ORIGINAL
E n d o c r i n e
ARTICLE
R e s e a r c h
Biological Variability of Plasma Intact and C-Terminal
FGF23 Measurements
Edward R. Smith, Michael M. Cai, Lawrence P. McMahon, and Stephen G. Holt
Department of Renal Medicine (E.R.S., M.M.C., L.P.M., S.G.H.), Eastern Health Clinical School, Faculty of
Medicine, Nursing and Health Sciences, Monash University, Box Hill, 3128 Victoria, Australia; and
Clinical Investigation and Research Unit (E.R.S., S.G.H.), Royal Sussex County Hospital, Brighton and
Sussex University Hospitals National Health Service Trust, Brighton, BN2 5BE United Kingdom
Context: FGF23 measurement may have a role in the management of patients with chronic kidney
disease (CKD).
Objective: Our objective was to study the biological variability of plasma intact FGF23 (iFGF23) and
C-terminal FGF23 (cFGF23) concentrations.
Design: Plasma samples were taken from 12 healthy adults at multiple time points on 2 consecutive
days to assess diurnal variability of FGF23 concentrations. Early-morning fasting and nonfasting
samples were also taken over a 6-wk period to estimate components of biological variation. Samples from 170 volunteers were used to define reference intervals. FGF23 concentrations were
measured by commercial ELISA. Western blotting was used to analyze FGF23 species from the
plasma of healthy adults and patients with predialysis CKD and those undergoing dialysis.
Participants and Setting: A total of 180 healthy adults and 18 adults with stage 3–5D CKD participated in this study at a hospital research unit.
Main Outcome Measure: Estimates were made of the biological variability of plasma FGF23
concentrations.
Results: iFGF23, but not cFGF23, showed significant diurnal variation. cFGF23 had a significantly
lower intra-individual variation than iFGF23 (8.3 vs. 18.3%) but higher inter-individual variation
than iFGF23 (28.9 vs. 19.2%). Fourteen samples would be needed to estimate an individual’s
homeostatic set point (within 10%) for iFGF23 compared with only three samples for cFGF23. Using
Western blotting, C-terminal FGF23 fragments were detected in the plasma of individuals with
normal renal function and at all stages of renal disease. The percent cFGF23 was significantly higher
in those without renal impairment (P ⬍ 0.001) and was positively correlated with plasma phosphate
concentration in those with normal renal function.
Conclusions: The high intra-individual biological variability of iFGF23 may limit its clinical use
as a diagnostic or management tool. Risk-related thresholds may be more appropriate for
clinical decision making based on cFGF23 measurements than conventional reference intervals.
FGF23 cleavage pathways may be an important natural regulatory mechanism for phosphate
control. (J Clin Endocrinol Metab 97: 3357–3365, 2012)
ibroblast growth factor 23 (FGF23) is a bone-derived
regulator of phosphate and vitamin D metabolism
(1). Plasma FGF23 concentrations are elevated in some
F
patients with hereditary or acquired phosphate-wasting
disorders (e.g. autosomal dominant hypophosphatemic
rickets or tumor-induced osteomalacia) and in individuals
ISSN Print 0021-972X ISSN Online 1945-7197
Printed in U.S.A.
Copyright © 2012 by The Endocrine Society
doi: 10.1210/jc.2012-1811 Received March 29, 2012. Accepted May 18, 2012.
First Published Online June 11, 2012
Abbreviations: cFGF23, C-terminal FGF23; CI, confidence interval; CKD, chronic kidney
disease; CVA, coefficient of variation for analytical imprecision; CVI, intra-individual CV;
CVG, inter-individual CV; FGF23, fibroblast growth factor 23; iFGF23, intact FGF23; II, index
of individuality; RCV, reference change value; RU, relative units.
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
jcem.endojournals.org
3357
3358
Smith et al.
FGF23 Biological Variability
with chronic kidney disease (CKD) (2, 3). In the setting of
CKD, FGF23 secretion is thought to increase as an adaptive response to phosphate retention, acting to reduce dietary phosphate absorption and increase urinary phosphate excretion to maintain normophosphataemia (3).
However, chronic suppression of renal 1,25(OH)2D3 production may augment the development of secondary hyperparathyroidism and associated sequelae (3). FGF23
measurement has been advocated to help identify CKD
patients that might benefit most from aggressive management of their phosphate balance (4, 5).
Some of the FGF23 synthesized by osteocytes is processed by furin, a type I precursor convertase, cleaving the
nascent polypeptide at R179 to yield N- and C-terminal
protein fragments (6). Two types of commercial ELISA are
currently available for the determination of FGF23 in human plasma; intact FGF23 (iFGF23) assays from Kainos
and Immutopics, which detect only full-length FGF23
(⬃32 kDa) and, the C-terminal FGF23 (cFGF23) assay,
also from Immutopics, which detects both iFGF23 and
C-terminal fragments (⬃14 kDa). There is uncertainty
about which assay strategy is optimal for patient studies.
Although results from both types of FGF23 assay have
been consistently associated with mortality risk and decline in renal function in those with CKD (7–9), direct
comparison of measurements made with these kits has
shown marked disagreement, particularly in individuals
with normal or mild renal impairment (10 –14). We have
also demonstrated that iFGF23 is highly susceptible to ex
vivo proteolytic degradation (14).
Thus, it was apparent that additional work was needed
to investigate the biological variability of plasma FGF23
measurements while minimizing sample instability. In this
study we measured plasma FGF23 concentrations in samples from a cohort of healthy adults collected with a commercial plasma protein preservation system. The aims of
this study were to 1) investigate the diurnal variation of
plasma FGF23, 2) estimate intra- and inter-individual
components of biological variation, 3) define reference
intervals for plasma FGF23 using both assays, and 4) characterize the FGF23 species detected by the C-terminal
assay.
