SEARCH

SEARCH BY CITATION

Keywords:

  • FGF-23;
  • FRACTURES;
  • BONE MINERAL DENSITY;
  • BMD;
  • VITAMIN D;
  • CALCITRIOL;
  • RICKETS;
  • OSTEOMALACIA

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

A normal mineral metabolism is integral for skeletal development and preservation of bone integrity. Fibroblast growth factor 23 (FGF-23) is a bone-derived circulating factor that decreases serum concentrations of inorganic phosphorous (Pi) and 1,25-dihydroxyvitamin D3 [1,25(OH)2D3]. Increased FGF-23 expression is a direct or indirect culprit in several skeletal disorders; however, the relation between FGF-23 and fracture risk remains undetermined. We evaluated the prospective relation between serum intact FGF-23 (measured by a two-site monoclonal antibody ELISA) and fracture risk employing the Swedish part of the population-based Osteoporotic Fractures in Men Study (MrOS; n = 2868; mean age 75.4 ± 3.2 years; median follow-up period 3.35 years). The incidence of at least one validated fracture after baseline was 20.4 per 1000 person-years. FGF-23 was directly related to the overall fracture risk [age-adjusted hazard ratio (HR) per SD increase = 1.20, 95% confidence interval (CI) 1.03–1.40] and vertebral fracture risk (HR = 1.33, 95% CI 1.02–1.75). Spline models revealed a nonlinear relation between FGF-23 and fracture risk, with the strongest relation at FGF-23 levels above 55.7 pg/mL. FGF-23 levels above 55.7 pg/mL also were associated with an increased risk for hip and nonvertebral fractures (HR = 2.30, 95% CI 1.16–4.58, and HR = 1.63, 95% CI 1.01–2.63, respectively). These relations remained essentially unaltered after adjustment for bodymass index (BMI), bone mineral density (BMD), glomerular filtration rate, 25(OH)2D3, parathyroid hormone (PTH), and other fracture risk factors. In conclusion, FGF-23 is a novel predictor of fracture risk in elderly men. © 2011 American Society for Bone and Mineral Research.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Fibroblast growth factor 23 (FGF-23) is a circulating factor expressed predominantly in osteoblasts and osteocytes1, 2 that negatively regulates serum levels of inorganic phosphorous (Pi) and 1,25-dihydroxyvitamin D3 [1,25(OH)2D3].3 Overexpression of FGF-23 in bone is an important pathogenic factor in several human skeletal disorders such as autosomal dominant hypophosphataemic rickets (ADHR),4 tumor-induced osteomalacia (TIO),5 fibrous dysplasia/McCune-Albright syndrome (FD/MAS),1 X-linked hypophosphataemic rickets (XLH),6 and autosomal recessive hypophosphatemic rickets (ARHR).7, 8 Many of these patients suffer from generalized osteomalacia and skeletal deformities and are at high risk for having insufficiency fractures. In concert, Hyp mice (a murine model of XLH) and mice overexpressing FGF-23 (transgenic Fgf23 mice) similarly suffer from demineralization and reduced bone mineral content (BMC), as well as growth retardation and decreased cortical and trabecular bone mineral density (BMD).9, 10

We previously failed to identify an independent relation between circulating FGF-23 and BMD in elderly men.11 However, given the essential role of FGF-23 in mineral metabolism and its putative suppressive effect on bone matrix mineralization,12 we hypothesized that FGF-23, independent of BMD and renal function, may affect fracture risk. Accordingly, we explored the association between a single serum FGF-23 measurement and fracture risk in a large prospective population-based cohort of elderly Swedish men.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Study subjects

The MrOS study is a multicenter prospective study including older men in Sweden, Hong Kong, and the United States. The Swedish MrOS cohort (n = 3014) consists of three subcohorts from three different Swedish cities (n = 1005 in Malmö, n = 1010 in Göteborg, and n = 999 in Uppsala).13 Study subjects (men aged 70 to 81 years) were randomly identified using national population registers, and a total of 45% of the subjects who were contacted participated in the study between October 2001 and December 2004. To be eligible for the study, the subjects had to be able to walk without aids. There were no other exclusion criteria. Informed consent was obtained for all subjects, and the study was approved by the local ethics committees at the Universities of Malmö (ethical approval number LU-693-00), Göteborg (Gbg M 014-01), and Uppsala (Ups 01-057) and conducted in accordance with the guidelines in the Declaration of Helsinki. For this study, 2868 subjects with complete measurements of FGF-23, fractures, and other parameters were included.

