June 2009
Feature Story

William Check, PhD

Winston Churchill famously said, “Democracy is the worst form of government except all the others that have been tried.” We could use the same formula to describe serum prostate-specific antigen screening for prostate cancer. This test is far from perfect, as demonstrated by two large clinical trials reported a few months ago. One found that “PSA-based screening reduced the rate of death from prostate cancer by 20 percent but was associated with a high risk of overdiagnosis” (Schroder EH, et al. N Engl J Med. 2009;360:1320–1328). In the second trial, the mortality rate was actually higher (though not significantly) in the PSA-screened group than in the control group (Andriole GL, et al. N Engl J Med. 2009;360:1310–1319). The authors of the second trial emphasized that mortality from prostate cancer was “very low” in both groups. An editorial notes that any benefit of PSA screening “comes at the cost of substantial overdiagnosis and overtreatment” (Barry MJ. N Engl J Med. 2009;360:1351–1354).

It is easy to understand why PSA screening leads to these marginal results. In a review article published two years ago, the authors pointed out that PSA is “prostate-specific, but not prostate-cancer specific” (Wright JL, PH Lange. Rev in Urol. 2007;9:207–213). They cited a meta-analysis that found the positive predictive value of a PSA above 4.0 ng/mL to be only 25 percent, which means that the greatest proportion of the approximately 1 million biopsies performed in the U.S. annually are unnecessary.

Despite these drawbacks, the PSA assay has become the gold standard for detecting prostate cancer, along with digital rectal examination, because its perceived utility exceeds that of every other candidate assay proposed so far. “PSA testing has survived as one of the few biomarkers accepted for cancer detection, despite its obvious limitations and poor specificity,” the authors of the review article concluded.

“It is controversial whether serum PSA is useful for screening or not,” says Arul Chinnaiyan, MD, PhD, director of the Michigan Center for Translational Pathology and professor of urology at the University of Michigan Medical Center. Whether it makes a difference in mortality is still in question. “As a biomarker PSA has a number of problems,” he notes. “It is relatively nonspecific—its specificity is only about 20 percent. Often, elevated serum PSA is due to benign inflammation of the prostate or benign prostatic hyperplasia. We need newer biomarkers to supplement PSA for greater specificity.”

Mark A. Rubin, MD, worked for several years with Dr. Chinnaiyan at the University of Michigan seeking newer biomarkers for prostate cancer. He points to the two recently published PSA screening trials as evidence of the drawbacks of PSA screening. “We are highly aware that PSA is a flawed test,” says Dr. Rubin, who is currently Homer T. Hirst professor of oncology in pathology in the Department of Pathology and Laboratory Medicine at Weill Cornell Medical College and vice chair for experimental pathology. “It is sensitive but not cancer specific. It looks like as many as 75 percent of biopsies in the U.S. don’t have cancer. We need to reduce the number of unnecessary biopsies and better predict which ones do have cancer and should be appropriately treated.” He and Dr. Chinnaiyan stress, too, the need to be able to identify which men have aggressive disease and which have indolent disease.

“We really need something better than PSA,” Dr. Rubin says. “We need the next-generation PSA test.” He identifies three biomarkers now in development as interesting candidates for this role: TMPRSS2:ETS gene fusions, prostate cancer antigen 3 (PCA3), and early prostate cancer antigen 2 (EPCA-2). All of these tests still need to be validated in a clinical setting. “A number of clinical trials are underway trying to demonstrate that these tests can help us evaluate prostate cancer better than PSA alone,” Dr. Rubin says.

Drs. Chinnaiyan and Rubin are two of the many clinical researchers who are looking for a better screening test. Four years ago they and their coworkers reported an important advance in this search—the first recurrent gene rearrangements seen in prostate cancer (Tomlins SA, et al. Science. 2005;310:644–648). These consisted of fusions between the 5’ untranslated region of the gene for androgen-regulated transmembrane protease serine 2 (TMPRSS2) and the E26 transforming sequence (ETS) family of transcription factors. Fusions of this type were found in most prostate cancers, with the most common combination, TMPRSS2:ERG, present in about 50 percent of PSA-screened localized prostate cancers and in 15 percent to 35 percent of population-based cohorts. This finding demonstrated that “dormant oncogenes can be activated in prostate cancer by juxtaposition to tissue-specific or ubiquitously active genomic loci,” the investigators wrote (Tomlins SA, et al. Nature. 2007;448 595–599; Tomlins SA, et al. Eur Urol. 2009;Apr 24. [Epub ahead of print]).