Materials and Methods
Study samples
For the assessment of diurnal variation, blood was taken from
12 healthy volunteers (eight male, four female; mean age 45 yr,
range 25– 62 yr) at 0800, 1200, 1600, and 2000 h on two consecutive weekdays. Dietary intake and activity were unrestricted.
For the analysis of intra- and inter-individual biological variability, weekly nonfasting blood samples were collected from the
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
same 12 healthy volunteers between 0800 and 1000 h for 6
successive weeks. In eight subjects (five male, three female; mean
age 42 yr, range 25–54 yr) weekly, early-morning samples were
obtained for 6 successive weeks after a 10-h overnight fast. For
the determination of reference intervals, nonfasting, early-morning (0800 –1000 h) samples were taken from 176 healthy
volunteers.
Healthy volunteers included laboratory workers, clinical and
departmental staff, or individuals recruited from local screening
clinics. These individuals were not obese and reported no history
of anemia, vitamin D deficiency, diabetes, cardiovascular, cerebrovascular disease, or malignancy. None were receiving vitamin D, antihypertensive, or lipid-lowering therapies. All had an
estimated glomerular filtration rate over 60 ml/min 䡠 1.73 m2
and urine total protein to creatinine ratio below 10 mg/mmol.
For the analysis of endogenous FGF23 species, blood samples
were also taken from patients with predialysis CKD (n ⫽ 8)
attending local outpatient clinics and patients undergoing hemodialysis (n ⫽ 10).
The study was approved by the local research and ethics committee (reference LR31/1112), and all participants provided
written informed consent. The study was conducted in accordance with the Helsinki Declaration.
Sample collection and analysis
Blood samples were taken using standard phlebotomy procedures with Vacutainer Safety-Lok blood collection sets into 8.5
ml serum or P100 tubes (Becton Dickinson Ltd., North Ryde,
New South Wales, Australia). P100 tubes contain spray-dried
K2-EDTA as an anticoagulant and proprietary additives (protease inhibitors) for the preservation of plasma proteins. Samples
were centrifuged for 10 min at 4 C and 2500 ⫻ g, and plasma was
aliquoted into cryovials for storage at ⫺80 C until batched analysis. The time from venipuncture to freezing was less than 45 min
in all cases.
Before analysis, samples were thawed at room temperature
(for 2 h) and mixed by gentle inversion. Plasma FGF23 was
measured using intact and second-generation C-terminal ELISA
kits from Immutopics (San Clemente, CA) following the manufacturer’s instructions. Both kits are calibrated using recombinant human FGF23 expressed in a mouse myeloma cell line.
Coefficient of variation for analytical imprecision (CVA) was
determined according to Clinical Laboratory and Standards Institute guidelines (15). CVA for the iFGF23 assay was 6.8 and
6.5% at 25 and 120 pg/ml, respectively. CVA for the cFGF23
assay was 3.4 and 3.8% at 36 and 164 relative units (RU)/ml,
respectively. Plate washing was automated using a BioTek
EL406 microplate washer dispenser (Winooski, VT). All measurements were made in duplicate, in a random order, by a single
operator (E.R.S.) using a single lot of reagents, standards, and
controls. Serum phosphate was measured using the QuantiChrom colorimetric assay (malachite green/molybdate) from
BioAssay Systems (Hayward, CA). CVA for this assay was 3.8
and 4.1% at 1.10 and 1.78 mmol/liter, respectively. Plasma PTH
was measured using Elecsys immunoassay reagents on a fully
automated Modular Analytics E170 platform (Roche Diagnostics, Castle Hill, New South Wales, Australia). Local reference
interval was 1.6 – 6.9 pmol/liter. CVA for the PTH assay was 2.5
and 2.9% at 4.6 and 30.1 pmol/liter, respectively.
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
jcem.endojournals.org
3359
Immunoprecipitation and Western blotting
Affinity-purified goat polyclonal antibody, recognizing
epitopes within the human FGF23 C-terminal region (amino
acids 186 –206) was covalently immobilized on an agarose resin
(ThermoFisher Scientific, Scoresby, Victoria, Australia), and
500 ␮l of each plasma sample or recombinant human FGF23
standard (R&D Systems, Minneapolis, MN) was incubated with
the agarose resin in a microspin column for 12 h at 4 C with gentle
mixing. Bound proteins were separated on 10 –20% Novex Trisglycine precast gels (Invitrogen, Carlsbad, CA) and transferred
onto polyvinylidene difluoride using the iBlot dry blotting system
(Invitrogen). Membranes were then incubated with an affinitypurified goat polyclonal antibody, recognizing a different portion of the human FGF23 C-terminal region (amino acids
225–244). The signal was visualized with horseradish peroxidase-conjugated antigoat IgG (eBioscience, San Diego, CA) and
Immu-Star chemiluminescence Western C kit (Bio-Rad, Hercules, CA) on a VersaDoc 4000 MP imaging platform (Bio-Rad).
Both antihuman FGF23 antibodies were a kind gift from Jeffery
Lavinge (Vice President, Immutopics Inc.).
