Dead ends point the way: Finding robust biomarkers in limited proteomics data

Skip to Navigation


  • Published: Feb 7, 2011
  • Author: Jon Evans
  • Channels: Laboratory Informatics
thumbnail image: Dead ends point the way: Finding robust biomarkers in limited proteomics data

Must be robust

The search for disease biomarkers is plagued with dead ends. When scientists compare diseased and healthy samples of blood or tissue, they usually find a whole host of proteins or metabolites that are expressed differently in the two samples and thus could be potential biomarkers for the disease.

The problem is that most of these potential biomarkers don't stand up to detailed scrutiny, with the biomarkers discovered by one group often failing to be confirmed by other groups. This has been attributed both to analytical errors (see No laboratory for old men?) and the possibility that many of the differently expressed proteins and metabolites are not actually specific to the disease being investigated but are a response to general cellular stress (see Sick from stress).

One way to discover more robust biomarkers that are able to stand up to scrutiny is simply to work with larger data sets, but this is often not possible for diseases that aren't particularly common. Now, however, a team of medical researchers from the Washington University School of Medicine in St Louis, Missouri, has come up with another way, which utilises a network-based scoring function and some of the less-than-robust biomarkers.

'A typical proteomics approach would require large datasets, which may not be clinically feasible and might be cost prohibitive,' lead researcher Issam El Naqa told separationsNOW. 'Therefore we developed this bioinformatics approach based on graph theory and utilization of prior available knowledge to address this sample size issue while still getting a robust biomarker.'

Don't stray from the path 

The approach is based on the idea that although previously discovered biomarkers may not themselves be robust enough to distinguish between healthy and diseased samples, they may be able to point the way to more robust biomakers. This is because the discovered biomarkers are probably involved in disease-related processes, such as inflammation or the immune response, it's just that they're not specific enough to the disease to act as unambiguous markers for it. But other proteins or metabolites involved in similar processes might be.

To uncover such unambiguous biomarkers, El Naqa and his team developed a process in which proteins highlighted by liquid chromatography-mass spectrometry (LC-MS) as potential biomarkers for a specific disease are compared with previously discovered biomarkers. This involves finding metabolic pathways that both types of biomarker take part in, as revealed by previous studies.

Using these shared pathways, El Naqa and his team build up a network of the relationships between the biomarkers, in which each biomarker is linked to one or more other biomarkers. The closer two biomarkers are to each other in this network, the more likely they are to be involved in similar metabolic pathways.

Scoring these new biomarkers according to how close they are to the previously discovered biomarkers, and then weighting this score according to how many studies have confirmed close links between them, should reveal those biomarkers most likely to be specifically associated with the disease in question.

Radiation pneumonitis

El Naqa and his team applied this approach to finding biomarkers for radiation pneumonitis, which is potentially fatal damage to the lungs caused by radiotherapy for lung cancer. This involved using LC-MS to compare proteins in the blood serum of 26 patients undergoing radiotherapy for lung cancer, three of whom developed severe radiation pneumonitis.

After finding 22 proteins that seemed to be associated with radiation pneumonitis, El Naqa and his team built up a network based on the four previously discovered biomarkers for this condition. This network was able to incorporate 10 of the 22 proteins, allowing El Naqa and his team to produce scores for these 10 proteins.

This revealed that an immune system-modulating protein called alpha-2-macroglobulin appeared to be most closely linked with radiation pneumonitis. And this finding was then confirmed using an enzyme-linked immunosorbent assay (ELISA), which showed that alpha-2-macroglobulin is present at much higher concentrations in those suffering from radiation pneumonitis.

Following this success, El Naqa and his team are now planning use this approach to find robust biomarkers for other diseases.

The views represented in this article are solely those of the author and do not necessarily represent those of John Wiley and Sons, Ltd.

Dead ends point the way: finding robust biomarkers in limited proteomics data

Social Links

Share This Links

Bookmark and Share


Suppliers Selection
Societies Selection

Banner Ad

Click here to see
all job opportunities

Most Viewed

Copyright Information

Interested in separation science? Visit our sister site

Copyright © 2018 John Wiley & Sons, Inc. All Rights Reserved