Assessment of covariates

All measurements were carried out by the same trained staff at the baseline visit. Height was measured using a Harpender stadiometer, and weight was measured by an electric scale to the nearest 0.1 kg. Body mass index (BMI) was calculated as weight/height2 (kg/m2). Grip strength (kg) was analyzed using Baseline equipment (Baseline, Inc., Chattanooga, TN, USA). Time to complete a “narrow walk” (6 m × 20 cm) and ability to rise from a chair without arms were assessed.14 Smoking, occurrence of fracture after age 50, fall occurrence of the latest 12 months, and overall health rate were determined at baseline by a questionnaire.

Assessment of fractures

Participants were followed for a median of 3.35 years after the baseline examination. The follow-up time was recorded from the date of the baseline visit to the date of the first fracture or the date of death. Central registers covering all Swedish citizens were used to identify the subjects who experienced fractures and the time of death for subjects who died during the study. X-ray archives in Malmö, Göteborg, and Uppsala were searched for new fractures occurring after the baseline visit using the unique personal citizenship registration number of the subjects.15 All fractures that were reported by the study subjects after the baseline visit were confirmed by physician review of radiology reports. Fractures reported by the study subjects that were not possible to confirm by the X-ray report were not included in this study. All validated fractures were included in the main analyses, followed by exploratory subanalyses of fracture type. In the latter, we studied the associations between serum intact FGF-23 and validated fractures, divided into three main groups: (1) X-ray-verified clinical vertebral fractures, (2) nonvertebral osteoporosis fractures at the major osteoporosis-related locations (defined as hip, distal radius, proximal humerus, and pelvis), and (3) other fractures (ie, radius/ulna, hand, fingers, humerus, elbow, skull, cervical vertebrae, clavicle, scapula, rib, femoral shaft, patella, upper tibia, ankle, foot, and toes).15

Assessment of BMD and body composition (DXA)

Areal BMD (aBMD, g/cm2) of the femoral neck was assessed using the Lunar Prodigy DXA (n = 2004 from the Uppsala and Malmö cohorts; GE Lunar Corp., Madison, WI, USA) or Hologic QDR 4500/A-Delphi (n = 1010 from the Göteborg cohort; Hologic, Waltham, MA, USA). The coefficients of variation for the aBMD measurements ranged from 0.5% to 3%. To be able to use DXA measurements performed with equipment from two different manufacturers, a standardized BMD (sBMD) was calculated, as described previously.13 Furthermore, total-body fat and lean mass were assessed as allowed by the DXA measurement.

Serum analyses

All plasma and serum samples were collected at 8 a.m. after at least 10 hours of fasting and nonsmoking. Samples were frozen immediately and stored at −80°C. Cystatin C was analyzed with polyclonal antibodies against human cystatin C and measured by immunoturbidimetry (Cystatin C Immunoparticles, Dako Denmark A/S, Glostrup, Denmark). Estimated glomerular filtration rate (eGFR) was calculated using the following estimate: GFR = 79.901 × (cystatin C)−1.4389. This proxy for GFR has good precision, good linearity, and strong correlation with iohexol clearance (R2− = 0.956).16 Serum parathyroid hormone (PTH) was analyzed using the Immulite 2000 Intact PTH Assay (Diagnostic Products Corporation, Los Angeles, CA, USA), and 25(OH)2D3 was measured on the Nichols Advantage automated assay system (Nichols Institute Diagnostics, San Juan Capistrano, CA, USA). All other serum biochemistries were measured by routine techniques at each site.

