Abstract

BACKGROUND:

Mass spectrometry (MS)-based profiling was used to determine whether ion fingerprints could distinguish women with stage 1 breast cancer from women without breast cancer.

METHODS:

The initial study population consisted of 310 subjects: 155 women with yearly negative breast examination and negative mammography findings for at least 4 years, and 155 women undergoing surgery for pathology-proven stage 1 invasive ductal carcinoma. High-resolution SELDI-TOF (surface-enhanced laser desorption ionization-time of flight) analysis was performed on serum obtained from blood samples collected before mammography in controls, and before surgery in patients with breast cancer. Samples were divided into a training (109 controls and 109 cancers) and blinded (46 controls and 46 cancers) testing set; each group had similar age demographics. In addition, an independent study set of 46 serum samples was analyzed 14 months after the initial study to validate the robustness of the classifier.

RESULTS:

A discriminatory profile consisting of seven ion peaks found in the training set, when applied to the blinded test set, achieved a sensitivity and specificity of 95.6% and 86.5%, respectively. This same seven-peak profile achieved a 96.5% sensitivity and 85.7% specificity, with correct identification of all of 17 T1a tumors when applied to the validation study set.

CONCLUSIONS:

Mass spectrometry profiling of human serum generated a robust classifier composed of seven low-molecular-weight ions that yielded a highly sensitive and specific diagnostic procedure for the discrimination of women with stage 1 breast cancer compared with women without breast cancer in this research study set.