MALDI-TOF Serum Protein Profiling for the Detection of Breast Cancer

Abstract

Proteomic expression profiling has beensuggested as a potential tool for the early diagnosis ofcancer and other diseases. The objective of our studywas to assess the feasibility of this approach for the detectionof breast cancer. Materials and Methods: In a randomizedblock design pre-operative serum samples obtainedfrom 78 breast cancer patients and 29 controlswere used to generate high-resolution MALDI-TOF proteinprofiles. The spectra generated using C8 magneticbeads assisted mass spectrometry were smoothed,binned and normalized after baseline correction. Lineardiscriminant analysis with double cross-validation, basedon principal component analysis, was used to classifythe protein profiles. Results: A total recognition rate of99%, a sensitivity of 100%, and a specificity of 97.0% forthe detection of breast cancer were shown. The areaunder the curve of the classifier was 98.3%, whichdemonstrates the separation power of the classifier. Thefirst 2 principal components account for most of the between-group separation. Conclusions: Double cross-validationshowed that classification could be attributed toactual information in the protein profiles rather than tochance. Although preliminary, the high sensitivity andspecificity indicate the potential usefulness of serum proteinprofiles for the detection of breast cancer.