Fibroblast growth factor 23 (FGF-23)

Intact FGF-23 was measured using an ELISA according to the manufacturer's protocol (Kainos Laboratories International, Tokyo, Japan).17 This second-generation two-site monoclonal antibody ELISA has been shown previously to recognize the biologically active intact FGF-23.17 The Kainos Intact FGF-23 assay has a lower limit of detection of 3 pg/mL and intraassay and interassay coefficients of variation of less than 5%. The Kainos Intact assay was the most sensitive among three different two-site enzyme-linked immunosorbent assays for FGF-23 measurements.18

Statistical analysis

Initially, the distributional properties of all baseline variables were examined. Data are presented as mean ± SD, except for the nonnormally distributed variables, which are presented as median (10th to 90th percentiles). Fracture rates were expressed as the number of subjects with a validated first fracture per 1000 person-years and 95% confidence intervals (CIs). Cox proportional-hazards regression models with age during follow-up as the underlying time scale19 were used to study the associations between serum intact FGF-23 and fracture outcomes. Proportional-hazard assumptions were confirmed by inspecting Schoenfeld residuals and linearity assumptions by inspecting Martingale residuals. All validated fractures were included in the main analyses, followed by exploratory subanalyses of fracture type. Age and BMI were included as covariates in the first multivariate model, followed by a model in which BMD also was included. Additional potential confounders then were added and retained in the model if addition changed the HR for FGF-23 by more than 10%. Confounders were serum biochemistries [ie, phosphate, calcium, PTH, and 25(OH)2D3], eGFR, grip strength, time to complete a “narrow walk,” ability to rise from a chair without aid, smoking, occurrence of fracture after age 50, previous 12 months fall occurrence, overall health rate, and body composition. All models were stratified by the study center.

Three methods were used to evaluate the association between FGF-23 and time to first fracture. FGF-23 first was modeled as a continuous independent variable (per SD change) and also as quartiles based on the FGF-23 distribution in the cohort. Because subjects in the first, second, and third FGF-23 quartiles had similar risks of fracture, we created a dichotomous variable where the highest FGF-23 quartile was compared with the other three quartiles. Second, cubic splines with three knots placed at the 5th, 50th, and 95th percentiles were used to allow for nonlinear effects of FGF-23. The splines are restricted to be linear below the first knot point and above the last knot point.20 Third, we performed exploratory cutpoint analysis. We dichotomized FGF-23 at various quantiles using log likelihoods of Cox proportional-hazard models. The cutpoint at which FGF-23 was dichotomized to produce the highest profile log-likelihood was considered the best value for further dichotomizing. The cutpoint analyses supported the use of the highest quartile as a cutpoint. All analyses were performed using SAS Version 9.2 for Windows (SAS Institute, Cary, NC, USA).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Baseline examination

In total, 2868 subjects with serum FGF-23 analyses available were included and followed prospectively for an average of 3.3 ± 1.0 years (median 3.35 years, ranging from 6 days to 6.15 years) after the baseline examination. Baseline characteristics are shown in Table 1. Median serum FGF-23 concentration at the baseline visit was 43.6 pg/mL (10th and 90th percentiles, 23.5 and 74.6 pg/mL, respectively), and the follow-up time was recorded from the date of the baseline visit to the date of the first fracture (depending on the fracture type) or the date of death. Baseline characteristics over FGF-23 quartiles are shown in Table 2.

Table 1. Baseline Characteristics of the Study Subjects in MrOS Sweden
 MrOS Sweden (n = 3014)
  1. Note: Values are mean ± SD for normally distributed continuous variables and median (10th to 90th percentiles) for nonnormally distributed variables. For fractures, the number of subjects with a validated first fracture is given with the incidence per 1000 person-years and 95% CI within parentheses. Some subjects included in the group of “All fractures” had more than one type of first fracture, and therefore, these subjects were included in more than one of the different subtypes of fractures. Nonvertebral osteoporosis fractures are defined as fractures in the hip, distal radius, proximal humerus, and pelvis.15 Other fractures include all validated fractures minus nonvertebral osteoporosis fractures and clinical vertebral fractures.15