They concluded that androgen- responsive promoter elements of TMPRSS2 mediate the overexpression of the ETS family of oncogenic transcription factors when these two genes are fused in prostate cancer. Work in transgenic mice supported the notion that the fusions mediate invasion, “consistent with the defining histologic distinction between PIN [prostatic intraepithelial neoplasia] and prostate cancer” (Tomlins SA, et al. Neoplasia. 2008;10:177–188).

Dr. Chinnaiyan calls this fundamental discovery “actually serendipitous.”

“We were exploring bioinformatic and computational approaches to analyze DNA microarray data,” he says. In a set of prostate cancer tissue samples, expression anomalies were observed. Sequencing revealed genes composed of oncogenic transcription factors fused with androgen-regulated genes. The transcription factors are normally not expressed in prostate tissue. “But when they get joined with active genes, they get turned on,” Dr. Chinnaiyan explains. “This was a surprising finding. At the time, gene fusion was classically associated with leukemia and lymphoma. We discovered it as being present in a common solid tumor.”

Extensive surveys have showed that TMPRSS2: ERG gene fusions are specific to prostate cancer. “Specificity is in the high 90 percent range, even 100 percent,” according to Dr. Rubin. “So if you have a positive test for a gene fusion, right away you know that individual has prostate cancer.” His “best guess” for sensitivity is 45 percent to 50 percent: About half of men screened with PSA who have prostate cancer will have this fusion. “So a fusion assay will be negative in some individuals with cancer, but when you see it, it is definitive,” Dr. Rubin says. “It offers the possibility of a really specific way of detecting prostate cancer.”

Dr. Rubin says data reported at the May meeting of the American Urological Association showed that the level of transcription of the fusion gene correlates with whether an individual has indolent or significant disease based on Epstein criteria. “So it may have some prognostic value, too,” he adds.

It is possible to detect TMPRSS2:ETS fusions noninvasively in the urine of men with prostate cancer, with a specificity rate in PSA-screened cohorts of over 90 percent (Laxman B, et al. Neoplasia. 2006;8:885–888). “The advantage of a urine assay,” Dr. Chinnaiyan says, “is that it doesn’t require biopsy and DNA profiling.” Work by a British group published earlier this year showed that TMPRSS2:ETS fusions can be detected in circulating tumor cells from patients with castration-resistant prostate cancer (Attard G, et al. Cancer Res. 2009;69:2912–2918). Dr. Chinnai­yan’s group had dem­onstrated previously that TMPRSS2:ERG fusions are very common in androgen-independent disease and that the same molecular fusion subtype was present in both primary and metastatic sites, “suggesting clonal expansion of disease” (Mehra R, et al. Cancer Res. 2008;68:3584–3590).

Among men with prostate cancer followed with watchful waiting, a TMPRSS2:ERG fusion was associated with a statistically higher rate of prostate cancer-specific mortality (DeMichelis F, et al. Oncogene. 2007;26:4596–4599). “Having a fusion is associated with a more aggressive form of disease,” Dr. Chinnaiyan says. However, he cautions, “This is a somewhat controversial area. We have studied a number of watchful waiting cohorts and some prostatectomy cohorts. We found that once patients are treated we don’t see much of a prognostic difference whether they have a fusion or not. Treatment selects patients with more aggressive cancer.” Other groups have not confirmed the association between fusion and worse prognosis (Winnes M, et al. Oncol Rep. 2007;17:1033–1036; Rouzier C, et al. Cancer Genet Cytogenet. 2008;183:21–27).

In the simplest case, fusion assays may be useful to help decide which patients should be biopsied. “If you have a high PSA level plus a positive result in the fusion assay, it would be wise to push to get the biopsy done sooner,” Dr. Chinnaiyan says. A positive fusion test may also suggest the presence of clinically significant—as opposed to indolent—prostate cancer.

If, on the other hand, the TMPRSS2:ERG test is negative, a biopsy may not be so urgent. “Some patients are already willing not to have a biopsy on the basis of only the PSA result,” Dr. Chinnaiyan says. “I would say if PSA is positive and the fusion assay is negative, it is not as difficult to choose a watchful waiting protocol.” The fusion assay now in development only detects the most common fusion partner of the ETS family, ERG. Making a decision to defer a biopsy will be easier when the assay monitors all five primary fusion partners.

Dr. Rubin identifies one further potential utility of the fusion assay, which he calls “an exciting development.” It has recently been demonstrated that men with fusion-positive prostate cancer respond better to a new drug, abiraterone acetate, a highly specific CYP17 inhibitor that ablates the synthesis of androgens and estrogens, which drive proliferation of cells containing TMPRSS2-ERG fusions (Attard G, et al. Cancer Res. 2009;69:2912–2918). “So this assay may have utility to predict which men would respond to specific therapies,” Dr. Rubin says.

