November 1, 2005 (Vol. 25, No. 19)

Incorporating Mass Spectrometry into Diagnostic Tests

Ready availability and easy collection of plasma, which contains a huge variety of protein classes, make it attractive for prospecting for protein biomarkers.

However, plasma is also one of the most complex sources to work with. Sample-to-sample variations complicate the effort to trace and compare peptides. Separating “noise” from a true signal with the pharmaceutical level of statistical significance requires processing thousands of samples. Multiple steps may be required to resolve individual components, which always carry the risk of losing peptides of interest.

Moreover, clinically important plasma peptides may differ in abundance by a factor of 1010 presenting a challenge for detection of multiple analytes. Only 13% of the peptides reproducibly change in abundance with the changes in patient’s disease state. This represents a few thousand peptides, most of them less than 20 kDa. Such peptides are considered to be putative biomarkers.

Mass spectrometry (MS) seems to be perfectly positioned to deal with the complexities of the plasma proteome. Various combinations of MS and affinity chromatography (AC) or liquid chromatography (LC) strive to produce unbiased identification and quantitation of signature peptides. The specific advantage of MS in diagnostics is its ability to track complex biomarker sets virtually without limitation on the number of components. MS can detect subtle protein changes that cannot be detected using current techniques. (For in depth information on plasma biomarkers see www.

The instrumentation is less expensive and more commonplace every year (see GEN, September 15, 2005, p.1). Nevertheless, the rate of introduction of new diagnostic tests based on plasma biomarkers is low. Diagnostic companies seem in no hurry to embrace proteomics technologies for routine testing.

Could MS-based tools and discoveries be integrated into diagnostics worldwide?

Reliable Results Start With Reliable Samples

“The whole biomarker field is very MS-centered,” says Bruce Haywood, business development leader for clinical proteomics, preanalytical systems, BD Diagnostics ( “But overall method reproducibility will be limited by the weakest link in the workflow. Even during the blood collection process, biomarkers are already subject to at least 20 different mechanisms of degradation.”

BD Diagnostics used its considerable understanding of blood to develop BD P100 tube for plasma collection, specifically geared for discovery of plasma biomarkers. The tube contains a proprietary cocktail of protease inhibitors, EDTA anticoagulants, and other stabilizing components.

“P100 aims to preserve in vivo-like plasma proteome samples,” continues Craig Gelfand, Ph.D., director of growth technologies. “Standardization helps to eliminate preanalytical variables. In effect, P100 may become an enabling technology for the entire plasma biomarker field.”

For example, the elevated level of brain natriuretic peptide biomarker (BNP) is commonly used to diagnose congestive heart failure. BNP is rapidly cleaved by several proteases, which may lead to false-negative diagnosis. Blood collection with BD P100 would provide more reliable analysis.

BioVision ( ) reportedly secured a position in the field of differential peptide display with the patent “Process for Determining the Status of an Organism by Peptide Measurement,” issued by USPTO in April. The company uses the term Peptidomics to describe the discovery process specifically related to native peptides.

Differential Peptide Display (DPD) technology analyzes peptides in both a qualitative and quantitative way. At the core of the DPD discovery process is the observation that each native peptide can be consistently mapped according to its mass-to-charge ratio (m/z), retention time on RP-HPLC and peak intensity. Differentially expressed peptides are sequenced by tandem MS and matched to the protein database. Quantitation of peptides is achieved without spiking with isotope-labeled peptides.

“We try to start with a well qualified population to minimize biological variations based on gender, ethnicity, age etc.,” says Peter Schulz-Knappe, M.D., founder and CSO of BioVisioN. “We have achieved excellent levels of sensitivity, being able to detect peptides at concentrations less than 100 pmol/L and using less than 1 ml of sample.”

To discover truly differentially expressed peptides, BioVisioN has to eliminate those peptides that fluctuate randomly across the population. To do that, the company processes up to a thousand samples per study, and can detect over 5,000 peptides at the same time.

In a feasibility study, plasma peptides were analyzed for changes resulting from glucose consumption (glucose tolerance test). Over 2000 peptides were detected but only 15 showed significant changes. Sequencing revealed known biomarkers, such as insulin and C-peptide, as well as previously undescribed peptides.