Data analysis
Data are expressed as mean (SD) or median (interquartile
range) as appropriate. Analyses were conducted after inspection
for outliers using Cochran and Reed tests (16). No outliers were
identified among duplicate measurements, and only a single
iFGF23 outlier was identified for one volunteer using the Reed
test. Linear regression analysis was used to assess the data for
constant trends in analyte concentrations over time. The partial
F test was used to test whether gradient of the regression line
differed significantly from zero. A single volunteer showed a
significant trend in serum phosphate concentration over the
study period, but exclusion of that individual from the calculation of variance components gave similar results. Biological intra-individual CV (CVI), inter-individual CV (CVG), and reference change values (RCV) were determined using nested
ANOVA as previously described by Fraser and Harris (16). CV
were compared using the Wald test (17). The index of individuality (II) (defined as CVI/CVG) and the number of specimens
needed to estimate the homeostatic set point of an individual
within 10 and 20% limits were also determined (16). Reference
intervals for plasma iFGF23 and cFGF23 concentrations were
established according to international guidelines (18). We tested
for normality using the Shapiro-Wilk test. Outliers were excluded (n ⫽ 6) using the aforementioned tests. Nonparametric
methods were used to derive 95% reference limits (corresponding to 2.5th and 97.5th centiles). Passing-Bablok regression was
used to compare method readout. Spearman’s correlation coefficients were calculated to assess the association between variables, and unpaired or paired t tests (as appropriate) were used
to test for differences in mean between groups. Analysis was
performed using Analyze-it add-in for MS Excel (Analyze-it software Ltd., Leeds, UK). P ⬍ 0.05 was considered significant.
Results
Diurnal variation of plasma FGF23 in healthy
individuals
Figure 1 shows the variation in serum phosphate and
plasma FGF23 and PTH concentrations in 12 healthy vol-
FIG. 1. Variation of plasma PTH, serum phosphate, and plasma
iFGF23 and cFGF23 concentrations in 12 healthy adult volunteers,
measured at 0800, 1200, 1600, and 2000 over 2 consecutive days.
Mean values are plotted. Error bars indicate SEM.
unteers measured at 0800, 1200, 1600, and 2000 h on 2
successive days. Plasma iFGF23 concentrations showed
substantially greater variation during the day than concentrations determined using the cFGF23 assay. iFGF23
peaked in the early morning and reached their nadir in the
evening with a mean decrease of 30%. Serum phosphate
and plasma PTH concentrations also showed significant
fluctuations during the day, with concentrations at their
highest in the late afternoon and lowest in the early morning, with a mean increase of 31% for phosphate and 34%
for PTH from 0800 –2000 h. Plasma cFGF23 concentrations were at their highest levels in the late afternoon/
evening (mean increase of 6%), although these changes
did not reach statistical significance (paired t test: 0800 vs.
2000 h, P ⫽ 0.10).
Intra- and inter-individual variation of plasma
FGF23 in healthy individuals
Figure 2, A and B, shows the week-to-week variability
of nonfasting early-morning (0800 –1000 h) plasma
FGF23 concentrations in 12 healthy volunteers over a period of 6 wk using iFGF23 and cFGF23 assays, respectively. Components of intra- and inter-individual biological variation are summarized in Table 1. Plasma FGF23
measurements made with the cFGF23 assay demonstrated
better analytical precision (CVA) and significantly lower
intra-individual variation (CVI) than for the iFGF23 assay
(Wald test, P ⬍ 0.001). As a result, fewer samples would
be needed to determine the homeostatic set point for
plasma FGF23 concentration using the cFGF23 assay.
Similarly, use of the cFGF23 assay yielded a lower RCV.
However, inter-individual variability (CVG) was significantly higher using the cFGF23, resulting in a lower II for
cFGF23 measurements. Plasma PTH showed comparable
variance characteristics to iFGF23 measurements. Serum
3360
Smith et al.
FGF23 Biological Variability
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
played a slightly flattened (platykurtotic) distribution with a median concentration of 53.7 RU/ml. Neither
iFGF23 nor cFGF23 concentrations
were significantly associated with gender or age (either as a continuous variable or when categorized by decade).
For plasma iFGF23 concentration, the
lower reference limit [90% confidence
interval (CI)] was 11.7 (9.2–12.6) pg/
ml, and the upper limit was 48.6 (46.551.3) pg/ml. For plasma cFGF23 concentration, the lower reference limit
was 21.6 (19.1–22.8) RU/ml, and the
upper limit was 91.0 (87.5–93.2) RU/
ml. Figure 3C shows the poor agreement between plasma iFGF23 and
cFGF23 concentrations in this reference population, as indicated by the
substantial scatter around the regression line. Analysis by Passing-Bablok
regression gave a slope of 2.10 (95%
CI ⫽ 1.73–2.58) and an intercept of
⫺2.90 (95% CI ⫽ ⫺14.3–7.3). Earlymorning plasma cFGF23 concentraFIG. 2. A and B, Week-to-week variation of nonfasting, early-morning (0800 –1000) plasma
iFGF23 (A) and cFGF23 (B) concentrations in 12 healthy adult volunteers; C and D, weekly
tions (natural log-transformed) were
variation in early-morning (0800 –1000) plasma iFGF23 (C) and cFGF23 (D) concentrations
strongly associated with serum phosafter a 10-h overnight fast in eight healthy adult volunteers over 6-wk periods. Box plots
phate concentration (r ⫽ 0.422; P ⬍
show median (horizontal line), upper and lower quartile (box), and range (whiskers) of values
for each participant.
0.001) but only relativity weakly correlated with plasma iFGF23 concenphosphate concentrations showed a similar CVI to plasma tration (r ⫽ 0.138; P ⫽ 0.022).
cFGF23 measurements but had a significantly lower CVG,
Immunoprecipitation and Western blotting studies
giving a higher II.
