As reported by the Journal of the National Cancer Institute, October 8, 2010.

PSA Kinetics: No Substitute for Annual Biopsy in Prostate Cancer Surveillance?

By Charlie Schmidt

Medicine treats more and more men who have early-stage prostate cancer with active surveillance, whereby clinicians monitor their illness and intervene only when the pathology report indicates a worsening risk profile. But while doctors have some criteria for putting men on active surveillance, criteria for going off surveillance—i.e., deciding when the cancer needs treatment—remain ambiguous.

In the June 10 Journal of Clinical Oncology, Ashley Ross, M.D., Ph.D., and colleagues from the Johns Hopkins University School of Medicine in Baltimore reject one promising biomarker—prostate-specific antigen kinetics (PSAK)—as a substitute for annual biopsies. If the study had been positive, it might have allowed nearly a quarter of men treated under active surveillance protocols to avoid these biopsies.

“According to our findings, the best way to see if someone needs treatment is to conduct a surveillance biopsy, and at this point, PSAK can’t supplant that,” said Ross.

Ross and his colleagues limited their analysis to very-low-risk patients, which the National Comprehensive Cancer Network (NCCN) defines as having stage T1c tumors, Gleason scores of 6 or below, PSA levels less than 10 ng/mL, fewer than three positive biopsy cores containing at most 50% cancer, and a PSA density (i.e., the amount of PSA per unit volume of prostate tissue) of less than 0.15 ng/mL/g. The researchers analyzed data from 290 patients on active surveillance from 1994 to 2008, looking for associations between biopsy results and PSAK—a measure including PSA doubling time (the time needed to double a baseline value) and PSA velocity (the change in levels from year to year). They found no association between pathology results and PSA doubling time. For PSA velocity, they report only a “marginally significant” association (P = 0.06). The authors concluded that PSAK can’t predict adverse pathology and therefore shouldn’t replace annual biopsies for very-low-risk men on active surveillance protocols.

The finding affects a large population. Very-low-risk patients account for about 23% of patients eligible for active surveillance, who along with a separate “low risk” NCCN category, make up 60% of all new prostate cancer diagnoses, according to Ross.

Not the Last Word

These results are among the first to emerge from long-term studies of cancer progression in patients under active surveillance. Peter Carroll, M.D., professor and chair of urology at the University of California, San Francisco, Medical Center, who co-authored a commentary about the Ross report, said that all these studies suffer from methodological shortcomings. In the Johns Hopkins study, for instance, patients were taken off surveillance with the appearance of Gleason pattern 4 (leading to Gleason scores of 3x3 = 7). But this is an endpoint that Carroll says is arbitrary for cancer progression. “PSAK might associate with other important biological endpoints, but these weren’t evaluated in the study,” he said.

In Carroll's view, the Johns Hopkins finding can’t be taken as the last word. Peter Albertsen, a urologist and professor of surgery at the University of Connecticut Health Center in Farmington, agreed. According to Albertsen, Gleason scores have become “inflated” by pathologists who increasingly err on the side of caution. Patterns that were once graded routinely as 1 or 2 are increasingly graded as 3 or 4, he said, with the result that Gleason 7 scores are becoming more common. “But does the grade 7 tumor always lead to a bad outcome?” he asked. “That's an open question.”

That PSAK didn’t correlate with low-grade progression in tumor biopsy isn’t entirely surprising, adds Peter Nelson, M.D., a professor at the University of Washington School of Medicine in Seattle. A range of factors influence the level of PSA protein, which leaks from the prostate into blood under clinical scenarios that include benign prostate hyperplasia in addition to cancer.

The marker works best after radical prostatectomy, when PSA levels can fall to zero after the prostate's removal. Postsurgical detection of PSA reveals the presence of runaway prostate cancer cells somewhere in the body. “That's where PSA really shines,” Nelson said. “It's a fantastic indicator for metastasis.”

The Case for PSAK

Speculation that PSAK might predict cancer progression grew after Anthony V. D’Amico, M.D., Ph.D., a professor in radiation oncology at Harvard Medical School in Boston, published findings in 2004 showing that a prediagnostic PSA velocity of 2.0 ng/mL/year was associated with prostate cancer mortality, even after radical prostatectomy. (Ballentine Carter, M.D., a professor of urology at Johns Hopkins and a coauthor on Ross’ article, later found that an even lower PSA velocity of 0.35 ng/mL/year predicts prostate cancer mortality.)