Differential Peptide Expression

Caprion Pharmaceuticals (www. pursues biomarker discovery by using differential peptide expression of trypsin-digested samples, followed by peptide sequencing. Caprion’s CellCarta platform generates 3-D maps of peptide isotopes. Even co-eluting or closely related peptides can be differentiated based on their signature ions. The sophisticated pattern matching algorithms are used to compare the peptide maps between normal and affected populations. Simultaneous tracking and measuring of almost 40,000 peptides per sample and across thousands of samples may result in only a handful of clinically relevant biomarkers.

To reach a statistically significant level of data, Caprion has to evaluate a large number of samples in a very standardized manner. “We industrialized proteomics,” says Greg Opiteck, Ph.D., director of protein analysis at Caprion. “We go to great lengths to standardize sample acquisition and sample preparation. We learned how to remove random effects and to achieve a great consistency at every step of our process.”

The company’s platform is geared to toward early clinical development, providing insights into drug efficacy and toxicity. It can also be used in late pre-clinical studies to help in the compound nomination process. In a study with Wyeth (, quantitation of dose-dependent changes in classical pharmacodynamic markers was used to confirm anti-inflammatory effects of ERB-041, a selective agonist of estrogen-receptor . The study demonstrated that the compound reverses 80% of plasma peptides upregulated in adjuvant-induced arthritis.

Predicant Biosciences (www. is also looking for peptide signatures, but chose an approach that analyzes undigested proteins and peptides. After separation on a proprietary microfluidics chip, the output is fed directly and continuously into a mass spectrometer, which takes a “snap-shot” of the proteome every second. The company’s signal processing system filters the data and reduces the number of analytes to about a thousand.

“Our separation process takes only 12 minutes, and can serve both discovery and routine assaying,” says Shawn Becker, M.D., vp of market development. “Rigorous sample collection standards combined with the complete system integration allow us to achieve highly reproducible results, which is critical to our focus of developing diagnostic tests intended for large-scale clinical use.”

In a technical feasibility pilot study presented at the National Cancer Institute’s Specialized Programs of Research Excellence (SPORE) meeting, Predicant described identification of a three-biomarker pattern for prostate cancer in men whose prostate-specific antigen (PSA) levels are not specific enough for use as a screening tool. The pattern had a positive predictive value of 40% and negative predictive value of 90%. The company is in the early stages of validating these results and is exploring other diagnostic applications as well.

Immunoaffinity Step Increases Sensitivity

A common error in biomarker discovery, according to Randall Nelson, Ph.D., president and CEO of Intrinsic Bioprobes (IBI;, is that a protein exists as the same single entity in every individual.

“For example, troponin I is used as a biomarker for detecting heart attack. But troponin in plasma exists not as a single protein but as a family of breakdown products. Knowing how these species correlate with various presentations of heart disease would help to provide the correct diagnosis and to choose the correct line of treatment.”

IBI finds “true biomarkers” within candidate biomarkers, he says. IBI’s MASSAY platform identifies and quantifies structural changes (point mutations, post-translational modifications, breakdown products) that show a statistically significant correlation with a disease or a physiological condition, continues Dr. Nelson.

Since the level of structural diversity of plasma proteins is quite high even in normal population, all variant species have to be detected in the same analysis with the high degree of sensitivity (see Nedelkov, D. et al, “Investigating Diversity in Human Plasma Proteins”, PNAS (2005) 102, 1085210857). The company utilizes proprietary pipette tips with immobilized polyclonal antibodies to sample low-level proteins directly from plasma samples. Using this immunoaffinity approach, several proteins can be sampled from the same well. When needed, isolated proteins are characterized further using trypsin-activated MALDI-TOF targets.

“We are able to detect protein modifications, such as point mutations and PTMs, which ELISA would not be able to differentiate,” notes Dr. Nelson.

Plasma Proteome Institute (PPI; is attempting to perfect its own immunoaffinity method of enriching for specific tryptic peptides in plasma digests prior to quantitation by MS. The Institute aims to achieve sensitivity levels of five orders of magnitude over any current proteomics platform.

