At A Glance
Name: Brad Guild
Position: Director, protein sciences, Millennium Pharmaceuticals, since July 2003.
Background: Director, discovery proteomics, Millennium Pharmaceuticals, 2002-03.
Director, protein biochemistry, Millennium Predictive Medicine, 1999-02.
Director, functional genomics and other positions, Genome Therapeutics, 1995-99.
Director, molecular biology, ImmunoGen, 1989-94.
Founding scientist and head of gene expression, Somatix, 1988-89.
Post-doc, Whitehead Institute for Biomedical Research, 1985-88.
Post-doc, Center for Cancer Research, MIT, 1984-85.
PhD in biochemistry and molecular, cellular, developmental biology, Harvard University, 1984.
BS in biology, Northeastern University, 1976.
How did you get involved with proteomics?
I was trained as a protein biochemist, and about four years ago, this interesting opportunity came up at Millennium to work at what was then a subsidiary known as Millennium Predictive Medicine. It specifically called for someone to come in and build a mass spectrometry group to look at protein biomarkers. That was exactly what I wanted to do, although I must say I only knew mass spectrometry by word when I first started.
What made you interested in protein biomarkers?
Being trained as a protein biochemist, I really wanted to use mass spectrometry to try to begin to characterize in patient body fluids proteins that could be used for better diagnostics. We did not have the deal with Roche Diagnostics in place when I started; it was particularly exciting when that came along. What Roche and Millennium wanted to do was look at synovial fluid and ask, ‘if you go to the site of the joint destruction, would you have a better chance of finding protein markers of disease than if you just went in the peripheral blood to look for those markers? Would they be in higher concentration in synovial fluid?’ This was a great test of many questions: Can you find such markers in synovial fluid; can you do that using mass spectrometry; and if you do find prospective candidate markers there, then can you take that information and translate it into a good candidate marker that will appear in the peripheral blood where most physicians want to perform a diagnostic test? There were no clear-cut answers to those three questions when we started the project with Roche.
When you arrived at Millennium, you were charged with building the mass spec proteomics group?
Yes. The beginnings were a key scientist who was a mass spectrometrist, [and] a couple of pieces of instrumentation — an ion trap, a triple quad. Then once we started thinking about how we would structure deals in this area, [we thought about] bringing more of these specialists on board, and complementing that mass spec effort with some good protein biochemists as well. What we perceived then and have found to be true, is that your ability to get very precise and quantitative information on protein biomarkers in body fluids is complemented by the quality of biochemistry of the sample preparation that you do up front. So it was my charge to try to build a small but efficient team of biochemists and mass spectrometrists who could give this an element of throughput. We’d like to be able to analyze a sufficient number of clinical samples to be confident that [what] we identified in this biomarker discovery portion of the study would ultimately be something we could validate in statistically meaningful populations of patients. We [wanted to] take the information we gathered from discovery using mass spec, and then turn it into ELISAs and immunoassays that you could use on hundreds of patient samples.
So is it better to aim for an ELISA test as a diagnostic rather than use a mass spec?
I think mass spec may be the tool of choice in the coming years, but right now there is so much power and confidence in using immunoassays and ELISAs, that I suspect it’s the most practical way to go for the next five years and maybe longer. Ultimately I believe that mass spec is a very powerful tool, but I think developments need to be made in increasing the sensitivity of the instruments, increasing the reliability, and making sure that we can make mass spectrometry as quantitative as possible, in the sense of absolute quantification. You can get an absolute quantification with an ELISA — it’s much harder to do that by mass spectrometry.
How many people and what infrastructure do you have now?
When we did the Roche work, [we had] about seven people. Now my team is larger because I run protein sciences, and we have over 20 people.
At one point, we had five ion traps that we were using in candidate marker discovery. We complemented that with two triple quad mass specs. We found that with the seven people [on our team], we could do a lot in terms of sample throughput. We didn’t need dozens of samples to get pertinent information about prospective candidates. We could work with very few, because the total amount of information that came out of any one run was so rich, that we could quickly take information on candidates from the ion trap data and convert that into validation or verification experiments on the triple quad. That’s taking the jump from synovial fluid now to serum. So we can say, ‘we see this particular marker candidate coming up again and again in synovial fluid samples; let’s go in and start looking at serum samples.’ And using what we call this MRM mass spec approach, [we could] try to get a semi-quantitative measurement of those proteins in the peripheral blood of patients.
Tell me about the study with Roche.
We looked at a couple dozen synovial fluid samples from patients who had rheumatoid arthritis. Those samples were divided roughly into two groups: patients who had erosive RA, and patients who had non-erosive RA. We characterized these two groups of samples against one another in a differential profiling analysis. We added up all of the proteins that we identified in the synovial fluid proteome of erosive patients, versus all of the proteins that we found in the synovial fluid proteome of the non-erosive patients and we said, ‘Where are the differences?’ When we saw markers that were distinctly abundant in patients with erosive disease, those were the ones that we put on a list of candidate markers. So having built this list of candidate markers from synovial fluid studies, we could then move on and ask another critical set of questions which were, not only are these markers expressed abundantly in the serum of people with erosive RA, but, are they less abundant in patients with non-erosive disease, and are they absent or very low in serum of healthy individuals? That’s where we introduced the important control of healthy individuals because we did not have access to healthy synovial fluid when we were doing the initial profiling.
So what were your initial conclusions?
What we came up with from that was remarkable — we found about 36 candidate markers in the synovial fluid of patients with RA. A number of those were in patients with erosive disease and a few of them were in patients with non-erosive disease. Then we began to work our way through that candidate list and asked, ‘which ones appear to be the most abundant in serum?’ We found a group of markers that were clearly enriched in the serum of patients with erosive RA or aggressive RA over patients who were categorized as having non-erosive disease, or healthy individuals. So what we now are excited about is the idea that hopefully, some of these markers will stand up as markers of disease severity. When people get RA, a lot of damage can occur in the first two years of the disease — there can be a tremendous amount of cartilage lost, and then bone erosions. So if one could identify very early on the patients at the highest risk for an aggressive or more severe form of RA, those are the patients you would want to give often the most expensive and most risky but efficacious agents.
So you’re currently validating the markers?
We’re validating them. We’ve done at Millennium a small-scale validation in a limited set of serum samples to convince ourselves as a research team that this subset of markers looks to be selective for erosive RA. What will take place in the coming years at Roche will be a further development of these ideas by building immuno-assays that will allow our collaborators to look at hundreds of patient samples, and then to truly add statistical validation [on] whether any of these markers can hold up as good markers of disease severity. That’s pretty much where we stand with the relationship at this point.
Will you apply this method to other diseases as well?
Yes. We took all of the methodology that we built around marker discovery in synovial fluid and began to apply it to serum. And we found we could also do protein biomarker discovery in the serum of patients. So we’re now using it at Millennium in other programs, specifically in inflammation and oncology.
What questions in biomarker work need answering?
Do we use a single marker, or do we work with multiple markers? What is a good biomarker and how do you measure that? How careful have you been to make sure that the variation that you’re seeing isn’t just due to polymorphisms? I think it’s going to be very important for pharma to work with the FDA in defining what FDA would like us to provide them in the way of information on how we’ve gone about characterizing this biomarker.