Background: Understanding the biological milieu associated with disease states has important implications for biobehavioral research. Cytokines, signaling molecules that mediate and regulate immunity, inflammation, and hematopoiesis, are an important component of the biological milieu associated with breast cancer. Cytokines have been used as biomarkers in research for prognosis and have been associated with symptoms and adverse outcomes in multiple conditions, including breast cancer. To date, however, the examination of cytokine patterns has been limited by traditional laboratory methods. Advances in proteomic technology now permit the characterization of a broader array of cytokines in a single specimen. Because cytokines operate in integrated networks, a more complete understanding will be gained as multiple cytokines can be examined for patterns of response that may be associated with symptoms and prognosis.
Objectives: To use proteomic technology (a) to examine whether there was a difference in cytokine levels and patterns in women with breast cancer compared with controls, (b) to define and compare the receiver operator characteristic curves for standard cytokine classifications, and (c) to identify the best-fitting empirical model of cytokines to distinguish groups of women found to have breast cancer from those with negative biopsies.
Methods: The cytokine levels of 35 women who had been diagnosed recently with breast cancer were compared with 24 women with a suspicious breast mass who were found subsequently to have a negative breast biopsy. Multiplex bead array assays permitted the simultaneous measure of multiple markers in a small volume of serum. Nonparametric procedures were used to determine differences in the median values and the distributions for each cytokine. The receiver operator characteristic curves were defined to identify patterns of cytokines.
Results: There were significantly higher systemic cytokine values in women with cancer in comparison with those in women without cancer for all cytokines measured, with the exception of granulocyte colony-stimulating factor and interferon-gamma. The only significant associations found between cytokines and age or race were increased levels of interleukin-8 (r = .53) and macrophage inflammatory protein-1 beta (r = .45) with increased age in women with a negative biopsy. Three cytokines (granulocyte colony-stimulating factor, interleukin-6, and interleukin-17) distinguished between the breast cancer and no-cancer groups with an exceptionally high areas under the curve (0.981; SE = 0.017).
Discussion: Levels of cytokines and their patterns were markedly different in women with breast cancer as compared with those in women who did not have breast cancer. Results from this study highlight the need for further research to examine the levels and patterns of cytokines that may serve as biomarkers in clinical research. Innovations in proteomic technology have implications for expanding biobehavioral research.