Abstract
Serum cytokine and chemokine networks may reflect the complex systemic immunological interactions in cancer patients. Studying groups of cytokines and their networks may help to understand their clinical biology. A total of 178 cases of ovarian cancer were analyzed in this study, including 73 high-grade serous (HGSC), 66 clear cell (CCC) and 39 endometrioid carcinomas. Suspension cytokine arrays were performed with the patients' sera taken before the primary surgery. Associations between each cytokine and clinicopathological factors were analyzed in all patients using multivariate linear regression models, and cluster analyses were performed for each histotype. In the multivariate analyses, twelve of 27 cytokines were correlated with histotypes. Cluster analyses in each histotype revealed 2 cytokine signatures S1 and S2 in HGSC, and similarly C1 and C2 in CCC. Twenty-two of 27 cytokines were commonly clustered in HGSC and CCC. Signature S1 and C1 included IL-2,6,8,15, chemokines and angiogenic factors, whereas signature S2 and C2 included IL-4,5,9,10,13, TNF-α and G-CSF. Four subgroups based on a high or low level for each signature were identified, and this cluster-based classification demonstrated significantly different progression-free and overall survivals for CCC patients (P = 0.00097 and P = 0.017).