Basic and clinical proteomics researchers - ACS Publications

---

Potential misunderstandings

Every situation is different, but according to many researchers who have participated in collaborations that include basic and clinical participants, initial expectations are often unrealistic. “I think that many clinicians have expectations that the basic scientists cannot deliver,” says Denis Hochstrasser, a clinician and a basic proteomics scientist at Geneva University and University Hospital. He explains that clinicians who don’t understand what is involved in a proteomics study will be disappointed with how long a discovery study takes or with the paucity of biomarker candidates that pan out. Several scientists on both sides of the equation point out, however, that clinicians are not the only ones who think that proteomics projects will have quick 2278

Time

Redundancy

Instrumentation

Funding Validation

Patient’s perspective

Data analysis

Basic researchers Clinicians

Inappropriate testing Quick fix

The experience was enough to give someone whiplash. At a disease-oriented national conference a few years ago, leaders in the proteomics field organized sessions to showcase the potential of the science. During one session, which was heavily attended by clinical researchers, speakers and members of the audience had nothing but praise for a certain proteomics method. Listening to the exchanges, you’d think that finding protein biomarkers was a simple task. Just a few hours later, however, analytical chemists and basic proteomics scientists held a session in which they pointed out the numerous limitations of this technique and presented methods that they contended were more accurate, albeit more difficult to perform and with lower throughput. Are basic and clinical researchers really so far apart on the issues? Are clinicians naive about the capabilities of methods and instrumentation? Are there some aspects of biomarker research that basic scientists don’t understand? Do the two groups talk to each other?

Expectations

Basic and clinical proteomics researchers: the great divide?

Sample limitations

news

Data context

Collaboration Accurate biomarkers

payoffs. Gil Omenn at the University of Michigan, who trained as a clinician and protein chemist, remembers that only days after the mapping of the human genome sequence was celebrated in 2001, the media already were shifting gears and touting the promise of proteomics. It’s easy for the public and scientists who are not involved in proteomics research to become too excited when the media oversell the capabilities of the technology, he says. Clinicians might expect quick fixes, but Robert Cotter, an analytical chemist at Johns Hopkins University School of Medicine, points out, “You have to remember that their goal or focus is different from ours. [Clinicians] are always hoping that they can get something that, even if they don’t understand the mechanism, will actually alleviate pain or reverse a condition.” Although collaborations between the two types of researchers can work very well, basic scientists find that clinical teams may misunderstand the roles that analytical chemists or biochemists play. “The clinicians have their specific aims, but they have to understand that we are not just technicians who are going to run samples for them either. So, there is a give-and-take here,” says David Lubman, an analytical chemist who recently has joined the surgery department at the University of Michigan. Basic scientists also have some unrealistic or impractical notions. Because

A n a ly t i c a l C h e m i s t r y / A p r i l 1 , 2 0 0 8

they were trained as developers and users of analytical instrumentation to solve basic biological and chemical problems, they typically do not know what types of tests will be appropriate in a clinical situation. Clinicians “bring reality into the game because they are the ones that know exactly what is needed in the clinic, so you don’t go and try to identify markers or set up diagnostics that perhaps are never going to be used,” says Julio Celis, a basic scientist at the Institute of Cancer Biology and the Danish Cancer Society. Unfortunately, many basic scientists working on their own waste time and money pursuing biomarker candidates that will be worthless to physicians, say researchers. For example, some recent proteomics studies report possible protein biomarkers for myocardial infarction, but Hochstrasser points out that good diagnostics for this condition already exist, so insurance companies are not likely to pay for yet another diagnostic technique. In addition, basic scientists can overlook the patient perspective when planning the sample collection protocol. Lubman says that he and his team want as much sample as they can get, but the clinicians they work with remind them that these samples are coming from real people, so only limited amounts of sample can be obtained. Louise Alldridge, a basic scientist at the Helen Rollason Heal Cancer Charity Research Laboratory of Anglia Ruskin University (U.K.),

