Chronic kidney disease (CKD) is associated with significantly elevated mortality rates, predominantly driven by cardiovascular (CV) complications, which remain the principal cause of death among these patients.1,2 Therefore, the early identification of reliable biomarkers capable of predicting adverse outcomes is paramount for improving risk stratification and implementing effective therapeutic strategies to halt disease progression and reduce mortality.

Disruptions in mineral metabolism are a hallmark of CKD, often emerging at its earliest stages.3,4 Fibroblast growth factor 23 (FGF-23) has garnered considerable attention among the key players. The FGF-23 levels rise before the onset of overt hyperphosphatemia, acting as an early compensatory mechanism to counter phosphate (Pi) retention. Through its interaction with Klotho-FGF receptor complexes in the kidneys, FGF-23 promotes urinary Pi excretion and suppresses the synthesis of 1,25-dihydroxyvitamin D (1,25(OH)2D).5 Activating Klotho and increasing fractional excretion of phosphate (FEp), serves as a compensatory mechanism to maintain normal serum Pi levels to match dietary intake.6 This response precedes elevations in serum Pi and parathyroid hormone (PTH).7 Experimental and clinical investigations have implicated FGF-23 in left ventricular hypertrophy (LVH), myocardial fibrosis, soft tissue calcifications, and heart failure, independent of traditional CV risk factors.8,9 Persistent hyperphosphatemia, secondary hyperparathyroidism, and vitamin D deficiency have been implicated as risk factors for both all-cause and CV mortality disease.10–12 Nevertheless, the causality of these associations remains a subject of ongoing debate. Although elevated FGF-23 and Pi levels are consistently linked to higher mortality risk in CKD, the precise thresholds at which these factors transition from compensatory to maladaptive remain undefined.13 Recent research suggests that the pathogenic potential of FGF-23 may depend on the co-occurrence of hyperphosphatemia and decreased Klotho expression, a combination frequently observed in advanced CKD.13 Importantly, the early rise in FGF-23 levels—even among normophosphatemic patients—may reflect an underlying inability to effectively regulate Pi homeostasis. This “maladaptive” elevation of FGF-23 could serve as a critical signal, identifying individuals with early-stage CKD who are at higher risk of adverse CV outcomes and who may benefit from intensified Pi management strategies. In addition, prior studies have highlighted that even Pi concentrations within the upper normal reference range are associated with adverse outcomes, suggesting that subtle disturbances in Pi metabolism may be clinically relevant.14

However, associations are not causality. Although hyperphosphatemia and FGF-23 are associated with increased mortality in CKD15,16 it is not clear whether compensatory increases in FGF-23 secretion induce or protect against the increased CV morbidity and mortality observed in CKD. Furthermore, the precise reference values for FGF-23 and Pi, beyond which adverse effects start to be manifested, remain unclear, and the comprehensive impact of each factor involved is not fully understood. In this study, we aimed to examine the relationship of FGF-23 and serum Pi with other biomarkers of kidney disease and CV risk in a non-dialysis CKD population and sought to explore their role as biomarkers of adverse outcomes and mortality during a five-year follow-up period.

In the present study, we examined, in a non-dialysis CKD population under outpatient nephrological care, with renal function spanning all five CKD stages, the associations between the serum levels of iFGF-23 and Pi with CV disease biomarkers and explored prospectively their role as predictors of adverse events and mortality in a 5-year follow-up period. In adjusted models, only serum Pi (independent of other confounders except GFR) remained significantly associated with all-cause mortality. In unadjusted analysis, we found that both iFGF-23 and Pi levels were associated with baseline CV risk factors as well as with a composite CV outcome during follow-up. However, in adjusted analysis, we found that serum Pi levels, but not iFGF-23 were associated with mortality among CKD patients, independent of other confounding factors except GFR. Segmentation of the population into terciles according to serum Pi levels at baseline showed a significant increase in mortality during follow-up among patients with serum Pi levels within the high-“normal” range but not in patients with serum Pi levels below “normal” targets. These findings reinforce the view that high-normal Pi concentrations may carry clinical significance in CKD and are supported by previous studies in both the general population18 and non-dialysis CKD cohorts.19 They also suggest that current reference values for serum Pi in non-dialysis CKD patients may benefit from re-evaluation.20