In a separate line of research, Dr. Chinnaiyan’s group has been exploring metabolomics—monitoring metabolites in an unbiased way. “A lot has been done looking at genes and proteins in cancer, but there has been very little unbiased work on small molecules,” Dr. Chinnaiyan says. In a proof-of-principle demonstration, he and his colleagues mapped metabolic alterations that occur in prostate cancer progression and identified a number of metabolites that are perturbed. “In this initial effort, we focused on sarcosine, a methylated derivative of glycine that is highly elevated in metastatic prostate cancer,” Dr. Chinnaiyan says. “We found that sarcosine could be detected in the urine of men with prostate cancer and that the sarcosine pathway itself appeared to be involved in prostate cancer progression, as a mediator of invasion and aggressivity.” Finally, they found that gene fusions themselves can induce higher sarcosine levels when introduced into prostate cancer cells (Sreekumar A, et al. Nature. 2009;457:910–914).

Further along in development is the assay for the PCA3 gene, which was discovered about 10 years ago by Marion J. G. Bussemakers, PhD, while she was doing a postdoctoral fellowship in the laboratory of William B. Isaacs, PhD, at Johns Hopkins Medical Institutions. Dr. Bussemakers and colleagues found that PCA3 is 66-fold upregulated in about 95 percent of prostate tumors. The PCA3 gene is not translated into protein, so the assay measures PCA3 mRNA. Upon returning to Radboud University in Nijmegen, the Netherlands, Dr. Bussemakers and Jack A. Schalken, PhD, worked together to apply this analyte to urine sediment from patients with and without prostate cancer and obtained a sensitivity of 67 percent and specificity of 83 percent for predicting biopsy outcome (de Kok JB, et al. Cancer Res. 2002;62:2695–2698; Hessels D, et al. Eur Urol. 2003;44:8–15).

Gen-Probe has licensed the rights to PCA3 from the Canadian company DiagnoCure. “We focus on detecting RNA for infectious diseases and blood screening,” says Harry Rittenhouse, PhD, Gen-Probe’s senior director of research and development. “So the PCA3 assay is a good fit for us.” Gen-Probe’s assay for PCA3 is based on transcription-mediated amplification, or TMA, similar to its clinical diagnostic Aptima Combo 2 assay for chlamydia/gonorrhea.

In Gen-Probe-sponsored studies, sever­al positive attributes of the PCA3 assay have been found, says Jack Gros­kopf, PhD, Gen-Probe’s director of R&D cancer diagnostics. “In samples taken prior to prostatectomy, PCA3 score has been correlated with tumor volume, extracapsular extension, and pathologic grade,” Dr. Groskopf says, “and larger, more invasive tumors may be more likely to shed cells into urine” (Nak­anishi H, et al. J Urol. 2008;179:1804–809; Whitman EJ, et al. J Urol. 2008:180:1975–1978). Other work has demonstrated that the PCA3 score predicts the outcome of the first prostate biopsy as well as of repeat biopsies in men who have already had one or two negative biopsies (Deras IL, et al. J Urol. 2008;179:1587–1592; Haese A, et al. Eur Urol. 2008;54:1081–1088). In the second study, the authors noted that the PCA3 score had greater diagnostic accuracy than percent-free PSA and that the “PCA3 score was independent of the number of previous biopsies, age, prostate volume, and total PSA level.”

“All of these studies suggest that the PCA3 assay may help select men for active surveillance, which is an increasingly important need in prostate cancer diagnosis today,” Dr. Groskopf says. An initial target for the assay could be men with elevated PSA (=4.0 ng/mL) but one or more negative biopsies. “As many as 10 million men in the U.S. fall into this category,” Dr. Groskopf says. “Most positive serum PSA results are falsely positive. That leaves patients and physicians in a dilemma.” Should further biopsies be done? Or should monitoring be limited to serum PSA measurement? If the PCA3 score is high, repeat biopsy is more likely to be productive. If it is low, noninvasive monitoring could be the better choice.

Dr. Rittenhouse says that feedback from Europe, where the PCA3 test has been in use longer, indicates that with a low PCA3 score some urologists will delay biopsy. “Since most prostate cancers are slow-developing, a delay of six months to one year is not as meaningful as for breast cancer,” he points out. “It’s an individual decision by physician and patient. We are hearing that delaying biopsy has helped some patients who have had many negative biopsies and are reluctant to have another.”

