Patients with chronic kidney disease (CKD) have a significantly increased risk of cardiovascular disease (CVD) compared with age-matched individuals with normal kidney function.1 Mineral abnormalities complicating CKD such as hyperphosphatemia, calcitriol deficiency and secondary hyperparathyroidism (SHPT) are associated with increased cardiovascular (CV) and overall mortality.2–4 Proposed mechanisms for this relationship include endothelial dysfunction, arterial stiffness, left ventricular hypertrophy (LVH) and vascular calcification.5 The term ‘Chronic Kidney Disease-Mineral Bone Disorder’ (CKD-MBD) has been developed to highlight the intimate relationship between abnormalities of mineral metabolism, renal bone disease and excessive tissue calcification. The recent characterization of fibroblast growth factor-23 (FGF-23) and its important role in CKD-MBD has challenged the traditional understanding of the pathophysiology of SHPT. With an increasing number of clinical studies linking FGF-23 to clinical outcomes, we review the physiology of FGF-23 and its potential role as a biomarker and therapeutic target in CKD.
Fibroblast growth factor 23 (FGF-23) is a recently discovered regulator of phosphate and mineral metabolism. Its main physiological function is the enhancement of renal phosphate excretion. FGF-23 levels are inversely related to renal function and in patients with chronic kidney disease (CKD) elevation in FGF-23 precedes the rise of serum phosphate. Studies have demonstrated an important role for FGF-23 in the development of secondary hyperparathyroidism through an effect on parathyroid hormone and calcitriol. In cross-sectional studies FGF-23 has been associated with surrogate markers of cardiovascular disease such as endothelial dysfunction and arterial stiffness. FGF-23 has also been associated with both progression of CKD and mortality in dialysis patients. The discovery of FGF-23 has provided a profound new insight into bone and mineral metabolism, and it may become an important biomarker and therapeutic target in CKD.
STRUCTURE AND PHYSIOLOGY
Structure and assay
The link between FGF-23 and phosphate regulation was first described in the rare inherited condition of autosomal dominant hypophosphatemic rickets, and soon after in the acquired condition of tumour-induced osteomalacia.6,7 These diseases are characterized by a common phenotype – hypophosphatemia, low or inappropriately normal calcitriol levels, urinary phosphate wasting and osteomalacia.8 The postulated phosphaturic circulating factor was subsequently identified as FGF-23 and the characteristic phenotypes in patients with conditions of FGF-23 excess or deficiency provided important early clues regarding its function.9,10
The FGFs are a family of polypeptides that share a common region of amino acids, and have variable terminal residues which allow for a diversity of physiological functions.11 FGF-23 is a 251 amino acid protein that is predominantly synthesized and secreted by cells from an osteoblast lineage,12,13 and has an estimated half-life in the circulation of 58 min.14 FGF-23 can be detected with an enzyme-linked immunosorbent assay, in which antibodies detect N-terminal and C-terminal portions. An alternative C-terminal assay recognizes only the C-terminal fragments of active and inactive FGF-23.15 Early debate focused on whether the circulating FGF-23 is biologically active or whether the available assays also detect inactive compounds. A recent study compared the immune-based and intact FGF-23 assays with an assessment of FGF-23 bioactivity and western blot characterization of circulating FGF-23.16 The assays strongly correlated with each other and with FGF-23 bioactivity. The western blot detected only intact FGF-23 suggesting that virtually all circulating FGF-23 is biologically active.