Several mechanisms may underlie the pathogenicity of elevated serum Pi in CKD, including promotion of vascular calcification via osteoblastic transformation of vascular smooth muscle cells and calcium phosphate deposition,21,22 LVH and myocardial fibrosis23 and statin-resistance due to altered lipid metabolism.24 Indeed, a secondary analysis of the AURORA trial revealed that higher levels of serum Pi may blunt the CV benefit of statin treatment in dialysis patients.24 Endothelial dysfunction has also been implicated in phosphate's pathogenicity. Translational models have shown that Pi loading can impair endothelial function, likely through disruption of nitric oxide signaling.25 Stevens et al. showed that sustained oral Pi loading in 19 healthy volunteers caused endothelial dysfunction, which was accompanied by significant increases in serum FGF-23 and urinary Pi excretion, without significant changes in serum Pi levels.26 Recent data confirm that glycerol-3 phosphate (G3P), secreted by the proximal renal tubular cells, increases in response to Pi loading, with a subsequent increase in FGF-23 before serum Pi concentration rises.27 The deleterious influence of Pi load on endothelial function may be reversible and assume relevance in the CKD population, in which the compromised renal excretion of ingested Pi is counteracted by an increase in FGF-23 levels from the earliest CKD stages. We previously reported in a group of CKD patients with a mean GFR of 49mL/min that the restriction of Pi intake for 14 days from a baseline intake of 1100mg/day to 700mg/day was accompanied by a significant improvement in endothelial function, going along with non-significant reductions in FGF-23, iPTH, and Pi levels.28

As shown in our cohort, the FEp increased significantly with declining renal function, consistent with compensatory responses to early Pi retention. This increase may reflect enhanced FGF-23-mediated phosphaturia in the setting of reduced nephron mass. However, despite this adaptation, serum Pi levels still predicted mortality, and prior studies have demonstrated that heart failure and CV mortality are independent of FEp, in patients with moderate to advanced stages of CKD.29 Therefore, we consider that Pi pathogenicity begins early, even at levels deemed normal, and that its management should be anticipated to prevent CV complications.

In the present study iFGF-23 levels increased with declining renal function and were significantly associated with CV disease and risk factors at baseline, thus reinforcing the role of FGF-23 as an indicator of established CV disease in CKD.30,31 However, iFGF-23 levels were not independently associated with the composite CV outcome or mortality during follow-up. This is in line with findings from previous studies,32–34 and reinforces the view that iFGF-23 may function more as a biomarker of disease burden than as a causal factor.35 Indeed, many controversies exist regarding FGF-23 levels and its role as a predictor versus biomarker of CV disease.32 Elevated FGF-23 levels are not specific to CKD and were also associated with various off-target effects on multiple organs, including fractures,36 infections,37 inflammation,38 hereditary hypophosphatemic rickets39 and even prostate cancer.40 These associations across a wide range of disease states could reflect the pleiotropy of FGF-23 in disease causation. Hereditary disorders of FGF-23 excess share a common biochemical phenotype, including low Pi and normal or low-1,25-(OH)VitD that can be treated with a human monoclonal antibody targeted against FGF-23. Notably, it has been reported that treatment with an anti-FGF-23 monoclonal antibody, although correcting rickets, increased serum Pi to the lower end of the normal range and elevated 1,25(OH)2D levels, was also associated with the development of cardiac calcification.41,42

