This story originally ran on April 8.
By Tony Fong
Name: Pierre Lescuyer
Position: Head of Routine & Quality and Clinical Proteomics Laboratories, Laboratory Medicine Service, department of genetics and laboratory medicine, Geneva University Hospitals, 2006 to current
Background Clinical chemist, Central Clinical Chemistry Laboratory, Geneva University Hospitals, 2005 to 2006; Post-doctoral scientist, Biomedical Proteomics Research Group, department of bioinformatics and structural biology, faculty of medicine, Geneva University, 2002 to 2005; PhD in cellular and molecular biology, Joseph Fourier University
Despite years of effort to bring proteomics into the clinical environment, clinical proteomics remains elusive.
There has been considerable progress on some fronts, such as ovarian cancer diagnostics. In September, Vermillion's OVA1 triage test for ovarian cancer became the first proteomic in vitro diagnostic multivariate index assay to receive clearance from the US Food and Drug Administration. Meantime, several other proteomic tests for ovarian cancer are under consideration for FDA clearance or are being readied for FDA submission.
In addition, as a step toward leveraging the potential of proteomics for clinical applications, especially in the field of cancer research, the National Cancer Institute and the American Association of Clinical Chemistry entered into a memorandum to co-promote proteomics technologies and standards in the clinical chemistry community (PM 03/12/10).
Yet translating proteomics work into the clinic is still rife with obstacles and chances for missteps. At the annual meeting of the National Cancer Institute's Clinical Proteomics Technologies for Cancer in the fall, several proteomics researchers acknowledged that clinical proteomics has not met expectations from a decade ago (PM 10/09/09).
In an article published March 30 in the online version of Trends in Biotechnology, researchers in Switzerland highlighted recent developments in clinical proteomics and the problems that still need to be overcome for proteomics to become a technology with widespread use in the clinic.
In the paper, the authors wrote that proteomic methods "are not yet ready for implementation in routine clinical chemistry laboratories, except for simpler samples, such as microbial colonies, but the goal seems attainable in the near future."
While technologies such as mass spectrometry are increasingly finding their way into clinical settings, they are primarily being used for drug screening and bacterial proteins, which are orders of magnitude less complex than human proteins.
"Important developments in workflows and instrumentation are necessary before various proteomic methods can compete with protein immunoassays performed on high-throughput immuno-analyzers," the researchers said.
This week, ProteoMonitor spoke about these issues with Pierre Lescuyer, the lead author of the article and head of Routine and Quality and Clinical Proteomics Laboratories at Geneva University Hospitals.
Below is an edited version of the conversation.
Why did you decide it was necessary to write this article?
Prof. [Dennis] Hochstrasser [the senior author on the paper] has been working for more than 20 years in the proteomics field is also a physician and he is in charge of the department of laboratory medicine at [Geneva University Hospitals].
[ pagebreak ]
I [have been] in charge for three years now of what is called the Clinical Proteomics Laboratory and the goal is to try to make the link between the hospital and the research group that we have at the Faculty of Medicine [at Geneva University] and to evaluate proteomic techniques and to evaluate the possibility [of transferring] these types of techniques to the clinical lab.
That's exactly the focus of the article. I think that for us, it was to point out all the potential problems that we have identified.
For myself, in addition to being in charge of the Clinical Proteomics Laboratory with, for the moment, a focus mainly on problems of biomarker discovery — which is classical proteomics analysis research [but not applicable] to the clinical field — I am also in charge of the routine clinical chemistry lab.
So … I have both missions and when I was in the routine lab, I was thinking of all the problems I face regarding quality control, regarding maintenance of the analyzers we are using, and all the problems we have to deal with in the routine lab.
And [I tried] to imagine what kind of program can be created if we try to transfer what we are doing now in the proteomic lab to the routine laboratory. And there are many issues that [still need] to be addressed. And that was the goal of this paper, just to underline what we need now to think about before transferring some techniques to the routine lab.
What do you hope other proteomics researchers will do as a result of this article? Do you hope that it will lead to more discussion or actual action?
I think that that's important. One example for me is that most single-reaction monitoring papers that are published are targeting proteins that are already analyzed in routine labs using immunoassays.
These are proteins, for example, PTH — parathyroid hormone — [and] troponin, [that] we can measure very easily in the lab. There are very well established assays. This can be done with very short turnaround time, and the idea is that if the goal of proteomics is to replace these very simple methods with much more complicated workflows, I am not sure it will work because there are still a number of issues that are still to be solved.
[The article] was written to point out to people working in the research field and [have them] think about what problems they have to address before they put forward the tests they are developing.
