Moving beyond PSA

Finding prostate cancer has traditionally been a two-step process: First a simple blood test identifies men with an increased risk for the disease. For those at risk, a prostate biopsy follows. The main goal of the biopsy is to detect cancer, if it’s there. If it is, doctors also examine the tissue to try to distinguish an aggressive cancer that needs treatment from a slow-growing one that may never spread.

Determining who is at risk

Since 1986, the prostate-specific antigen (PSA) blood test has become a routine part of a doctor’s visit. The test measures levels of a protein produced by the prostate and can predict whether a man has cancer. It has enabled doctors to diagnose more than 80% of prostate cancers before they spread to lymph nodes, bones, or other tissues. In short, PSA testing helps detect tumors years — perhaps even a decade or more — before they cause symptoms. If you’re over 50, you’d think that having the test would be a no-brainer.

But even the strongest advocates of PSA testing admit that it has significant shortcomings. For one thing, PSA isn’t specific for cancer. Although the likelihood of cancer increases with greater elevations in PSA, values considered abnormal — in the range of 4 to 10 ng/ml or even higher — occur in men with benign conditions such as an enlarged prostate. In fact, about 75% of the patients who fall into this category — roughly 1.3 million men a year in the United States who have undergone a prostate biopsy (a procedure that removes snippets of tissue for examination under the microscope) — don’t have cancer. Adding to the confusion: 15% of men with a PSA below 4 ng/ml who have a biopsy actually do have prostate cancer, according to a 2004 study. That’s right: men with a “normal” PSA test result may have cancer, and most with an “abnormal” result don’t.

The PSA test also falls short in its ability to distinguish potentially deadly cancers from insignificant ones. Some cancers spread rapidly, but many grow so slowly that they may never cause problems. For the slow-growing cancers, the side effects of treatment — which include impotence and incontinence — can be worse than the disease itself.

“No one is entirely happy with PSA, including me,” says William Catalona, M.D., a prostate cancer surgeon at Northwestern University who helped pioneer PSA testing. “Because of these deficiencies, there’s a crying need to improve PSA testing.”

Researchers have spent the last several years combing through blood, tissue, and urine samples in the hopes of finding some sort of biological change, or biomarker, to more accurately diagnose prostate cancer and predict its behavior. Their efforts have recently begun to bear fruit (see Table 1 below).

Catalona has been studying a form of PSA called “free” PSA — as opposed to “bound” PSA, which binds to proteins in the blood. Today’s PSA tests measure total PSA, the sum of free and bound PSA. Catalona’s research has shown that a subcategory of free PSA called proPSA is superior to PSA in discriminating cancer from benign conditions. Its greater diagnostic accuracy likely stems from the fact that proPSA is produced in the prostate’s outer zone, the area where most cancers arise.

Subsequent studies of proPSA in 2,000 men confirmed Catalona’s initial results, and a San Diego company has developed an automated method for detecting it. The company plans to seek FDA approval for the test, which Catalona estimates could be used in clinical practice in as little as three years. While it could one day supplant PSA, experts say it will initially be an adjunct to PSA, especially in cases where the total PSA reading might prompt a biopsy.

“It’s far from perfect,” says Catalona, “but it’s definitely better than what we have now.”

Another test that may soon be available in the United States checks for the presence of an RNA dubbed PCA3. When prostate cells become cancerous, their PCA3 genes kick into overdrive, producing massive amounts of this cancer-specific nucleic acid. If a doctor massages a cancerous prostate gland, PCA3 is shed into the urine, where it can be detected with a sophisticated molecular test. PCA3 levels don’t rise if a man has an inflamed or enlarged noncancerous prostate, so this protein more closely correlates with cancer than PSA does. As a result, men with a PSA level that might normally warrant a biopsy could have a PCA3 test first. Those with a slightly elevated PSA level but a low PCA3 level could be spared a biopsy.

In February 2008, researchers at the University of Michigan announced that they had built upon the PCA3 test by screening for it and six additional biomarkers in the urine of 234 patients (see “PCA3 and other biomarkers” below). By correlating biopsy data with urine test results, researchers found that four of the biomarkers were strong predictors of prostate cancer: GOLPH2, SPINK1, PCA3, and TMPRSS2:ERG, which is a combination of two genes. In fact, the four together proved more accurate than either PSA or PCA3 alone, correctly identifying more than 75% of patients who were later found to have prostate cancer. The researchers’ next step is to prove that these initial findings hold up in tests of more men at multiple institutions.

Perhaps the most promising biomarker is EPCA-2, discovered by Robert Getzenberg, Ph.D., and other researchers at Johns Hopkins Hospital. Examining tissue samples, the team found that EPCA-2 was present in prostate cancer cells but not in normal tissue. Since biopsies are not a practical screening tool, the team tried to detect EPCA-2 in blood from 330 people.

The results were striking: healthy men and women, those with other types of cancer, and most men with benign prostate disorders had lower levels of EPCA-2 than men with prostate cancer. Only a few samples from men with prostate enlargement had elevated EPCA-2 readings, meaning the test was highly specific for cancer. Statistically speaking, the test detected 94% of prostate cancers — much better than the 65% of cases detected with PSA in this same group of people. Even better, among those who tested positive for cancer, results also correlated well with disease stage: the test correctly detected 36 of 40 men with localized cancer and 39 of 40 men with cancer that had spread (see “EPCA-2 biomarker” below). If the test performs as well in a larger group of men, it could augment or even replace PSA testing.

