Intraindividual variability over time in plasma biomarkers of inflammation and effects of long-term storage

Systemic measures of chronic inflammation, often based on a single blood draw, are frequently used to study the associations between inflammation and chronic diseases such as cancer. However, more information is needed on the measurement error in these markers due to laboratory error, within-person variation over time, and long-term storage. We investigated the intraindividual variability of inflammation markers C-reactive protein (CRP), interleukin-6 (IL-6), and soluble tumor necrosis factor receptors I and II (sTNFRI and II) in a subsample of the Seattle Barrett's esophagus study cohort. Two fasting blood samples were collected between 1995 and 2009 from 360 participants on average 1.8 years apart. CRP, IL-6, and sTNF receptor levels were measured by immunonephelometry, ELISA, and multiplex assays, respectively. Intra- and inter-batch coefficients of variation (CV) were estimated using blinded pooled samples within each batch. Intraclass correlations (ICCs) were computed using random effects ANOVA. Intra- and inter-batch CVs for the pooled plasma aliquots were low (2.4-8.9 %), suggesting little laboratory variability. Reliability over time was excellent for sTNF receptors (ICCsTNF-RI = 0.89, ICCsTNF-RII = 0.85) and fair-to-good for CRP and IL-6 (ICCCRP = 0.55, ICCIL-6 = 0.57). For samples stored for over 13 years, the ICCs for CRP and IL-6 were decreased but those for sTNF receptors were unaffected. sTNF receptor levels are more stable within person over time than CRP or IL-6. Long-term storage of samples appears to increase the variability of CRP and IL-6 measures, while the reliability of soluble TNF receptor measures was not affected by storage time.