” is defined as a measurable analyte that correlates with a specific phenotype, such as a normal biological condition, a pathological process, or a pharmacological response to a therapeutic intervention (Group, 2001
). A biomarker is unique in that it distinguishes and discriminates between comparative biological conditions such as cancer and non-cancer. “Cancer proteomics
” is defined as molecular profiling of cancer-associated proteins approached by analyzing the global protein expression patterns of tumor cells or extracellular fluids such as serum from cancer patients (Alaiya et al. 2000b
). To date there have been many candidate proteomic biomarkers suggested for various types of cancer. However, none has yet met the stringent requirements for clinical use, such as reproducibility, specificity and sensitivity. A true biomarker should be confirmable across laboratories and technology platforms (Aebersold et al. 2005
), in its ability of predicting the clinical state. Today the greatest need in cancer proteomics research lies in improvement in the methodology and technology for the recognition and identification of reliable protein biomarkers specific for the disease or biology of the disease.
Historically, biomarker discovery was dominated by targeted approaches, in which candidates derived from biological knowledge were selectively evaluated for their correlations with clinical conditions (Gillette et al. 2005
). However, in the past two decades, the advances in new mass spectrometric ionization techniques for discovering macromolecules has revolutionized biomarker research (Karas and Hillenkamp, 1988
). Now for the first time it is possible to display complex protein profiles in convenient one- and two-dimensional formats, and at the same time identify the proteins. More and more studies are employing mass spectrometry technology to search for novel cancer biomarkers. Reflecting this trend, in March 2007, a MedLine search for the words “cancer biomarker” listed in 119,912 publications, while a similar simultaneous search that included the word “protein” resulted in 96,598 articles. A selection of published representative studies on cancer biomarkers using mass spectrometry technique are summarized in . However, as reported by Alaiya et al. (2005)
, the lack of proteomics markers for cancer stands in sharp contrast to the significant progress that has been made in the discovery of gene-based biomarkers during the last 20 years. This disparity also reflects the increased difficulty when working with the proteome as opposed to the genome. Nevertheless, the potential offered by proteomic technology in biomarker discovery draws research support from both government funding agencies and private sector.
Summary of selected MS-based cancer biomarker studies.
A protein biomarker detected by mass spectrometry can be a single entity (with protein identification, therefore identity-based) or a suite of entities (protein signatures, therefore pattern-based). For the biomarker to be clinically useful, the abundance or change in abundance of specific proteins must reflect some aspects of normal physiology/biochemistry or a disease process. Combined liquid chromatography-mass spectrometry (LCMS) is the technique to which most cancer proteomic research is now gravitating. Because of the direct on-line link between chromatography and mass spectrometry, this technique can more easily accommodate complex biological samples such as serum and plasma. In contrast, the enthusiasm which originally greeted other techniques that do not use direct on-line chromatography, such as surface enhanced laser desorption mass spectrometry (SELDI), has now waned. This has occurred as the need for chromatography when dealing with complex samples, such as serum and plasma, has become obvious. In this context then, as defined here, a biomarker may be a peak or peaks detected on a chromatogram, a signal or signals in a mass spectrum, or a feature or features in a three dimensional representation of peak retention times, signal mass to charge (m/z) ratios and intensities. These are three different representations of LCMS data that are routinely acquired in proteomic analyses.
It has been pointed out that the utility of a protein biomarker for disease diagnosis or detection does not necessarily require knowledge of the identity of the protein or proteins involved (Petricoin et al. 2002
; Villanueva et al. 2005
), so long as the pattern of peaks, signals or features are sufficiently reproducible to reflect the present disease. Strictly speaking, this could be the case. However, without knowing the identity of the biomarker proteins, information related to the underlying disease process will not be forthcoming. Furthermore, the identity of the protein or proteins involved is necessary for independent validation of the biomarker by different technologies, and for the future development of faster, cheaper and more reliable assays (which probably will not rely on chromatography or mass spectrometry). Consequently, disease-based protein biomarker research involves both recognition of the biomarker signature and identification of the proteins involved.