FGF23 Concentrations Vary With Disease Status in Autosomal Dominant Hypophosphatemic Rickets

Authors

  • Erik A Imel,

    1. Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    2. Department of Pediatrics, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Siu L Hui,

    1. Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    Search for more papers by this author
  • Michael J Ecibs MD

    Corresponding author
    1. Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA
    2. Medical and Molecular Genetics, Indiana University School of Medicine, Indianapolis, Indiana, USA
    • 541 North Clinical Drive, Clinical Building 459, Indianapolis, IN 46202, USA
    Search for more papers by this author

  • Dr Econs has a patent pending for FGF23, and receives royalties from Kirin Brewery. All other authors state that they have no conflicts of interest.

Abstract

We measured FGF23 concentrations in subjects with ADHR. FGF23 and serum phosphate correlated positively in controls and negatively in subjects with ADHR. Elevated FGF23 concentrations were associated with lower phosphate values. The variable phenotype in ADHR may be caused by fluctuations in FGF23.

Introduction: Autosomal dominant hypophosphatemic rickets (ADHR) is a rare disorder of phosphaturia, hypophosphatemia, inappropriately low/normal calcitriol, and rickets/osteomalacia. ADHR is caused by mutations in a circulating peptide, fibroblast growth factor 23 (FGF23). We present the first report of FGF23 concentrations in subjects with ADHR. The aim was to test the hypotheses that subjects with ADHR have elevated FGF23 concentrations and that FGF23 concentrations are associated with disease severity.

Materials and Methods: This was an observational study at a tertiary referral center. Subjects were from three kindreds with FGF23 mutations causing ADHR (n = 42). Controls were participants from these families without mutations (n = 55). Fasting blood and urine samples were obtained. Biochemistries were determined, and FGF23 concentrations were measured using two ELISAs.

Results: Three cases are presented illustrating activity of disease and FGF23 concentrations. One case shows persistent hypophosphatemia and elevation of FGF23 concentrations, whereas another shows remission of hypophosphatemia corresponding to a decrease in previously elevated FGF23 concentrations. Overall cross-sectional group differences were nonsignificant for serum phosphate and FGF23 concentrations. C-terminal FGF23 concentration in controls was 61.0 ± 28.6 (SD) RU/ml (median, 52.5 RU/ml), and in subjects with mutations was 148.8 ± 374.5 RU/ml (median, 63.1 RU/ml). Mean intact FGF23 concentration in controls was 44.7 ± 14.9 pg/ml (median, 40.4 pg/ml), and in subjects with mutations was 83.2 ± 233.0 pg/ml (median, 39.0 pg/ml). C-terminal FGF23 concentrations were at least +2 SD in 10/42(24%), and intact FGF23 concentrations were at least +2 SD in 3/34(9%). Phosphate correlated positively with C-terminal and intact FGF23 in both controls and in subjects with mutations with phosphate >2.5 mg/dl but correlated significantly negatively with C-terminal and intact FGF23 in ADHR subjects with phosphate ≤2.5 mg/dl.

Conclusions: Elevated FGF23 concentrations are associated with hypophosphatemia in ADHR, and remission of the phenotype is associated with lower FGF23 concentrations. FGF23 has an opposite relationship with phosphate in ADHR compared with controls. We conclude that ADHR symptoms and disease severity likely fluctuate with FGF23 concentrations.