To investigate the effect of fasting on plasma FGF23 of plasma FGF23
We hypothesized that the lack of agreement between
variability, weekly early-morning concentrations were determined in eight volunteers for a period of 6 wk, as before, cFGF23 and iFGF23 measurements made in healthy inbut after an overnight (10 h) fast (Fig. 2, C and D). Mean dividuals and patients with predialysis CKD (14) was due,
fasting and nonfasting concentrations did not differ sig- at least in part, to the presence of circulating C-terminal
nificantly using either assay [iFGF23: nonfasting vs. fast- fragments. Figure 4A shows the blot of FGF23 species
ing, 26.1 (6.4) vs. 25.9 (5.2) pg/ml; cFGF23: 49.0 (8.1) vs. detected in 10 healthy individuals with normal renal func52.7 (9.0)], and estimates of CVI were similar to those tion. C-terminal FGF23 fragments (⬃14 kDa) were present in each sample but accounted for a variable proportion
determined in the nonfasting state (data not shown).
of the total FGF23 (21–56%) determined by volumetric
Reference intervals for plasma FGF23
quantitation of the blot. Serum phosphate concentration
The distribution of nonfasting plasma iFGF23 and was significantly correlated with percent iFGF23 (SupplecFGF23 concentrations measured in 170 healthy individ- mental Fig. 1, published on The Endocrine Society’s Jouruals [mean age 56 (13) yr, 60% male, mean albumin-ad- nals Online web site at http://jcem.endojournals.org; ␳ ⫽
justed serum calcium 2.36 (0.10) mmol/liter, mean serum 0.424; P ⫽ 0.003) but not with cFGF23 or iFGF23 by
phosphate 1.01 (0.12) mmol/liter, mean plasma PTH 3.8 ELISA. C-terminal FGF23 fragments were also detected in
(2.1) pmol/liter] are shown in Fig. 3, A and B, respectively. the plasma of patients with predialysis CKD and those on
Plasma iFGF23 concentrations showed a right-skewed dialysis (Fig. 4B) but proportionally less in hemodialysis
distribution with a median concentration of 24.7 pg/ml. patients (mean 5.2% of total) than predialysis patients
Concentrations measured using the cFGF23 assay dis- (mean 18.7% of total) (Fig. 5).
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
jcem.endojournals.org
3361
TABLE 1. Variance components and test parameters of plasma FGF23, PTH, and serum phosphate concentrations in
healthy adult individuals
Number of
samplesa
Assay
iFGF23
cFGF23
PTH
Phosphate
a
Mean
26.1 pg/ml
49.0 RU/ml
3.7 pmol/liter
0.91 mmol/liter
CVA (%)
6.2
3.6
2.5
3.8
CVI (%)
18.3
8.3
20.2
10.8
RCVa (%)
54
25
56
32
CVG (%)
19.2
28.9
21.8
11.1
II
0.95
0.29
0.93
0.97
ⴞ10%
14
3
16
5
ⴞ20%
4
1
4
1
P ⬍ 0.05.
Discussion
In the present study, we have investigated the biological
variability of plasma FGF23 measurement alongside other
markers of mineral metabolism in a cohort of healthy
FIG. 3. A and B, Distribution of plasma cFGF23 (A) and cFGF23 (B)
concentrations in a reference population of 170 healthy adults; C, scatter
plot with Passing-Bablok regression showing method agreement.
adults. iFGF23, but not cFGF23, showed significant diurnal variation, reaching their highest levels in the early
morning and their nadir in the evening. These fluctuations
were of a similar magnitude to that of plasma PTH and
serum phosphate concentration. The diurnal variation of
phosphate and PTH are well described and consistent with
FIG. 4. Western blot (WB) of FGF23 immunoprecipitated from plasma
of 10 healthy volunteers (A) and five patients with predialysis CKD and
five hemodialysis patients (B) using polyclonal antihuman FGF23
antibodies from Immutopics C-terminal ELISA kit. Arrows indicate
position of intact (⬃32 kDa) and C-terminal (⬃14 kDa) bands. Plasma
cFGF23 and iFGF23 (Immutopics ELISA), serum phosphate
concentrations, and percent total FGF23 as intact (from quantitation of
blot) is given in the table below. C, Pre-dialysis CKD patient; H,
hemodialysis patient; IP, Immunoprecipitation; N, healthy controls.
3362
Smith et al.
FGF23 Biological Variability
FIG. 5. Differences in the proportion of FGF23 present as C-terminal
fragments (by Western blotting) in healthy controls and in predialysis
CKD and end-stage renal disease patients undergoing hemodialysis.
***, P ⬍ 0.001; **, P ⬍ 0.01.
the patterns observed here (19, 20). Also consistent with
previous reports of acute phosphate loading (21), iFGF23
concentrations peaked 8 –12 h after phosphate, suggesting
a lag in response to endogenous changes in serum phosphate. This could explain why iFGF23 was only weakly
associated with serum phosphate concentration in the reference population, because measurement at the same time
failed to show the same association. The time of sampling
is therefore an important consideration in the interpretation of iFGF23 concentrations. Future studies should consider a phase-shifted relationship between iFGF23 and
phosphate concentrations.
iFGF23 secretion may also follow changes in the circadian rhythm of bone turnover. Indeed, various studies
have shown that bone markers generally peak in the early
morning (22). FGF23 expression may be regulated by local factors in bone [e.g. DMP1 (dentin matrix protein 1)
and PHEX (phosphate-regulating gene with homologies
to endopeptidases on the X chromosome)], allowing integration of FGF23 synthesis with bone matrix metabolism (23). Earlier attempts to define diurnal variation of
iFGF23 have been inconsistent (24, 25) and were possibly
affected by stability issues (14). Previous studies have also
shown no evidence of significant diurnal variation in
cFGF23 concentration (26, 27). However, as a composite
measure of intact and C-terminal FGF23 species, cFGF23
measurements may obscure differences in the metabolism
and clearance pathways of intact FGF23 protein and
fragments.
