College of American Pathologists
Printable Version

  Plugging the gaps in PSA testing





cap today



April 2007
Feature Story

Anne Paxton

The prostate-specific antigen test, one of the most successful but also most controversial cancer screening tools in history, remains one of the least accurate clinical diagnostic tests, with a specificity of only about 20 percent to 40 percent. Because of false-positive PSA results, hundreds of thousands of men each year in the U.S. undergo prostate biopsies that are unnecessary.

But research is showing that several variations of the PSA measurement, plus newly available genetic tests, are starting to make a dent in those numbers.

William J. Catalona, MD, professor of urology at Northwestern University’s Feinberg School of Medicine, Chicago, first started studying PSA in the late 1980s.“We just picked a cut point that would be clinically useful. But right from the start there was a lot of concern about false-positives and false-negatives,” he says.

PSA was approved in 1986 as a marker for monitoring the response of prostate cancer patients to treatment. But it wasn’t until 1994 that the FDA approved the use of PSA for early detection of prostate cancer. Because of concerns largely from the public health community about false alarms and biopsies of men without prostate cancer,“there was a big push to make PSA testing more accurate.”

The earliest refinement was PSA velocity—the rate of change in PSA levels over time. The assumption was that if PSA levels were going up faster, it was more likely that the patient had prostate cancer, he says.

“But right off the bat, when you were starting with men who already had a PSA higher than 4 ng/mL, you didn’t want to say, ‘Wait until next year to see if it goes to six.’ The other thing was there are natural fluctuations in PSA from week to week, so if the PSA was a little higher you wouldn’t know if it was just a day-to-day variation or a real finding. And the third thing—and it’s still a problem that hasn’t been resolved—is that different manufacturers’ assays are standardized differently, so you can test the same blood sample in different manufacturers’ assays and get different PSA values.”

“So PSA velocity was introduced, initially accepted, and rejected—but it’s really made a strong comeback in recent years, especially in patients with PSA levels lower than 4 ng/mL.”

A second variation was PSA density—the PSA value divided by the man’s prostate size.“So if you had two men and both had high PSAs of six, and one had a normal prostate size of 30 ccs and the other a prostate of 60 ccs, then the man with the smaller prostate was more likely to have prostate cancer,” Dr. Catalona says.

The downside of that was that“it’s hard to accurately measure the size of a man’s prostate without doing an ultrasound scan, and once you did an ultrasound scan, most doctors went ahead and did a biopsy anyway.”

The third variation on PSA testing was the use of different PSA isoforms.“The existence of different PSA isoforms was appreciated in the early 1990s, but not much was done about it initially,” Dr. Catalona says. In men with prostate cancer, PSA exists in two main forms. One form, bound to protein, is called complexed PSA. The other is the free-floating form in serum.“In men with cancer, almost 90 percent of their PSA is complexed form and 10 percent is free. So by measuring the free-to-total PSA ratio, you could get better discrimination.” In practice, that variation has proved to be fairly useful, he says.

More recently, researchers have intensified their study of the three different forms of free-floating PSA—including pro-PSA, which is higher for men with prostate cancer, and BPSA, which is more common in men with benign prostatic hyperplasia, Dr. Catalona says.“So we’ve done preliminary studies, and we’re about to launch a larger one that will probably show pro-PSA and BPSA will give better discrimination than anything now commercially available.”

Even though it’s the gold standard for prostate cancer diagnosis, the prostate tissue biopsy is not 100 percent correct, says Daniel Chan, PhD, professor of pathology, oncology, urology, and radiology at Johns Hopkins School of Medicine, Baltimore.“Typically, you would use a needle to poke into the prostate gland and pull out some of the tissue cells. In the old days, we’d have the needle go in six times and take out six little cores of tissue to look at under the microscope and see if a tumor was present.”

“ But if the tumor was small, you might not find it, so the six-core biopsy was increased to nine-, 10-, or 12-core. It’s painful to do, but that’s what we have right now.”

A PSA of four to 10 is in a“diagnostic grey zone,” Dr. Chan notes, because the man may have many benign conditions in that range, so the chance of finding cancer is even smaller. For the last seven or eight years, however, urologists have used the free PSA percentage to decide whether they should more aggressively look for the cancer in an initial or repeat biopsy.

