The first cohort study focusing on FGF23 was the Mild-to-Moderate Kidney Disease (MMKD) Study that enrolled 227 patients [
5]. This is a study showing an independent relationship between plasma FGF23 levels and renal outcome [doubling of serum creatinine and/or end-stage renal failure (ESRD)]. However, this study enrolled only nondiabetic patients with CKD. Next, a cohort study called the Osaka Vitamin D Study in Patients with CKD (The OVIDS-CKD study) was conducted in Japan [
6]. The study population consisted of 738 predialysis outpatients from nephrology departments including patients with diabetes. Because C-terminal fragments are circulating in serum from patients with ESRD [
7], we measured intact FGF23 levels by a sandwich enzyme-linked immunosorbent assay system (Kainos Laboratories, Inc., Tokyo, Japan). This was in sharp contrast to the fact that C-terminal assays, mostly Immunotopics assay, were employed in most well-known cohort studies including the Chronic Renal Insufficiency (CRIC) study [
8]. Because FGF23 levels as determined with the C-terminal assay are elevated by iron deficiency [Arrow (14) in Fig.
5] [
9], which is often observed with erythropoiesis stimulating agent use, our study may have offered a more appropriate setting in which the association between the bioactive form of FGF23 and clinical outcomes can be explored. We examined the relationship between estimated GFR (eGFR) and levels of MBD markers including intact FGF23, whole parathyroid hormone (PTH), serum phosphate, calcium, 25D, and 1,25D using baseline data. Figure
1 shows the order of the change in these markers with eGFR decline. The earliest change in MBD markers was observed for intact FGF23, and the latest change was serum calcium levels. Circulating 25D levels were relatively constant across eGFR categories. In contrast, levels of 1,25D decreased linearly as renal function deteriorated, which is attributed to atrophy or dysfunction of proximal tubules and downregulation of 1alpha-hydroxylase activity by elevated FGF23. Whole PTH levels were stable between eGFR points A and B (Fig.
1) despite the decrease in serum 1,25D levels. This is compatible with the fact that FGF23 inhibits the secretion of PTH from parathyroid glands [
10]. In our study, the rise in FGF23 preceded the rise in PTH levels. This order of changes in parameters is consistent with a previous animal study using rats treated with antibodies against glomerular basement membranes [
11]. However, this sequence is currently reported to be dependent on vitamin D status [
12]. In patients with vitamin D deficiency, PTH elevation precedes FGF23 elevation, whereas the opposite is true in vitamin D-repleted patients. The common finding is that the last marker to change regarding phosphate metabolism is serum phosphate. The explanation is that these compensatory increases in two phosphaturic hormones augment the fractional excretion of phosphate, leading to normalization of serum phosphate concentrations in early and moderate CKD. Clinical evidence for this explanation is seen in renal transplant recipients with persistent hyperparathyroidism. Serum phosphate levels are elevated by cinacalcet therapy, which reduces both PTH and FGF23 levels [
13].
Next, we analyzed the risk factors for poor vitamin D status (defined as 25(OH)D < 23 ng/mL, median value in this cohort) with multivariate analysis. In addition to female sex and high PTH levels, moderate or high proteinuria (2 + or greater with a dipstick test) and diabetes were risk factors (Fig.
2) [
14]. This is consistent with our basic science data showing that D-binding protein is lost with vitamin D in the puromycin aminonucleoside (PAN)-nephrosis rat model in which podocyte function is impaired [
15]. Our data are also consistent with a clinical finding showing impaired megalin function in diabetic patients, even those with microalbuminuria [
16], because megalin is indispensable for vitamin D reabsorption by proximal tubules.