INTRODUCTION

Autosomal dominant hypophosphatemic rickets (ADHR, OMIM no. 193100), initially described by Bianchine et al.,(1) is a rare disorder characterized by renal phosphate wasting, hypophosphatemia, inappropriately low or normal calcitriol concentrations, and rickets/osteomalacia.(2) We previously evaluated a large ADHR kindred with many affected individuals.(3) This kindred provided us with an opportunity to explore the phenotypic variability of this disease in a large number of individuals with the same mutation. Similar to the more common disorder, X-linked hypophosphatemic rickets (XLH), we found no evidence of genetic anticipation or imprinting. In contrast to XLH, ADHR displays incomplete penetrance and variability in age of onset of clinical features.(3) The family contains two subgroups of affected individuals. One subgroup consists of patients who presented during childhood with phosphate wasting, rickets, and lower extremity deformity in a pattern similar to the classic presentation of XLH. The second subgroup consists of individuals who presented as adolescents or adults. These individuals complained of bone pain, weakness, and insufficiency fractures, but did not have lower extremity deformities.(3) Their clinical presentations were essentially identical to patients who present with tumor-induced osteomalacia, but the disease is not a paraneoplastic process. To date, delayed onset of clinically evident disease has only been observed in women. In addition, two men in that report showed evidence of past hypophosphatemia and rickets in childhood, but were able to discontinue therapy because of resolution of their phosphate and bone defects.

Positional cloning studies led to identification of the gene responsible for ADHR, fibroblast growth factor 23 (FGF23).(4) Three different mutations were found in four kindreds, each resulting in amino acid substitutions in FGF23: R176Q, R179Q, and R179W. These mutations occur at a protease cleavage site (RXXR), and the resulting mutant FGF23 molecules are resistant to cleavage.(5–7) FGF23 is a secreted peptide hormone that is detectable in the circulation of normal individuals and is responsive to dietary phosphate.(8–11) Both mutant and wildtype FGF23 decrease renal tubular phosphate reabsorption,(6) and mouse models of FGF23 overexpression manifest rickets and hypophosphatemia.(12,13) However, only full-length FGF23 reduced serum phosphate in vivo.(6)

FGF23 concentrations are frequently elevated in subjects with the renal phosphate wasting disorders tumor-induced osteomalacia (TIO) and XLH.(14–16) Although in vitro data predict that the ADHR phenotype is caused by elevated circulating concentrations of cleavage-resistant FGF23, data regarding circulating FGF23 concentrations in human subjects with ADHR have not been previously described. The aims of this study were to test the hypothesis that elevated FGF23 concentrations are associated with disease activity and phosphate concentration in ADHR. Here, we report the FGF23 concentrations in members of a large previously described ADHR kindred (family 1406)(3,4) and in members of two smaller kindreds. Additionally, our data suggest that the clinical manifestations of ADHR are even more variable than previously noted.

MATERIALS AND METHODS

Subjects

The study was approved by the Indiana University-Purdue University Indianapolis (IUPUI)/Clarian Institutional Review Board, and written informed consent was obtained from all subjects before participation. For this study, we included subjects 17–83 years of age from three white American ADHR kindreds. Study subjects have FGF23 mutations that were previously shown to cause ADHR, whereas controls were family members without FGF23 mutations.

We measured FGF23 concentrations in single samples in a total of 42 subjects with FGF23 mutations and in 55 control subjects without mutations. Family 1406 has been previously described.(3,4) Affected subjects from this family have an R176Q mutation in FGF23. Samples from family 1406 included 39 subjects with R176Q mutations and 51 subjects without mutations. Family 2401 includes two adults with the R179Q mutation in FGF23 and three adults without mutations. Family 2501 includes one subject with the FGF23 mutation at R179Q and one subject without a mutation in FGF23. Mutation detection was determined using PCR and sequencing gels for FGF23 exon 3 as previously described.(4) In addition, three cases are presented, showing correlation with clinical features; two include longitudinal data.

Biochemical analysis

Fasting blood samples were obtained on all subjects in serum separator and EDTA tubes. Fasting second morning urine samples were collected on some patients. Plasma and serum were separated and stored at −80°C until analysis. Serum calcium, phosphate, creatinine, and alkaline phosphatase, and urine calcium, creatinine, and phosphate were determined by standard clinical methods. Tubular maximum phosphate reabsorption per deciliter glomerular filtration rate (TmP/GFR) was estimated using the nomogram method of Walton and Bijvoet.(17) Because hypophosphatemia is a hallmark of the disease, serum phosphate was used as an indicator of active clinical disease at the time of sampling.

