New methods for analyzing cytokine levels in chronic diseases

Abstract

Perturbations in blood cytokine levels have been suggested as potential biomarkers in chronic diseases. Several technological and computational barriers have stood in the way of fully exploiting this source of information, however. We evaluated two commonly used laboratory methods, ELISA and fluorescent bead based multiplex assays, on two commonly used sources, serum and plasma, for measuring resting cytokine levels in a pool of 43 people with chronic fatigue syndrome (CFS) and 60 not ill controls (age, sex, race, and BMI matched) in a population based study of Wichita, KS. Our main finding is that cytokine concentrations in serum and plasma samples best fit a Weibull distribution, not a normal or lognormal distribution, although this is not uniformly true. A model is proposed for the dynamics underlying cytokine production in the blood which could explain these distributions, and could also explain the lack of correlation between serum and plasma measurements observed by our group and other groups. We also develop a symmetric form of the Kullback-Leibler distance to measure the separation between Weibull distributions and a bootstrap method to estimate p-values for the same. Based on our analysis of the Weibull distributions, we found a statistically significant difference between CFS cases and controls in IL-6 levels, in keeping with the allostatic load hypothesis for CFS. Though we used different reagents and different methods in different laboratories for these measurements, the plasma IL-6 measurements yielded good correlation between luminex and ELISA based methods, although the scales differed by more than an order of magnitude for unknown reasons. This research was supported by the Centers for Disease Control and Prevention.