Normal physiology and function
About 80% of total body phosphate is present in bone, 9% in skeletal muscle and only 0.1% in extracellular fluid.17 The distal duodenum is responsible for most phosphate absorption, a process actively mediated by calcitriol.18,19 In the kidneys about 95% of filtered phosphate is reabsorbed in the proximal tubular cells, a process driven by a high extracellular sodium gradient that is actively maintained by a Na+-K+-ATPase.18 This is further facilitated by Na-P co-transporters on the luminal side of the tubular cells, which are modulated by parathyroid hormone (PTH) and calcitriol.20 FGF-23 induces phosphaturia by reducing the number of co-transporters on the renal tubular cells, as well as mitigating the effects of calcitriol on intestinal absorption.21 PTH can stimulate phosphaturia in a similar manner; however, studies from transgenic mice suggest that FGF-23 induced phosphaturia is not PTH dependent.22
The biological effects of FGF-23 are exerted through activation of FGF receptors (FGF-R). Klotho is a trans-membrane protein originally described in mice with a phenotype of accelerated ageing and atherosclerosis.23 Klotho directly interacts with FGF-R, allowing it to bind FGF-23 with a higher affinity and increased specificity.13,24 The activation of FGF-R therefore occurs in a Klotho-dependent manner.24 Klotho-deficient mice manifest a similar phenotype to FGF-23 deficient mice despite high circulating levels of the FGF-23.8 The tissue selectivity of FGF-23 may be conferred by Klotho expression in the renal tubule and parathyroid glands.25 The expression of FGF-R and Klotho in the parathyroid glands also supports a regulatory effect of FGF-23 on PTH secretion.26
The main known physiological role of FGF-23 is to regulate urinary phosphate excretion and maintain a stable serum phosphate (Fig. 1).27 An important secondary role is the counter-regulation (against PTH) of vitamin D biosynthesis. The main stimuli for increased expression of FGF-23 are high dietary phosphate, calcitriol and persistent hyperphosphatemia,28–30 while phosphate restriction has the opposing effect. Animal studies have demonstrated a linear association between FGF-23 and phosphate; however, human trials have reported a variable rise in FGF-23 levels following phosphate-loading.31–33 This highlights the complexity of phosphate regulation in humans. It is likely that FGF-23 is not the only mediator of increasing phosphate excretion, and that other phosphatonins (frizzled-related protein-4, fibroblast growth factor-7, matrix extracellular phosphoglycoprotein)34 play an additional role which is currently poorly understood. The stimulation of FGF-23 by phosphate may be dependent on its dose, duration of exposure, bone derived co-factors and the severity and chronicity of CKD. It is also unclear as to whether serum or local phosphate concentrations provide the primary stimulus for FGF-23 secretion. FGF-23 has an inhibitory effect on PTH secretion; however, FGF-23 secretion may also occur in response to PTH levels. It is not known whether this occurs through a negative feedback loop mechanism or is conferred by the effects of PTH on calcitriol and serum phosphate (Fig. 1).26
The interaction between FGF-23 and Klotho may be necessary for normal phosphate metabolism. However, it is possible that high levels of FGF-23, as seen in CKD patients can exert a Klotho-independent effect, and bind to FGF-R with low affinity.13 This is supported by decreased expression of Klotho in renal biopsies from CKD patients.35 The expression of Klotho occurs predominantly in the distal tubules, and the signalling sequence that leads to decreased phosphate absorption in the proximal tubules remains unclear.36
Changes in CKD
FGF-23 levels are increased early in CKD and cross-sectional studies involving patients with a wide range of glomerular filtration rates (GFR), demonstrate an inverse relationship with renal function.37–39 The increase in FGF-23 levels observed in CKD may in part be a physiological response to restore normal serum phosphate levels. Proposed mechanisms include reducing renal tubular phosphate re-absorption, as well as decreasing circulating calcitriol levels (by downregulation of 1α-hydroxylase and upregulation of 24-hydroxylase) with resultant decreased intestinal phosphate absorption.40 Calcitriol is involved in a feedback loop, via liganded vitamin D receptor (VDR) binding to the FGF-23 promoter.41 It is therefore increasingly likely that early FGF-23 release, rather than decreasing renal mass and subsequent reduced 1α-hydroxylase function, constitutes the main mechanism leading to the biochemical changes that characterize SHPT.
Recently reported clinical studies support a phosphate-centric, FGF-23-mediated pathogenesis of SHPT (Fig. 2). One study involving 125 CKD stage 1–3 patients reported elevated FGF-23 and PTH levels inversely associated with estimated GFR (eGFR), and positively associated with increased urinary fractional excretion of phosphate. A greater proportion of patients had elevated FGF-23 compared with PTH (60.8% vs 9.8%).42 In another cross-sectional study of 80 CKD patients, FGF-23 levels were significantly associated with deteriorating renal function and decreased calcitriol levels.43 FGF-23 levels were elevated at an earlier stage of CKD compared with serum phosphate, which was more likely to be elevated in advanced CKD. An analysis of 792 patients with stable CVD demonstrated a continuous rise in FGF-23 levels at an eGFR < 90 mL/min.37 The recent Study for the Evaluation of Early Kidney Disease (SEEK), which involved 1814 Canadian participants, demonstrated calcitriol deficiency in 12% of patients with an eGFR > 80 mL/min, higher than at previously reported eGFR.