FGF23 assay

FGF23 concentrations were assessed on stored plasma samples using a C-terminal ELISA (Immunotopics, San Clemente, CA, USA) and an intact ELISA (Kainos Laboratories, Tokyo, Japan), according to the manufacturer's specifications. All subjects were measured with the C-terminal FGF23 assay. However, because of limitations in sample volumes, we were unable to measure intact FGF23 on all subjects. We measured intact FGF23 concentrations on 34 subjects with mutations and 33 subjects without mutations.

Statistical analysis

Descriptive statistics (means and SDs) were calculated. The distributions of the FGF23 measurements were highly skewed in the ADHR group, so medians were also reported. The normal limits for each FGF23 assay were derived from the mean ± 2 SD in the control group. Two-sample t-tests were used to test the difference in means between the ADHR and control groups. Correlations were calculated between each FGF23 assay and the phosphate levels for each group, and regression analysis was used to test whether the slope of the regression of FGF23 on phosphate was the same in both the ADHR and control groups. All tests were based on a 5% significance level.

RESULTS

Cases

Three cases are presented from family 1406. These cases illustrate the clinical features in three female subjects with ADHR and show remission of the phenotype in ADHR and the relationship of phosphate and clinical symptoms to FGF23 concentrations in individual patients.

Figure Figure 1.

Bone scan from proband showing increased uptake in multiple areas indicating stress fractures and pseudofractures. Used with permission.(3) Copyright 1997, The Endocrine Society.

Case 1-proband

Case 1 is the proband (referred to as patient VI-51 in reference 3) from family 1406.(3) She had a normal serum phosphorus concentration at screening at 3 months of age and grew normally. However, at 14.5 years of age, she developed back pain and renal phosphate wasting with phosphorous of 1.2 mg/dl and was treated with high-dose vitamin D with symptomatic improvement. At 19.5 years of age, she developed fatigue, ankle pain and back pain with recurrent hypophosphatemia (1.4 mg/dl), low TmP/GFR (1.2 mg/dl), PTH of 24 pg/ml, and elevated alkaline phosphatase (229 IU/liter). Bone scan revealed multiple areas of increased tracer uptake (Fig. 1), whereas both pseudofractures and stress fractures were evident on radiographs. Osteomalacia was present on bone biopsy. She was treated with high-dose phosphate and calcitriol. After several years, she noted marked improvement in her symptoms and gradually stopped her medications. She is currently asymptomatic and normophosphatemic (phosphate, 3.1 mg/dl; TmP/GFR, 3.5 mg/dl). At the time of her initial presentation, FGF23 had not been discovered. Later, FGF23 concentrations with normophosphatemia and absence of symptoms were 151.1 RU/ml (mildly elevated) with the C-terminal assay and 50.8 pg/ml (normal) with the intact assay.

Case 2

Case 2 is a recently identified distant relative of case 1. She presented at age 51 years with complaints of intermittent hip pain, generalized weakness, and fatigue. Although biochemistries are not available, she gave a history of several years of waxing and waning symptoms. She did not have a history of rickets but had multiple tooth abscesses as a young adult. In her 40s, she developed fatigue and debilitating muscle pains and cramps. On our initial evaluation, she had severe hypophosphatemia (0.7 mg/dl) with a highly elevated FGF23 concentration (C-terminal, 841.8 RU/ml; intact FGF23, 126.7 pg/ml; Fig. 2). Physical examination showed a height of 160.4 cm and no weakness, kyphosis, or lower extremity deformities. She wore upper plate dentures and was missing several lower teeth. Treatment with high-dose phosphate and calcitriol has been complicated by abdominal discomfort with medications, leading to poor compliance. She has remained hypophosphatemic and is severely limited by her symptoms, requiring a cane or crutches at times for ambulation. Her FGF23 concentrations have remained elevated, accompanied by low phosphate and low TmP/GFR (Fig. 2).