Demographics
 Age (years)75.4 ± 3.2
 Weight (kg)80.7 ± 12.1
 Height (cm)174.8 ± 6.5
 Body mass index (kg/m2)26.4 ± 3.6
Serum biochemistry
 Serum phosphate (mmol/L)1.07 ± 0.16
 Serum calcium (mmol/L)2.3 (2.2–2.5)
 Serum 25(OH)2D3 (nmol/mL)69.6 ± 23.6
 Serum intact PTH (pmol/mL)4.4 (2.2–7.7)
 Serum FGF-23 (pg/mL)43.6 (23.5–74.6)
 Estimated GFR (mL/min/1.73 m2)72.0 ± 20.5
Fractures
 All fractures194 (20.4, 95% CI 17.7–23.4)
 Nonvertebral osteoporosis fractures78 (8.1, 95% CI 6.5–10.1)
 Hip35 (3.6, 95% CI 2.6–5.0)
 Clinical vertebral fractures63 (6.4, 95% CI 5.0–8.3)
 Other fractures71 (7.3, 95% CI 5.8–9.2)
Table 2. Baseline Characteristics of the Study Subjects in MrOS Sweden Over FGF-23 Quartiles
 FGF-23 quartile 1FGF-23 quartile 2FGF-23 quartile 3FGF-23 quartile 4
  1. Note: Values are mean ± SD for normally distributed continuous variables and median (10th to 90th percentiles) for nonnormally distributed variables. For fractures, the number of subjects with a validated first fracture is given with the incidence per 1000 person-years and 95% CI within parentheses.

Age (years)75.4 ± 3.175.3 ± 3.275.4 ± 3.275.6 ± 3.2
Weight (kg)77.9 ± 11.380.7 ± 12.381.1 ± 11.583.0 ± 12.7
Height (cm)174.3 ± 6.3174.6 ± 6.6175.0 ± 6.5175.1 ± 6.8
Body mass index (kg/m2)25.6 ± 3.426.4 ± 3.626.5 ± 3.327.0 ± 3.8
Serum phosphate (mmol/L)1.08 ± 0.171.06 ± 0.171.07 ± 0.161.08 ± 0.16
Serum calcium (mmol/L)2.4 (2.2–2.6)2.3 (2.2–2.6)2.3 (2.2–2.5)2.4 (2.2–2.5)
Serum 25(OH)2D3 (nmol/mL)70.1 ± 23.067.9 ± 23.168.4 ± 22.772.1 ± 25.3
Serum intact PTH (pmol/mL)3.8 (1.9–6.6)4.4 (2.2–7.5)4.4 (2.4–7.7)4.9 (2.4–9.5)
Serum FGF-23 (pg/mL)25.3 (13.8–30.9)37.8 (33.2–42.1)49.5 (44.5–55.4)69.9 (59.3–110.7)
Estimated GFR (mL/min/1.73 m2)80.1 ± 18.875.3 ± 18.471.0 ± 17.461.8 ± 22.6
All fractures41 (16.4, 95% CI 12.0–22.2)43 (17.5, 95% CI 12.9–23.5)47 (20.0, 95% CI 15.0–26.6)63 (28.5, 95% CI 22.3–36.5)
Nonvertebral osteoporosis fractures19 (7.5, 95% CI 4.8–11.8)17 (6.8, 95% CI 4.2–10.9)16 (6.7, 95% CI 4.1–10.9)26 (11.6, 95% CI 7.9–17.1)
Hip8 (3.1, 95% CI 1.6–6.2)7 (2.8, 95% CI 1.3–5.8)5 (2.1, 95% CI 0.9–5.0)15 (6.6, 95% CI 4.0–10.9)
Distal radius8 (3.1, 95% CI 1.6–6.2)2 (0.8, 95% CI 0.2–3.2)7 (2.9, 95% CI 1.4–6.1)8 (3.5, 95% CI 1.7–7.0)
Proximal humerus4 (1.6, 95% CI 0.6–4.1)4 (1.6, 95% CI 0.6–4.2)5 (2.1, 95% CI 0.9–5.0)4 (1.7, 95% CI 0.7–4.6)
Pelvis1 (0.4, 95% CI 0.1–2.8)5 (2.0, 95% CI 0.8–4.8)1 (0.4, 95% CI 0.1–2.9)3 (1.3, 95% CI 0.4–4.0)
Clinical vertebral fractures13 (5.1, 95% CI 2.9–8.7)12 (4.8, 95% CI 2.7–8.4)13 (5.4, 95% CI 3.1–9.3)25 (11.0, 95% CI 7.4–16.3)
Other fractures14 (5.5, 95% CI 3.2–9.3)18 (7.2, 95% CI 4.5–11.5)21 (8.8, 95% CI 5.7–13.5)18 (7.9, 95% CI 5.0–12.5)

Relation of FGF-23 to any validated incident fracture

The impact of baseline serum FGF-23 level on the risk of any fracture was evaluated in the main analyses. During follow-up, 194 subjects (6.8%) had at least one validated incident fracture, and the incidence rate for any validated incident fracture was 20.4 (95% CI 17.7–23.4) per 1000 person-years (Table 1).

