Proteomics to Diagnose Human Tumors and Provide Prognostic Information

April 1, 2004

Biomedical research is in themidst of unprecedented transformationstemming from theoverall impact of molecular biologyon medical research, including theemerging high-throughput genomicsbasedtechnologies. These new paradigmsare leading to better definitionof the disease state as well as moreprecise and less toxic therapeutic strategies.But even as we begin to understandthe implications of gene-basedinformation on the genesis, pathophysiology,and progression of disease andon the development of novel therapeuticapproaches, the dawn of theera of proteomics is heralding evenmore radical changes.

Biomedical research is in themidst of unprecedented transformationstemming from theoverall impact of molecular biologyon medical research, including theemerging high-throughput genomicsbasedtechnologies. These new paradigmsare leading to better definitionof the disease state as well as moreprecise and less toxic therapeutic strategies.But even as we begin to understandthe implications of gene-basedinformation on the genesis, pathophysiology,and progression of disease andon the development of novel therapeuticapproaches, the dawn of theera of proteomics is heralding evenmore radical changes.Drs. Ornstein and Petricoin have produceda concise yet excellent review ofcurrently existing proteomics technologiesthat aim to define and characterizethe various proteins and peptidesassociated with biologic processes and/or disease states. These new proteomicstechnologies hold promise for providingdisease biomarkers, includingcancer markers that would be valuableto both patients and clinicians.Proteomic Analytic Tools
The authors discuss both classicaland emerging gel-based proteomicanalytic tools such as two-dimensionalpolyacrylamide gel electrophoresis(2D-PAGE) and digital in-gel electrophoresis(DIGE). Although improvementssuch as immobilized pHgradients, image analysis software,and better fluorescent-based dyes haveimproved reproducibility, 2D-PAGEremains a comparatively low-throughputmethod of proteome interrogationrequiring a relatively large amount ofbiologic sample when clinical assaydevelopment is concerned.[1]The importance and limitations ofrelatively novel technologies such asisotope-coded affinity tagging, lasercapture microdissection, tissue microarrays,and reverse phase proteinarrays are also adequately discussed.Protein-based arrays have a clear advantageover cDNA arrays in that thelatter cannot address the various featuresof protein alterations and posttranslationalmodifications that occcurduring the disease process. However,the dynamic range exhibited by the humanproteome presents an equally formidablechallenge for arrays that relyon traditional antibody-staining protocols.[1] The limitation of antibodyspecificity has sparked the developmentof novel arrays employing newclasses of protein-capture agents thatinclude aptamers,[2] ribozymes,[3]modified binding proteins, and partial-molecule imprints.[4]Role of Mass Spectrometry
Mass spectrometry-based proteomicsis providing the impetus for advancementand is proving to be anindispensable tool for molecular andcellular protein studies, making it themethod of choice for analyzing complexbiologic samples.[5] The constantinvention of higher platforms of massspectrometers exhibiting an exponentialincrease in dynamic range andresolution further increases the functionalityof this technology, not only indiscovery but in translational research.Aside from the utility illustratedin the article by Ornstein and Petricoin,other mass spectrometry-basedconcepts are being developed. Forexample, mass spectrometric tissueimaging[6] employs tissue sectionssubjected to matrix-assisted laser desorption/ionization (MALDI), and thediagnostic protein profiles containedin the tissue sections are generatedby "imaging" the sample with an arrayof mass spectra.The authors have emphasized theutility of surface-enhanced laserdesorption/ionization time-of-flight(SELDI-TOF) mass spectrometry usingCiphergen's ProteinChip BiologySystem II (with its ProteinChip Reader)and Applied Biosystems' QSTARPulsar LC/MS/MS System. The diagnosticutility of this technology in recognizingthe serum protein patternsdiagnostic of ovarian cancer or prostatecancer has been highlighted.[7-10]Although mass spectrometry equipmenthas the potential to provide desirablediagnostic biomarkers, the cost and levelof expertise required to perform thespectrometric analysis is of concern.The high dimensionality of data generatedfrom proteomic investigationsrequires powerful bioinformatics toolsto uncover biomarkers and hidden patternsthat are characteristic of the diseasestate. Genetic algorithms, self-organizedcluster analysis, and decision treealgorithms described by the authorsrepresent some of the analytic toolsemployed. Artificial neural networks[11-12] and discrete wavelettransform data reduction[13] are othermethods currently under investigation.Recent successes have emphasized theimportance of bioinformatics. However,key issues affecting data analysis,such as biologic variability,consideration of preanalytic variables,and analytic reproducibility, need tobe addressed.[14]Conclusions
The issues surrounding both thepromising potential and evident limitationsof proteomics and its medicalapplications are testaments to the developmentalstage of this emergingtechnology. Investigators in cancerresearch or other disease-focused researchwill increasingly need to relyon technologic advancements in genomicsand proteomics to achievetheir goals; ie, discovering the pathwaysthat lead to malignancy and takingadvantage of gene or proteintargets for diagnosis, prognostication,therapy, or prevention of disease. Thefurther development of the proteomicinnovations discussed here and theinvention of novel ones will certainlyprove beneficial in overcoming presentlimitations and hopefully will providerobust, practical platforms for the clinicalsetting.