Determining which prostate cancers need treatment

Many men diagnosed with prostate cancer have other illnesses that may cause serious problems, even death, before the prostate cancer ever causes problems. They die with but not from prostate cancer. For that reason, doctors have been seeking clues to help determine who needs more immediate treatment and who can pursue active surveillance.

One obvious place to look, in this “post-genomic era,” is in the genes. Research has proven that several cancers can be caused by mutations in single genes and by the fusion of two genes, a specific type of mutation. Would the same hold true in prostate cancer?

In 1985, Lewis Cantley, Ph.D., a researcher at Harvard’s Beth Israel Deaconess Medical Center, discovered PI3K, a gene that acted like a switch for tumor growth. When flipped on, it drove the development of tumors in animals. But the relevance of his discovery for human cancers remained unclear until the late 1990s, when researchers at Columbia University discovered PTEN, a gene that can help prevent cancers, including those caused by PI3K. It turns out that PTEN is a brake to keep cancer-causing PI3K in check; when PTEN is deficient, PI3K is uncontrolled and drives prostate cancer. Subsequent studies showed that prostate cancer that had spread harbored more PTEN mutations or abnormalities than localized cancer — suggesting that an assay for defects in PTEN in primary prostate cancers may reveal patients who are most likely to develop metastatic cancers.

More recently, researchers have found that a category of small molecules once considered useless are emerging as key players in the development of cancers, including prostate cancer. Some of these molecules, called microRNAs, or miRNAs, interfere with the production and regulation of various proteins, which can play a role in tumor development or suppression. Pier Paolo Pandolfi, M.D., Ph.D., a geneticist at Beth Israel Deaconess Medical Center, says that drugs could be made to attack the offending miRNAs in prostate cancer.

Scientists have also discovered that the abnormal fusion of two genes — TMPRSS2 and ERG — increases the probability that prostate cancer will return and prove fatal, echoing the experience with fusion genes in other cancers. (Gene fusions occur when pieces of genetic information on a chromosome trade places with each other, altering their sequence.) The activity and location of this fused gene in the cell is being studied in prostate biopsy samples by Massimo Loda, M.D., a researcher at Harvard’s Brigham and Women’s Hospital. Loda says that the fusions may override molecular switches that prevent excess cellular growth, potentially making some prostate cancers more aggressive.

The good news: in other cancers, the discovery of fusion genes has led to the birth of powerful new treatments. After researchers found that the BCR:ABL gene fusion drives the development of chronic myeloid leukemia, they discovered that the drug imitanib (Gleevec) could send the disease into remission. To date, however, drugs that target the TMPRSS2:ERG gene fusion are not available.

Most cancers — indeed, most diseases — result not from single genes but from multiple genes. New technologies allow laboratories to identify which of the approximately 25,000 human genes are turned on and which are turned off in a cancer. For example, in a particular type of breast cancer, a test called Oncotype-DX identifies such a “gene signature” from tumor tissue and helps doctors decide whether chemotherapy is likely to prolong survival when other predictors of tumor behavior are favorable.

Researchers haven’t yet had quite the same success in finding gene signatures for prostate cancer, but they are getting closer. In 2003, one team reported a set of 17 genes in prostate tumors that predicted whether a cancer would spread to other parts of the body. It also reliably predicted poor prognosis. Another team uncovered a set of 11 genes in 2005 that seem to forecast disease-free survival after treatment. But these studies involved only a small number of prostate cancer cases — just 100 in all.

A much larger 2008 study uncovered five genetic variations that dramatically raise prostate cancer risk (see “Five genetic variations” below). By examining DNA from 2,893 men with the disease and 1,781 healthy subjects, researchers found that men with four of the five genetic variants were more than four times as likely to develop prostate cancer as those with none of the markers. Men with at least four of the five markers and a family history of the disease were more than nine times as likely to develop the disease. Interestingly, the combination of genetic variants did not correlate with PSA levels.

Molecular studies designed to improve diagnosis and help physicians judge who needs treatment may also point to effective new therapies. For example, Beth Israel Deaconess’ Cantley and Pandolfi are studying drugs that might enhance PTEN’s protective effect, as well as agents that might inhibit the cancer-promoting molecule PI3K. Drugs that target PI3K have been approved for some cancers, including renal cell carcinoma. Four drugs that might inhibit PI3K-fueled prostate cancer are currently in clinical trials. Other agents affect PTEN’s switching ability and have stopped the growth of tumors in mice. But whether they might rein in prostate cancer in humans remains to be studied.

While PSA testing currently remains the best way to predict prostate cancer risk, and the cancer’s appearance under the microscope remains the best way to guess how it will behave, researchers are convinced that ongoing biomarker research and genetic analyses will lead to significant improvements in prostate cancer detection and treatment. Each individual could have a panel of tests: one for diagnosing the cancer; a second to determine if the cancer needs to be treated; and a third to determine the best treatment. That’s what the hope of “personalized medicine” is all about.

Originally published June 2009; last reviewed March 17, 2011.