Low serum iron is associated with high serum intact FGF23 in elderly men: The Swedish MrOS study

Fibroblast growth factor (FGF23) is a protein that is produced by osteoblasts and osteocytes. Increased serum levels of FGF23 have been associated with increased risks of osteoporotic fractures and cardiovascular disease, particularly in participants with poor renal function. Serum iron (Fe) has been suggested as a regulator of FGF23 homeostasis.

To determine whether Fe and iron status are determinants of the levels of intact FGF23 (iFGF23) in elderly men.

The MrOS study is a population-based study of elderly men (N=1010; mean age, 75.3years; range, 69-81years). The levels of Fe, transferrin saturation (TS), and ferritin were evaluated in relation to the serum concentrations of iFGF23 before and after adjustments for confounders.

TS <15% was found in 3.5% (34/977) of the participants, who had a higher median level iFGF23 compared with the remaining subjects (47.4μmol/L vs. 41.9μmol/L, p=0.008). The levels of iFGF23 correlated negatively (un-adjusted) with the levels of Fe (r=-0.17, p<0.001), TS (r=-0.16, p<0.001) and serum ferritin (r=-0.07, p=0.022). In addition, in participants with estimated glomerular filtration rate eGFRCystatin C>60mL/min, the levels of iFGF23 correlated (age-adjusted) negatively with the levels of Fe (r=-0.15, p<0.001) and TS (r=-0.17, p<0.001). The level of iFGF23 correlated positively (un-adjusted) with lumbar spine bone mineral density (BMD) (r=0.14, p<0.001), total body BMD (r=0.11, p=0.001), and total hip BMD (r=0.09, p=0.004). The corresponding correlations, when adjusted for age, weight, and height were: r=0.08, p=0.018; r=0.05, p=0.120; and r=0.02, p=0.624, respectively. No associations were found between BMD and the levels of Fe or TS. Multiple step-wise linear regression analyses [adjusting for age, body mass index (BMI), comorbidity index, cystatin C, C-reactive protein (hs-CRP), serum vitamin D 25-OH (25OHD), phosphate, calcium, parathyroid hormone (PTH), erythropoietin, hemoglobin, lumbar spine BMD, apolipoprotein B/A1 ratio] were performed in three separate models with Fe, TS or ferritin as potential explanatory variables. Fe and TS, but not ferritin, were independent predictors of iFGF23 level (standardized β-values: -0.10, p<0.001; -0.10, p<0.001; and -0.05, p=0.062, respectively).

Low levels of Fe in elderly men are associated with high levels of iFGF23, independently of markers of inflammation and renal function, suggesting an iron-related pathway for FGF23 regulation.