Treatment of X-Linked Hypophosphatemia with Calcitriol and Phosphate Increases Circulating Fibroblast Growth Factor 23 Concentrations

Abstract

Context: X-Linked hypophosphatemia (XLH) is characterized by renal phosphate wasting, with inappropriately low or normal serum 1,25-dihydroxyvitamin D concentrations causing rickets and osteomalacia. Mutations in PHEX result in increased fibroblast growth factor 23 (FGF23) expression, elevating circulating FGF23 concentrations. Treating XLH with phosphate and calcitriol may further increase FGF23 concentrations, based on in vitro and in vivo models.

Objective: The aim of the study was to investigate whether current standard XLH therapies increase circulating FGF23 concentrations.

Design and Setting: We conducted a prospective observational study of XLH subjects during routine clinical management at two tertiary referral centers.

Patients: The study included 10 XLH patients (seven children, three adults; age, 2–30 yr) initiating therapy and five XLH patients (age, 18–41 yr) electing not to undergo therapy.

Intervention(s): Oral calcitriol and phosphate were administered.

Main Outcome Measures: We measured circulating intact FGF23 concentrations.

Results: Baseline circulating FGF23 concentrations were elevated in 14 of 15 subjects, increasing after treatment in most subjects. Follow-up was 14.4 ± 11.7 months (treatment cohort) and 25 ± 32 months (nontreatment cohort). FGF23 concentrations increased 132.7 ± 202.4% from pretreatment to peak during therapy but did not change significantly over time in the nontreatment cohort. FGF23 concentrations were related to phosphate doses (P = 0.04) and nonsignificantly to calcitriol doses (P = 0.06).

Conclusions: Treating XLH with phosphate and calcitriol was associated with concurrent increases in circulating FGF23 concentrations, which may diminish therapeutic effect or contribute to complications of therapy. Because it is unknown whether the degree of FGF23 elevation correlates with disease severity in XLH, further study is needed to determine whether adjusting therapy to minimize effects on FGF23 concentration is warranted.

X-linked hypophosphatemia is mediated by elevated circulating FGF23 concentrations; the current standard of care with calcitriol and phosphate further increases FGF23 concentrations in treated patients.

The most common inherited form of rickets is X-linked hypophosphatemia (XLH), biochemically characterized by hypophosphatemia due to renal phosphate wasting, with inappropriately low or normal 1,25-dihydroxyvitamin D concentrations. XLH results from mutations in PHEX (phosphate-regulating gene with homology to endopeptidases on the X-chromosome) causing abnormalities of osteocyte, osteoblast, and odontoblast function (1), including increased bone expression of fibroblast growth factor 23 (FGF23) (2). FGF23 inhibits renal tubule phosphate reabsorption and down-regulates 1α-hydroxylase (3).

The murine equivalent of XLH, the Hyp mouse has serum FGF23 concentrations 10-fold higher than wild-type mice (4). Although published FGF23 concentrations in humans with XLH range from normal to 10-fold elevations, the majority of XLH patients have elevated FGF23 concentrations (5,6). The significance of this wide variation is not known, but may represent an effect of medical treatment.

Current standard therapy for XLH involves treatment with high-dose phosphate and calcitriol, which improves the osteomalacia (7). However, growth remains suboptimal in many patients (8), and orthopedic procedures are often required to correct leg deformities. Furthermore, therapy with calcitriol and phosphate is associated with complications of nephrocalcinosis and hyperparathyroidism (8,9). Due to these complications, adult patients are sometimes not treated unless affected by pseudofractures or active bone pain with osteomalacia.

Calcitriol and phosphate stimulate FGF23 expression both in vitro and in animal models (10,11). In healthy humans, FGF23 levels increase over several days of phosphate loading and decrease with phosphate restriction (12). The purpose of this study was to investigate whether current standard XLH therapies increase circulating FGF23 concentrations in both adult and pediatric subjects with XLH.

Subjects and Methods

Subjects

Pediatric and adult patients with clinical and biochemical evidence of XLH were recruited by four clinicians at two tertiary referral centers. The diagnosis of XLH was based upon the presence of hypophosphatemia, renal phosphate wasting, biochemical and radiographic evidence of rickets and/or osteomalacia, with or without a family history of hypophosphatemic rickets. Subjects were excluded if they demonstrated clinical evidence of other acquired or inherited renal phosphate wasting disorders. PHEX mutations were confirmatory, but not required.

Subjects were recruited before initiating treatment. Two cohorts of patients were obtained: treatment (children and adults) and nontreatment (adults only). Baseline samples were obtained before initiating therapy. Subjects were followed longitudinally while either receiving standard medical therapy for XLH or not receiving therapy. Three children were recruited from Yale University; all other subjects were recruited from Indiana University.