Case 3

Case 3 is the daughter of case 2. She presented at 26 years of age with hip pain and stiffness, weakness, fatigue, and hypophosphatemia that developed after the birth of her second child. She had no childhood history of rickets or hypophosphatemia; although she did have surgery for scoliosis at 14 years of age. She had a history of one tooth abscess. On physical examination, her height was 166 cm, and she had normal dentition, a surgical scar on her back from scoliosis surgery, and no lower extremity deformity or weakness. She had a slight decrease in hip external rotation bilaterally. Her FGF23 concentrations were quite elevated (C-terminal, 898.3 RU/ml; intact, 190.2 pg/ml) in the setting of low serum phosphate (1.6 mg/dl) and TmP/GFR (0.9 mg/dl). She was treated with high-dose phosphate and calcitriol. She gradually noted a decrease in her symptoms, and after ∼1 year of therapy, she reported taking her medications intermittently, because of complete resolution of her symptoms. She eventually stopped all medications and had resolution of her hypophosphatemia with normalization of her Tmp/GFR. This improvement coincided with decreasing FGF23 concentrations into the normal range (Fig. 2).

Comparison between ADHR and control groups

ADHR and control subjects came from the same families. Demographic data for ADHR and control subjects are listed in Table 1. Groups were similar in age range. Only 10/42 subjects with ADHR mutations were on treatment (with calcitriol and phosphate) at the time of sampling for FGF23.

Biochemistries

Table 1 also indicates the biochemical indices of subjects with FGF23 mutations and controls. Most subjects with mutations were not hypophosphatemic, indicating that many subjects did not have active disease at the time of the evaluation. Only 16/42 subjects with ADHR mutations had phosphate ≤2.5 mg/dl at the time of FGF23 measurement. There were no significant differences between control subjects and subjects with FGF23 mutations in terms of serum phosphate, calcium, creatinine, or alkaline phosphatase.

Figure Figure 2.

Phosphate, TmP/GFR, C-terminal FGF23, and intact FGF23 concentrations over time in case 2 (Δ) and case 3 (•). Dashed lines are used to indicate phosphate and Tmp/GFR of 2.5 mg/dl and the upper limit of normal for FGF23 assays. Both subjects had elevated FGF23 concentrations at presentation (time = 0 months). Therapy with phosphate and calcitriol began after the second data point for each subject (arrows). Both C-terminal and intact FGF23 concentrations rose initially after treatment, but in case 3, the FGF23, phosphate, and TmP/GFR all subsequently normalized, whereas these remained abnormal in case 2. Open arrows indicate when case 3 stopped taking phosphate and calcitriol, after normalization of FGF23 concentrations.

Table Table 1.. Demographic and Laboratory Values for Subjects and Controls
original image

FGF23 concentrations

A total of 55 subjects without mutations and 42 subjects with FGF23 mutations were assessed for C-terminal FGF23 concentrations. A total of 33 subjects without mutations and 34 subjects with FGF23 mutations were assessed for intact FGF23 concentrations. Figure 3 shows the C-terminal and intact FGF23 concentrations in control subjects and individuals with FGF23 mutations. Although the plots are skewed toward higher values in subjects with FGF23 mutations, the medians were similar, and there was no significant difference between the overall groups for C-terminal and intact FGF23 concentrations (Table 1). Most subjects were nontreated. Both pretreatment and post-treatment values were available only for the subjects in cases 2 and 3. In both cases, FGF23 concentrations increased initially after treatment began (Fig. 2). Subsequent wide variation in the FGF23 concentrations in case 2 may be associated with inconsistency in taking therapy because of gastrointestinal intolerance. Of note, case 3's FGF23 concentration became normal before cessation of therapy and before her phosphate normalized.

A small number of serial samples in normal controls were available. For C-terminal and intact FGF23, the within-subject SD was 38% and 54% of the mean, respectively, in eight controls over 3–10 years. However, this change occurred largely within the normal range.