news

adds that working in the breast clinic with clinicians has made her realize what the patients have to endure. Currently, her team is analyzing blood samples from chemotherapy patients. “We are tracking the progress of these patients, and sometimes you find out that they can’t have their chemotherapy because they’re not very well. You get a lot closer to it, and you also get a feel for what’s doable,” she says. You learn that “these people are having so many nasty things being done to them, and you eventually tend to start thinking of things that are going to be less disruptive to what they’re already going through.” Clinicians note that their colleagues in basic science sometimes get prematurely excited about their research results. However, these studies often compare only healthy controls with patients who have a disease. Hochstrasser points out that biomarkers identified in this type of study are not yet ready for clinical use and should be validated further. “If you compare someone who has pulmonary cancer with a healthy person, you will immediately find biomarkers, but they are biomarkers showing differences between sick and healthy,” he explains. That doesn’t help, he says, because the doctor already knows the patient is sick. The question is, What does the patient have? Many diseases have common elements, such as inflammation, so specificity is important. “If you want to find a specific biomarker for small-cell carcinoma of the lung, then you should compare it versus adenocarcinoma, versus mesothelioma,” he says.

Working it out

For a collaboration of clinical and basic researchers to be a success, scientists say that communication is key. Serhiy Souchelnytskyi, a basic researcher at Karolinska University Hospital (Sweden), says that differences between clinicians and basic researchers quickly melt away after they discuss the issues. Barry Karger, an analytical chemist at the Barnett Institute at Northeastern University, has noticed that these interactions

‘No’ can be better than ‘yes’ Most people think that the purpose of a diagnostic test is to identify a person’s illness, and commercial tests are designed to yield such positive-predictive results. But when a patient comes into the emergency room with symptoms, a physician often wants to rule out certain conditions with tests that have negative-predictive value. For example, to rule out that a patient has a pulmonary embolism, a doctor performs a blood test to detect the levels of fibrin dimers, which are released from a dissolving blood clot. If the levels are low, a patient probably does not have a pulmonary embolism. However, if levels are high, the test is inconclusive. “As clinicians, we really like a test with negative-predictive value because it helps us to exclude something and [possibly allows us to] send the patient back home,” says Hochstrasser.

are becoming even easier over time. He says that ~30 years ago, clinical and basic researchers mostly kept to themselves. Now, the situation has changed and more scientists from diverse fields are coming together to share ideas. An important way to get everyone on the same page is for all of the collaborators to understand everybody’s role in the project from the start, says Karger. “The clinician brings obvious expertise, and the analytical people bring obvious expertise, but they really have to blend in with each other,” he adds. In his opinion, all of the participants should know why the project is being done, where the samples originated, and how the data will be interpreted. “If we understand why we’re doing it and what it means, then we will generate better data, and it will be more interesting and more exciting,” he points out. It is important for both parties to work out the procedures for obtaining and handling samples before the project begins, say veterans of such collaborations. Sometimes, initial misunderstandings can result in comical discussions. For example, Lubman says the clinicians have had to remind his team that the samples from patients are available in limited quantities: obtaining a 100 mL blood sample from each patient is out of the question. “The clinicians have to explain to students and postdocs that only ~0.5 mL is available for the discovery,” he says. “Then they may

come back to us and ask if we can run 450 of these samples, and we have to explain to them that we are not ready to run 450 samples at this time. We each have to understand where the other is coming from, but if you communicate, that seems to be resolvable.” On some occasions, basic and clinical scientists even work together to figure out details, such as the best buffers to use or the best way to get the samples from the operating room (OR) to the lab. Fresh samples are best for pro­te­o­ mics experiments, so Celis and his team actually wait in the pathology lab, which is downstairs from the OR at a nearby hospital. The surgeon removes the sample from the patient, puts it in a tube, and sends it through a chute to the pathology lab. Similarly, at the beginning of Alldridge’s collaboration, her team would stand beside the surgeons and nurses with an ice bucket. Arranging the sample collection required negotiation, because the long-established practices in the OR and in pathology had to change. For example, she notes that the surgery team was accustomed to putting the samples directly into preservative for pathology analysis, but this step is not ideal for proteomics analysis. (Currently, samples are collected directly from the OR and managed by a medical lab assistant in the pathology department who is funded by the same charity that funds Alldridge’s group.) Likewise, Cotter’s team worked closely with clinicians to