“If the free PSA is high, maybe 30 percent, then the probability of cancer is actually quite small and maybe you don’t need a repeat biopsy. Maybe you can wait another year or two.”

It makes sense, he explains, because PSA is a serine protease that is supposed to take proteins and chop them into fragments or cut peptide bonds.“If you think about it, you would not have a molecule like this running around in circulation because it has the potential to chop up all your proteins. That’s why the majority of PSA that are free form are complexed with alpha 1 antichymotrypsin [ACT] which acts as an inhibitor.”

Free PSA itself, however, occupies a diagnostic grey zone when it is 10 percent to 25 percent. Pro-PSA and BPSA are inactive, and they are better indicators than the free PSA percentage by itself, Dr. Chan says.“Now you can lower the cutoff of total PSA. Instead of cutting it off at four, which we know will miss as much as 30 percent of the cancers, maybe you can lower the cutoff to three or 2.5, then use free PSA and pro-PSA and BPSA to help you minimize the false-positives.”

Dr. Chan, who also directs the Proteomics Biomarkers Discovery Center at Johns Hopkins, says the center is working to develop a panel of biomarkers, proteins that could be readily measured in serum or plasma.“PSA is one molecule, one test, and I often tell people maybe we are asking too much for one test. If you had a multiplex test—a panel of, say, five protein biomarkers in combination—then that might let you make better decisions about what to do with patients.”

PSA was never meant to be a marker to tell someone whether they have cancer or not, because it’s not cancer-specific, says Richard J. Babaian, MD, professor of urology at the University of Texas, Houston, and director of the Prostate Cancer Detection Center at M.D. Anderson Cancer Center.“Its role was to stratify patients’ risk, and it was never really interpreted correctly. It’s the best tumor marker out there, but I think we’ve beaten PSA to death and we need to move on.”

M.D. Anderson has done extensive work with complexed PSA and, in his opinion, it’s superior to total PSA and free PSA, which require two tests.“Complexed PSA has never gotten its market share or its due, because people have just been married to PSA and can’t think in another language, and that’s what it takes to think about another test.”

Percent-free PSA is a measure that mainly benefits patients with PSAs of four to 10, Dr. Babaian points out.“But complexed PSA seems to be better in the range of 2.5 to six total PSA—and since that’s where most patients are today, it has a wider applicability.”

At M.D. Anderson itself, Dr. Babaian says practices among urologists vary:“Basically for our screening effort we use complexed PSA, but some physicians are still using total PSA.”

When Dr. Catalona was at Washington University in St. Louis, Mo., he led a large PSA study that enrolled 36,000 men and followed them for 12 years (Smith DS, Catalona WJ. J Urol. 1994;152: 1732– 1736).“What really became apparent to me was that a lot of these men would start out with PSA values that were very, very low—less than one. They would come in every six months to be screened, and often they would come together either on buses or little groups for the study. And for some men their PSA would be .6, .6, .6, .7, while others would show more variation. But we weren’t allowed to order a biopsy until their PSA hit 4 ng/mL. By that time we’d find the cancer had already spread to the margin of the prostate gland or beyond in about 30 percent of them.”

“Men in the study would call and say, ‘Listen, my PSA has gone to 3.2 from .6, and my friend who rides with me on the bus, his PSA has stayed .6 all along.&8217; I’d tell them we were not allowed to recommend a biopsy until the PSA hits four, then they’d be diagnosed with prostate cancer and they’d be very upset with me. So it became apparent to me the four cutoff was really too high.”

In 1995 he lowered the cut point in the study to 2.5.“When we did that we picked up cancer earlier, more of the cancers were contained within the prostate gland, and we had better 10-year survival rates.” Now people tend to accept that number, and the National Comprehensive Cancer Network (NCCN) guidelines likewise recommend a biopsy for patients with a PSA higher than 2.5.

In the meantime, PSA velocity has become one of the hottest parameters in the field, Dr. Catalona says. A study he carried out in conjunction with Harvard University showed that if PSA velocity is higher than 2 ng/mL/yr in the year before diagnosis, there is a very high cancer death rate (D’Amico AV, et al. N Engl J Med. 2004;351:125–135).“That’s almost a lethal velocity, and now another study has shown a PSA velocity of .35 over the long term is also associated with increased risk of ultimately dying from prostate cancer.”