Analytical characteristics of Gen-Probe’s assay for PCA3 mRNA have been published (Groskopf J, et al. Clin Chem. 2006;52;1089–1095). Sensitivity was 69 percent and specificity 79 percent. In contrast, serum PSA assay specificity was 28 percent for this same group. In Europe the assay, called Progensa, is CE-marked. In the U.S. Gen-Probe is selling the reagents as ASRs, mostly to reference laboratories that are offering it as their own lab-developed test for PCA3 mRNA. After speaking with FDA officials, Gen-Probe recently announced it will initiate a U.S. clinical trial of the Progensa PCA3 assay later this year.

Gen-Probe has also licensed the rights to the TMPRSS2:ETS gene fusions. “We are approaching that analyte on two fronts,” Dr. Groskopf says. First, they have devised a research assay in the same format as the Progensa PCA3 assay that quantitates the amount of TMPRSS2:ERG mRNA in the urine. “With that assay we hope to be able to prove prognostic utility,” Dr. Groskopf says. The goal is to demonstrate that measuring fusion mRNA helps distinguish cancers that need to be managed aggressively from those for which noninvasive monitoring is acceptable. Already they have obtained preliminary data correlating TMPRSS2:ERG mRNA in the urine with some indicators of cancer aggressiveness available at the time of biopsy. “This assay can also be performed on blood samples and may have utility for predicting and monitoring response to hormonal therapies in advanced cancer patients,” Dr. Groskopf says.

In a second approach, Gen-Probe is collaborating with Ventana Medical Systems to develop a FISH test for the gene fusion. “Ventana’s goal with the FISH assay is to be able to detect the fusion DNA in biopsy tissue,” Dr. Groskopf says. Population-based studies have shown that if fusion-positive tumors are left untreated, they have a more aggressive phenotype, he says.

“Dr. Chinnaiyan has hypothesized that gene fusion is an early event in the development of prostate cancer,” Dr. Rittenhouse says, “and that high-grade PIN with gene fusion is a progression from high-grade PIN without gene fusion.” Evidence for this idea comes from analysis of high-grade PIN tissue, which is considered pre-malignant. Cells in high-grade PIN are thought of as precursor cells for prostate cancer. A subpopulation of high-grade PIN lesions, perhaps 15 percent to 20 percent, when found in close proximity to or in contact with prostate cancer tissue, contain the same gene fusion that is found in the prostate cancer itself. Dr. Rittenhouse calls the hypothesis “provocative.”

“This biomarker, considering it was only discovered in 2005, has already generated a remarkable amount of interest and literature,” he says.

Dr. Chinnaiyan has also been developing a new assay: a first-generation multiplex assay performed on urine sediment that could be useful for early detection of prostate cancer. Dr. Chinnaiyan calls published work an “initial characterization” of the multiplex panel. “One gene fusion, TMPRSS2:ERG, when measured in urine accounts for about 50 percent of prostate cancer patients,” he explains. “We thought we would explore multiplexing it with other urine markers to see if we could develop a multiplex assay.” Indeed, this fusion combined with PCA3 and two other markers—GOLPH2 and SPINK1—provided better sensitivity for prostate cancer.

In 234 patients presenting for biopsy or radical prostatectomy, Dr. Chinnaiyan and his colleagues found that “increased GOLPH2, SPINK1, and PCA3 transcript expression and TMPRSS2:ERG fusion status were significant predictors of prostate cancer” (Laxman B, et al. Cancer Res. 2008;68:645–649). More important, they reported, “Multivariate regression analysis showed that a multiplexed model, including these biomarkers, outperformed serum PSA or PCA3 alone in detecting prostate cancer.”

“Taken together,” they concluded, “these results provide the framework for the development of highly optimized, multiplex urine biomarker tests for more accurate detection of prostate cancer.”

In other words, optimal prostate cancer screening will not be an either-or proposition, but a both-and solution. Optimal sensitivity and specificity for the multiplexed model were 65.9 percent and 76.0 percent, respectively, whereas those figures for PCA3 alone were 75.4 percent and 56.3 percent. At these values, positive and negative predictive values for the multiplex model were 79.8 percent and 60.8 percent, respectively, compared with the corresponding numbers of 71.2 percent and 61.4 percent for PCA3 alone.

“Combining multiple markers has tremendous appeal,” Dr. Rubin says. “Not all individual markers are sensitive in all individuals.” In the multiplex model, SPINK1 was seen in less than 10 percent of prostate cancers, he says, yet when added to the other markers it significantly improved sensitivity.