Most samples were only thawed once for each assay, and the order of the assays varied. There was no decrease with repeat assays after freeze-thaw cycles in 23 random samples.

Relationship of FGF23 with phosphate

FGF23 concentrations were +2 SD above the mean of the controls in 10/42 (24%) with C-terminal FGF23 assay and 3/34(9%) with intact FGF23 assay. Most subjects with a prior history of clinically significant hypophosphatemia, rickets, and osteomalacia had FGF23 concentrations within the range of control subjects. At the time of sampling, the majority of subjects were asymptomatic, and only 16/42 subjects with ADHR mutations had phosphate concentrations ≤2.5 mg/dl.

Figure Figure 3.

FGF23 concentrations using two assays for controls and subjects with ADHR mutations. The median for each group is indicated by the horizontal bar. Differences are not statistically significant.

Figure 4 shows the relationship between FGF23 and phosphate concentrations in these kindreds. In normal subjects, C-terminal FGF23 correlated positively with serum phosphate (r = +0.321). Subjects with ADHR mutations had a negative correlation between C-terminal FGF23 concentrations and serum phosphate. However, this effect was entirely caused by the subgroup of subjects with ADHR mutations who had serum phosphate ≤2.5 mg/dl (r = −0.769). Those with phosphate >2.5 mg/dl had a positive correlation between C-terminal FGF23 and phosphate (r = +0.316). The slope of the relationship was significantly different between controls and ADHR subjects with low phosphate (p = 0.005) but not between controls and ADHR subjects with normal phosphate.

A similar relationship was observed for the intact FGF23 assay. In normal subjects, intact FGF23 correlated positively with serum phosphate (r = +0.276), whereas in subjects with ADHR mutations, intact FGF23 correlated negatively with serum phosphate in the low-phosphate group (r = −0.825). The slope of the relationship was different between the two groups (p = 0.023). In ADHR subjects with normal phosphate, intact FGF23 and phosphate correlated positively (r = +0.147), and the slope was not significantly different from controls.

DISCUSSION

We present the first report of measurement of FGF23 concentrations in human subjects with FGF23 mutations causing ADHR. In vitro data predict that the mutation in FGF23 would result in elevated concentrations of circulating bioactive FGF23. We showed that FGF23 concentrations were not consistently elevated in patients with FGF23 gene mutations, even among those with a prior history of clinical manifestations. However, FGF23 concentrations seemed to be elevated only in the presence of active disease.

Three cases illustrated the relationship of the clinical features to FGF23 concentrations. The proband was well documented to have severe disease, but over time, her symptoms resolved and her phosphate concentrations were normal once FGF23 measurement became available. While normophosphatemic, her intact FGF23 was normal, and her C-terminal FGF23 was slightly elevated. Case 2 has had persistent severe hypophosphatemia and osteomalacia. Her FGF23 concentrations have remained elevated. In addition, case 2, as well as the overall data on hypophosphatemic subjects, showed wide variability in FGF23 concentrations. It is possible that this variability is caused by a threshold effect of FGF23 on phosphate concentration, beyond which further increases in FGF23 have little additive effect. However, this finding may also be caused by changes in other physiologic factors acting in combination on phosphate metabolism. In contrast, case 3 strikingly showed concurrent resolution of both hypophosphatemia and abnormal TmP/GFR with normalization of FGF23 concentrations. In the total study population, FGF23 and phosphate had an opposite relationship in normal subjects versus those with mutations. In those without mutations, the FGF23 concentration was positively correlated with serum phosphate concentration, similarly to that shown by other investigators.(9,18) However, among subjects with FGF23 mutations, FGF23 correlated negatively with serum phosphate concentrations in subjects with hypophosphatemia but positively in subjects with normal phosphate concentrations. This relationship was present for both assays. Most subjects with FGF23 mutations were asymptomatic and normophosphatemic at the time of study and not on treatment. Interestingly, only 24% and 9% of subjects with FGF23 mutations had elevated C-terminal and intact FGF23 concentrations, respectively, whereas 38% of subjects with FGF23 mutations had phosphate ≤2.5 mg/dl. Similar to other endocrine disorders, it may be necessary to interpret FGF23 concentrations in relation to the concentration of the modified target (i.e., phosphate).