In crude models, 1 SD increase in the baseline serum FGF-23 level was associated with a 21% (CI 3%–41%) increased risk for any type of fracture. After further adjustment for BMI, 1 SD increase in baseline FGF-23 was associated with a 23% increased fracture risk (Table 3). Subjects in the highest FGF-23 quartile (>57.4 pg/mL) were at a 62% increased risk for fractures compared with those in the lowest three quartiles (HR = 1.62, CI 1.19–2.20). Kaplan-Meier survival curves according to this cutoff value are shown in Fig. 1A. The association remained consistent after adjustment for BMD with a slightly increased hazard ratio (Table 3).

Table 3. FGF-23 and the Risk for All and Clinical Vertebral Fractures
 Adjusted for ageAdjusted for age and BMIAdjusted for age, BMI, and BMDAdjusted for age, BMI, BMD, GFR, vitamin D, and PTH
  • Note: Values are HR (95% CI), Q1–Q4: FGF-23 quartiles 1 to 4.

  • *

    p < .05;

  • **

    p < .01;

  • ***

    p < .001.

Relation between FGF-23 and all fractures
 1 SD increase1.20* (1.03–1.40)1.23** (1.06–1.43)1.26** (1.08–1.47)1.24* (1.05–1.47)
 Q4 versus Q1–Q3 (>57.4 pg/mL)1.54** (1.14–2.09)1.62** (1.19–2.20)1.64** (1.20–2.25)1.56* (1.11–2.20)
Relation between FGF-23 and clinical vertebral fractures
 1 SD increase1.33* (1.02–1.75)1.39* (1.08–1.80)1.44** (1.10–1.88)1.56** (1.18–2.06)
 Q4 versus Q1–Q3 (>57.4 pg/mL)2.02** (1.21–3.38)2.26** (1.35–3.80)2.30** (1.36–3.90)2.72*** (1.54–4.79)
Relation between FGF-23 and nonvertebral osteoporosis fractures
 1 SD increase1.09 (0.86–1.38)1.11 (0.87–1.40)1.16 (0.91–1.47)1.06 (0.82–1.38)
 Q4 versus Q1–Q3 (>57.4 pg/mL)1.63* (1.01–2.63)1.68* (1.04–2.72)1.73* (1.06–2.83)1.50 (0.87–2.59)
Relation between FGF-23 and hip fractures
 1 SD increase1.18 (0.83–1.38)1.18 (0.82–1.69)1.29 (0.89–1.85)1.16 (0.78–1.72)
 Q4 versus. Q1–Q3 (>57.4 pg/mL)2.30* (1.16–4.58)2.30* (1.15–4.60)2.47* (1.22–5.01)2.18* (1.00–4.73)
thumbnail image

Figure 1. FGF-23 and the risk for all and clinical vertebral fractures. Kaplan-Meier survival curves according to the concentration of FGF-23 above and below the cutoff value (57.4 pg/mL) and the outcome of all fractures (A), clinical vertebral fractures (B), nonvertebral osteoporosis fractures (C), and hip fractures (D).

Download figure to PowerPoint

Relation of FGF-23 to different types of validated incident fractures

Next, we performed explanatory subanalyses of the predictive value of FGF-23 for different fracture types. Fractures were divided into three major categories: (1) nonvertebral osteoporosis-related fractures (hip, proximal humerus, distal radius, and pelvis), (2) clinical vertebral fractures, and (3) other fractures (Table 1).