Our estimates of the CVI for plasma PTH and serum
phosphate concentration were in close agreement with
previous studies (28, 29). Plasma cFGF23 concentrations
had a lower CVI compared with iFGF23 concentrations
and, combined with lower analytical imprecision, yielded
a lower RCV (25 vs. 54%). Fasting does not appear to
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
affect the variability of iFGF23 or cFGF23 concentrations.
This finding is consistent with a previous study in which no
postprandial changes in plasma cFGF23 were observed
(30). We calculated that only three samples would be required to determine an individual’s homeostatic set point
(within ⫾10%) using the cFGF23 assay, whereas 14 samples would be needed using the iFGF23 assay. The need for
multiple serial measurements to obtain an accurate estimate of a patient’s iFGF23 concentration certainly limits
the utility of this test. Overall, the problems of iFGF23
analysis are reminiscent of the issues encountered with
PTH: high intra-individual variability, lack of betweenmethod agreement, and depending on sample type, analyte instability. Just as for PTH (28), these characteristics
greatly undermine the use of plasma iFGF23 measurements as a management tool.
Conversely, cFGF23 concentrations demonstrated a
much higher CVG than for iFGF23, PTH, or phosphate.
This is consistent with a recent publication by Isakova et
al. (31), which showed that between-subject variance was
the main component of variation in cFGF23 concentration of peritoneal dialysis patients. The resultant low II for
cFGF23 (CVG ⬎⬎ CVI) suggests that conventional population-based reference intervals may not be the most sensitive method for detecting abnormal results, particularly
in early disease (16). We determined 95% reference limits
for a healthy cohort of volunteers who were free from
renal, cardiovascular, metabolic, and hematological abnormalities, minimizing the risk of confounding. The reference intervals defined here are in keeping with concentrations previously observed in control groups, although
in the case of iFGF23 were slightly higher and possibly due
to our use of protease inhibitors, minimizing loss of intact
hormone after collection. With respect to these limits, it is
apparent that for a given individual, concentrations of
iFGF23 span a much larger fraction of the reference interval than for cFGF23. Using these reference intervals,
retrospective analysis of a previous study conducted by us
of 200 patients with stages 3 and 4 CKD (32) showed that
65% of individuals had iFGF23 concentrations above the
upper reference limit, whereas only 52% had cFGF23 concentrations greater than the upper limit for the cFGF23
assay. This supports our assertion that despite superior
analytical precision and lower intra-individual variability,
the use of population-based reference intervals for
cFGF23 may result in a loss of clinical sensitivity. Riskrelated thresholds may be more appropriate for clinical
decision making based on cFGF23 measurements. These
have yet to be clearly defined.
Although iFGF23 and cFGF23 concentrations are invariably correlated when measured in the same study,
agreement is poor (10, 12–14), irrespective of whether the
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
Immutopics or Kainos iFGF23 assay is used. Judging by
the scatter around the regression line, we again found a
marked lack of analytical agreement between results from
Immutopics iFGF23 and cFGF23 assays. Such lack of
agreement is surprising for assays that purport to detect
the same analyte. Investigators have reported not only a
lack of agreement between iFGF23 and cFGF23 measurements in similar cohorts but also highly significant differences in their association with other biochemical parameters (10, 13, 33, 34). Indeed, here we found the strength
of association between serum phosphate and plasma
FGF23 concentration to be substantially different with
iFGF23 and cFGF23 assays. Elimination rates and plasma
clearance profiles have also been found to differ depending
on which FGF23 assay is used, suggesting that there may
be alternate clearance pathways for intact protein and Cterminal fragments (35). These findings cannot be readily
explained by differences in assay calibration and analytical imprecision. On the contrary, the disparity in patterns
of diurnal variation, intra- and inter-individual variability
and distribution of iFGF23 and cFGF23 concentrations
points strongly to differences in the analytical specificity of
these assays.
Initially, it was thought that the very high concentrations of FGF23 encountered in patients with renal failure
may have been due to the accumulation of C-terminal
fragments. This appeared not to be the case when Shimada
et al. (36) subsequently showed that only intact, fulllength FGF23 was present in the plasma of patients with
end-stage renal disease on peritoneal dialysis (36). However, this study did not conclusively demonstrate that Cterminal fragments were not present in predialysis CKD.
A major finding of the present study is that C-terminal
FGF23 fragments are present in the plasma of healthy
individuals and in those with predialysis CKD. Unlike Shimada et al. (36), we used both antibodies from the cFGF23
assay in our immunoprecipitation and Western blotting
protocol. C-terminal fragments formed a variable proportion of the total FGF23 detected in this system, and they
were less abundant in CKD patients than in controls and,
consistent with Shimada et al. (36), at very low levels in the
plasma of dialysis patients. In addition, we found that the
serum phosphate concentration was more closely associated with the proportion of intact hormone by Western
blotting than cFGF23 concentration by ELISA. This is
consistent with data that show that phosphate augments
the expression of GALNT3 (UDP-N-acetyl-␣-D-galactosamine:polypeptide N-acetylgalactosaminyl-transferase
3), which O-glycosylates iFGF23 and reduces its susceptibility to proteolysis by furin (37). This is perhaps where
cFGF23 measurement is unhelpful; it does not capture this
regulatory process. It is possible that, as renal disease pro-
jcem.endojournals.org
3363
gresses and FGF23 secretion is increasingly stimulated by
phosphate retention and other factors, the physiological
counterregulatory cleavage mechanism becomes overwhelmed. Bhattacharyya et al. (38) have recently found
differences in the ratio of cFGF23 to iFGF23 in different
disease states, which points toward C-terminal fragments/
proteolytic pathways having pathophysiological significance. Because isolated C-terminal FGF23 fragments appear to compete with iFGF23 for receptor binding and
antagonize phosphaturic activity (39), it suggests that
measurement of iFGF23 rather than cFGF23 may be more
physiologically relevant. Indeed, some studies show a significant association only between iFGF23 concentrations
and other parameters, e.g. phosphate (10) and IGF-I (33),
but not cFGF23, whereas other studies show a significant
correlation only with cFGF23 but not iFGF23 concentrations, e.g. iron (13) and phosphate (33). It is possible that
important biological signals may be missed or underestimated by measuring plasma FGF23 concentrations with
one assay alone.
