Background: Over-diagnosis of breast lesions represents a significant problem in detection and screening of breast cancer, especially in women under the age of 50. Despite this issue, few new approaches have been developed to augment standard of care in the more precise detection of breast cancer. The combination of breast imaging, which provides anatomical evidence, with a robust protein signature that would detect biochemical cues of breast cancer offers an attractive approach to this problem. We have recently identified a protein signature composed of immune-regulatory cytokines, growth factors and tumor derived autoantibodies. Here, we test the hypothesis that a protein signature, combined with standard of care can greatly increase the precision of breast cancer diagnosis in women under the age of 50 in a randomized and blinded study.

Methods: Provista-001 enrolled 351 patients from 10 sites across the US and will follow patients for 6 months following the first blood draw under IRB approval. Patients were consented after first assessment of a BIRADS 3 or 4 and considered eligible if they were under the age of 50, no history of cancer, no prior breast biopsy, and were assessed as BIRADS 3 or 4 within 21 days of blood draw. Upon enrollment, patients were randomized to either training or validation groups. Clinical truth was set at equal to or greater than 80% sensitivity/specificity. Serum protein biomarkers and autoantibodies identified in prior proteomic screens were measured in serum samples collected prior to biopsy. Individual biomarker (22 serum protein and autoantibodies) concentrations, together with specific patient data were evaluated using various logistic regression models developed from prior retrospective studies. A training set, comprised of 200 patients, was used to develop and refine new models, which were then validated in the remaining 151 subjects. Clinical findings were compared to biopsy (largely BIRADS 4) or will be followed for 6 months and re-assessed (BIRADS 3).

Results: The novel algorithm utilizing patient data, serum protein and autoantibody concentrations combined with regression models was able to distinguish benign from breast cancer lesions in a statistically significant manner. Importantly, the serum protein biomarkers alone were unable to adequately distinguish benign lesions, consistent with prior work. However, the addition of tumor autoantibodies markedly increased both the sensitivity and specificity of the assay in this group of women. The use of the algorithm in conjunction with imaging was more accurate than imaging alone in this population.

Conclusions: Our current findings suggest that when used in combination, the protein signature developed here and breast imaging provides a more precise detection methodology than either alone. This is particularly important in women under the age of 50 where a low prevalence of disease makes detection difficult. The follow-up data at six months (BIRADS 3) will yield additional data in this understudied group of women. Additional studies utilizing the protein signature with women over the age of 50 are currently underway.

Citation Format: David E Reese, Michael Silver, Susan Yeh, Sherri Borman, Henderson C Meredith. Provista-001 a multi-center prospective study of protein signature used in the differentiation of benign breast lesions from invasive breast cancer in women under the age of 50 with a BI-RADS 3 or 4 [abstract]. In: Proceedings of the Thirty-Seventh Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2014 Dec 9-13; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2015;75(9 Suppl):Abstract nr P5-03-05.