Those studies were accompanied by others suggesting that PSAK and cancer progression are statistically correlated. Laurence Klotz, M.D., a professor of surgery at the University of Toronto, found that among 450 patients monitored on active surveillance for up to 13 years, a PSA doubling time of 3 years or less predicted biochemical failure—that is, PSA elevations after treatment with surgery or radiation. NCCN's guidelines for active surveillance later incorporated Klotz's findings, which state that if doubling time is 3 years or less, cancer progression is possible.

In Carter's view, however, these results were biased because PSA doubling time was a trigger for intervention. Klotz removed patients from active surveillance and treated them on the basis of PSA doubling time (in addition to adverse biopsy). So although the doubling time may have predicted biochemical failure after treatment, it couldn't be said to directly predict the need for treatment.

To avoid that problem, the Johns Hopkins team deliberately excluded PSAK as a criterion for going off active surveillance. Patients underwent both PSA testing and digital rectal examination every 6 months, but they were treated only if their annual biopsy revealed an increase in cancer grade, regardless of PSA changes. According to Ross, that approach removed the bias of using PSAK as an intervention trigger. It also ultimately revealed that PSA changes were not related to cancer progression in these very-low-risk patients.

Asked why that might be so, Ross responded that benign tissues in the prostate generate “a noisy background,” that is, considerable variation in PSA results, which obscures potential indicators of worsening disease. Ross wouldn't comment on how the results might apply to low-risk patients, who weren't included in the study.

Yet Carroll emphasized that by limiting their analysis to the very-low-risk group, the Hopkins researchers excluded other patients among whom PSA and cancer burden might be linked. More specifically: Carroll criticized using PSA densities lower than 0.15 ng/mL/g as a criterion for inclusion in the Hopkins study, pointing out that values less than that would be considered normal in patients without prostate cancer.

Klotz, whose cohort included both very-low-risk and low-risk patients, sticks by his approach. He acknowledged that he can’t validate whether PSAK reliably predicts the need for treatment. “Because, as they state, we use PSAK as a trigger for intervention,” he said. Nonetheless, Klotz pointed out that in the nearly 13 years since his study began, virtually none of the subjects has died of prostate cancer. “The limitation is that you wind up treating roughly 20% of patients on the basis of PSAK, and some of those guys probably didn’t need to be treated,” he said. “So it's not perfect, but for us PSAK is working well.”

Looking for New Clues

Still, PSA's shortcomings—namely, that it's not cancer specific and that it's influenced by other factors, such as hormonal status—frustrate scientists looking for better clues into cancer progression. “What the [Hopkins study] tells you is that PSA just isn’t a great biomarker,” Nelson said. “It gives you some rough insights from a diagnostic or treatment standpoint, but it falters in its ability to give you reliable information on a per-patient basis. It's OK for research with large groups, but not for individuals.”

Nelson leads the Prostate Active Surveillance Study, a multi-institutional investigation funded in part by the NCI, which aims to find better progression markers for prostate cancer. Eight institutions now participate in the study, he said, all collecting a wide variety of biological specimens—DNA, urine, blood, and biopsy tissue—from patients on active surveillance. He said two new biomarkers look promising: a genetic marker in urine called prostate cancer gene 3 (PCA3), which only prostate cancer cells produce and is unaffected by the size of the organ, and the TMPRSS2–ERG gene fusion, which occurs in roughly half of prostate cancers and which is detectable in tissue and urine.

“The nice thing about urine markers is that they run right through the prostate,” Nelson said. “So it's a good readout of what's going on in the whole organ rather than just the biopsy sample.” Similarly, imaging biomarkers might also offer noninvasive views of the entire gland, Nelson said, and thus limit the possibility of missing clinically meaningful cancers during biopsy.

Meanwhile, PSAK and cancer progression remains an evolving story. “We’re still not sure how to follow men on active surveillance,” Ross said. “It's a matter of figuring out the ground rules.”