The importance of sample collection when using single cytokine levels and systemic cytokine profiles as biomarkers — a comparative study of serum versus plasma samples

Cytokines, soluble adhesion molecules and metalloproteinases can be detected in human serum or plasma samples. Such systemic levels are widely used as biomarkers in epidemiological and clinical studies. We prepared serum samples and three types of plasma samples (EDTA, heparin, citric acid) from 20 healthy individuals. The levels of 31 cytokines, four soluble adhesion molecules and eight matrix metalloproteinases were analyzed by Luminex technology. Most mediators showed detectable levels in both plasma and serum. Several mediators that can be released by platelets showed increased serum levels, especially CCL5 and CD40L, but for the other mediators the serum levels did not correlate with peripheral blood platelet counts and for these last mediators serum and plasma levels often showed strong correlations. The use of bivalirudin for anticoagulation significantly increased and citric acid combined with platelet inhibitors (ticagrelor, acetylsalicylic acid plus prostaglandin E2) did not alter plasma levels of platelet-store mediators compared with citric acid alone. The impact of sample preparation differed between mediators; for many mediators strong correlations were seen between serum and plasma levels even when absolute levels differed. Soluble adhesion molecule levels showed only minor differences between samples. Unsupervised hierarchical clustering suggested that the effect of sampling/preparation was strongest for serum and heparin plasma samples. Careful standardization of sample preparation is usually necessary when analyzing systemic mediator levels, and differences caused by sample preparation should be considered as a possible explanation if studies show conflicting results.