Figure Figure 4.

(A) C-terminal FGF23 and (B) intact FGF23 plotted against phosphate values show a positive correlation in controls (□, gray regression lines) and in ADHR subjects with phosphate >2.5 mg/dl (•, dashed regression lines). FGF23 concentrations show a negative correlation in subjects with ADHR mutations with phosphate ≤2.5 mg/dl (⋄, black regression lines). Phosphate 2.5 mg/dl is indicated by a horizontal line. FGF23 concentrations are plotted on a logarithmic scale on the x-axis. Listed p values are for the difference in slopes between ADHR subjects with phosphate ≤2.5 mg/dl and controls.

A comparison of other specific variables with FGF23 concentrations requires measurement of simultaneous samples. Although many subjects had measures of 1,25-dihydroxyvitamin D, PTH, and TmP/GFR in the past, these did not correspond in time with the available samples for assessment of FGF23 concentrations. Thus, we were unable to compare other potential measures of disease activity to FGF23 concentrations. Previously Econs and McEnery(3) observed that the penetrance was incomplete, and in some cases delayed, and that resolution of the phosphate wasting defect occurred in some male patients. This study expands previous data to confirm that remission of the phosphate wasting defect occurs in other subjects with this mutation. Our data indicate that patients with FGF23 mutations who seem to resolve their phosphate metabolism defect are doing so by modulating their FGF23 concentration rather than compensating for the effects of excess FGF23. Consequently, this modulation is responsible for the variable phenotype seen in ADHR. The means of this regulation is unclear. Patients may be decreasing FGF23 expression, altering post-translational modification, or may be increasing degradation of FGF23 by an alternative pathway. We were unable to assess effects on FGF23 expression, because this would require bone biopsy data from affected subjects both before and after resolution of their phosphate defect. In addition, one might expect that degradation by an alternate pathway might still result in an increase in C-terminal FGF23 compared with intact FGF23 in those subjects who resolve their hypophosphatemia. This was not observed. In fact, both C-terminal and intact FGF23 assays provide evidence of decreased circulating FGF23 in normophosphatemic subjects. Further study is necessary to determine the mechanism of regulation, and dysregulation, of FGF23 in subjects with ADHR, that lead to the waxing/waning phenotype.

Whereas there is evidence that treatment with high-dose calcitriol and phosphate are necessary to heal rickets and osteomalacia in subjects with phosphate wasting disorders,(19,20) there is no evidence to suggest that treatment with calcitriol and/or phosphate would improve the FGF23 concentrations. On the contrary, increased dietary phosphate increases serum FGF23,(8,10,11) and both in vitro and in vivo data indicate that current treatment strategies for these disorders may actually increase the production of FGF23.(21–23) Thus, the normalization of FGF23 concentrations that occurs in some patients is unlikely to be the result of current therapy, and involves an intrinsic physiologic alteration in the patient's metabolism of FGF23, leading to resolution of ADHR features.

We have shown that FGF23 concentrations are not universally elevated in ADHR but have an opposite relationship with phosphate compared with normal subjects. Elevated FGF23 concentrations are associated with lower phosphate concentrations, and remission of the phenotype is associated with lower FGF23 concentrations. We conclude that ADHR symptoms and disease severity likely fluctuate with the concentration of FGF23 and that the ability to alter the FGF23 concentration changes the clinical phenotype, resulting in the observed delayed penetrance, and also in, at least temporary, resolution of the phenotype in some individuals.

Acknowledgements

We are indebted to the families with ADHR for participation in this study This work was supported by National Institutes of Health Grants R01AR42228, T32 AR 07581-09, and MO1 RR00750.

Ancillary