Disclosures:

The authors have nosignificant financial interest or other relationshipwith the manufacturers of any productsor providers of any service mentioned in thisarticle.

References:

1.

Hanash S: Disease proteomics. Nature422:226-232, 2003.

2.

Smith D, Collins BD, Heil J, et al: Sensitivityand specificity of photoaptamer probes.Mol Cell Proteomics 2:11-18, 2003.

3.

McCauley TG, Hamaguchi N, Stanton M:Aptamer-based biosensor arrays for detectionand quantification of biological macromolecules.Anal Biochem 319:244-250, 2003.

4.

Xu L, Aha P, Gu K, et al: Directed evolutionof high-affinity antibody mimics usingmRNA display. Chem Biol 9:933-942, 2002.

5.

Aebersold R, Mann M: Mass spectrometry-based proteomics. Nature 422:198-207,2003.

6.

Stoeckli M, Chaurand P, Hallahan DE, etal: Imaging mass spectrometry: A new technologyfor the analysis of protein expressionin mammalian tissues. Nat Med 7:493-496,2001.

7.

Petricoin EF III, Ardekani AM, Hitt BA,et al: Use of proteomic patterns in serum toidentify ovarian cancer. Lancet 359:572-577,2002.

8.

Adam BL, Qu Y, Davis JW, et al: Serumprotein fingerprinting coupled with a patternmatchingalgorithm distinguishes prostate cancerfrom benign hyperplasia and healthy men.Cancer Res 62:3609-3614, 2002.

9.

Petricoin EF III, Ornstein DK, PaweletzCP, et al: Serum proteomic patterns for detectionof prostate cancer. J Natl Cancer Inst94:1576-1578, 2002.

10.

Bañez LL, Prasanna P, Sun L, et al: Diagnosticpotential of serum proteomic patternsin prostate cancer. J Urol 170:442-446, 2003.

11.

Rogers MA, Clarke P, Noble J, et al:Proteomic profiling of urinary proteins in renalcancer by surface enhanced laser desorptionionization and neural-network analysis:Identification of key issues affecting potentialclinical utility. Cancer Res 63:6971-6983,2003.

12.

Mian S, Ball G, Hornbuckle J, et al: Aprototype methodology combining surfaceenhancedlaser desorption/ionization proteinchip technology and artificial neural networkalgorithms to predict the chemoresponsivenessof breast cancer cell lines exposed to paclitaxeland doxorubicin under in vitro conditions.Proteomics 3:1725-1737, 2003.

13.

Qu Y, Adam BL, Thornquist M, et al:Data reduction using a discrete wavelet transformin discriminant analysis of very high dimensionalitydata. Biometrics 59:143-151,2003.

14.

Boguski MS, McIntosh MW: Biomedicalinformatics for proteomics. Nature 422:233-237, 2003.