October 2009
Feature Story

Karen Lusky

Avoiding over- or undertreatment of prostate cancer can be difficult. Screen older men and you pick up many tumors that might best be left alone. Don’t screen and you miss the smaller percentage of tumors that will kill. And even at diagnosis, there is no good way to settle on which is which and the optimal treatment approach.

Overall, about 12 percent of men with prostate cancer will die from it, said urologist Markian Iwaszko, MD, of Mayo Clinic in Rochester, Minn., who presented on the topic in a workshop in July at the AACC annual meeting. “The million dollar question,” he says: “Who does early intervention really help?”

The answer may become clearer as more clinicians begin to use currently available tissue-based testing at diagnosis to help identify prostate cancers in need of more aggressive treatment. In fact, James D. Faix, MD, director of clinical chemistry and immunology at Stanford University Medical Center, says “prostate cancer could end up being a poster child for the new approach to individualizing health care delivery.”

For now, on the screening side of the care equation, clinicians and laboratories are faced with what to do with the not-so-cancer-specific prostatespecific antigen test, which, as Dr. Faix notes, is great for detecting benign prostatic hypertrophy but not so great for cancer.

“PSA,” says Peter Albertsen, MD, a urologist at the University of Connecticut in Farmington, “can lower prostate cancer mortality but it comes at a significant price,” which is unnecessary biopsies and treatment. That’s the key finding of the highly publicized Dartmouth study conducted by H. Gilbert Welch, MD, of the Dartmouth Institute for Health Policy and Clinical Practice and the Veterans Affairs Outcomes Group, and Dr. Albertsen, published in the Aug. 31 issue of the Journal of the National Cancer Institute (Welch HG, Albertsen PC. J Natl Cancer Inst. Aug. 31, 2009, Epub ahead of print).

Although the Dart­mouth study raises questions about the risk-benefit ratio of widespread PSA screening, as has other recent research, PSA testing isn’t going away, in the view of Amy Saenger, PhD, director of central clinical laboratory/ central processing at Mayo Clinic, who co-presented with Dr. Iwaszko at the AACC meeting. “Even though PSA is controversial, it has universal acceptability as a tool for screening, and it’s reimbursed by the government,” she says. Which is why “anything we can do to help with the sensitivity and specificity issues [for PSA] is beneficial until we can offer something better.”

Labs and clinicians can take a number of steps to avoid the false-positives associated with PSA testing without allowing cancer to slip through too loose a detection net. For one, Dr. Iwaszko cautioned, clinicians should realize that PSA is “not a surrogate for digital rectal exam,” which he says detects 20 percent of prostate cancers. PSA and DRE act in a synergistic way in detecting prostate cancer, he emphasized.

A lab’s choice of a PSA assay can also affect prostate cancer diagnosis based on the clinical cutoff for ­referring men for biopsy. The widely accepted cutoff of 4 ng/mL, Dr. Saenger says, was developed based on a study using the Tandem-R PSA assay (Hybritech), which uses a unique calibration standard. A majority of PSA assays in use are calibrated based on the World Health Organization reference standard and may produce patient results that are about 20 percent lower than the Hybritech-calibrated assays, she says. This may lead to confusion if cutoffs of 4.0 or 2.5 ng/mL are used; they were not derived using a WHO-calibrated PSA.

Yet even among the WHO-calibrated assays, differences exist among manufacturers, Dr. Saenger cautions. “If you look at CAP proficiency testing, you will notice a wide range of comparability. The Roche WHO-calibrated assay is comparable to the Beckman Coulter Hybritech assay, while other WHO-calibrated assays are 10 to 25 percent lower than the Beckman Hybritech, and some run higher than the Hybritech. Clearly, further standardization efforts are needed if PSA screening remains a common practice.”

Mayo Clinic uses a Hybritech-calibrated PSA assay and provides clinicians with “interpretive information about the assay, platform, and calibration” the lab uses—a strategy Dr. Saenger advocates for all laboratories, if feasible.