In crude models, an increase in the baseline serum FGF-23 level was associated with a 34% (CI 2%–75%) increased risk for clinical vertebral fractures. After adjustment for age and BMI, 1 SD increase of baseline FGF-23 was associated with a 39% increased risk for clinical vertebral fractures (Table 3). Furthermore, subjects in the highest FGF-23 quartile (>57.4 pg/mL) were at a higher risk for clinical vertebral fractures than subjects in the lowest three quartiles (HR = 2.26, 95% CI 1.35–3.80). Kaplan-Meier survival curves for clinical vertebral fractures according to serum concentrations of FGF-23 above and below this cutoff value are shown in Fig. 1B. Again, the association remained consistent after adjustment for BMD with a slightly increased hazard ratio (Table 3).

Similarly, subjects in the highest FGF-23 quartile (>57.4 pg/mL) were at a higher risk for nonvertebral osteoporosis fractures (HR = 1.64, 95% CI 1.02–2.66) and hip fractures (HR = 2.34, 95% CI 1.18–4.65). Kaplan-Meier survival curves for nonvertebral osteoporosis fractures and hip fractures according to serum concentrations of FGF-23 above and below this cutoff value are shown in Fig. 1C, D. These associations remained significant in the multivariate adjusted model but did not reach statistical significance when modeling FGF-23 as a continuous variable (data not shown). We found no significant evidence for any association between FGF-23 and other fracture types (data not shown).

Consideration of covariates

The association between FGF-23 and the fracture risk remained consistent and essentially unaltered after adjustment for other potential confounders, including serum biochemistries [ie, Pi, calcium, PTH, 25(OH)2D3], grip strength, time to complete a “narrow walk,” ability to rise from a chair without aid, smoking, occurrence of fracture after age 50, previous 12 months fall occurrence, overall health rate, and body composition (data not shown).

Consideration of renal function

The majority of subjects had normal renal function (mean eGFR 72.0 ± 20.5 mL/min/1.73m2). eGFR was only marginally associated with all fracture risk [odds ratio (OR) = 0.85, 95% CI 0.73–0.98 for 1 SD increase in eGFR), and the association was blunted when adjusting for age, BMD, and FGF-23. To further eliminate the possibility that the association between FGF-23 and fracture risk is due to decreased renal function, we stratified our sample based on median eGFR (71.5 mL/min/1.73m2) into two groups—subjects above and below median eGFR. Importantly, whereas the association between FGF-23 and all fracture risk was statistically significant in subjects with eGFR above median (OR = 1.31, 95% CI 1.06–1.63), it did not reach statistical significance for subjects below median eGFR (OR = 1.14, 95% CI 0.91–1.44). Thus FGF-23 predicts fracture risk exclusively in subjects with a higher eGFR in this elderly population (Fig. 2). Finally, we did not find any significant evidence for an interaction between FGF-23 and eGFR (p > .05 for the interaction term).

thumbnail image

Figure 2. FFG-23 and the risk for all fractures stratified over median eGFR (71.5 mL/min/1.73m2). Kaplan-Meier survival curves according to the concentration of FGF-23 above and below the cutoff value (57.4 pg/mL) and the outcome of all fractures for subjects with eGFR below the median (A) and above the median (B).

Download figure to PowerPoint

Nonlinearity and threshold analyses

Spline analyses showed a nonlinear association between serum FGF-23 and risk for any fracture (p = .014 for nonlinearity; Fig. 3A) and at the threshold of significance for clinical vertebral fractures (p = .063 for nonlinearity; Fig. 3B). Log-likelihood cutpoint analysis showed that dichotomizing FGF-23 at 55.7 pg/mL maximized the model fit for all fractures. This threshold concentration is similar to that associated with increased fracture risk in the highest FGF-23 quartile (>57.4 pg/mL).

thumbnail image

Figure 3. Spline models for the detection of any nonlinear relationships between FGF-23 and the risk for all and vertebral fractures risk, respectively. The curves show the estimated log hazard ratios over the entire range of observed FGF-23 values. All analyses were adjusted for age, BMI, and BMD. (A) All fractures; linear FGF-23 effect (Wald χ2 = 6.01, d.f. = 1, p = .014). (B) Vertebral fractures; linear FGF-23 effect (Wald χ2 = 3.46, d.f. = 1, p = .062).