“ The NCCN guidelines recommend .5 as the PSA velocity threshold. But I think that’s a little high, and it will eventually come down to .35 or .4,” Dr. Catalona says.

The Prostate Cancer Prevention Trial also found that arraying men with PSA levels less than four at different levels (for example, one to two, two to three, and three to four) showed there was a linear increase in the chance of finding cancer in biopsies.“At the lowest level—0 to 0.5—it was seven percent, but in the three to four range there was a 27 percent chance, and 20 percent of those cancers were high-grade, potentially lethal cancers. This shocked a lot of people, but I had seen this many times before—just because a man’s PSA is less than four didn’t mean he did not have aggressive prostate cancer.”

“The bottom line of the study was that there is no PSA value where you can rule out prostate cancer. So people are drifting a little more away from PSA cutoffs and looking more at the rate of change.”

Dr. Catalona now recommends that men start getting PSA measured at age 40, to get a baseline they can compare against every year.“If the PSA stays flat, then the risk is low, but if it starts rising rapidly then they shouldn’t wait until it’s four,” he advises.

The first urine-based prostate cancer genetic test, the uPM3 (DiagnoCure, Quebec), became available in the U.S. in 2003. Unlike the serum-based PSA, the uPM3 tested urine for the RNA of prostate cancer gene three (PCA3) in prostatic cells. But in 2006, the next generation of this test became available from Gen-Probe: the Aptima PCA3.

“Both the uPM3 and PCA3 tests are more specific for detection of prostate cancer than PSA,” says Junqi Qian, MD, director of molecular diagnostics for Bostwick Laboratories, Glen Allen, Va., which had an exclusive contract with DiagnoCure to offer uPM3.“We were the only laboratory offering the uPM3. But the problem with it was its high non-informative rate of 15 percent to 25 percent; in other words, we could only make a definitive diagnosis on 75 percent to 85 percent of samples, because we could not get enough prostatic cells from the urine.”

Then the PCA3 came on the scene, employing three new technologies:“target capture” —a technique used for purification of the target mRNA, transcription-mediated amplification, and a hybridization protection assay for detection.

According to Scott B. Shappell, MD, PhD, medical director of MOD Laboratory in Dallas,“Target capture actually involves having oligonucleotides, specific for the RNA you’re trying to measure—in this case PCA3, coupled to little metal beads, and that allows a magnetic purification system that helps get rid of things that might be interfering with the subsequent amplification reaction.” Together, the new technologies brought a test that has a diagnostic informative rate of approximately 98 percent.“It’s a striking improvement in the current generation of tests,” Dr. Shappell says.

When the uPM3 test was launched, Bostwick Laboratories charged $275 per test. But with the PCA3, which is eligible for a Medicare reimbursement of a lower price,“we were able to drop the price because the reagent produced by Gen-Probe is less expensive and test volume is higher,” Dr. Qian says. He believes the cost will continue to come down as more people use the test.

Could PCA3 ever be used as a screening test for prostate cancer? The current indication for the test is a negative biopsy after a positive PSA.“But I already see some urologists order it as a screening test, without PSA,” Dr. Qian says, noting that a good number of markers are being tested as potential prostate cancer screening tools, and PCA3 is one of the most prostate cancer-specific genes to come along so far.

The usefulness of a screening test, however, hinges on its sensitivity.“Right now, the cutoff for a positive PCA3 result is set at 35 or higher. If you had a lower cutoff the sensitivity could be improved, but sometimes for a screening test we have to strike a balance between sensitivity and specificity.”

Or as Dr. Shappell explains:“You can always tweak the cutoffs to have the highest levels, but to make the test more sensitive you lower the specificity, which limits its main application anyway.”

An additional factor is that because of the widespread use of PSA screening,“nowadays prostate cancers are getting smaller and smaller, and most require a 10- to 12-core biopsy to detect,” Dr. Qian says. Even with that procedure, “urologists miss a good number of cancers by biopsy.”

MOD Laboratory in Dallas recently became the third in the nation to begin performing PCA3.“We have been doing it for about two or three months,” says Dr. Shappell.“We officially get it as an ASR [analyte-specific reagent] test. ... We’re responsible for validating its performance in order to offer it as a clinical diagnostic test.”

