With Ciphergen’s announcement that it’s selling part of its business, does this affect your relationship with the company?
Not really. What they’re going to do is sell off their instruments, the tools division. What they’re keeping is the diagnostics part. And basically they have licensed my [discovery of biomarkers for ovarian cancer]. And the stuff that they are working on is pretty much all my work. Basically they licensed that particular invention and they’re going to make it into a diagnostic test with some new improvements.
Is your advisory role at Ciphergen specifically about that device then?
No, no. I’m on the scientific advisory board so I advise them on any issue related to the company’s developments of products and anything related to scientific aspects.
When do you think the ovarian cancer diagnostic could hit the market?
It depends on how you define ‘hits the market.’ There are two ways to provide diagnostic tests for clinical use. One would be the more traditional way — you make a product, then you sell a kit, a reagent kit. For that often you do a clinical trial and get FDA approval and then you market it as a product so a clinical lab such as mine can purchase the product.
The other way to do it is to work with a commercial reference laboratory. Ciphergen has been working with Quest. And Quest is interested in making the test available to clinicians directly rather than selling it as a diagnostic test. So those two things may be going on at the same time as the company develops this particular test. If I had to predict, I would say the diagnostic test from Quest would be available first and then the product that you can buy would be a little later.
Johns Hopkins recently entered into a deal with PerkinElmer. What is your role in that partnership and what exactly will you be looking for and looking at?
I will be the primary scientist. What we are doing is looking at their new mass spec, the prOTOF which is a new generation of MALDI-TOF mass spectrometer. It has different capabilities than mass specs that Ciphergen has. I wanted to see if [PerkinElmer’s] mass spec has additional capability and improvements in terms of biomarker discovery and validation. For example, you can run the MALDI plate where you can put the patient’s specimen and then you can use the mass spectrometry to weed out the proteins.
And the high-resolution mass spec has advantages and disadvantages. The advantage is that you get better resolution. You get better peak detection, you have better sensitivity. The trade-off is that you will not be able to read high molecular-weight proteins. Most mass specs concentrate on low molecular-weight proteins or fragments or peptides. So we want to find ways that allow us to see more of the proteins, [has] better sensitivity, better resolution. At the same, we hope to be able to look at some proteins with higher mass as well.
Is prostate and ovarian cancer the two diseases where you’re focusing most of your research?
Prostate cancer is the number one cancer in men so we wanted to see what we can do about that. In the case of women, breast cancer actually has many more cases, so we are working on breast cancer as well. But the reason we are working on ovarian cancer first is we think we can make an impact on ovarian cancer because often ovarian cancer is detected in very late stage.
Breast cancer, at least you have mammography. It’s not perfect but you have kind of a screening device. In ovarian cancer, we don’t. Even though CA-125 is used as a blood test, it’s mostly for monitoring the patient’s disease. We’re hoping to find a new marker for ovarian cancer early on, and if we can detect ovarian cancer at Stage 1, the five-year survival rate for ovarian cancer is greater than 80 percent. But right now, most of the cancer is detected at Stage 3 or Stage 4, where the five-year survival rate is less than 20 percent.
Prostate cancer is a different issue because we have PSA. Again, it’s not perfect. A lot of men get PSA screening, so we are able to detect prostate cancer early. The problem is this: We detect too many cancers, in a sense, because you detect all this small amount of cancer. And you don’t know which of those men are going to die actually of prostate cancer. When you become old, when you become 80 or 90 years old, chances are that you will have prostate cancer. You just won’t know [that you have it]. And the cancer can be very small. Maybe you will die a month later or in an auto accident or something. The prostate cancer is not going to kill you.
What we’re hoping to do is to find those [incidents of] prostate cancer that likely [are] going to progress and kill the patient versus those types of prostate cancer that we don’t have to treat at all. We’re hoping that we can find panels of biomarkers, not just one, but a group of biomarkers that can tell you which kind of cancers that are maybe more aggressive, which are less aggressive.
Do these panels actually exist, it’s just a matter of finding them, or do they theoretically exist?
I believe they exist. It’s just how difficult will it be to find them and how long will it take to find them. I don’t think it’s theory because if you think about it, each type of cancer, even within the same cancer, there has to be some differences. Otherwise, how would one kind of cancer kill the patient while the other one, the patient lives for many years? There has to be some difference.
How would you assess the proteomic field in terms of breakthrough research and movement toward clinical applications?
Last month I gave a [speech] at the AACC. They asked me to talk about the current state of clinical proteomics. The first slide I showed was [a picture of ‘Where is Waldo?’]. And I showed it to the audience, saying, ‘Where in this picture is the biomarker for clinical use?’
To summarize, that’s the current problem that we have. You can talk to many people. They will say, ‘I improved the technology, I improved the process. What’s wrong with this person’s study, that person’s study?’ But at the end of the day, [what matters is] what biomarker did you find that will make an impact on patients. I think we are lacking those biomarkers right now. And that’s why the challenge for me and for all of us in the area of clinical proteomics is to find clinical use for biomarkers.
We are limited by number one, technology; number two, the kind of clinical specimens that we use; and number three, knowledge. But the knowledge is not going to come unless you try it and do experiments.
I would say that in the beginning, [the problem] was technology issues —technologies that were not sensitive enough, not reproducible. You have to be able to find the same markers. So we’re trying to improve the technology such as mass spectrometry, protein chips, and the process so that we can always consistently detect certain biomarkers.
Another technology that we’re trying to improve is bioinformatics. The problem with proteomics and mass spec is that we see many peaks in a mass spec. Or you look for many proteins, but which one? Which one is the real biomarker? What we’re trying to do is develop bioinformatics tools that will allow us to select those peaks in a mass spec or the proteins and to narrow it down to a small subset that has clinical application.
Specimen-wise, fortunately, since I also run a clinical lab, I have a quite a bit of patients at John Hopkins [from whom] we can get specimens. But we also learned in the last few years that in order to discover biomarkers, you have to treat all the specimens identically so that there are no artifacts, there are no variations.
So I would say yes, we’ve been making progress in the last few years, and I think we have a better understanding about this whole process and the potential problems associated with it and how to minimize false discoveries.