Available data supports a correlation between FGF-23, decreased eGFR and the biochemical changes of SHPT. However, prospective, longitudinal data and time-specific correlation between FGF-23 levels and biochemical parameters of SHPT are needed. The significance of the extremely elevated FGF-23 levels seen in CKD patients on dialysis remains poorly understood. It has been postulated that this process may be mediated by a change in Klotho expression resulting in relative resistance to FGF-23, along with as yet unrecognized factors. There is also a lack of conclusive data about the short- and long-term effects of phosphate intake on elevated FGF-23 levels in CKD.
Recent research into the metabolic and bone complications of CKD has focused on local, bone-derived factors that may modulate these changes. The relationship between bone turnover and serum FGF-23 was studied in several mouse models, where bone turnover was altered by a variety of exogenous and endogenous factors.44 The administration of osteoprotegerin (OPG), a potent anti-resportive agent, resulted in a rise in serum FGF-23, which occurred after reduction in bone turnover and was proportionate to the degree of suppression. The converse was observed after administration of exogenous PTH, with increased osteoblastic activity and reduced serum FGF-23. These findings suggest that bone remodelling and the rate of bone formation may modulate FGF-23 synthesis and release. In a recent study of 32 patients with CKD stages 2–5, plasma FGF-23 levels were inversely related to eGFR; however, the amount of bone FGF-23 expression was not related to the degree of renal impairment.45 These findings reflect the complexity of FGF-23 metabolism in normal and CKD patients and highlight the deficiencies in our understanding of FGF-23 and its relationship to CKD-MBD.
FGF-23 AND CLINICAL OUTCOMES
The various biochemical markers of CKD-MBD have all been variably associated with clinical outcomes in CKD. Elevated serum phosphate and to a lesser extent deficiency of 25-hydroxyvitamin D and calcitriol have been associated with adverse outcomes,2–4,46–51 although much of this evidence is from observational studies. The association between PTH and outcomes has remained controversial with some studies showing an association between high and low PTH levels, while others have failed to reproduce these findings.49–51 It remains uncertain as to whether it is the treatment of SHPT or the achieved PTH level that confer the greatest benefit. This uncertainty is reflected in the recent international Kidney Disease Improving Global Outcomes (KDIGO) clinical guidelines which recommend a PTH range of 2–9 times the upper limit of the normal level in patients with CKD 5 on dialysis.52
FGF-23 and mortality
A greater understanding of FGF-23 physiology, its role in CKD-MBD and elevated levels seen in CKD, have focused research on the potential role of FGF-23 as a prognostic marker (Table 1). FGF-23 has been correlated with phosphate in clinical studies.43 In a nested case–control sample of 400 patients in the Accelerated Mortality on Renal Replacement (ArMMOR) study, high FGF-23 levels were shown to predict 1 year mortality independent of phosphate levels.53 FGF-23 levels were also associated with higher mortality in patients with near normal levels of phosphate. A prospective cohort study of 219 dialysis patients undergoing 5–8 h dialysis also demonstrated an association between FGF-23 levels and mortality, again independent of phosphate.38 Although FGF-23 levels in these two studies did not demonstrate additional prognostic information when compared with phosphate levels, the possibility of using FGF-23 as a biomarker in patients with normal phosphate levels is of interest and needs to be prospectively assessed.