The study was approved by the Indiana University-Purdue University Indianapolis/Clarian and Yale University institutional review boards; written informed consent was obtained from all subjects (or guardians) with assent from children 7 yr or older.

XLH therapy

Calcitriol and phosphate salts are the standard medical therapy for XLH (13). Doses of phosphate salts were titrated upward to minimize gastrointestinal side effects. Phosphate was given three to five times a day in doses titrated up to 20–35 mg elemental phosphorus/kg/d. Calcitriol was simultaneously titrated up to 20–30 ng/kg/d in two divided doses. Therapeutic decisions and dose adjustments were made by the treating physician without knowledge of FGF23 concentrations.

Measurements

Blood and urine samples were obtained during clinical visits. Serum and urine biochemistries were measured in standard clinical laboratories at Clarian and Yale. At each visit, serum phosphorous, calcium, creatinine, and alkaline phosphatase were measured. At baseline, PTH and urine calcium, creatinine, and phosphate were also measured. Although fasting samples were preferred, fasting samples were not always obtainable due to the timing of clinical visits.

Blood samples were stored at −80 C before measurement of intact FGF23. All samples from an individual were measured on the same FGF23 assay to minimize assay variability effects. We measured plasma (serum for three patients) intact FGF23 using a sandwich ELISA with monoclonal antibodies binding epitopes on either side of the RXXR cleavage motif of FGF23 (Kainos Laboratories, Tokyo, Japan) (14). The intraassay coefficient of variation was 5.4%.

Statistical analysis

Descriptive data were summarized for pretreatment and posttreatment periods. When multiple pretreatment samples were obtained (range, 1–3), these were averaged. Differences between groups (children vs. adults, treated vs. untreated) were tested using two-sample t tests on the original as well as normal-transformed data. Correlations between calcitriol and phosphate doses were calculated for individuals who had at least three distinct sets of dosages. Mixed effects models were used to fit FGF23 concentrations as functions of treatment doses (which were zero in the pretreatment period), with doses as fixed effects and subjects as random effects to take into account the correlations due to repeated measurements.

Results

The treatment cohort included 10 subjects: six female children, one male child, two male adults, and one female adult. Five children were naive to therapy at the time of the baseline assessment. Two children and all adults had a history of prior treatment with calcitriol and phosphate, but all had stopped treatment at least 1 yr before recruitment (at approximately 17–19 yr of age for the adults). All children were part of the treatment cohort. The adults in the treatment group began treatment due to ongoing bone pain (all three) and elevated serum alkaline phosphatase activity with pseudofracture (in two). The children ranged in age from 1.9–10.4 yr (mean, 4.9 ± 3.5 yr), and the treatment cohort adults ranged in age from 27.7–30.8 yr (mean, 29.2 ± 1.5 yr). Five adult female XLH subjects aged 18.6–41.4 yr (mean age, 30.7 ± 10.1 yr) were recruited in the nontreatment cohort.

Biochemistry measurements are summarized in Table 1. At baseline all subjects were hypophosphatemic for age. Adults in the treatment and nontreatment cohorts did not demonstrate significant differences in biochemistries, except for alkaline phosphatase (P = 0.03). PHEX mutations were confirmed in six of 10 subjects in the treatment cohort (two c1721C>G, three c1848delA, one c.2030_2031dupCA), and in four of five in the nontreatment cohort (one each, c1586 + 6T>C, c328_330delAAT, and 1735G>A; and one with both c58C>T and c281A>G). An additional treatment cohort subject was from a large kindred with a clear X-linked pedigree. This is consistent with reported mutation detection rates between 60 and 65%, even in subjects from clear X-linked dominant pedigrees (15).

Table 1.

Biochemistries in subjects with XLH

	Age (yr)	Serum phosphorus (mg/dl)	Serum creatinine (mg/dl)	Serum calcium (mg/dl)	PTH (pg/ml)	Alkaline phosphatase (U/liter)	TmP/GFRb (mg/dl)	FGF23 (pg/ml)	% Change FGF23
Children (n = 7)
Pretreatment	4.9 ± 3.5	2.7 ± 0.7	0.5 ± 0.1	9.5 ± 0.3	57 ± 35	487 ± 136	2.2 ± 0.5	329 ± 456
Posttreatmenta		3.4 ± 0.8	0.4 ± 0.2	9.9 ± 0.6	47 ± 34	425 ± 143		625 ± 1168	38.6 ± 48.6
Adult (n = 3)
Pretreatment	29.2 ± 1.5	1.8 ± 0.1	0.7 ± 0.2	9.0 ± 0.1	131 ± 49	177 ± 56	1.5 ± 0.0	216 ± 67
Posttreatmenta		2.3 ± 0.3	0.8 ± 0.3	9.0 ± 0.5		193 ± 13		1096 ± 848	352.4 ± 271.1
Total (n = 10)
Pretreatment		2.4 ± 0.7	0.5 ± 0.2	9.4 ± 0.4	80 ± 51	394 ± 188	2.0 ± 0.6	295 ± 378
Posttreatmenta		3.1 ± 0.9	0.5 ± 0.3	9.6 ± 0.7		374 ± 161		767 ± 1059	132.7 ± 202.4
Adult (n = 5)
Nontreatment	30.7 ± 10.1	2.0 ± 0.5	0.8 ± 0.2	9.1 ± 0.1	90 ± 35	72 ± 21c	1.8 ± 0.7	92 ± 12c
Follow-up		2.2 ± 0.3	0.8 ± 0.2	9.1 ± 0.2		91 ± 42		81 ± 26	−12.9 ± 19.2

