Chief of clinical chemistry, Department of Laboratory Medicine
National Institutes of Health
At A Glance
Name: Glen Hortin
Position: Chief of clinical chemistry, Department of Laboratory Medicine at the National Institutes of Health, since 2000. Acting, assistant chief of clinical chemistry, since 1997.
Background: Associate professor of pathology, University of Alabama at Birmingham's clinical chemistry section, 1993-1997.
Assistant professor in pediatrics and pathology, Washington University, 1989-1992.
Last week, Glen Hortin gave a talk on "Translating New Biomarkers into Clinical Laboratory Tests" during an American Association for Clinical Chemistry meeting in Washington, DC. ProteoMonitor caught up with Hortin to find out about his advice on this issue, and about his research background.
How did you get into developing biomarkers and clinical tests?
Originally, I was interested mostly in protein chemistry. I started in the 1970s looking at protein biosynthesis. As an undergraduate and graduate student, I was looking at processing events like the processing of secretory proteins and glycosylation of proteins. As a graduate student I worked on the basic aspects of a signal hypothesis in terms of how secretory proteins are processed.
Then I was interested generally in the issues of protein processing post-translational modifications and molecular recognition between proteins, how various proteases or substrates would find their right substrates or protease to inhibit how they got together. Simultaneously I was getting interested in clinical chemistry. Those are interests of mine that go back about 20 years.
Which laboratory were you working in?
I was working with Irving Boime at Washington University in St. Louis. I continued on as a faculty person. Arnie Strauss helped me get going he had done a lot of work in the protein processing area. He was a pediatric cardiologist, but also did basic science.
Before I got started on my graduate work also, I had the good fortune to work with Donald Steiner at the University of Chicago he was the fellow who discovered proinsulin. At the time, he was studying the precursors of proinsulin. I spent a couple of summers in his lab as a student. It was an interesting entré into protein chemistry.
I went to medical school, and I did a residency from 1983 to 1987 in clinical pathology. My main interest was in the clinical chemistry end of the laboratory.
What types of problems were you working on?
In clinical chemistry, you tend to get interested in a wide range of problems whatever problems come up in the laboratory. It's really kind of problem solving and problem focused. In the clinical chemistry area, probably half of the assays we deal with relate to protein measurement some of them are enzyme assays, some of them are immunoassays, some of them are simple electrophoretic assays. Basically, in the clinical laboratory, a major portion of the protein assays are really kind of within the subspecialty of clinical chemistry. Another section is the immunology section.
Were you working on enzyme assays?
Well, I did a fair amount of work in looking at protease specificity. From a practical standpoint, you're often in a problem solving mode in the clinical lab to address the patient needs. In general, I've had a more basic research lab effort that's kind of gone in parallel. Part of that effort was directed at studying post-translational processing of proteins. I spent a fair amount of work on sulfation of proteins.
Where did you go after Washington University?
I stayed on as faculty there for about five years, and then I went to the University of Alabama in Birmingham. A greater amount of my time there was spent doing clinical work helping to run a clinical laboratory. I got involved fairly actively in terms of point-of-care testing, which is a lot of the testing done at the patient's bedside.
The things done at the bedside are usually things like glucose testing, electrolytes, things like that.
Then eight years ago, I came to the NIH, where I had a little bit more time for doing research work. I really became interested in some of the issues with protein analysis. With new developments in mass spectrometry and new tools, I could go back and look at some of the issues I had been interested in years before. I suppose I was basically interested in proteomics 15 or 20 years ago, but we didn't have the tools to do it. I remember 15 years or so ago being interested in 2D electrophoresis and how you might use it for diagnostic purposes. But it wasn't well enough developed then. You would get lots of spots, but you really couldn't identify what most of the spots were.
It's really over about the last five years or so that with new mass spectrometry tools, and with completion of the human genome that basically defines all the protein sequences, you can go back and identify whatever protein or peptide you want, and you have a lot more of a solid basis for doing these things. Another thing is that detection limits have become much more practical to work with. The scale of samples that you would get in a clinical lab before you might have had to draw much larger volumes of blood than would be routinely practical, and kind of go through many steps to get down to what you need. This is starting to really enable things to be more approachable on true clinical samples.
Did you start to work with biomarkers at the NIH?
