As nature did not predict dialysis--what we can learn from FGF23 in end-stage renal disease?

Abstract

The fibroblast growth factor 23 (FGF23)/klotho system has raised a lot of attention in the past 10 years. Genetic knockout models demonstrated that failure of this system produces a phenotype of premature ageing and especially a failure to excrete phosphate resulting in significant hyperphosphataemia and vascular calcification [1,2]. FGF23 is synthesized in the bone, and osteocytes increase FGF23 production in response to elevated phosphate and calcitriol [3]. Therefore, FGF23 may be a key adaptive factor preventing early hyperphosphataemia in progressive chronic kidney disease (CKD). In the preterminal phases of CKD, FGF23 may become a valuable biomarker of phosphate load and phosphate exposure, perhaps analogous to the predictive value of HbA1C in the evaluation of diabetes control. The central target organ of FGF23 appears to be the kidney, where tubular phosphate reabsorption and 1-alphahydroxylase expression are suppressed. These features raised the question which role FGF23 might play in dialysis patients (CKD stage 5D), the CKD stage where end-stage kidney failure is firmly established and neither substantial phosphaturic effects can be caused nor an already almost lost calcitriol synthesis can be substantially further suppressed. The European study by Jean and colleagues in this issue of the journal came to similar results as the US analysis by Gutierrez et al., using a somewhat, but not substantially, different clinical study approach [4,5]. While the latter study had exclusively focussed on incident dialysis patients not on active vitamin D treatment employing a ‘nested’ analysis, the current cross-sectional study reports on a 2-year follow-up of prevalent dialysis patients independent of their active vitamin D therapy. Both investigations found FGF23 serum levels to be extremely high and remaining closely correlated to serum phosphate levels. Thus, FGF23 regulation is apparently regulated in a similar way in dialysis patients as in progressive non-terminal CKD, despite an obvious loss of its intended biological actions on kidney tissue. However, this does not come as a total surprise, since nature could not predict the invention of renal replacement therapy when establishing such a potent antiphosphataemic protection system to protect against premature hyperphosphataemia on the route to renal failure. The key finding, however, is that in the course of both studies FGF23 has proven to be a strong and independent predictor of mortality, even in serum phosphate ranges which are presently considered relatively safe, at least according to KDOQI target levels for stage CKD 5D (<5.5 mg/dl/<1.78 mmol/l). The apparent link between serum phosphate and FGF23 in this dialysis population raises the question of whether FGF23 may reflect more appropriately the biological harm of individual phosphate levels than the actual phosphate levels themselves (illustrated in Figure 1)? In favour of this theory is the observation by Jean et al. that FGF23 was also apparently related to vascular calcification. Gutierrez et al. just published a link between FGF23 levels and left ventricular hypertrophy in CKD patients [6]—a phenomenon that could potentially be caused by arterial stiffness and increased pulse wave velocity. Alternatively, we should possibly examine whether such enormously elevated FGF23 concentrations, as those present in dialysis patients, are capable of inflicting damage at any tissue sites other than the kidney (such as the myocardium), perhaps by actions on other FGF receptors (FGFR) than on the prototypical dimeric FGFR:klotho receptor. There are two observations in the study by Jean et al. that deserve close attention as hypothesis-generating features: