Chief Scientific Officer
Windber Research Institute
At A Glance
Name: Michael Liebman
Position: Chief Scientific Officer Windber Research Institute
Background: Director for Biomedical Informatics University of Pennsylvania Abramson Cancer Center
Education: PhD in Physical Chemistry Michigan State University, 1977 Post-doctorate University of Wisconsin, 1977-1978
When Pharmacogenomics Reporter spoke to Michael Liebman, he was attending the Innogen Evolution of the Life Sciences Industries Conference conference in Edinburgh, Scotland, explaining the system that he and others have set up to catalogue samples and microarray data at the Windber Research Institute. A bioinformatics expert Liebman is the author of more than 60 journal articles and several books, and sits on 20 advisory boards and three editorial boards. Windber and its partner, Walter Reed Army Hospital, are currently shepherding a breast cancer vaccine through clinical trials
You said you'd been talking to a lot of cancer researchers at the Innogen conference. What about the Windber are they interested in?
We're having a discussion about what we're doing at Windber being a template for the cancer centers in England. They have a group of about 14 cancer centers, and we're talking about sharing our capabilities in informatics and tissue collection to set up some of the databases that they need for integrating genomics, proteomics, and clinical data.
How do they hope to use that?
Well, the idea is to really be able to focus in on the perspective that we have at Windber is a little bit different from most institutions, because we worry about the patient first, and so we're really stratifying the disease as it presents in the patient, and then using that to reduce the "noise" in the grouping, so when you add the genomics and the proteomics, you get a better signal as to what some of the mechanistic parameters are.
Only in cancer?
In cancer, but we're working across multiple diseases now, but our main effort right now is in breast cancer, cardiovascular disease, obesity, and diabetes. But we've been expanding it and generalizing it, and that's one of the reasons they're interested.
How many patients are in the Windber database?
Right now we've got about 2,000 patients, 14,000 samples, and the only thing that's restricting what we have in the database is the complications of all the data entry right now. So, we work closely with Walter Reed Army Medical Center, and they see about 10,000 patients with breast disease a year, and we work with them to try to bring all of that data in, as well as all the samples.
This is all directed toward treatment and research, I assume.
Well, it's directed at improving patient care, which means, 'How do you get to earlier detection? How do you get to better patient stratification? How do you understand what subtype of disease the patient has? How do you stage the patient more accurately?' And all of that leads to better treatment, of course, and better quality of life.
How long was this system at Windber in the making?
The Institute really started in 2001 with a Congressional relationship with Walter Reed that was established with Congressman Murtha [?]. This is an ongoing process. We continue to evolve it today, because science when it's done right should generate questions, not just answers. And so we're always learning new things we should include.
What's different about how things work and Windber and what the UK wants to do?
We take a very strong patient focus. And we're very, very focused on the proposition that translational medicine, which has always been from the bench to the bedside, has to really occur in both directions. So, the basic scientists, who don't really have a good understanding of what the clinical problems are really need to have that translated to them, so that they know what they should be working on not just working on things that don't translate well to the clinic.
So, we work very closely with our Walter Reed partners, who are all clinicians - surgeons, oncologists, and pathologists - to feed back to the basic scientists what the questions are that they have in their daily practice. What are the things they have the least confidence in - whether it's treatment, or diagnosis, or outcome, or recurrence, or whatever - and then, 'how can we apply our technologies and our science to try to address those problems?'
So, because of that, we're very much more 'mission oriented' than you normally have in an academic department. And we're focused on problem solving by whatever combination of tools is necessary to accomplish that. So, we don't necessarily tackle each problem with the same set of components - although we have a very broad range of them.
And that's something that is of interest [to UK health authorities], because it is different from the way that most academic groups perform.
What are your major missions at the Windber right now?