A p r i l 1 , 2 0 0 8 / A n a ly t i c a l C h e m i s t r y

2279

news

Biobank warning The selection of the most appropriate samples for a retrospective clinical study can be a herculean task and can take a lot of time, so clinicians and basic researchers sometimes resort to samples of convenience that are stored in large repositories, says Thompson. However, such samples may have biases that could lead a scientist down the wrong path. For example, lung cancer samples might have been taken from patients while they were under anesthesia during surgery, whereas the control samples might have been taken from people who visited a hypertension clinic. A proteomics study comparing these samples might uncover a marker that seems to indicate the presence or absence of lung cancer. “Then when you go out to a group of smokers and use these markers, they may do a wonderful job telling you who’s under anesthesia and who has hypertension,” cautions Thompson. In other cases, the clinical annotation for samples in biobanks is not complete, says Celis. Therefore, even if a researcher analyzes the descriptions that are included, unknown biases may be present. Omenn notes that some scientists do not even specify whether a blood-derived sample is actually serum or plasma or record which anticoagulant was used. In addition, samples collected in the past and stored in large biobanks weren’t necessarily collected with standard procedures that meet current requirements. To address these problems, the U.S. National Cancer Institute (NCI) held town meetings on the topic and, in June 2007, officials developed the NCI Best Practices for Biospecimen Resources document, which is available at http://biospecimens. cancer.gov/practices.

2280

figure out the best way to handle samples of cerebrospinal fluid (CSF). “The clinicians knew better how to do this, because it turns out that trying to get CSF without getting blood contamination is really important,” he says. Data analysis is another key step where input from both types of researchers is vital, says Karger. Clinicians won’t necessarily be able to interpret mass spectra or protein microarray results, but they need to understand the challenges (such as dynamic-range issues) that are involved with obtaining the data. On the other hand, clinicians have access to detailed patient histories, which typically are off-limits to basic scientists, and this information helps put proteomics data into context. Without such close collaboration in the data analysis and validation steps, investigators run the risk of rushing biomarkerbased assays to the clinic too soon. “We have a tremendous obligation to the public to not push out clinical tests that could potentially [do] harm,” says Ian Thompson, a clinical researcher at the University of Texas Health Science Center at San Antonio. He points out that false-positive and false-negative tests can have enormous ramifications for patients—either a disease is not detected or a patient who really doesn’t have a disease is given unnecessary treatment or surgery. “The probability of false positives rises dramatically when such a test is used to screen healthy or even highrisk populations, rather than patients,” says Omenn. “The familiar PSA test for prostate cancer is a startling example. It is very useful to monitor treated patients for recurrence of tumor, but its predictive value for screening is negligible.”

Making a match

Additional challenges lie ahead for basic and clinical researchers. Several scientists say that promotion and tenure committees at various institutions still do not value collaborative research. “There continues to be this focus at the national level on an investigator in a laboratory cranking out identifications

A n a ly t i c a l C h e m i s t r y / A p r i l 1 , 2 0 0 8

of markers,” says Thompson. Celis notes that funding mechanisms must change to reward large collaborative projects, which may take longer than a few years to produce concrete results. “So, you need some sort of security,” he says. A team cannot set up a venture and then tell the patients after 2 years that the project has stopped because we “just didn’t get the grant,” he says. In the U.S., the National Institutes of Health (NIH) already has begun to shift some of its funding strategies to cover larger collaborations, according to Karger. An example of NIH’s progress on this front is the creation of the Early Detection Research Network (EDRN), led by Sudhir Srivastava of the National Cancer Institute. To achieve EDRN’s mission of translating biomarkers from the bench to the bedside, the project brings together basic scientists and clinicians to rigorously validate potential biomarkers. “It is a monumental task to bring together disparate groups of people with varied agendas and cultural values,” notes Srivastava. Thompson, who conducts research on prostate, bladder, and kidney cancer, says, “The fascinating thing is that when you’re sitting day in and day out with a breast cancer researcher, for example, you realize the issues and potential opportunities are incredibly similar.” To find collaborators, Sou­chel­nyt­skyi pounded the pavement. “I was taking time to go to different places and talk to as many people as possible,” he says. When he spoke to potential col­lab­o­ra­ tors, he told them what his group could do and asked the clinicians whether they would be a good fit. Lubman finds collaborators closer to home because his department encourages such interactions. He often meets with his clinician colleagues at seminars. Sometimes, a colleague in his department will refer another clinician to him for a project. Celis has found that clinicians are a busy group, so he does some of the initial pathology analysis before getting others involved in a project. “I’m a basic

