The stability and variability of serum and plasma fibroblast growth factor-23 levels in a haemodialysis cohort

BACKGROUND: Serum fibroblast growth factor 23 (FGF-23) levels are markedly elevated in haemodialysis patients and have been linked to mortality outcomes. Small studies in health and chronic kidney disease, have demonstrated marked intra- and inter-individual variability in measured FGF-23 levels, and variable degradation in serum as compared to plasma samples. In end-stage kidney disease (ESKD), the intra- and inter-individual variability of FGF-23 levels, and the optimal collection methods remain poorly characterized. In this study we assessed the variability of FGF-23 levels in a cohort of stable haemodialysis patients. Secondly, in a subset of patients, we assessed the effects of different collection methods on measured FGF-23 levels. METHODS: To assess the variability of FGF-23, pre-dialysis blood samples were collected over 3 consecutive weeks from 75 haemodialysis patients. The effects of different specimen collection methods were examined in a subset of patients (n = 23), with pre-dialysis blood collected into different tubes: plain (serum), EDTA (plasma) and EDTA with the addition of a protease inhibitor (EDTA-PI). All analyses were performed in the main cohort and repeated in each subgroup. Variability over a 3-week period was assessed using repeated measures ANOVA and random effects linear regression models. Intra-class correlation coefficients were calculated to assess agreement, and coefficients of variation were calculated to assess intra- and inter-individual variability. RESULTS: Over the 3-week study period the mean FGF-23 levels were not significantly different in the serum (p = 0.26), EDTA (p = 0.62) and EDTA-PI (p = 0.55) groups. FGF-23 levels demonstrated marked intra- and inter-individual variability with a CV of 36 and 203.2%, respectively. In the subgroup analysis, the mean serum FGF-23 levels were significantly lower than the EDTA (p < 0.001) or EDTA-PI (p < 0.001) groups, however there was no difference in mean FGF-23 levels between EDTA and EDTA-PI (p = 0.54). CONCLUSIONS: The measured FGF-23 levels were significantly lower in serum as compared to plasma, and the addition of a protease inhibitor did not confer an additional benefit. Importantly in this cohort of ESKD patients, FGF-23 levels showed marked intra- and inter-individual variability. The routine measurement of FGF-23 in ESKD remains challenging, however this study suggests the plasma is the optimal collection method for FGF-23 analysis.