|Outcome||Study||No.||Patient population||Study design||Primary end point||Comment|
|Mortality||Gutierrez et al. (2008)53||400||HD||Nested, case–control study||FGF-23 association with mortality||OR for mortality across all FGF-23 levels 1.5 (CI, 1.2–1.8)|
|Jean et al. (2009)38||219||HD||Prospective observational study||FGF-23 association with mortality||Two-year mortality higher in patients with FGF-23 levels in the higher quartile, HR 2.5 (1.3–5), than in the first quartile|
|CKD progression||Fliser et al. (2007)39||177||CKD||Prospective cohort study||Doubling of serum creatinine or ESRF||Mean time to progression 46.9 months versus 72.5 moths if FGF-23 > 104 rU/mL|
|SHPT||Nakanishi et al. (2005)54||103||HD||Prospective cohort study||Association of FGF-23 with PTH levels||FGF-23 > 7500 associated with significantly higher PTH levels at 2 years|
|CVD||Parker et al. (2010)55||833||Mostly non-CKD, 22% eGFR < 60||Observational study||FGF-23 and mortality and CVD events||Compared with lowest FGF-23 tertile, patients in the highest tertile, had mortality HR 2.15 (CI 1.43–3.24) and CVD events HR 1.83 (CI 1.15–2.91)|
|Gutierrez et al. (2009)56||162||CKD, 36% eGFR > 60||Cross-sectional study||Association of FGF-23 with LVH and CAC||In CKD OR of LVH per 1-SD increase in log FGF-23 2.3 (CI 1.2–4.2)|
|Mirza et al. (2009)57||795||Mostly non-CKD, 20% eGFR < 60||Cross-sectional study||Association of FGF-23 with LVH and LVMI||Increased FGF-23 levels associated with LVH and LVMI|
FGF-23 associations with CVD
Increased mortality associated with biomarkers of CKD-MBD is predominantly attributed to an increased CV risk. The effects of FGF-23 on the incidence and mechanisms of CVD in the CKD population have been explored. In an observational study of 833 patients with early CKD and stable coronary artery disease, elevated FGF-23 was independently associated with mortality and CV events.55 Another cohort study of 967 patients with early CKD reported elevated FGF-23 levels correlated with arterial stiffness and endothelial dysfunction.57 In a subset of these patients, FGF-23 was associated with a greater atherosclerotic burden as measured by whole body magnetic resonance angiography.58 FGF23 has also been variably associated with vascular calcification, although a likely association may be obscured by the differences in diagnostic techniques and reporting of calcification scores.38,59 In a study of 162 CKD patients and 58 non-CKD patients where LVH was assessed by echocardiogram and computed tomography, FGF-23 was found to be independently and significantly associated with LVH and left ventricular mass index.56 A study of 795 Swedish patients also reported that FGF23 levels were independently associated with concentric LVH (odds ratio (OR) 1.45, 95% confidence interval (CI) 1.19–1.77) and left ventricular mass index. The association was stronger in those with eGFR < 60 mL/min (OR 1.83, CI 1.17–2.85).60 The significance of these associations remains unclear. Further studies are needed to demonstrate whether FGF-23 is more closely correlated with surrogate CVD markers than the existing serum markers of CKD-MBD, and whether potential therapies that lower FGF-23 may have an effect on reducing CV risk.
FGF-23 and CKD progression
Despite the lack of longitudinal data, multiple cross-sectional studies show an inverse association between renal function and FGF-23. A few studies have examined the potential of FGF-23 as a prognostic marker of CKD progression. The Mild to Moderate Kidney Disease (MMKD) study examined a prospective cohort of 177 patients with mild to moderate, non-diabetic CKD for a median of 53 months.39 FGF23 was inversely associated with baseline eGFR, and baseline FGF-23 levels were a predictor of progression of CKD when adjusted for phosphate and PTH. The lack of longitudinal measurement of FGF-23 and the biomarkers of CKD-MBD, however, was a major limitation of this study.
The significance of the extremely high FGF-23 levels in dialysis patients has also been examined. In 103 non-diabetic haemodialysis (HD) patients serum FGF-23 levels of 7500 ng/L predicted the future development of refractory SHPT.54 This may be due to a relative resistance of the hyperplastic glands to FGF-23. High circulating levels of biologically active FGF-23 led to speculation of a direct, non-Klotho mediated toxic effect on FGF-R; however, Klotho independent activation of the FGF-R has not been conclusively demonstrated.26 The effect of FGF-23 on the activity of extra-renal 1α-hydroxylase and local tissue calcitriol synthesis and levels remains unknown.
FGF-23-A therapeutic target?
Despite numerous studies showing the association between biochemical markers of CKD-MBD and FGF-23, only a few pilot studies have explored the effect of available treatments of SHPT on FGF-23 levels.