Beckman Coulter is recommending that labs set a lower cutoff when using a WHO-calibrated assay—if the assay is truly 20 percent lower than the Hybritech-calibrated assay, says Beckman Coulter’s Bernard Cook, PhD, of scientific and professional relations. “We are telling customers that if they calibrate to WHO, they should use a 3.1 ng/mL [rather than a 4.0 ng/mL] cutoff. A lower cutoff of 2.5 ng/mL on Hybritech should be 2.0 ng/mL with a WHO-calibrated method.”

Dr. Faix, however, says his laboratory at Stanford is using a WHO-calibrated assay and has purposefully maintained the 4 ng/mL cutoff, with the support of the urologists. “We know this makes the screening less sensitive,” he says, “but that may be a good thing if it reduces the number of false-positives. We are careful to add a comment to the effect that no single PSA cutoff can accurately identify patients with prostate cancer.”

Columbia University urologist Franklin Lowe, MD, uses a cutoff of 2.5 ng/mL for men under 60, which is based on a study he and colleagues did on more than 36,000 patients (Gilbert SM, et al. Urology. 2005;65: 549–553). The research found that men who had PSA levels between 2.5 and 4 had a 27.5 percent chance of having prostate cancer, he says. “Between 4 and 10, the likelihood was 30 percent. Only when you got below 2.5 did it drop to 22 percent. Below 2 was 18.7 percent.”

For men with elevated total PSA, the percentage of free PSA, which is a subset of the total, can serve as a triage test of sorts to quantify the odds that the person has prostate cancer.

Total PSA, Dr. Saenger noted in her presentation, consists of free PSA and PSA bound primarily to alpha-1-antichymotrypsin. A ratio of free to total PSA (%fPSA) greater than 25 percent reduces the likelihood of prostate cancer and may imply benign prostatic hypertrophy. A low ratio (<10 percent) suggests a greater probability of prostate cancer, and a %fPSA ratio between 10 to 25 percent falls in a “gray zone.”

Typically, said Dr. Iwaszko, if the person has a free PSA of greater than 40 percent, he has less than a 10 percent risk of prostate cancer on biopsy. If the fractionation is less than 10 percent, however, the risk of cancer on biopsy is usually close to 50 percent.

At Mayo, “most clinicians order a PSA screen to receive just the total PSA,” Dr. Saenger told CAP TODAY. They can request to add on a free PSA to get the %fPSA, but this will be performed only if the total PSA is between 2 and 10 ng/mL. “Mayo Clinic does not automatically reflex to perform a free PSA based on the total PSA result, which minimizes the payment and reimbursement impact on the patient and clinic for an extra test,” she says.

Measuring PSA velocity over time can also improve specificity for detecting cancer. Men with benign prostate growth, Dr. Iwaszko explained, don’t show a rapid increase with time. Usually less than 0.75 ng/mL over a year for men with PSAs between 4 and 10 would be more related to benign tissue growth, he said. “Typically, we like to see a few points along the graph. Once you have the ability to place multiple plots along the graph, that allows us to understand the kinetics and mechanics of that PSA better.”

In Dr. Faix’s view, looking at PSA velocity over time is an approach that “hasn’t been given its due.”

“A PSA of 6 ng/mL may not necessarily require a biopsy immediately. If [the PSA value] continues to increase, then you revisit the issue of doing the biopsy. Biopsies, especially when you are looking for a very small lesion, don’t have a very high sensitivity,” he says. “The urologist does focus on certain parts of the prostate where the incidence is greatest or you may see something on the ultrasound, but it is still often a blind stab in the dark.”

To improve the specificity of prostate cancer screening at Mayo, Dr. Saenger says, there’s been discussion about performing PCA3 (prostate cancer antigen 3), a molecular test. Gen-Probe has worldwide licensing rights to the test, which is performed on urine collected after digital man­ipulation of the prostate gland. PCA3 mRNA is more than 60 times overexpressed in prostate cancer when compared with benign prostatic tissue, Dr. Saenger said in her presentation. While PCA3 isn’t an orderable test at Mayo, it is offered by a “handful of [reference] labs in the United States,” she notes. (See “New prostate markers waiting in the wings,” CAP TODAY, June 2009.) Gen-Probe announced in August that it has begun a clinical trial as a step toward gaining FDA approval of its PCA3 assay.

