Skip to main content
Premium Trial:

Request an Annual Quote

Predictive Biosciences Adds PCR to Protein Assay to Improve Performance of Bladder Cancer Test


By Bernadette Toner

Predictive Biosciences has found that adding a real-time PCR assay to its protein-based test for bladder cancer can improve its negative predictive value without reducing its specificity.

The company, which presented this finding at a poster at the 10th Annual Meeting of the Society of Urologic Oncology earlier this month, found that it could improve the negative predictive value of the test — the probability that a patient who has tested negative for cancer is actually free of the disease — from 94 percent with the protein-based assay alone to 97 percent with the combined approach.

The company is seeking CLIA certification for its lab, which it expects to receive by the end of January, and then plans to launch the test commercially "no later than the second quarter of 2010," Ellen Sheets, chief medical officer, told PCR Insider.

Predictive Biosciences, founded in 2006, has for several years been developing a non-invasive diagnostic assay to monitor disease-free status and cancer in high-risk bladder cancer populations. The test is based on matrix metalloproteinases, or MMPs, a family of zinc-dependent endopeptidases that have been associated with certain cancers.

Looking to further increase the negative predictive value of the test, the company recently added a real-time PCR-based assay for detecting mutations in fibroblast growth factor receptor 3 in urine samples. While MMPs are associated with high-grade, invasive tumors, FGFR3 mutations "are particularly associated with low-stage non-invasive tumors where sensitivity reaches approximately 70 percent," the firm said in its poster.

The combined approach, which the company calls Multi-Analyte Diagnostic Readout, or MADR, is an offshoot of its broader diagnostic-development strategy, called Clinical Intervention Determining Diagnostic, or CIDD. CIDD is an algorithmic method that the company has developed to identify biomarker cutoffs that maximize negative predictive value in order to distinguish patients who require no further intervention with very high certainty.

The company believes that with MADR, patients below a certain MMP cutoff would be excluded from unnecessary intervention, while those patients who are positive for an FGFR3 mutation would receive accelerated intervention.

Accentuate the Negative

Tony Shuber, chief technology officer at Predictive Biosciences, told PCR Insider that the company is focused on negative predictive value because it addresses a key issue in cancer treatment. "The reality in a clinical population is that the majority of people at any time will not have cancer. But yet we spend a lot of our healthcare dollars looking for that needle in the haystack," he said.

Predictive's products "are really going to initially focus on negative predictive value — and that is essentially debulking the population of those individuals who absolutely do not have cancer, and allowing the physician to truly triage that population into those that they can wait and see what happens, or not even treat because the result tells them with very high negative predictive value … that the patient doesn't have cancer," Shuber added.

Shuber noted that the company is targeting the test for two particular patient populations: those who have exhibited symptoms of bladder cancer, such as blood in the urine; and bladder cancer survivors. Predictive is currently conducting separate clinical trials for the test in the two high-risk patient populations, which it expects to wrap up next year.

The test currently has a specificity of 38 percent — meaning that 62 percent of true negatives would not be flagged as being cancer-free. Shuber noted that the high negative predictive value means that those patients who are flagged are almost certain to be true negatives, which is of great value in determining which of those high-risk patients will not require further intervention. He stressed that the test would not be of value for screening in a general population, and is not intended for that use.

[ pagebreak ]

The company plans to improve the specificity of the test, as well as its positive predictive value, by adding more biomarkers. Shuber added, however, that "we believe it's commercially viable today because of the patient population where we're applying the test."

He added that Predictive has already done some work to improve the positive predictive value of the test, but declined to provide further details because the firm plans to publish the study.

Quantitative And Binary

Shuber explained that the combination of the protein and DNA markers "takes advantage of the inherent properties of the analytes themselves — proteins being quantitative, and DNA being more binary."

Specifically, he said, "when one looks at proteins, you would set a quantitative cutoff to allow one to discriminate between certain clinical populations. But because DNA is binary — either there's the presence or absence of a mutation — if you layer DNA on top of the protein result, you can start to cherry pick … samples and move them out into the population of positives."

The concept of combining results from protein and DNA assays is not new. Clinicians often run both — say the NMP22 BladderChek protein test from Matritech and the UroVysion DNA-probe test from Quest — and determine whether the results are complementary.

However, "no one has combined them in the same assay," Shuber said. The benefit of a single assay is not only convenience for clinicians, but also improved results due to the company's CIDD approach, which generates "hybrid" ROC curves for the DNA and protein markers together.

With ROC curves, "usually you'll pick one spot that maximizes sensitivity and specificity," Shuber said. "We actually intend to take part of an ROC curve, which might result in extremely high sensitivities of one biomarker, and combine that with an ROC curve of another biomarker where there is extremely high specificity, and where there may be overlap between the two, we are actually calling that indeterminate."

Shuber said that the company is in discussions with several instrumentation vendors about developing a platform that would perform the protein and DNA assays concurrently on the same system, but Sheets, the CMO, noted that the company doesn't require anything other than an ELISA plate system and a real-time PCR system to conduct the assay in a CLIA setting.

Clinicians would split the urine sample into two parts — one untreated and one mixed with EDTA to reduce the breakdown of DNA, send it off to Predictive, and get "a combination result," Sheets said.

"So, yes, we'll be running two different types of assays on them, but the answer will give you a combination of the sensitivity of the protein biomarkers and the specificity of the DNA to be able to cover the entire gamut of bladder cancer — from low grade, low stage to high grade, high stage disease," Sheets said.

"If it's below our cutoffs for this test, the chance of a patient having cancer will be very low," she said. "If it's above the cutoff, for either the mutation or the protein that we look at, the chance of cancer being present will be markedly elevated compared to our competitors."

Another advantage of offering the test on existing platforms is that there are already CPT reimbursement codes for similar tests. "We're confident that the codes associated with our tests exist out there; they're already reimbursed," Sheets said. "They're not adding something new to the cost fee structure, and insurers should pay for this readily."

She estimated that the aggregate reimbursement for the combined test would likely be in the range of $300 to $350.

By comparison, UroVysion is in the $800 to $1,000 reimbursement range, while cystoscopy — the current standard of care for detecting bladder cancer — is also in the $300 to $500 range, but is invasive.

While the company is currently focusing on bringing the bladder cancer test to market, it also has several others in its pipeline that would also combine protein and DNA markers. Right now, Predictive is targeting a test for prostate cancer for some time in 2011 and one for breast cancer the following year, Sheets said.