A potential drawback of PSA, of course, is the fear that it leads to overdiagnosing of prostate cancer. But the good news, says Dr. Shappell, who is a molecular pathologist, is that fears of overdiagnosing are, well, overblown.“The data suggest that the cancers being detected are clinically significant, and we’re finding them earlier.”

The bad news, though, is that PSA is not a very specific prostate cancer marker and there are other things that cause mild elevations in PSA.“The problem is that if you have a slightly elevated PSA and then a negative biopsy, you don’t really know if you just missed a cancer, or did something else cause that elevation, such as BPH or mild inflammation?”

At least six studies have been published or presented, including a recent study in Clinical Chemistry (Groskopf J, et al. Clin Chem. 2006;52: 1089–1095), and all have shown PCA3 has markedly higher specificity than what is shown for PSA, Dr. Shappell says. Sensitivity for detection of prostate cancer or concurrent prostate biopsy has been in the 65 percent to 70 percent range, slightly lower than for serum PSA, but specificity has been in the 75 percent to 80 percent range, much higher than for PSA. The current goal of using PCA3 in conjunction with PSA in patient management—not to replace it—is proper, in his view.“Essentially when there is a negative biopsy, you would reflex to the PCA3 test. A positive test currently may indicate the need for timely repeat biopsy, but we are still looking to continually redefine the reference ranges to determine what a positive PCA3 test means for specific patient subsets.”

A biopsy of the prostate, he points out, has inherent sampling errors that do not crop up in biopsies of the colon or breast or other organ systems.“After a negative prostate biopsy, we’re not necessarily saying the patient is out of the woods; those patients still have to have repeat PSA testing and be followed.”

A negative PCA3 also has limits.“I caution urologists that for patients with a negative PCA3, that doesn’t rule out prostate cancer or the need for an initial biopsy or rebiopsy [if the initial biopsy is negative]; we’re a long way from saying that. I don’t want to be the guy who tells the man a year later, when he has a high-grade or advanced-stage prostate cancer, ‘Oh yeah, but your PCA3 test was negative.’ But that doesn’t mean we won’t potentially eventually generate data to indicate how PCA3 values could be used for that initial biopsy decision point.”

A cancer-specific marker like PCA3 has a lot of other possibilities too; it could potentially provide prognostic information to help indicate which patients with prostate cancer can be managed conservatively or help in following patients with prostate cancer treated by surgery or radiation treatment, he adds.

Urologists, for example, are now using PSA to follow post-radical prostatectomy patients.“Sometimes there is a small blip in PSA level, and we wonder whether it’s recurrent prostate cancer or a little bit of benign prostate tissue left. We don’t know what will happen. PCA3 could be useful in that setting,” Dr. Shappell says. Or many patients with prostate cancer are radiated and may also have a slight increase in PSA following treatment. Cancer-specific markers could be more useful than a serum protein (that is, PSA) that could be derived from both residual benign prostate or cancer, but a tremendous amount of research is needed to investigate any such applications.

Or, down the line, it could be used to augment the sensitivity of the PSA test. For initial and, it is hoped, early diagnosis, “one possibility would be using PCA3 to detect patients who have high PCA3, the cancer-specific gene, but for some reason they don’t have an elevated PSA, so that it may allow us to detect clinically significant and even aggressive prostate cancer in patients with PSAs of less than 2.5. But I don’t see it replacing PSA for a long, long time,” he says.

The search for better markers hasn’t stopped at PCA3.“Gen-Probe also has licensing rights to a few other very promising genetic alterations in prostate cancer genes, and we will always be trying to develop new markers, but this is the first time we’ve had an RNA test for a gene that’s very specific to prostate cancer.”

Dr. Shappell hopes the cost, which some laboratories are billing to private insurance at the $400 to $500 range, will come down.“I’m sure PSA [now about $30 per test] was more expensive when it first came out. PCA3 is essentially a secondary screening test, so we don’t want it priced as some extravagant, exotic genetic test that limits its use.”