Despite these differences, higher iFGF23 and cFGF23
concentrations have both been consistently and independently associated with increased mortality risk and poor
outcome in patients with CKD. Notably none of these
studies used protease inhibitors to stabilize their samples
(7, 9). Overall, therefore, in large epidemiological studies,
FGF23 concentration, determined by either assay, appears
to capture significant (but not necessarily the same) prognostic information, independent of other markers of mineral metabolism.
Limitations
The main limitation of this study is that we were unable
to evaluate the performance of the other, widely referenced, iFGF23 ELISA from Kainos. There is a paucity of
data on how Kainos and Immutopics iFGF23 assays compare with one another, although both show a similar lack
of agreement with the Immutopics cFGF23 assay (either
generation). However, we cannot discount the possibility
that the iFGF23 assay from Kainos may have different
performance characteristics, and the reference intervals
derived here using the Immutopics kit need to be validated
for use with this other assay. A further limitation is that we
have appraised the variability of FGF23 measurements
only in healthy adults and not in those with CKD or other
pathologies. Although biological variation is generally of
the same order of magnitude in health and disease, significant differences in variation may exist, which would further affect the interpretation of plasma FGF23 concentration. Finally, apart from the fasting analyses, we did not
restrict the dietary intake of our study participants. Conceivably, differences in diet may have inflated our esti-
3364
Smith et al.
FGF23 Biological Variability
mates of intra- and inter-variation because dietary phosphate intake has been reported, although not consistently,
to affect circulating FGF23 concentrations (40). However,
the lack of postprandial changes in plasma FGF23 concentration reported by others (30), together with our finding of very similar variance components after an overnight
fast, make this unlikely. Our estimates of biological variability are representative of what would be routinely encountered in clinical practice.
Conclusion
iFGF23 appears to be a more biologically relevant than
cFGF23. The combined measurement of iFGF23 and Cterminal fragments, which may differ in their function and
clearance, doesn’t represent a homogeneous signal. However, as is the case for PTH, attempting to measure a highly
dynamic hormone can generate highly variable results,
which may affect its use both diagnostically and as a management tool. Although intra-individual variation for
cFGF23 is comparatively small, the inter-individual variation is high, meaning that cFGF23 concentrations are
individualized. This presents a problem when conventional population-based reference intervals are used to
identify abnormal FGF23 concentrations and may lead to
loss of clinical sensitivity. The decision whether the
iFGF23 or cFGF23 assay is more clinically useful is therefore not straightforward. In dialysis patients where FGF23
is predominantly intact and the assays show good agreement, either method could conceivably be used. In patients
with normal renal function or predialysis CKD, serial assessment (either for monitoring disease progression or response to therapy) is perhaps best made with the cFGF23
assay, given the significantly lower intra-individual biological variability. If FGF23 measurement is to be used
diagnostically, iFGF23 ascertainment may be preferable
due to its lower individuality and likely improved clinical
sensitivity.
Acknowledgments
Address all correspondence and requests for reprints to: Edward
R. Smith, Department of Renal Medicine, Eastern Health Clinical School, Faculty of Medicine, Nursing and Health Sciences,
Monash University, Level 2, 5 Arnold Street, Box Hill, 3128
Victoria, Australia. E-mail: [email protected].
We gratefully acknowledge the funding from the Sussex Kidney Unit and the Clinical Investigation and Research Unit,
Brighton and Sussex University Hospitals, United Kingdom; the
Department of Renal Medicine, Eastern Health Clinical School,
Monash University, Australia; and an unrestricted research
grant from Amgen.
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
Disclosure Summary: S.G.H. has received honoraria from
Amgen, Baxter, and Shire. L.P.M. has received research grants
from Amgen. E.R.S. and M.M.C. have nothing to disclose.
References
1. Shimada T, Hasegawa H, Yamazaki Y, Muto T, Hino R, Takeuchi
Y, Fujita T, Nakahara K, Fukumoto S, Yamashita T 2004 FGF-23
is a potent regulator of vitamin D metabolism and phosphate homeostasis. J Bone Miner Res 19:429 – 435
2. Yamazaki Y, Okazaki R, Shibata M, Hasegawa Y, Satoh K, Tajima
T, Takeuchi Y, Fujita T, Nakahara K, Yamashita T, Fukumoto S
2002 Increased circulatory level of biologically active full-length
FGF-23 in patients with hypophosphatemic rickets/osteomalacia.
