Intact fibroblast growth factor 23 levels predict incident cardiovascular event before but not after the start of dialysis

Abstract

Purpose

Low 25-hydroxyvitamin D (25D), increased levels of fibroblast growth factor 23 (FGF23), parathyroid hormone (PTH), and alkaline phosphatase (ALP) were reported to be risk factors for mortality in chronic kidney disease (CKD). However, the independent associations of these factors with cardiovascular disease (CVD), the leading cause of death among CKD patients, remain unclear. Our purpose was to identify which of these factors predict incident CVD in CKD.

Methods

In this prospective cohort study, we enrolled 738 predialysis outpatients in the two nephrology departments. We employed Cox proportional hazards analyses to elucidate predictors of the endpoint, defined as fatal or non-fatal cardiovascular event requiring hospitalization. Multiple imputation was performed for missing values.

Results

Mean estimated glomerular filtration rate (eGFR) was 35 mL/min/1.73 m². During a median duration of 4.4 years, 86 patients developed the endpoint, of whom 62 patients achieved it before the initiation of dialysis. Multivariable analyses revealed that high serum intact FGF23 levels predicted the outcome preceding dialysis initiation (hazard ratio (HR) per lnFGF23 (SD), 1.64 (1.27–2.30)), while 25D, PTH, and bone-specific ALP did not. Adding FGF23 to the conventional model of age, sex, diabetes, prior CVD, pulse pressure, and eGFR, led to a net reclassification improvement of 6.87% (P = 0.04). Not censoring the patients at the start of dialysis and continuing follow-up even after dialysis, FGF23 levels did not predict the outcome (HR, 1.16 (0.91–1.48)). Complete case analyses yielded similar results.

Conclusions

Intact FGF23 levels in predialysis CKD predicted incident cardiovascular events requiring hospitalization before starting dialysis, but did not predict events during the entire follow-up period, including post dialysis initiation.

Introduction

CKD-MBD is an umbrella concept, originally established in dialysis patients, which covers laboratory abnormalities, bone abnormalities and ectopic calcifications [1]. In dialysis patients, studies have revealed that laboratory abnormalities in CKD-MBD are risk factors for incidence of CVD [2], [3]. However, data are still scarce among CKD patients not yet on dialysis. Candidates of laboratory abnormality as a risk factor include low concentrations of 25D and high levels of FGF23, PTH, and ALP or BSAP, in addition to hyperphosphatemia [4], [5].

Low circulating 25D [6], [7], [8] and high PTH [9] levels were reported to be related to mortality in CKD patients. Previous studies showed that approximately half of the deaths in CKD patients were from cardiovascular causes [10], [11]. In this regard, these laboratory abnormalities might be associated with cardiovascular events in CKD patients. With regard to PTH, the association with prevalent CVD was reported [12], but not for the association with incident CVD.

ALP is an enzyme that catalyzes the hydrolysis of pyrophosphate which has inhibitory effects on vascular calcification [13]. In fact, high ALP levels was reported to be associated with severe coronary artery calcification in dialysis patients, [14] and also reported to be a risk factor for mortality in both hemodialysis [15] and predialysis patients [16]. Among CKD patients without liver disease, ALP levels increase due to high turnover bone disease. Recently, it was found in dialysis patients [17] that the effect sizes related to cardiovascular and non-cardiovascular mortalities were much higher for BSAP in comparison with ALP. However, little data exists in CKD patients not on dialysis.

FGF23 is a bone-derived hormone that maintains phosphate homeostasis and vitamin D metabolism [18] and increases as renal function declines [19], [20]. Among patients with CKD, elevated FGF23 levels were shown to be associated with left ventricular hypertrophy [21] and endothelial dysfunction [22], which are known to be risk factors for cardiovascular events and deaths [23], [24], [25]. These results suggest a significant association between FGF23 and CVD in CKD. This association was recently reported by a small study [26], in which it did not adjust for important confounder PTH. Therefore, studies of the association between FGF23 and future CVD are still scarce and insufficient in those with CKD not on dialysis.

Observational studies among pre-dialysis CKD patients have constituted insufficient proof of the association of laboratory abnormalities with incident CVD and no study has measured all of these MBD-biomarkers in one study, to our knowledge. In the present study, we measured serum 25D, FGF23, PTH, and BSAP at the same time. The aims of this study were (1) to clarify which of the four markers predict incident cardiovascular events in patients with CKD and (2) to assess the extent to which the addition of significant MBD-biomarkers to the conventional model improves the accuracy of prediction.