Are We Closing in on a Blood Test for Breast Cancer?

NORFOLK, Virginia—Using an innovative technique called protein chip mass spectrometry, researchers at Eastern Virginia Medical School in Norfolk have identified specific serum protein profiles that may enhance the detection of breast cancer. Lori Wilson, MD, previously research associate at Eastern Virginia Medical School and now surgical oncology fellow at John Wayne Cancer Institute in Santa Monica, California, reported that in early testing, the biomarker profiles have shown a specificity and sensitivity that approaches that of mammography.

Protein chip mass spectrometry searches for multiple differentially expressed proteins to create unique protein profiles. For their study, the researchers used Surface-Enhanced Laser Desorption/Ionization (SELDI)/Time of Flight (TOF) mass spectrometry. They applied the SELDI/TOF technology to sera collected through the division of surgical oncology from both healthy women and women with breast cancer.

"Our samples were prospectively collected and were pretreatment as well as from normal healthy patients," Dr. Wilson said. "We analyzed 139 female sera by our SELDI technology, and the clinicopathologic information was recorded in our breast study data base."

Patient samples were classified as normal/benign or cancer. The median age was 46.5 years in the normal/benign patients and 59.3 years in the patients with breast cancer. The overall age range was 21 to 91 years. Eighty percent of the breast cancer patients were staged as ductal carcinoma in situ (DCIS) or stages I or II. Only 13% were stage III and 7% were stage IV.

Serum Sample Processing

"In serum sample processing for the SELDI analysis, we initially used the IMAC copper chip but decided to use the SAX chip to increase the number of proteins that were resolved with the mass spectrometry. We noted that with two chips there were peaks that were conserved but also peaks that we could isolate on one chip or the other," Dr. Wilson said.

Those samples were similarly processed through denaturing in the presence of urea buffer and were applied onto the chip in duplicate and randomized, she added. The nonspecific proteins were washed away with binding buffer, and then the data analysis and peak labeling were performed with Wizard Biomarker Software.

The researchers analyzed the initial IMAC-copper surface, looking at the learning set and then testing that learning set with a cross-validation analysis. When they attempted to discriminate between cancer and benign, the sensitivity and specificity for the learning set were 96% and 93%, respectively. With cross-validation, the sensitivity and specificity were 85% and 78%, respectively.

When discriminating cancer vs normal, the sensitivity and specificity for the learning set were 96% and 81%, respectively. On cross-validation, they were 82% and 74%, respectively.

IMAC Plus SAX

"We next tried to re-evaluate our detection rate. "We decided to use the SAX chip along with the IMAC surface to see if we could improve the resolution of our proteins and the sensitivity and specificity," she continued. "On our learning set, we improved both our sensitivity and specificity to 96.6% and our cross-validation to 93.3% on both."

No single protein peak completely separated the groups. In the single-chip analysis, seven protein peaks were able to classify cancer and benign. With the combination of IMAC and SAX chips, four protein peaks predicted 93% of the blinded test samples with a sensitivity and specificity of 96% for differentiating cancer from normal specimens.

Validation Needed

The next steps include validating the initial classifiers in a larger sample and having the results validated by outside institutions.

"The identification of multiple differentially expressed proteins," Dr. Wilson concluded, "increases the chance of discovering protein profile panels that, when validated and proven, will give us an opportunity to lead in the future to a blood test for the detection of breast cancer."