In this study, a significant proportion of our cohort received vitamin D supplementation, which may influence FGF-23 levels.43 Moreover, other authors have reported that elevated FGF-23 combined with low vitamin D were associated with adverse outcomes.44 Notably, subgroup analysis showed consistent results across treated and untreated groups. As with iFGF-23, the iFGF-23/25(OH)D ratio was also associated with the composite CV outcome only in univariate analysis, however, we acknowledge that FGF-23 more strongly influences 1,25(OH)2D than 25(OH)D levels, and that the generally adequate 25(OH)D levels in our cohort may help explain the absence of significant effects. Toward a deeper understanding of the pathophysiological role of FGF-23 in CKD-associated CV disease, further mechanistic studies may be warranted.

Overall, these findings reinforce the view that early elevation of FGF-23—even in normophosphatemic patients—may represent the impaired ability to maintain Pi balance. This “maladaptive” increase may help identify high-risk individuals who might benefit from intensified therapeutic strategies targeting Pi control.45 Parallel to this, a lower target for Pi levels may warrant additional benefit regarding CV outcomes, particularly in CKD.

We acknowledge some limitations of our study. First, it is a single-center study with a relatively small number of patients and outcomes; second, the results in our ethnically homogeneous population may not be generalizable to other ethnic groups; third, the “snapshot” evaluation of Pi and FGF-23 levels at baseline could underestimate the association with the outcomes during follow-up.

Our study has also strengths that should be emphasized: First, its prospective nature and the significant follow-up period of nearly five years; second, the inclusion of patients from both genders and spanning all five CKD stages; third, the assessment of the intact biologically active isoform of FGF-23 rather than the inactive C-terminal FGF-23 fragment. The data presented here offers valuable preliminary insights that can serve as a foundation for future, larger-scale and more targeted investigations.

In conclusion, our results agree with the view that serum Pi levels within the high-normal range, are a risk predictor for mortality in non-dialysis CKD patients. Our results also suggest that the contribution of iFGF-23 as a biomarker to CV outcomes and mortality in CKD may not outweigh the role of serum Pi levels within the high-“normal” range. Because serum Pi is a biomarker easily affordable, our findings reinforce the need to re-examine contemporary guidelines for serum Pi reference levels in the non-dialysis CKD population. This may allow the establishment of more timely interventions, whether through dietary adjustments or pharmacological measures, aimed at enhancing overall Clinical Outcomes in the CKD Population.

Conceptualization, A.C., J.Q.-S. and M.P.; methodology, A.C., J.Q.-S., N.P; software, A.C., J.Q.-S., and C.C.D.; validation, A.C., J.Q.-S. and C.C.D.; formal analysis, A.C., J.Q.-S. and C.C.D.; investigation, A.C., J.Q.-S., N.P.; C.C.D. and M.P.; resources, A.C., J.Q.-S., N.P.; C.C.D. and M.P.; data curation, A.C., J.Q.-S., N.P. and C.C.D.; writing—original draft preparation, A.C., J.Q.-S., and M.P.; writing—review and editing, A.C., J.Q.-S. and M.P.; visualization, A.C., J.Q.-S., N.P.; C.C.D. and M.P.; supervision, M.P. and J.Q.-S.; project administration, J.Q.-S. and M.P.; funding acquisition, M.P., J.Q.-S. and A.C.

All authors have read and agreed to the published version of the manuscript.

Informed consent was obtained from all subjects involved in the study.

This work was financed by FEDER-Fundo Europeu de Desenvolvimento Regional funds through the COMPETE 2020-Operacional Programme for Competitiveness and Internationalisation (POCI), Portugal 2020, and by Portuguese funds through FCT-Fundação para a Ciência e a Tecnologia/Ministério da Ciência, Tecnologia e Ensino Superior in the framework of the project “Institute for Research and Innovation in Health Sciences” (POCI-01-0145-FEDER-007274), and a grant from Portuguese Society of Nephrology.

All the authors declare that there is no conflict of interest.

The data presented in this study are available on request from the corresponding author.