It’s somewhat strange to think how prostate cancer cells and hence the PCA3 mRNA shows up in the urine, such as from tumors potentially way out in the periphery of the gland, he admits.“Some tumors could be very removed from normal communication with the prostatic ducts and prostatic urethra, but somehow it happens and it’s facilitated by a digital rectal exam, which is a component of the recommended test procedure.”

“If we can identify those patients who have a high likelihood of having cancer on a repeat biopsy, and bring them back in quickly compared to just following them for a couple of years while the cancer is growing, if urologists have that as an augmentation to their current practice, it’s going to be well received.”

Will it reduce the number of unnecessary surgeries? Dr. Shappell is not sure.“With the specificity already out there and the data from the trials already done, we can say the test can reduce unnecessary repeat biopsies. But right now, that 10 percent of patients who are detected with elevated PSA and get a positive biopsy and may not have needed treatment—we can’t predict that preoperatively. And we’re way away from using this or other markers to identify those patients.” Down the line, however, he thinks that may be possible.

M.D. Anderson is using PCA3 only as a research tool—not yet in clinical settings, Dr. Babaian says.“We just presented data that shows a relationship of PCA3 to tumor volume [Ochiai A, et al. J Urol. 2007;177:903–906], which may lend itself to being added to nomograms with predictive value in determining who needs therapy.” At this point there is no clinical data to substantiate the finding, but he believes PCA3 could possibly be used beyond its current role of helping decide who needs a second biopsy.

Dr. Catalona, however, is a bit skeptical about PCA3’s usefulness.“I think the PCA3 test fundamentally is a very good idea, and maybe we’ll find better markers in the future. But my feeling now is it will be helpful in some instances, but we really have not fully evaluated its performance characteristics at this point.”

“Let’s say this gene is upregulated 30-fold in patients with prostate cancer. If you find cells in the urine that have a lot of this gene, and the gene is absolutely specific for prostate cancer, then you can be pretty sure there’s a prostate cancer there.”

“But when it’s negative, I think this is going to be its Achilles heel. It could either mean the cancer is not there or that in this particular cancer the gene wasn’t upregulated, or there is a cancer there but the cells don’t really have access to the prostatic ductal system. But I don’t think the data is there to say you could skip the biopsy in someone who had a negative PCA3 test, if other factors were suspicious such as rising PSA or high PSA density, or low free-to-total PSA.”

Based on his clinical practice, Dr. Catalona does not believe there’s going to be a threshold above or below which you’ll have a high level of certainty.

“Last week I saw a man who was young and had a very low PSA, but he was concerned because his father and uncle had died of prostate cancer, and his PSA had jumped from .75 to 1.4. Because of his family history the urologist did a biopsy and found a Gleason grade seven prostate cancer. Both his total PSA and PSA velocity threshold many would consider modest, but in the context of his family history it triggered a biopsy.”

While there are small arguments about details, in the urology profession there is uniform consensus that PSA testing is a valuable tool, Dr. Catalona believes.“And the proof of the pudding is the Surveillance, Epidemiology, and End Results database,” the National Cancer Institute project that tracks incidence rates of prostate cancer for deaths and stage of disease.

Since Dr. Catalona’s first PSA paper came out in 1991 (N Engl J Med. 1991;324:1156–1161), the SEER database shows the percentage of men who had metastatic prostate cancer at the time of diagnosis has decreased by 70 percent, and prostate cancer death rates, which had continually gone up since they were first measured in 1930, leveled off in 1995, just four years after use of PSA was authorized for screening.“Since then there has been a 32.5 percent reduction in prostate cancer death rates,” he says.

Dr. Chan agrees.“ If you think about it, we have had PSA for the last 20 years and we don’t have patients, at least in the U.S., coming in with metastases from prostate cancer—very few. Most of them have early stages of prostate cancer, I think because of PSA. You can say how bad PSA is because it’s not specific, because it’s also elevated for benign conditions, but the fact is we don’t see late-stage prostate cancer, we only see early-stage, and men present with smaller tumors than ever before. That’s because of PSA screening.”

In Dr. Catalona’s view, informed physicians and patients are getting away from a cookie-cutter approach to prostate cancer diagnosis.“ They’re considering all parameters—race, family, percent-free PSA, etc. —and making a global risk assessment rather than relying on one single parameter. But of all the parameters, PSA is still by far the most powerful.”

Anne Paxton is a writer in Seattle.