Skip to main content
Premium Trial:

Request an Annual Quote

Emanuel Petricoin Searches For a Prince on The Road to Ovarian Biomarkers Clinical Trials


When it comes to developing an ovarian cancer biomarker test, Emanuel Petricoin will do anything it takes to find the cleanest, most consistent model — even if it means kissing frogs.

As the co-director of the FDA-NCI clinical proteomics program prepares to send his mass spec pattern-based diagnostic test, which first made a splash in the February 2002 issue of The Lancet (see PM 2-18-02), into two separate clinical trials at the end of this year, he and his colleagues are first “trying to find our prince among the frogs,” Petricoin told ProteoMonitor this week. That means identifying the best combination of instruments and protocols to produce a high level of accuracy and reproducibility across platforms. “You have to be absolutely sure that the spectra [making the patterns] are reproducible,” Petricoin said. “Or else your house of cards falls apart.”

That reproducibility is a central problem evident from some of Petricoin’s recent work. At the June meeting of the American Society for Mass Spectrometry, Petricoin’s colleague Thomas Conrad introduced a paper describing a repeat of the original study in which ABI Q-Star mass specs replaced Ciphergen’s SELDI mass specs to successfully detect more precise — but different — ovarian cancer proteomic patterns.

That reproducibility is a central problem evident from some of Petricoin’s recent work. At the June meeting of the American Society for Mass Spectrometry, Petricoin’s colleague Thomas Conrad introduced a paper describing a repeat of the original study in which ABI Q-Star mass specs replaced Ciphergen’s SELDI mass specs to successfully detect more precise — but different — ovarian cancer proteomic patterns.

Petricoin is determined to iron out these differences. “A tremendous amount of our work right now is quality control measures,” he said. Upgrading mass specs was an important first step. Since switching from SELDI to the Q-Star, the models “got much much better” with 100 percent accuracy achieved in some runs, Petricoin said. The Q-Star machine’s advantage is that it not only generates more peaks overall, leading to higher specificity, but it also generates fewer background peaks. “SELDI tends to generate a lot more fragments because the mass analyzer is directly coupled to the source,” Petricoin said. Still, Petricoin has not ruled out retaining the use of SELDI chips, which he used along with the Q-Star in the recent experiment. He is also interested in trying out electrospray, using Advion’s source.

For FDA approval, Petricoin will have to show that the results from a sample run on one machine will be classified the same way on at least two other machines, across multiple time periods. This doesn’t mean that there can’t be any variability across platforms, but that variability must be recognized and controlled for. And that involves more than just controlling for different machines’ sensitivities. It also requires developing effective control measures for every step of the process.

“It’s an onion of a problem — layers of problems at the clinic, chipping, software reliability that you have to address, the way the samples are handled and processed for the chip — every step introduces variability,” Petricoin said. “So basically, the least number of steps you have, the better. A bicycle works almost all the time because it has a lot less parts than a spaceship.”

Even after the kinks are ironed out, the ovarian biomarker pattern tests still will not be applied in a large scale, long-term general screening study anytime soon. This is not because the FDA treats pattern-based tests any differently than it would treat any other test using new technology, according to Petricoin. “A lot of people ask us [if that’s the case]. But the FDA looks at all things on a case by case basis,” he said. Given that, why not launch a big trial of thousands of people right away? “The knee jerk reaction is to say, ‘let’s start having big screening trials’,” Petricoin said. “But we don’t want to necessarily wait a decade to see if the technology has any promise.”

In one sense, they won’t have to wait at all. Petricoin’s partner, Correlogic Systems of Bethesda, Md., last November already licensed the pattern test to Quest Diagnostics and Laboratory Corporation of America Holdings for homebrew use. The laboratories can use any pattern test based on Correlogic’s algorithms and modeling techniques in their work, so long as they don’t sell the technology to anyone else.

The FDA will also consider approving Petricoin’s version of the test based on two small trials set to begin within four to six months from now. Both are expected to be completed within 18 months of their start dates. In one trial, Petricoin’s biomarker test will be used alongside the only existing ovarian cancer biomarker, the rather unreliable CA125, which has a positive predictive value of less than 10 percent. The trial will look at whether the pattern test as equally effective or better than CA125 at predicting whether cancer will recur among women who have already been successfully treated, and are now in danger of relapsing. Since 80 percent of ovarian cancer patients relapse within 18 months, this highly enriched population will provide very quick answers.

The second trial will couple the pattern test with an ultrasono-graphy imaging screen, again looking at patients at risk for recurrence. Patients who are suspected of having a malignancy based on the imaging test will all be biopsied, but Petricoin and his team will look retrospectively at whether the pattern test could have distinguished a benign growth from cancer. The idea is to eventually be able to prevent unnecessary biopsies, which can sometimes lead to morbidity. “This is not the slam dunk clinical benefit of detecting disease early, but this is an important baby step,” Petricoin said.

An obvious next baby step, for many scientists, would be to identify the proteins that make up the patterns. “I think in the end it would be better to have the identities of these proteins,” said David Springer, a staff scientist at Pacific Northwest National Laboratory who also works on the proteomics of ovarian cancer. “We should be able to get to the bottom line and understand where they’re coming from and what’s different.”

This is something Petricoin is working on, but he insists that identifying the proteins and finding antibodies to them is not the best way to think about cancer biomarkers. “We need to stop thinking that cancer is a foreign disease,” he said. “It is your own body cells acting like mini-terrorists — it doesn’t make a unique protein.”

Because the proteins that make up the pattern are mostly truncated versions of naturally occuring proteins, finding a specific enough antibody for diagnostic purposes is an “improbability that is not likely to happen.” For this reason, mass spec-produced patterns, Petricoin believes, are a better model for cancer biomarkers. Springer doesn’t disagree. “I think there’s truth to that,” he said.

In any case, the pattern-based approach is picking up steam. Last week at the American Association for Cancer Research meeting, data was presented by NCI-affiliated researchers showing that similar pattern-based serum tests, when applied to prostate cancer, could classify 70 to 80 percent of benign lesions as benign without doing an unnecessary biopsy. Unnecessary biopsies often occur when PSA, the prostate cancer biomarker in use now, is in the “diagnostic gray zone,” Petricoin said. Projects are also in the works to find biomarkers for breast and lung cancer that could be used to verify or refute image-based mammography or spiral CT tests. “This will probably be the next big thing to look at after ovarian cancer,” Petricoin said.


The Scan

Pig Organ Transplants Considered

The Wall Street Journal reports that the US Food and Drug Administration may soon allow clinical trials that involve transplanting pig organs into humans.

'Poo-Bank' Proposal

Harvard Medical School researchers suggest people should bank stool samples when they are young to transplant when they later develop age-related diseases.

Spurred to Develop Again

New Scientist reports that researchers may have uncovered why about 60 percent of in vitro fertilization embryos stop developing.

Science Papers Examine Breast Milk Cell Populations, Cerebral Cortex Cellular Diversity, Micronesia Population History

In Science this week: unique cell populations found within breast milk, 100 transcriptionally distinct cell populations uncovered in the cerebral cortex, and more.