Samples often have to be frozen and thawed prior to analysis for immunoassays. Reliable cytokine measurement from small-volume plasma and serum samples is critical for biomarker research and clinical Show more
Samples often have to be frozen and thawed prior to analysis for immunoassays. Reliable cytokine measurement from small-volume plasma and serum samples is critical for biomarker research and clinical studies. However, repeated freeze-thaw (F-T) cycles may alter analyte stability, introducing error and reducing reproducibility. Standard statistical methods often overlook donor-to-donor and matrix variability, leading to overestimation of F-T effects. We measured 80 cytokines across three donors, three matrices (EDTA-plasma (PL-EDTA), heparin-plasms (PL-heparin), serum), and four F-T cycles using an 80-plex Luminex immunoassay. Linear mixed-effects modeling was applied to partition donor, matrix, and F-T contributions, while principal component analysis (PCA) summarized global variance. Cytokines were classified as stable, decreasing, or matrix-specific based on within-matrix slopes and matrix×cycle interactions. Stability was defined as the absence of a statistically significant per-cycle change (p ≥ 0.05) within all matrices, corresponding to changes smaller than typical assay imprecision (CV%). PCA revealed that donor and matrix were the dominant sources of global variation, whereas F-T cycles contributed minimally. Most cytokines remained stable (no significant within-matrix slope across cycles; p ≥ 0.05 in all matrices) across four cycles, with only a minority showing monotonic decreases. Matrix context strongly influenced F-T effects: PL-heparin displayed both increases and decreases, PL-EDTA was largely stable, and serum showed decreases without increases. Representative analytes highlighted the three categories: IL17E/IL25 and IL27 decreased modestly across matrices, chemokines such as 6CKINE/CCL21 remained stable, and analytes like SDF1A + B/CXCL12 showed matrix-specific increases. Freeze-thaw cycling contributes far less to cytokine variability than donor or matrix effects. Most cytokines are robust across four cycles, and when F-T sensitivity occurs, it is largely matrix-dependent. These results provide evidence-based guidance for sample handling and highlight the importance of modeling donor and matrix effects in biomarker studies. Show less