, the physician may biopsy any suspicious lesions. If the biopsy is negative and total PSA is 4 to 10, free PSA may be ordered. "If percent free PSA is low, the urologist will do closer followup and may encourage the patient to undergo repeat biopsy," Dr. Markus says. "If it is high, the clinician might be a little less aggressive."

But if %fPSA comes back greater than 25 percent and the clinician decides to biopsy anyway, the increased specificity advantage of the %fPSA assay has been nullified. What has been gained? "Final decision on deferring biopsy has to be made between physician and patient," Dr. Catalona says. "They have to look at overall risk."

Dr. Slawin addresses this issue slightly differently. "It used to be that most patients who had abnormal total PSA got two sets of biopsies," he says. "In the past, a significant number of men who were eventually found to have cancer had a negative initial biopsy, so those who were negative on the first biopsy often got repeat biopsy." As a result, many men got two sets of negative biopsies. Now, with %fPSA, he says, "If a man is biopsy-negative and his free PSA is not in a suspicious range, we may defer the repeat biopsy." But he would also consider the man’s age and clinical picture.

Ultimately, the goal of measuring PSA is to detect prostate cancer early, while curative surgery can be performed. Yet whether radical prostatectomy is truly curative has been a topic of controversy. To support his affirmative view of this question, Dr. Catalona looks to historical record. Immediately after total PSA testing was introduced, he notes, the number of cases of prostate cancer rapidly increased; these were mostly likely "catch-up" diagnoses. "Starting around 1995," he continues, "the incidence rate has fallen dramatically. And mortality from prostate cancer, which had been increasing for the past 30 years, began to plateau."

Since then, there has been a seven to eight percent drop in the overall mortality rate and a 12 to 15 percent decline in the most favorable subpopulation, young white males. "Around Mayo Clinic," Dr. Catalona adds, "where screening is very prevalent, there has been about a 20 percent drop."

Authors of an epidemiological analysis concluded that PSA screening is at least partly responsible for a decline in incidence and mortality trends for prostate cancer in the United States. The investigators concluded: "The decline in IBM [incidence-based mortality] for distant stage disease and flat IBM trends for localized/regional disease provide some evidence of improved prognosis for screen-detected cases, although alternative interpretations are possible" (J Natl Cancer Inst. 1999; 91:1025-1032).

"I think the mortality rates have come down a bit quicker than I would expect based on PSA testing alone," Dr. Catalona says. "It takes seven to eight years for prostate cancer to kill a man. So if somebody came up with a perfect test and a perfect treatment today, we wouldn’t begin to see those effects for seven to eight years." Also contributing to the decline in mortality from prostate cancer, he believes, have been an improvement in surgical techniques and a five- to six-fold increase in the use of radical prostatectomy beginning in the early ’90s, about the same time PSA screening was adopted.

Rigorously proving the value of radical prostatectomy remains important because of the high incidence of serious side effects from this surgery. At 18 months post-surgery for clinically localized disease, impotence was reported by 65.6 percent of men who had conventional surgery and 56 percent of those with nerve-sparing procedures. Incontinence was reported by 8.4 percent of the subjects (JAMA. 2000;283:354-360). Dr. Catalona concedes that this is "a live issue."

"This debate has been likened to the evangelist versus the snail," says Dr. Slawin. "Evangelists are people who feel that screening should be performed, and snails expect to see clear evidence of reduced mortality before recommending it."

He notes that the recommendations of the U.S. Public Health Service Preventive Task Force (2nd ed., 1995) do not endorse PSA screening but that the American Cancer Society recommends that DRE and PSA be offered beginning at age 50 to all men with at least a 10-year life expectancy.

Two ongoing developments may augment the clinical value of PSA screening. One is the possibility of adding a marker called human kallikrein 2 (hK2) to the screening sequence. Employing neural networks for risk assessment is the second innovation. Using these computer algorithms to integrate all risk factors is equivalent to doing a multiple regression analysis on each patient. "In some preliminary work that I have done with two separate companies that are developing neural networks for this purpose," Dr. Catalona says, "when you throw in the free PSA, it adds substantial value."

In addition, Dr. Slawin is working on an isoform of PSA that is specific for benign prostate disease, called BPSA (Urology. 2000;55(1):41-45). "It appears when you include BPSA with total and free PSA, you get even better differentiation between BPH and prostate cancer," he says. "So as we get newer markers, we will improve on performance."

William Check is a freelance medical writer in Wilmette, Ill.