J Clin Endocrinol Metab 87:4957– 4960
3. Gutierrez O, Isakova T, Rhee E, Shah A, Holmes J, Collerone G,
Jüppner H, Wolf M 2005 Fibroblast growth factor-23 mitigates
hyperphosphatemia but accentuates calcitriol deficiency in chronic
kidney disease. J Am Soc Nephrol 16:2205–2215
4. Isakova T, Wahl P, Vargas GS, Gutiérrez OM, Scialla J, Xie H,
Appleby D, Nessel L, Bellovich K, Chen J, Hamm L, Gadegbeku C,
Horwitz E, Townsend RR, Anderson CA, Lash JP, Hsu CY, Leonard
MB, Wolf M 2011 Fibroblast growth factor 23 is elevated before
parathyroid hormone and phosphate in chronic kidney disease. Kidney Int 79:1370 –1378
5. Yilmaz MI, Sonmez A, Saglam M, Yaman H, Kilic S, Eyileten T,
Caglar K, Oguz Y, Vural A, Yenicesu M, Mallamaci F, Zoccali C
2012 Comparison of calcium acetate and sevelamer on vascular
function and fibroblast growth factor 23 in CKD patients: a randomized clinical trial. Am J Kidney Dis 59:177–185
6. Benet-Pagès A, Lorenz-Depiereux B, Zischka H, White KE, Econs
MJ, Strom TM 2004 FGF23 is processed by proprotein convertases
but not by PHEX. Bone 35:455– 462
7. Gutiérrez OM, Mannstadt M, Isakova T, Rauh-Hain JA, Tamez H,
Shah A, Smith K, Lee H, Thadhani R, Jüppner H, Wolf M 2008
Fibroblast growth factor 23 and mortality among patients undergoing hemodialysis. N Engl J Med 359:584 –592
8. Isakova T, Xie H, Yang W, Xie D, Anderson AH, Scialla J, Wahl P,
Gutiérrez OM, Steigerwalt S, He J, Schwartz S, Lo J, Ojo A, Sondheimer J, Hsu CY, Lash J, Leonard M, Kusek JW, Feldman HI, Wolf
M 2011 Fibroblast growth factor 23 and risks of mortality and
end-stage renal disease in patients with chronic kidney disease.
JAMA 305:2432–2439
9. Fliser D, Kollerits B, Neyer U, Ankerst DP, Lhotta K, Lingenhel A,
Ritz E, Kronenberg F, Kuen E, König P, Kraatz G, Mann JF, Müller
GA, Köhler H, Riegler P 2007 Fibroblast growth factor 23 (FGF23)
predicts progression of chronic kidney disease: the Mild to Moderate
Kidney Disease (MMKD) Study. J Am Soc Nephrol 18:2600 –2608
10. Burnett SM, Gunawardene SC, Bringhurst FR, Jüppner H, Lee H,
Finkelstein JS 2006 Regulation of C-terminal and intact FGF-23 by
dietary phosphate in men and women. J Bone Miner Res 21:1187–
1196
11. Ito N, Fukumoto S, Takeuchi Y, Yasuda T, Hasegawa Y, Takemoto
F, Tajima T, Dobashi K, Yamazaki Y, Yamashita T, Fujita T 2005
Comparison of two assays for fibroblast growth factor (FGF)-23.
J Bone Miner Metab 23:435– 440
12. Bacchetta J, Dubourg L, Harambat J, Ranchin B, Abou-Jaoude P,
Arnaud S, Carlier MC, Richard M, Cochat P 2010 The influence of
glomerular filtration rate and age on fibroblast growth factor 23
serum levels in pediatric chronic kidney disease. J Clin Endocrinol
Metab 95:1741–1748
13. Imel EA, Peacock M, Gray AK, Padgett LR, Hui SL, Econs MJ 2011
Iron modifies plasma FGF23 differently in autosomal dominant hypophosphatemic rickets and healthy humans. J Clin Endocrinol
Metab 96:3541–3549
J Clin Endocrinol Metab, September 2012, 97(9):3357–3365
14. Smith ER, Ford ML, Tomlinson LA, Weaving G, Rocks BF, Rajkumar C, Holt SG 2011 Instability of fibroblast growth factor-23
(FGF-23): implications for clinical studies. Clin Chim Acta 412:
1008 –1011
15. Clinical and Laboratory Standards Institute 2003 Estimation of total analytical error for clinical laboratory methods; approved guideline. 1st ed. Wayne, PA: Clinical and Laboratory Standards Institute
16. Fraser CG, Harris EK 1989 Generation and application of data on
biological variation in clinical chemistry. Crit Rev Clin Lab Sci 27:
409 – 437
17. Shoukri MM, Colak D, Kaya N, Donner A 2008 Comparison of two
dependent within subject coefficients of variation to evaluate the
reproducibility of measurement devices. BMC Med Res Methodol
8:24
18. Clinical and Laboratory Standards Institute 2008 Defining, establishing, and verifying reference intervals in the clinical laboratory;
approved guidelines. 3rd ed. Wayne, PA: Clinical and Laboratory
Standards Institute
19. Kemp GJ, Blumsohn A, Morris BW 1992 Circadian changes in
plasma phosphate concentration, urinary phosphate excretion, and
cellular phosphate shifts. Clin Chem 38:400 – 402
20. el-Hajj Fuleihan G, Klerman EB, Brown EN, Choe Y, Brown EM,
Czeisler CA 1997 The parathyroid hormone circadian rhythm is
truly endogenous: a general clinical research center study. J Clin
Endocrinol Metab 82:281–286
21. Nishida Y, Taketani Y, Yamanaka-Okumura H, Imamura F, Taniguchi A, Sato T, Shuto E, Nashiki K, Arai H, Yamamoto H, Takeda