Well, 'biomarkers' is in a sense a nonsense term. I've been working with biomarkers for 30 years, though people didn't call them that back then. All the lab tests that we deal with pretty much are measuring biomarkers. Biomarkers is such a vague term, it's practically a nonsense term. I don't know why people use it. It's just kind of a catchy term that for some reason people think means something special.
Every test that we use in the clinical lab is basically measuring a biomarker it may be an enzyme activity. If you look at the current definition, it's basically anything that you can currently measure that might be related to a physiological response that you might be interested in following. We've had biomarker labs for many years they're called clinical labs.
It is a very interesting time in that you can sort through possibilities much faster in terms of doing discovery work, and the whole pace of discovery has been speeded up a lot. Some of the techniques allow you to look at hundreds of components at a time. That's really a new dimension in terms of looking at things. In the past, most of the things that we've done in the clinical lab have been done one test at a time. That's not the case 100 percent [of the time], but the majority of them have been. A few things we've looked at [involved more than one thing at a time] a profile on electrophoresis, or some things looked at an entire pathway, or perhaps 10 or more components. But we now have a new tool set to look at these things, and I think it is opening up new possibilities, particularly to look at molecular variation and some of the post-translational modifications.
What kind of tests are you working on developing currently?
Mainly I'm interested at the moment in trying to characterize a diversity of smaller peptide components in biological fluids mainly plasma and urine samples. By traditional tools, that was one area that we did not see in very great detail. The new tool sets have really expanded quite a bit our ability to look at these. Mass spec is very good for looking at peptides. The older tools like 2D electrophoresis the peptides fall off the bottom, so they're kind of invisible. We probably had underaccounted for them some. So that's the area I'm interested in looking at the moment.
What issues do researchers have to deal with in translating a biomarker into a clinical test?
There are many steps in going from a basic biomarker to a useful laboratory test. Particularly in some of the profiling techniques that have been used for the MALDI-TOF or SELDI-TOF techniques have not applied the usual types of standards that we use in the clinical laboratory. To start to figure out how to standardize these types of assays better you have to define some of the things that you usually do for any clinical test in terms of evaluating reproducibility, linearity of responses. You have to figure out how the intensity of various peaks relate to the concentration of what you're trying to measure. I think that to move some of these techniques to clinical lab tests, you have to come up with approaches that deal with these things. Otherwise you would not be meeting the usual standards that we would apply in a clinical lab.
What's the biggest challenge in moving something into a clinical test?
There are so many steps in this that it's a little bit hard to know where to begin. There are really probably a half dozen or more important steps. There are whole issues in terms of having adequate specimens to validate the test in the first place, and to kind of serve as a database. There are so many issues related to the specimen selection, and how you perform those studies. And then people are still trying to grapple with specimen collection issues. Often times they haven't paid quite enough attention in terms of pre-analytical variables the patient preparation, and how to standardize the collection processes and things.
There's also the issue of marker selection. Many markers are not going to be suitable if they're too variable or unstable.
Most of what we're doing at the moment is method and technology development rather than specific test development. It's a little premature to try to develop a test until you really have your methods optimized more. A lot of our efforts now are really towards trying to get the most information possible about peptide repertoires in different samples and stability issues so that we can figure out how to handle the samples before we even get to the point of doing a test. Once we do that, I think then we'll be able to move forward fairly quickly and a little bit more rigorously in terms of doing tests.
We have analyzed some clinical sets and things, but those were a little bit more discovery efforts in terms of finding peptides that might be discriminatory for certain things. We weren't really to the point of doing tests yet.
Some people think as soon as they get some peaks showing up in a pattern, they have a test, but I think in general there's a fairly big gap in terms of finding a few peaks in a pattern and figuring out how you can use that meaningfully in a test.
What kind of advice would you give to someone who has found some biomarkers and is looking to develop them into a clinical test?
Generally, I feel it's important to be able to identify what your peaks are, and to understand the biology and what they represent. Is there really any physiological basis that they should be related to what you're trying to understand? The second thing is you have to optimize your methods so you get good reproducibility. Your ability to get accurate measurements in general kind of depends on your ability to get high signal to noise. If you're dealing with very weak signals, in general you're not going to be able to get very robust tests out of it.