In the prostate cancer screening neighborhood, the new kid on the block is early prostate cancer antigen 2, or EPCA-2. “There are two different molecules called EPCA,” explains Robert H. Getzenberg, PhD, who was instrumental in discovering both. “They are not structurally related, even though both are nuclear matrix proteins that are elevated in prostate cancer.”

“A number of years ago we found the original EPCA molecule. More recently we discovered EPCA-2,” says Dr. Getzenberg, who is Donald S. Coffey professor of urology and director of uro­logy research in the Brady Urological Institute, Johns Hopkins Hospital. Both turned up when Dr. Getzenberg addres­sed a fundamental question: Could he find something that underlies what the pathologist sees under the microscope, the changes in how the nucleus looks at the structural level?

The original EPCA was assayed by quantitative immunohistochemistry. Dr. Getzenberg now focuses on EPCA-2, which is a serum biomarker. He and his colleagues found that EPCA-2 has 92 percent specificity for healthy men and men with BPH and 94 percent sensitivity for overall prostate cancer. In contrast, the specificity for PSA in these selected groups of patients was 65 percent. Moreover, they reported that EPCA-2 “accurately differentiates between men with organ-confined and non-organ-confined disease” (Leman ES, et al. Urology. 2007;69:714–720).

An important aspect of this study, Dr. Getzenberg says, was that the EPCA-2 assay could differentiate individuals with prostate cancer from those with BPH and other benign diseases as well as those with other cancers, and could separate organ-confined cases from those that already had extracapsular extension at the time of surgery. Benign diseases included prostatitis, which Dr. Getzenberg calls “probably the most common reason for elevated PSA levels.”

His laboratory has been attempting further to validate the marker. “Our role is not that of a clinical chemistry lab, but more of a biomarker discovery effort that attempts to understand the marker better,” he says. To that end, they made a whole new ELISA against a separate epitope of the EPCA-2 protein molecule. “You need to develop separate regions for sandwich assays,” he says. “Work with a separate epitope would also independently confirm our findings with the original epitope.” With the ELISA against this second epitope he was able to replicate the separation seen with the first epitope between prostate cancer and noncancer samples, including BPH, and between prostate cancer and other types of cancers (Leman ES, et al. Prostate. 2009;May 5. [Epub ahead of print]).

Dr. Getzenberg hopes to get a company interested enough to sponsor a multi-institutional clinical study with EPCA-2.

To define the setting in which an EPCA-2 assay would be most helpful, Dr. Getzenberg notes that, as with the other biomarkers discussed, most data show that EPCA-2 is specific for prostate cancer. “This would allow us to narrow in on who needs prostate biopsy from the 80 percent or so of men who get biopsies who don’t need them,” he says. So, as with the other biomarkers described, EPCA-2’s initial application would be the man with an elevated PSA level, and the question would be: Can you help him determine whether to go to (initial or repeat) biopsy? He calls this application “the lowest-hanging fruit to which we can apply this assay.

“However, this is not the greatest need,” Dr. Getzenberg recognizes, as have other researchers. He sees the more important application as separating “bad” prostate cancers from “good” ones—that is, the more aggressive prostate cancers from the indolent variety. At Johns Hopkins there is a large, active surveillance program numbering about 700 men with elevated PSA levels who have biopsies annually but don’t undergo “curative” therapy—neither radiation nor surgery. “Who should be in these programs?” Dr. Getzenberg asks. “We need a marker to separate the different kinds of prostate cancer. We can tell somewhat from Gleason scores, but we need a better way to say, ‘Your prostate cancer needs to be treated.’ To me,” he continues, “that’s the biggest need we have in the field.”

To find out whether current biomarkers can meet this need, Dr. Getzenberg has put in a proposal for a large NIH grant to study the Hopkins surveillance cohort. Biomarkers would include EPCA-2, PCA3, and the tissue-based TMPRSS2:ETS fusions. The urine fusions, though, will not be analyzed. “I’m not sure that the detection of fusions in the urine would be helpful in delineating prognosis,” he says. The one study that did show a positive result with fusion genes was done with FISH on tissue, not on urine, he notes. And, in his view, that finding has been countered by several studies in the other direction.

Even the best prostate screening test is unlikely to replace the PSA assay, at least not at first, Dr. Rubin says. “Whether that is the best approach is another question,” he says. But Dr. Rubin is optimis­tic: He can imagine a future in which a multiplex test plus PSA plus clinical parameters provide a better indication of which men to treat and which to follow.

Dr. Getzenberg, too, is hopeful. “With the number of new markers coming along, we will get there,” he says. “We are finally asking the right questions. And I think there are some markers out there that have the potential to answer them.”