[Diagnostic application of serum protein pattern and artificial neural network software in breast cancer].

BACKGROUND & OBJECTIVE: The progress in proteomics provides a novel platform for early diagnosis of cancer, and screening for new tumor biomarkers. This study was designed to develop and evaluate a diagnostic model of breast cancer with surface enhanced laser desorption/ionization-time of flight-mass spectrometry (SELDI-TOF-MS) ProteinChip array technology and artificial neural network software.

METHODS

SELDI-TOF-MS ProteinChip was used to detect serum protein patterns of 49 patients with breast cancer, and 33 healthy women. Diagnostic model was developed, and validated using artificial neural network software.

RESULTS

An intact diagnostic model from all 253 discrepant protein peaks, and a terse model from the top-scored 4 peaks were built. The diagnostic sensitivity, and specificity of the intact model were 83.33% (15/18), and 88.89% (8/9)u the detection rates of breast cancer of stage I, and stage II-IV using the intact model were 90.00% (9/10), and 75.00% (6/8). The diagnostic sensitivity, and specificity of the terse model were 76.47% (13/17), and 90.00% (9/10)u the detection rates of breast cancer of stage I, and stage II-IV using the terse model were 100.00% (3/3), and 71.43% (10/14). The diagnostic values of these 2 models were similar (P>0.05). Their diagnostic abilities to breast cancer of stage I were not worse than those to breast cancer of stage II-IV (P>0.05).

CONCLUSION

High sensitivity and specificity achieved by this method show great potential for early diagnosis of breast cancer, and screening for new tumor biomarkers.