Are you pushing for new goals then, if you're saying that some of these tests in development are looking at proteins for which we already have good diagnostics?
Yes, I think that the problem is that there are a number of areas where there are probably no diagnostic solutions. I often am in contact with physicians and there are many questions that are not solved.
For example, we are working on a project for diagnostic applications of biomarkers for pancreatic cysts and biliary tumors, and I think there are many, many other areas where the current diagnostic assays are not bringing solutions because, first, if you want to commercialize the assay, there has to be a market.
That means diagnostic companies will focus on the main diseases, cardiac diseases, diabetes, things like that. But there are many other fields where currently nobody is trying to find a diagnostic solution. And … one of the examples that I cited in the paper … is the paper published by [Andreas Kistler] on kidney diseases.
When you are doing mass spec analysis, for example, with MRM you can measure, many, many biomarkers on the same platform, and I think that's where proteomics can prove [that it] is a useful technique for clinical diagnosis.
There is a lot of research in proteomics looking at diseases where sufficient tests don't exist. Are you saying that there still isn't enough of a focus on these areas?
That's always the problem with proteomics, that people, not only in proteomics, but outside of proteomics … that there are many, many publications, but in the end there are not so many results in terms of biomarkers [that are] defined.
[ pagebreak ]
I don't really have an answer for that. One month ago, I was at a very, very interesting conference … on the problem of translating new biomarkers into clinical practice.
It's not a problem only with proteomics. I think that … to get a test approved by the [US Food and Drug Administration] is a very, very difficult task because you have to have a large number of people involved in clinical studies.
I think people are focused on proteomics because at the beginning there was maybe too much advertising. There was in Nature an article, "Proteomics and Cancer: Running before we can walk."
[But] I think that's a problem for all biomarkers. If you look at transcriptomics, there are also a lot of problems. In fact, currently there are not so many tests based on transcriptomics that are applied in the clinical field.
What I've noticed also is that when you talk to people involved in routine clinical chemistry or a laboratory in general, they are not involved in a research field like transcriptomics or genomics. And … they are not really prone to support these types of technologies. … And I don't think these types of people are very easy to convince.
That was also one of the purposes of the paper: to underline that if we want to convince these people, we need to speak the same language as them. It was also to focus on the problem that these people would like to be addressed. If you come with a new biomarker identified with a proteomics tool, [you need] to convince someone working in a clinical chemistry lab. If you have no idea of turnaround time, of quality control, you won't be able to talk with these people because these will be the first questions they will ask.
Going back to what you said about proteomics being hyped a few years ago, do you think that proteomics is still hurt by some of the unrealistic expectations that came along with the hype?
I indeed believe that proteomics is still hurt by the fact that much emphasis was put on some techniques that were finally very disappointing. SELDI is a typical example.
I believe that the problem is that solutions were, and are still, proposed without taking into account the basic needs and requirements of the end-users, such as clinical chemists. The goal of our article was therefore to list some of these basic needs and requirements.
What I can imagine is that techniques like multiple-reaction monitoring could be quite easily applied in routine clinical chemistries because they work just like an immunoassay. It's just that the technical principle is different, but the concept, which is targeting specifically a protein, is the same.
Maybe in that specific field, there is a problem with trypsin digestion because it's clear that this is not reproducible. At the last Bergmeyer conference [in March, which focused on bringing novel biomarkers to the clinic], someone from the European Institute of Reference Materials and Measurements said that they did some tests and they clearly showed that trypsin digestion is not reproducible. That means that any assay involving trypsin digestion is not quantitative.
However, there are many, many small proteins [and] peptides that could be measured by MRM without involving trypsin digestion and I'm sure there are many applications that are possible in that field, for example, peptide C … a lot of hormones, a lot of cytokines that initially should be measured in this technique without digestion.
After, if we consider the problem of protein microarrays or MS profiling, that's a difficult question because the proposed concept is new.
[ pagebreak ]
I know that there was just a few months ago a test that was approved by the FDA, the OVA1 test [from Vermillion]. And I have to say that if someone had asked me, 'Do you believe in this test?' I would say, 'Probably not because this is a mixture of known proteins, most of them not being described as markers of cancer,' but obviously it's working.
Probably the very critical point here was that [OVA1] was validated for a very specific clinical question. Also this is a clear example showing that there are many, many questions in the clinic for physicians without clear answers. But proteomics probably could bring answers for this type of problem.
I also think that, depending on the question, the requirements that you have for analytical performance will be different. If you want to measure, for example, a protein such as troponin, you absolutely need to be highly sensitive, highly specific. You cannot tolerate any kind of analytical problem.