Dr. Saenger says Mayo would also be open to offering other biomarkers that could be used in conjunction with total and free PSA, as they become available, “to enhance the sensitivity and specificity of the current [PSA] assays.”

Beckman Coulter has one in the pipeline that fits that bill. It’s an isoform of free PSA, [-2]proPSA, which is a “more cancer-specific form of PSA,” Dr. Cook says. The immunoassay, which is called p2PSA, is beginning to be used in Europe, and Beckman Coulter plans to move forward with an FDA submission for the test in the United States.

Many experts suggest using a risk-based approach to screening as one way to avoid overdetecting more indolent cancers. Dr. Iwaszko noted that a future screening paradigm for prostate cancer might include family history and genetic testing to flag the men at highest risk for the disease, who could get more frequent PSA-based screening and perhaps prophylactic therapy. Men at low risk would receive only PSA-based screening.

“There is some evidence,” says pathologist Dean Troyer, MD, “that BRCA1 can identify men at higher risk for aggressive prostate cancer, although it’s a very small subset. It may mean something if you have it, but it’s not useful in the larger issue of how we handle prostate cancer screening from the public health perspective.” Dr. Troyer is a professor in the Department of Microbiology and Molecular Cell Biology at Eastern Virginia Medical School, Norfolk, and adjunct professor of pathology at the University of Texas Health Science Center San Antonio.

As for genetic testing that could have widespread implications, Dr. Troyer points to studies, including one within the past year in the New England Journal of Medicine, that look at the potential for SNPs (single nucleotide polymorphisms) to identify men at higher risk for prostate cancer.

In the NEJM study, U.S. and Swedish researchers estimated that SNPs “in five chromosomal regions—three at 8q24 and one each at 17q12 and 17q24.3”—combined with family history accounted for 46 percent of prostate cancer in a population of almost 3,000 Swedish men with prostate cancer. (The study included 2,893 subjects with prostate cancer and 1,781 control subjects.) “The cumulative effect of these variants and family history was independent of serum levels of prostate-specific antigen at diagnosis,” the researchers wrote in the abstract (Zheng S, et al. N Engl J Med. 2008;358:910–919).

The authors do not know how the SNPs they analyzed affect the risk of prostate cancer. They also said that the five SNPs in the study “appear to be associated with a risk of prostate cancer in general, rather than with a more or less aggressive form....”

Dr. Troyer says it’s reasonable to think that genetic testing would help identify men at risk. “But the question is, Would it change the urologists’ current approach to prostate cancer in terms of identifying people for biopsy? The state of the art with SNPs is probably not to the point that you can use them to the extent that they would negate or change current practices, particularly for widely prevalent cancers with lower death rates.” The patient, too, has to deal with the possible implications of a “positive” test indicating a significantly increased risk of developing the cancer.

The uncertainty of receiving a diagnosis of prostate cancer is, of course, also difficult, especially when trying to decide whether “watchful waiting” is a safe bet. “The Holy Grail for prostate cancer diagnosis,” says Dr. Faix, “is to differentiate cancers that are going to behave badly from those that aren’t—in all likelihood, this will occur on biopsy.”

Aureon Laboratories provides a personalized prognostic tissue-based test, the Prostate Px+, that the company says can improve on the accuracy of the widely used Gleason scoring in predicting patient outcomes. The test (one for prostatectomies and another for biopsies) does this by incorporating image-based features derived from biomarkers, including the quantitative expression values of the androgen receptor and Ki-67 with multiplex immunofluorescence, says Michael Donovan, PhD, MD, Aureon’s chief scientific officer.

The test also includes H&E and immunofluorescent-based morphometric profiles, which Dr. Donovan says are “themselves very strong biomarkers for assessing the in situ properties of tumor organization, including cellular distribution and differentiation.” The patient’s total PSA is also incorporated into the mathematical risk-prediction model.