“In a typical protocol, multiple fractionation steps before MS are necessary to detect even medium abundance proteins. This is a big limitation of MS as compared to ELISA,” says Leigh Anderson, Ph.D., CEO of PPI. “ELISA can very specifically detect proteins in small concentrations on the background of abundant proteins, but such assays are difficult to develop and multiplex.”

To streamline pre-MS sample preparation, PPI specifically enriches the sample by capturing the tryptic peptides on 100 nanoliter anti-peptide antibody columns. Binding and elution of peptides from the affinity supports provide on average 120 fold enrichment, increasing sensitivity of detection by MS. The columns can be reused a number of times with little loss in binding capacity.

MS-based Detection Platforms

Dr. Anderson hopes that this low cost, high throughput method will become a gold standard for biomarker validation.

However, for peptide biomarkers and their MS-based detection platforms, the gap between discovery and diagnostics is still wide. Some of the outstanding issues include: finding ways to reduce the costs associated with validation of biomarkers or proteomic platforms, demonstration of biomarkers’ clinical significance, minimizing technical complexity, gaining regulatory approvals, and reaching pharmaceutical levels of reproducibility and sensitivity.

“After we discover the markers, they could be easily tested in a routine IVD format, such as ELISA or RIA,” says Lloyd Segal, president and CEO of Caprion Pharmaceuticals. The clinical testing of the biomarkers could be done concurrently with pharmaceutical clinical trials. Biomarkers could be evaluated for their positive predictive value one-by-one or as a group, just as any other parameter in the study. Therefore, the companion diagnostics based on plasma biomarkers can be more quickly developed and tested since the protein sequence is available. But the unfortunate economics of clinical diagnostics precludes diagnostic companies in investing into biomarker discovery. They are not willing to take on significant upfront costs and risks of biomarker prospecting,” continues Segal.

“It is also hard to foresee that an MS-based system such as CellCarta would be used by clinical labs for routine diagnostics,” adds Dr. Opiteck. “The payoff period would be too long, especially if the tests are used for conditions with low-incidence of occurrence. ELISA-based assays are portable, cheap, sensitive, easy to use and require minimal training to perform. If one cannot raise antibodies against particular biomarkers, perhaps those biomarkers should not be used at all.”

“Tracking larger sets of biomarkers is more informative, but ELISA assays are limited to measuring just a few markers at the same time. It is more feasible to look at multiple markers simultaneously by using MS,” counters Dr. Becker. “If diagnostic labs believe that an MS-based test is important and that there is enough clinical data supporting the system’s sensitivity, specificity and its predictive value, then they will invest. Certain high-value tests may become an important source of revenue.”

Nevertheless, Predicant Biosciences makes its first steps into the diagnostics field not by selling their system, but via an internal reference lab, which the company will own and operate. “It will all come down to how strong your clinical results are,” adds Dr. Becker.

“Clinical diagnostic companies are generally risk-averse,” agrees Dr. Nelson. “However, they are not afraid to invest in the range of a quarter of a million dollars for biomarker discovery.” According to the company, IBI entered into research agreements with several diagnostic companies. At the present time diagnostic companies are after a protein biomarker. However, the proteomic platforms themselves may have a great potential in clinical diagnostics. Their acceptance in diagnostics would depend on changes in general misperception of their costs and technical complexity.

The solution may be in full automation, minimization of human involvement, and focused interpretation of the data. Dr. Nelson envisions that performing the diagnostics using the MASSAY platform would simply require loading of the plate into the robotic workstation. The MS output would show only changes related to the protein of interest, i.e., three peaks where before it was one. “It would be a disservice to the world not to introduce proteomics tools into clinical diagnostics,” continues Dr. Nelson.

“The equipment manufacturers have to start designing systems with the clinical, and not research, goals in mind. And the scientific community has to show a lot more cohesiveness and collaboration in order to develop an MS platforms for diagnostics use,” concludes Haywood. “Collaborations are absolutely critical to develop a reproducible and standardized work flow from the sample collection to the final result. Parallel studies, data sharing and tying in with systems biology are the main factors that will determine success of proteomics in the diagnostic world.”

Previous articleRecipe for Ensuring Process Control Success
Next articleAmerican Lyme Disease Foundation