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Since the population in the developed world is aging, the burden of fragility fractures is a constantly increasing problem. Despite the fact that potent bone-specific pharmaceutical agents have become available, the problem of how to identify patients with high fracture risk yet remains an enigma. The recently proposed FRAX module is a way forward to identify patients at risk for fracture by assessing the impact of independent risk factors such as BMD, previous fracture, glucocorticoid use, and family history on the 10-year risk for fracture.21, 22 Adding new variables such as measures of certain hormones or markers of bone turnover to the model may improve this tool and therefore eventually enable us to target potent and expensive bone-specific drugs to the proper set of patients. Also, identifying novel independent risk factors for fracture may indicate new potential drug targets.

In this study, we show that a higher serum FGF-23 level is a novel independent predictor of overall fracture risk, as well as vertebral fracture risk, in elderly men. As evidenced by spline models and loss of significance when modeling FGF-23 as a continuous variable, the relation between FGF-23 and fracture risk was nonlinear and stronger in individuals with FGF-23 above the cutpoint of 55.7 pg/mL. Indeed, FGF-23 levels above this cutpoint also were associated with an increased risk for hip and nonvertebral fractures.

Mechanistically, there are several possible explanations to support our findings. Since FGF-23 is produced in bone, it is likely that even a moderate local increase in FGF-23 expression will alter bone microarchitecture and mechanical properties, leading to increased fracture risk. This is in agreement with hereditary disorders of increased FGF-23 bone expression such as ADHR or XLH.4, 23 Indeed, several studies support that FGF-23 is a negative regulator of bone matrix mineralization both in vivo and in vitro. 12, 24 Increased FGF-23 expression also potentially could modulate the response of the osteocyte to mechanical loading. Alternatively, it is also possible that yet undetermined factor(s) may cause both high FGF-23 levels and fragility of bone.

FGF-23 has been shown to play a central role in the chronic kidney disease–mineral and bone disorder (CKD-MBD), which is encompassed by biochemical abnormalities in mineral metabolism, vascular calcification, and increased mortality and fractures. Previous studies clearly demonstrate that FGF-23 is associated with biochemical changes arising in CKD, such as hyperphosphatemia,25 as well as multiple cardiovascular risk factors26–30 and death.31, 32 This study linking FGF-23 to increased fracture risk fulfills the notion that FGF-23 is at the very least a biomarker of multiple dearrangements translating into adverse patient outcome.

Importantly, the earliest histologic abnormalities of bone are observed after a relatively mild reduction in GFR33 and are found in virtually all patients with end-stage renal disease.34 Because FGF-23 is inversely related to renal function and is one of the earliest biomarkers of CKD-MBD,35 it would be reasonable if FGF-23 association with fracture risk were more pronounced in patients with a lower GFR. As a matter of fact, an early rise in FGF-23 may be an explanatory factor for the observed increased fracture risk in CKD. In contrast, FGF-23 association with fracture risk in this study was independent of renal function and found primarily in individuals above the median of eGFR. These findings need to be corroborated in future studies. It similarly will be of interest to explore whether FGF-23 can predict fracture risk in other populations with markedly elevated FGF-23 levels, such as advanced CKD.

Advantages of this study are the prospective design and the large number of subjects in a well-characterized population-based cohort. Limitations are that the results depend on a single FGF-23 measurement, which could underestimate the true association between FGF-23 and fracture risk. We did not measure or adjust for 1,25(OH)2D3 and cannot exclude the possibility of residual confounding. Finally, the study cohort included elderly white males; thus caution should be used in generalizing the results to other ethnic groups or women.

In summary, circulating FGF-23 is a novel independent predictor of fracture risk in elderly men, further accentuating its important underlying role in bone biology.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

We are deeply grateful to Senior Professor Olof Johnell, who played a vital role during the initiation phase of the MrOS study. We also would like to thank Anna-Lena Johansson, Violja Mixhue, Karin Önnby, Inger Abrahamsson, Maud Petterson, Ann-Charlotte Adolfsson, and Marja Gustafsson for excellent technical assistance. This work was supported by the Swedish Research Council.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Disclosures
  8. Acknowledgements
  9. References
  10. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
jbmr_263_sm_SupplChinese_Translation.pdf711KSupplementary Information: Chinese Translation

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.