Are major missions are really: understanding what the issues are around the heterogeneity of breast disease; are we really diagnosing the patient correctly, when we look at severity, without looking at heterogeneity; what are we doing about recognizing that aging is a background to disease - especially chronic disease - and therefore disease at different ages may be very different, not just because the host is different, but it may be that the disease itself is also different, and therefore needs to be considered in coming up with treatments that are better tailored to what the disease presentation is; and then how do we take all of this and feed it back into coming up with better complements of diagnostics that are both biologicals, as well as things that are imaging - because we work very closely with General Electric Healthcare right now on all of their imaging modalities to find out what it is that is really appropriate at what age or what stage of disease or subtype of disease to give us the best component of information.
What is the Institute's next milestone, and when does that occur?
Well, our next milestone is probably the completion of our full data integration with all of our genomic and proteomic data. We're really very critically looking at some biomarker development, where we're taking and contrasting the genomic information from microarrays with the proteomic information, and we've established some very unique relationships to start to look at extremely low-abundance proteins, and try to identify, really, how the unique diagnostic information that we have - which comes from the fact that all of our samples are reviewed by a single pathologist, which is really unique for a sample collection that large - to develop a very detailed profile of what heterogeneity in breast disease really means.
So this effort focuses on breast cancer?
That's our prototype disease. We're doing it for breast cancer, but everything we're building is fully modular, so it's extensible to any disease, even non-cancers.
You'll move on to other diseases once you've finished with this integration? When will you be finished with this first step?
The integration is underway now. We're actually doing automated data capture from our molecular experiments with our clinical data, and then the next step that's pretty exciting is, we're just about to start adding our imaging data from mammography, ultrasound, PET-CT, and MRI exams of the patients. So, we're going to have a really, fully comprehensive picture of what a patient is from the clinical and the molecular perspective, to enable us to really start to reason about what the diseases are.
You use that information for research in-house and in collaborations, correct?
Absolutely. So, while our main collaborator is Walter Reed and Landstuhl [?] Air Force base in Germany, and Bowling Air Force Base in Washington [?], and the Joyce Murtha Breast Center [?] at Windber - we're in the process of expanding that to six additional military sites. And then we have non-military collaborators, like the University of California at San Francisco's breast center, and we have other projects going on with the University of Pittsburgh, Georgetown University, the University of Nevada [at] Las Vegas, the University of Hawaii.
So we have a number of other scientific programs going on, that try to optimize the information from each site.
Are there diagnostic companies involved?
We've been approached by a number of companies, but the thing that were very, very specific about is that we're not purely a repository for tissue to give to or sell to different sites. We only work with pharmaceutical or diagnostic companies in strategic partnerships. I spent 10 years in the pharma industry, and I know from what we used to get from when I was in pharma, that pharma isn't yet able to appreciate the depth of annotation we have. We get [more than] 500 fields of information from each patient. And because they're not in a position to really take advantage of that yet, we don't want to just give the tissue away for what I would call 'primitive experimentation.' And that's why we're pretty specific about working through partnerships.
What partnerships are going on right now?
We have four that are under discussion. Two with pharma, two of them with some major disease foundations - where we're in discussions with them about becoming their central repository for all of their research studies. And helping them develop intellectual property out of that as well.
Can you tell me who they are?
I can't, but I can tell you that one is in the breast area, and one isn't.
As far as genomics and proteomics technologies, what are you using?
What we've done in genomics is that we have both the Affymetrix platform and the CodeLink platform for gene expression, and we also do sequencing using the Megabase for genotyping and sequencing. But our proteomics is a little different, because when I joined, I took our proteomics group and constructed two programs out of that - one on separations, and the other on mass spectrometry and protein identification - because I saw it was critical. We get samples from tissue, serum, plasma, urine, and cell lines that we optimize for each kind of sample what the appropriate separation technology is to give us the right ranges of concentration, and then how to optimize for each of those subclassifications of the sample, the right combination of mass spectrometry. And right now we've got four mass spectrometers - so we have a wide range of technology - but we also have a collaboration with the University of Pittsburgh that's looking at a broader set of mass spectrometry across the region.
And then we are just in discussions with the Pacific Northwest National Laboratory [?], one of the [US Department of Energy], and that has extremely high-resolution, high-magnet field, very unique capabilities in mass spectrometry. And they were looking to see, 'What is the detection limit on the low-abundance proteins that may be critical?'