news

scientist, but through my collaborations with clinicians, I had to transform myself into a molecular pathologist,” he explains. He says that pathologists have to make a certain number of diagnoses every day, so adding even five new things to their schedules for a proteomics discovery study is a burden. Therefore, he and his team narrow down the candidate biomarker list, generate antibodies, and then perform additional immunohistochemistry validation steps themselves. To avoid wasting pathologists’ time, only the best candidates are given to them for assessment. Once a pathologist is interested in some promising candidates, he or she will likely bring more clinical colleagues to the study. Many clinicians who conduct research were formally trained in medicine and basic science, so they are interested in developing research collaborations. Dennis Sgroi, who is at Massachusetts General Hospital, earned M.D. and Ph.D. degrees and performed postdoctoral research in a genomics laboratory. As a translational scientist who works with patients, he says that he can identify clinical problems that could be solved with basic-science techniques. According to Omenn, NIH is revamping the clinical research training scheme for the U.S. “It’s a time of high expectations and what NIH Director [Elias] Zerhouni calls ‘transformation’ for clinical translational research,” he says. “NIH has invested billions and billions in basic science for medicine, and people want to see results.” Omenn predicts that in the future, even more collaborations between clinicians and basic researchers will develop. “The emergence of validated biomarkers will probably be what will bring clinicians, basic scientists, and informaticians closer together,” he says. “Talking about things in the abstract is much less capable of attracting people’s detailed attention than having a useful application that can actually make a difference.” a —Katie Cottingham (Reprinted from J. Proteome Res. 2008, 7, 840–843.)

2008 ACS national award winners

(Top, left to right) Daniel Chiu, Catherine Fenselau, Jack Freed, Adam Heller, Rustem Ismagilov. (Bottom, left to right) Susan Olesik, Susan Richardson, Frantisek Svec, Mark Wightman, Richard Zare.

Among the honors to be presented on April 8 at the 235th American Chemical Society (ACS) National Meeting and Exposition in New Orleans are: Daniel Chiu, a professor at the University of Washington, won the National Fresenius Award for his pioneering methods that probe complex biological processes at the single-molecule level with applications that advance the understanding of cellular biology and neuronal systems. Chiu’s contributions include counting the number of a particular type of protein on a single vesicle, trapping and combining aqueous droplets, removing and reinstalling organelles, and NIR uncaging. Catherine Fenselau, Analytical Chemistry’s associate editor for MS and professor at the University of Maryland College Park, won the Field and Franklin Award for Outstanding Achievement in Mass Spectrometry for her sustained excellence in the application of MS to fundamental biomedical problems and for service to her profession. Fenselau’s work has encompassed bacterial characterization, the “middle molecules” concept, isotopic distributions in larger molecules, and enzymatic proteolysis to

incorporate 18O labels into peptides. Jack Freed, a professor at Cornell University, won the E. Bright Wilson Award in Spectroscopy for his development of electron spin resonance (ESR) spectroscopy into a powerful methodology and his applications to problems of dynamics and structure in condensed phases. Freed’s body of work includes ESR fundamentals, the theory of slowmotional ESR and NMR spectra, 2D FTESR, the multifrequency approach, and double quantum coherence. Adam Heller, professor emeritus at the University of Texas, received the Award for Creative Invention for fundamental contributions to the development of technological products that improved the quality of life for millions across the globe. In addition to being a faculty member, Heller cofounded a company that designed and manufactured a blood glucose monitoring device that uses such a small quantity of blood that the assay is painless. Rustem Ismagilov, an assistant professor at the University of Chicago, earned the Award in Pure Chemistry for using chemical principles and developing new tools to elucidate mecha-

A p r i l 1 , 2 0 0 8 / A n a ly t i c a l C h e m i s t r y

2281