A study reported in the July 2009 Journal of Urology showed that the Aureon test has a concordance index of 0.74 and a hazard ratio of 5.124 in predicting which of 686 prostate cancer patients treated with surgery would have clinical progression within eight years (Donovan MJ, et al. J Urol. 2009;182: 125–132). Validation using an independent cohort of 341 patients’ biopsies showed a concordance index of 0.73 and a hazard ratio of 3.47. The authors wrote, “Increased androgen receptor in tumor cells in the biopsy highly significantly predicted resistance to therapy, i.e., androgen ablation with or without salvage radiotherapy, and clinical failure (P<0.0001).”

“The main thrust of the paper,” says the University of Connecticut’s Dr. Albertsen, a co-author of the study, “is that the test can separate low from high risk.”

But is it good enough from a practical standpoint? he asks. “Is someone flagged with low-risk disease willing to watch that tumor and do active surveillance? That’s one test [of whether it’s going to be good enough]. Obviously, the higher-risk tumors are the easiest call. If it’s high risk, everyone will recommend having the cancer removed. The question is, Is it really high risk? The test isn’t 100 percent. Some people are willing to accept probability estimates and others want absolute guarantees.”

Dr. Albertsen notes that urologists at the University of Connecticut are just now talking to the head of pathology to decide whether the test can help urologists recommend active surveillance. “We know what the [study] report says. The question is, How does it function in real life?” And the answer to that only “comes with trying it out,” he says.

Daniel Eun, MD, assistant professor of urology in surgery at the Hospital of the University of Pennsylvania and at Pennsylvania Hospital, is using the Aureon test for patients undergoing robotic prostatectomies. “If I know the patient is at higher risk for unexpected aggressive disease, I may do a wider resection and change my lymph node dissection template,” he explains.

He has had a number of cases in which patients undergo surprise upgrading and upstaging, which “unfortunately you don’t find out until after surgery,” he says. That’s one reason he welcomes any information that provides a heads-up before surgery that a patient may have “worse parameters” than expected.

“Using the Aureon test also may potentially identify cases in which watchful waiting is not the best option,” Dr. Eun says. “For similar reasons, the test can be helpful also for people with low-volume, Gleason 6 prostate cancer who are considering brachytherapy.”

Dr. Eun has not been doing the test for long, and he will follow his data to see if the approach turns out to be beneficial to his patients.

Combimatrix Molecular Diagnostics is offering UroScan, an array CGH test that classifies prostate cancers into those with favorable, intermediate, and poor prognoses. Tumors with genomic profiles showing no deletion of PTEN or 21q have a favorable prognosis. Those with minimal genomic instability and 21q fusion-associated deletion have an intermediate risk of recurrence. The poor-prognosis tumors show 21q fusion-associated deletion and PTEN gene deletion, a poor prognosis.

UroScan has been validated on prostatectomy samples. Now Combimatrix Molecular Diagnostics is getting ready to begin a shared study with the University of Texas Health Science Center San Antonio using UroScan to identify genomic signatures associated with a bad prognosis in prostate cancer biopsy samples from men who died from the disease, reports Mercedes Gorre, PhD, vice president of scientific affairs at Combimatrix.

UroScan and the Aureon test “can be complementary,” says Mansoor Mohammed, PhD, president and CEO of Combimatrix, “because our test provides the genomic equivalent of a chromosome preparation from the prostate tumor,” which is “a cornerstone of the genetic testing consideration of any tumor workup.”

Dr. Troyer notes that the approach afforded by tests that predict prognosis “does change things by allowing you to identify several different flavors in a cancer. The public has the notion that there’s only one kind of cancer and it’s always bad.”

“Unfortunately, this type of testing is often done ‘in a box’ with the answer coming out the door,” Dr. Troyer says, adding that he’s not in any way questioning the integrity of those who run the companies. “The software is also sort of in-house, so it’s hard to take the lid off it and do the hard science in some ways.” But the in-house testing approach for prognostication is being done for breast cancer, he notes.

Use of prognostic tests can help figure into treatment decisions, but clinicians typically rely on PSA to monitor for prostate-cancer recurrence after prostatectomy or radiation. “Biochemical failure,” says Columbia’s Dr. Lowe, defined as PSA increasing after treatment, “provides an early marker for the disease still being active.”