Maybe if you are investigating other type of disease where you are looking for very clear differences among different patients or where the analysis is [done] in parallel with [other diagnostic methods] like histology, maybe the analytical requirements will be different or less critical.
So proteomics could find applications even if we know that there are some issues that are not addressed yet regarding analytical quality.
The paper mentions a need for standards and proof of reproducibility across labs of different methods. There are efforts underway by the Human Proteome Organization and the National Cancer Institute on both fronts. What is your assessment of those efforts? Do they carry enough weight behind them to effect some change, or are they too limited?
No, I think these are very important initiatives, but I think proteomics is still young and the idea of standardization in proteomics is just beginning. I think these initiatives are important to at least reveal what could be the problems, and I think that is very important.
For me, an example is the study by Terri Addona on the [precision and reproducibility] of MRM assays. This is very interesting because it clearly shows where the strong aspects of proteomics and where the weak points are.
And this is exactly the objective of this type of quality control. There is another method that I use in clinical chemistry today where we know that depending on the lab or the test that you are using, you will have different results. In some cases, you have to deal with that.
For example, for typical cancer biomarkers, depending on the immunoassay you use, you will have different results. The goal of quality standards initiatives is to reveal this type of thing. But just to be able to [do] that, it means that the maturity of proteomics is growing.
Is proteomics too young to think about translating the research into the clinic?
I don't think that it is too young. It's time to think about translation into the clinic, but it's also the time to really look directly at what the problems are that need to be solved, and it's time to go out of the research lab and address not only analytical problems of proteomic laboratories but also analytical and methodological problems in clinical laboratories.
For this, it is just the beginning, I think.
There are some efforts underway to address this disconnect between the proteomics researcher and the clinical chemist. What do you think needs to be done to bridge this gap between the two fields?
I think that probably it's to maximize collaborations between people working in the research field and people working in clinical chemistry labs and also physicians [such as] initiating a collaborative project that will allow [us] to get the answers to the problems we have now.
[ pagebreak ]
What about the area of mass spec-based assays? For proteomics, this is just starting to happen. Is movement into the clinic going to be hampered until we create more of these proteomics mass spec assays?
That's now what we are trying to set up in the lab. Our goal for the next five years in Geneva, for example, is to be able to have MRM assays not maybe for direct diagnostics but at least for clinical studies and to help physicians.
As I said before … you are measuring sophisticated proteins. There are some issues about specificity, sensitivity, and I think that's the easiest thing to apply because the principle behind that is simple — one test for one biomarker.
What about this issue of instruments? In the paper you very broadly say that instrumentation remains a problem.
The problem is the fact that the mass spec is not designed for routine analysis. It's often a complex instrument and someone who is not an expert will have some difficulties to run this type of instrument.
I know that the people you have in a routine lab in a hospital are not specialized chemists; they are not people [who have been] doing mass spec for 10 to 20 years. They are just some technicians who are [running] some immuno-analyzers, some HPLCs.
You need to have protocols and instruments that are quite easy to use and with very user-friendly software. And when you need to apply a diagnostic test every day of the year, you need an instrument [that will work reliably].
If you have something that is not working for weeks … that's a problem.
One of my colleagues has set up an LC-MS/MS system for toxicological testing. And that is exactly the problem that they have. They have an instrument that is quite difficult to use. … Now the problem is to have the system working [every] week, even on the weekends.
I've heard that in hospitals, mass specs are becoming more of an everyday instrument, though. Are you seeing the same thing?
Yes, yes. Just now [in my own hospital] there are many LC-MS [platforms] that are based on electrospray ionization that are being used in the toxicology lab, also an ion trap and a triple quadrupole that are used for toxicological screening.
And in the microbiology lab, they are using a MALDI instrument for screening of pathogens. These are established techniques … that are working really well.
But for drug [screening] and toxicology, the complexity is much lower than what we have for proteins in humans. Also, you don't need to perform digestion, so you can directly analyze [the sample]. There are also a lot of databases that have been developed for recognition. On the other side, for microbes, for pathogens, the proteome is much simpler than one of a human.
Are you seeing progress that we are moving from pure research into clinical applications, or are we stuck in the same place that we were a few years ago?
For me there is a lot of work that is very impressive … just what is being done by Ruedi Aebersold [of the Swiss Federal Institute of Technology], by Matthias Mann [of the Max Planck Institute for Biochemistry with] an emphasis on biology. Also, what is being done by a group like Steve Carr's [of the Broad Institute] on the development of methods for measurements of specific proteins.
So yes, there are a lot of people who are doing huge work, and it's both on the technological side and the methodological side, and I think that the fact that it's now not successful doesn't mean that people are not doing a good job.
It means the problem and the difficulty of the task is great.