Proteomics in cancer

Abstract

Serum protein profiling by mass spectrometry (MS) is a prom-ising direction for clinical and scientific advancement in diag-nosis and treatment. This technology has been reported bymany investigators [1–11] to achieve higher ovarian, breastand prostate cancer diagnostic sensitivity and specificity.Newer results show sensitivity and specificity approaching100% in some cases, compared with conventional and otherinvestigational cancer biomarkers [12]. The predecessor to MSfor discovering disease-associated proteins, two-dimensionalpolyacrylamide gel electrophoresis, is laborious, requires largequantities of protein and is not easily converted into adiagnostic test. The development of an MS and bioinformaticscoupled approach has largely overcome many of theselimitations, proving to be a fast, potentially cost-efficient,minimally invasive, highly sensitive and accurate diagnostictool.Surface-enhanced laser desorption and ionization (SELDI)or matrix-assisted laser desorption and ionization (MALDI)with time-of-flight (TOF) MS detection, coupled with artifi-cial intelligence-based informatics algorithms, have become apowerful tool with which to identify biomarker disease pro-files [3, 13–15]. Mass spectrometry has been used success-fully to detect several disease-associated proteins in samplesof <10ml from cell lysates, seminal plasma, urine, cerebro-spinal fluid, nipple aspirate fluid and serum. The SELDItechnology procedure involves application of samples to pro-tein capture chips composed of chemically modified surfaces,causing the proteins to be selectively adsorbed to the surface.These protein chips have multiple spots containing variedsurfaces, including hydrophobic, ion exchange, metal affinitybinding surfaces or normal phase chromatographic surface.The use of multiple capture chip matrices can provide differ-ent views of the proteome. The sample protein-bound chipsare introduced into the MS unit and separation resultsthrough ionization of the proteins with laser energy(Figure 1). Detection is in direct proportion to the size andnet electrical charge of the protein (m/z). The mass spec-trometry output is shown as a chromatographic patternwherein the peak amplitude is represented on the y-axis at agiven mass/charge assignment (m/z or x-axis). The resolutionof the mass spectrometry unit directly reflects the sensitivityof this technique; the datastream can contain 15000–350000data points in the region below 20000 Da/charge ratio,depending on the type of machine and its mass accuracy[14]. Different investigative groups have found that differentproteomic chip surfaces may be optimal for their diseasediagnosis [3, 14, 16–20]. Electrospray and MALDI are alsounder investigation.Higher-order analytical bioinformatics approaches are usedto define optimal discriminatory signature proteomic patternsto distinguish cancer versus non-cancer patients. The signaturepattern developed from the initial genetic algorithm bioinfor-matics program used identified a set of five key features, m/zspecies, which as an event in N-space fully segregated ovariancancer from non-cancer within a supervised teaching set ofdefined populations [3]. This was then applied to a blindedseries of samples from which its true discriminative abilitywas evaluated. Mass spectra from the blinded unknownsamples were classified by likeness to the pattern created bythe key features in the discriminative signature. This techno-logy and similar other high order bioinformatics has achievedsignature patterns by scientific investigators that are 99–100%sensitive and 99–100% specific, proving the concept to bepromising as a new biomarker tool [3]. Eventually, withextensive validation by independent research groups, massspectra proteomic pattern analysis could be applied to medicalscreening clinics as a diagnostic test.