Ultrasensitive PSA assays, which can accurately and precisely quantitate very low concentrations of PSA, can be used in monitoring for recurrence of prostate cancer. Some companies, Dr. Saenger said in her presentation, market the ultrasensitive approach as providing a longer lead time (up to 18 months) compared with conventional assays for detecting biochemical failure. An ultrasensitive PSA of less than 0.01 ng/mL for two years implies a “cure” requiring minimal followup, she added (Witherspoon L. AACC Expert Access Forum, July 1, 2008).

Yet Dr. Saenger expresses reservations about clinicians using the test in this way because it can generate undue patient anxiety. “Patients are very aware of their PSA values and worry if it’s even slightly elevated. And they don’t often take into account things like biological or analytical variability, which can cause very minor changes. Patients and urologists tend to be very focused on the number, and this can lead to misinterpretation of PSA.” Thus, “just because you can measure and report a concentration out to two decimal places or down to the limit of detection doesn’t, in my view, necessarily mean you should.”

Dr. Lowe agrees that using ultrasensitive PSA has its plusses and drawbacks. “If you believe that everyone who has a rising PSA after definitive surgery should have adjunctive radiotherapy, then it makes sense to do the testing. But if you are in more a watch and wait [mode] where you say, ‘Okay, the guy had surgery and we are going to see if his PSA gets up to 0.5,’ then it doesn’t make much sense to use the ultrasensitive test.”

If a patient does have biochemical failure, PSA doubling time can be prognostic, according to a Mayo Clinic study (Tollefson, MK, et al. Mayo Clin Proc. 2007;82:422–427). In the study, patients’ PSA doubling time was monitored along with their bone and CAT scans and biopsies. A doubling time of less than or equal to three months indicated a poor outcome, Dr. Saenger said in her AACC talk. The lowest-risk patients took 10 years to double their PSA.

The good news about biochemical failure if it does occur, Dr. Troyer says, is that “you can’t really say ‘the cat is out of the bag’ in terms of having a chance to prevent cancer-specific mortality.” He notes that “only a subset of men with PSA recurrence will die of prostate cancer, as the PSA recurrence often signals a local recurrence as opposed to meta­static disease.”

Some treatment centers show relatively low prostate cancer death rates related to biochemical failure, Dr. Troyer adds. “The data we have here at Eastern Virginia Medical School indicates that only 10 to 15 percent of men, at most, with PSA failure will die of PCA.” The differences in death rates in different cohorts of patients with PSA failure may reflect biases in populations, treatment, and the nature of practices, such as community versus tertiary academic or cancer centers.

In terms of new treatments coming down the pike for patients who do relapse, Charles Sawyers, MD, investigator at the Howard Hughes Medical Institute at Memorial Sloan-Kettering Cancer Center, points to two new anti-androgen drugs (MDV­3100 and abiraterone) in phase III trials. The drugs have been shown to work when patients’ tumors become resistant to the current generation of anti-androgen therapy, he says. Dr. Sawyers’ research group has been studying why resistance occurs. And “the answer is that the tumors express higher levels of androgen receptor and are able to overcome the drugs.”

There’s interest in using the new drugs for neoadjuvant therapy before surgery on patients with a large, high-risk prostate cancer that hasn’t yet spread outside the prostate. “The idea is to shrink it before surgery or even make it disappear to induce a pathologic complete remission”—analogous to the approach used now to treat some breast cancers.

While the experimental anti-androgen drugs are being tested now in all patients with late-stage prostate cancer, Dr. Sawyers surmises that the benefit of the drugs may be seen only in tumors that have a higher level of androgen receptor expression.

Like many others, Dr. Sawyers is certain that prostate cancer treatment will become more personalized. But in terms of determining what type of test will provide the clinical outcomes prediction needed to drive therapy, the problem has yet to be “cracked open,” he says.

As always, there’s work to be done. As Dr. Iwaszko puts it, “We need to decide how best to screen, risk stratify, and treat.”