Secondary analysis of the ACHIEVE trial, comparing the effect on PTH suppression of the calcimimetic agent cinacalcet plus low-dose calcitriol analogues to calcitriol analogues alone, examined the effect on FGF-23 in 91 HD patients.61 The study reported a significant 9.7% decrease in FGF-23 levels in the cinacalcet group, with these changes significantly related to alterations in calcium and phosphate concentrations but not PTH. Effects on FGF-23 were also studied in 40 normo-phosphatemic patients with CKD stages 3–4 and elevated PTH, when comparing phosphate binder treatment with calcium acetate or sevelamer therapy over a 6 week period.62 FGF-23 levels decreased from 107 to 54 pg/mL in the sevelamer group (P < 0.05), with non-significant reduction in the calcium carbonate group, and a decrease in PTH was reported in both groups. Another prospective study of 46 HD patients assessed the effect of sevelamer and calcium carbonate compared with calcium carbonate alone,63 reporting that after four weeks of treatment phosphate and FGF-23 levels were significantly lower in the combination group. Analysis of the ArMMOR study involving 10 044 HD patients, revealed a benefit of phosphate lowering which extended to patients whose phosphate levels did not meet current treatment guidelines, and was only marginally attenuated upon stratification for phosphate levels.64 These findings raise the possibility that the benefit of phosphate binders may extend beyond lowering phosphate alone, and could be mediated through a decrease in FGF-23 levels.
FGF-23 and transplantation
The impact of renal transplantation on FGF-23 levels has also been studied. FGF-23 levels are reported to remain elevated in the first few months post-renal transplantation compared with matched controls with a similar eGFR; however, this effect diminishes after 12 months.65–67 High FGF-23 prior to transplantation is independently associated with post-transplant hypophosphatemia and low calcitriol.66 Excess FGF-23, in addition to elevated PTH levels and calcineurin inhibitors, may therefore be another mechanism for post-transplantation hypophosphatemia. In a small study of 10 transplant recipients with persisting SHPT, cinacalcet was associated with a significant decrease in PTH and FGF-23 levels, although the reduction in phosphaturia was more strongly correlated with a reduction in PTH levels.68
FGF-23 has the potential to influence how and when we treat patients with CKD-MBD. The temptation to integrate FGF-23 measurements into current clinical practice should be cautioned by the many questions that still remain unanswered. The exact role of FGF-23, the determination of its ‘normal’ range and variation, and the association of FGF-23 with dietary phosphate intake and mediators that affect its secretion all need to be further delineated. It is clear that FGF-23 plays a significant role in mineral metabolism and mediates changes that lead to SHPT in CKD; however, we have a fragmented understanding of the factors that mediate the elevation of FGF-23 in CKD. The effects of bone-derived FGF-23 regulators and local tissue phosphate and calcitriol concentrations on FGF-23 levels are of particular interest. With the recognition that the activity of extra-renal 1α-hydroxylase activity is important in CKD patients, the need to understand the effects of FGF-23 on this enzyme is paramount.
The plethora of studies linking FGF-23 with various biochemical and clinical outcomes are largely observational. There still remains a paucity of data outlining FGF-23 measurements in the various CKD subgroups, and prospective clinical studies are lacking. The postulated direct, toxic effects of FGF-23 on tissues, in particular the CV system, remain largely theoretical. The association between FGF-23 and phosphate also raises the question of treating phosphate levels within the currently accepted ‘normal range’. The clinical utility of FGF-23 in CKD may be as a diagnostic and prognostic biomarker; however, its use as a ‘universal’ therapeutic target for the various CKD-MBD treatments needs further evaluation. The use of FGF-23 in this capacity may parallel some of the controversies associated with PTH measurements. Mouse models and inherited disorders of phosphate metabolism highlight the detrimental effects of FGF-23 levels at the extremes of normal; however, debate will continue as to what constitutes an appropriate and optimal target level in CKD.
The FGF-23 holds some promise as a novel marker of CKD-MBD, particularly in early CKD, and as a potential tool to monitor the efficacy for therapies used to treat this disorder. The significance and potential role of FGF-23 in clinical practice needs to be established, with large, prospective, clinical trials. These will determine whether FGF-23 is a more useful biomarker of CKD-MBD when compared with phosphate or PTH.
MD would like to acknowledge the support of the Royal Australasian College of Physicians Research Foundation and the Jacquot Awards.