E 2006 Acute effect of oral phosphate loading on serum fibroblast
growth factor 23 levels in healthy men. Kidney Int 70:2141–2147
22. Hannon R, Eastell R 2000 Preanalytical variability of biochemical
markers of bone turnover. Osteoporos Int 11(Suppl 6):S30 –S44
23. Strom TM, Jüppner H 2008 PHEX, FGF23, DMP1 and beyond.
Curr Opin Nephrol Hypertens 17:357–362
24. Vervloet MG, van Ittersum FJ, Büttler RM, Heijboer AC, Blankenstein MA, ter Wee PM 2011 Effects of dietary phosphate and calcium intake on fibroblast growth factor-23. Clin J Am Soc Nephrol
6:383–389
25. Carpenter TO, Insogna KL, Zhang JH, Ellis B, Nieman S, Simpson
C, Olear E, Gundberg CM 2010 Circulating levels of soluble klotho
and FGF23 in X-linked hypophosphatemia: circadian variance, effects of treatment, and relationship to parathyroid status. J Clin
Endocrinol Metab 95:E352–E357
26. Larsson T, Nisbeth U, Ljunggren O, Jüppner H, Jonsson KB 2003
Circulating concentration of FGF-23 increases as renal function declines in patients with chronic kidney disease, but does not change
in response to variation in phosphate intake in healthy volunteers.
Kidney Int 64:2272–2279
27. Isakova T, Xie H, Barchi-Chung A, Smith K, Sowden N, Epstein M,
Collerone G, Keating L, Jüppner H, Wolf M 2012 Daily variability
in mineral metabolites in CKD and effects of dietary calcium and
calcitriol. Clin J Am Soc Nephrol 7:820 – 828
28. Gardham C, Stevens PE, Delaney MP, LeRoux M, Coleman A,
Lamb EJ 2010 Variability of parathyroid hormone and other mark-
jcem.endojournals.org
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
3365
ers of bone mineral metabolism in patients receiving hemodialysis.
Clin J Am Soc Nephrol 5:1261–1267
Winkel P, Statland BE, Bokelund H 1974 Factors contributing to
intra-individual variation of serum constituents: 5. Short-term dayto-day and within-hour variation of serum constituents in healthy
subjects. Clin Chem 20:1520 –1527
Isakova T, Gutierrez O, Shah A, Castaldo L, Holmes J, Lee H, Wolf
M 2008 Postprandial mineral metabolism and secondary hyperparathyroidism in early CKD. J Am Soc Nephrol 19:615– 623
Isakova T, Xie H, Barchi-Chung A, Vargas G, Sowden N, Houston
J, Wahl P, Lundquist A, Epstein M, Smith K, Contreras G, Ortega
L, Lenz O, Briones P, Egbert P, Ikizler TA, Jueppner H, Wolf M 2011
Fibroblast growth factor 23 in patients undergoing peritoneal dialysis. Clin J Am Soc Nephrol 6:2688 –2695
Ford ML, Smith ER, Tomlinson LA, Chatterjee PK, Rajkumar C,
Holt SG 2012 FGF-23 and osteoprotegerin are independently associated with myocardial damage in chronic kidney disease stages 3
and 4. Another link between chronic kidney disease-mineral bone
disorder and the heart. Nephrol Dial Transplant 27:727–733
Bacchetta J, Cochat P, Salusky IB, Wesseling-Perry K 5 February
2012 Uric acid and IGF1 as possible determinants of FGF23 metabolism in children with normal renal function. Pediatr Nephrol
10.1007/s00467-012-2110-3
Durham BH, Joseph F, Bailey LM, Fraser WD 2007 The association
of circulating ferritin with serum concentrations of fibroblast
growth factor-23 measured by three commercial assays. Ann Clin
Biochem 44:463– 466
Khosravi A, Cutler CM, Kelly MH, Chang R, Royal RE, Sherry RM,
Wodajo FM, Fedarko NS, Collins MT 2007 Determination of the
elimination half-life of fibroblast growth factor-23. J Clin Endocrinol Metab 92:2374 –2377
Shimada T, Urakawa I, Isakova T, Yamazaki Y, Epstein M, Wesseling-Perry K, Wolf M, Salusky IB, Jüppner H 2010 Circulating
fibroblast growth factor 23 in patients with end-stage renal disease
treated by peritoneal dialysis is intact and biologically active. J Clin
Endocrinol Metab 95:578 –585
Chefetz I, Kohno K, Izumi H, Uitto J, Richard G, Sprecher E 2009
GALNT3, a gene associated with hyperphosphatemic familial tumoral calcinosis, is transcriptionally regulated by extracellular
phosphate and modulates matrix metalloproteinase activity.
Biochim Biophys Acta 1792:61– 67
Bhattacharyya N, Wiench M, Dumitrescu C, Connolly BM, Bugge
TH, Patel HV, Gafni RI, Cherman N, Cho M, Hager GL, Collins
MT 13 January 2012 Mechanism of FGF23 processing in fibrous
dysplasia. J Bone Miner Res 10.1002/jbmr.1546
Goetz R, Nakada Y, Hu MC, Kurosu H, Wang L, Nakatani T, Shi
M, Eliseenkova AV, Razzaque MS, Moe OW, Kuro-o M, Mohammadi M 2010 Isolated C-terminal tail of FGF23 alleviates hypophosphatemia by inhibiting FGF23-FGFR-Klotho complex formation.
Proc Natl Acad Sci USA 107:407– 412
Antoniucci DM, Yamashita T, Portale AA 2006 Dietary phosphorus regulates serum fibroblast growth factor-23 concentrations in
healthy men. J Clin Endocrinol Metab 91:3144 –3149