Data are expressed as mean ± sd. Normal mean intact FGF23 on this assay is 30 pg/ml, with an upper limit of normal of 71 pg/ml in our laboratory (16). SI unit conversions: to convert the values for calcium to mmol/liter, multiply by 0.250; to convert the values for phosphorus or TmP/GFR to mmol/liter, multiply by 0.323; to convert the values for creatinine to μmol/liter, multiply by 76.26. 

^aPosttreatment laboratory values correspond to sample with the peak FGF23 concentration. 

^bTmP/GFR, Tubular maximum reabsorption of phosphate per deciliter glomerular filtrate. For this value, n = 4 children and n = 2 adults in the treatment cohort, and five adults in the nontreatment cohort. 

^cP = 0.03 for the baseline difference between the treated adults and the nontreated adults. 

Follow-up ranged from 2.5–35.3 months (mean, 14.4 ± 11.7 months) on therapy in the treatment cohort, and from 2.5–80 months (25 ± 32 months) in the nontreatment cohort. One treated subject was found to have nephrocalcinosis after starting treatment. No subject in either cohort developed hypercalcemic hyperparathyroidism.

In the treated cohort, pretreatment intact FGF23 concentrations were elevated in nine of 10 treatment cohort subjects compared with our previously reported controls (<71 pg/ml) (16). Mean pretreatment FGF23 concentrations did not differ significantly between adults and children. Baseline FGF23 concentrations were elevated in all nontreatment adult cohort subjects, but were significantly lower than in the treatment cohort adults (P = 0.03).

During therapy, FGF23 concentrations increased in most subjects and did not decrease in any subject. FGF23 concentrations increased by more than 20% in seven of 10 subjects and more than 100% in three of 10 subjects. Pediatric and adult posttreatment FGF23 concentrations did not differ significantly. In two treated subjects, FGF23 concentrations declined 1–2 months after temporary treatment cessation. Furthermore, FGF23 concentrations did not increase over time in the five untreated adult subjects (Fig. 1).

Figure 1.

Mean percentage of baseline FGF23 at peak during treatment (n = 10) and during follow-up of untreated subjects (n = 5). P = 0.007 for difference between treated and untreated groups.

Treatment doses ranged from 7–38 mg of elemental phosphorus/kg/d and 7–31 ng/kg/d of calcitriol, including the titration phases. Because doses of phosphate and calcitriol were generally increased together, there was a high correlation between treatment doses of these two agents (of seven subjects with at least three different doses, r > 0.96 in five subjects and r > 0.55 in the other two). FGF23 concentrations were positively related to phosphate (P = 0.04) and calcitriol doses (P = 0.06). However, the doses of calcitriol and phosphate were too highly correlated to determine their separate effects on FGF23.

Discussion

Clinical severity of XLH varies greatly, even among members of the same family with the same genotype (17). In our study, both baseline and posttreatment FGF23 varied widely among both pediatric and adult XLH subjects. However, FGF23 levels were comparable between adult and pediatric subjects. Therapy with phosphate and calcitriol was associated with concurrent increases in FGF23 concentrations. Furthermore, cessation of treatment in two subjects resulted in declining FGF23 concentrations. Although it is likely that patients in the nontreatment cohort were likely clinically less severe, the findings of lower baseline FGF23 concentrations and the absence of FGF23 increases in nontreated subjects also support a treatment effect on FGF23 concentrations.

A weakness of our study is the small sample number, representing the limited number of treatment-naive patients encountered. Furthermore, we cannot exclude patient compliance as a confounder, although poor compliance is likely to lessen the impact of therapy on FGF23 concentrations.

Strengths of our study include its longitudinal design, length of follow-up, and inclusion of patients naive to therapy. In addition, we have included both adults and children with XLH and identified similar effects of treatment on FGF23. These features allow us to demonstrate FGF23 changes in a real-world clinical setting, with generalizability to both pediatric and adult patients.

Our results are consistent with previous animal and human studies. Hyp mice respond to changes in phosphate and calcitriol intake, altering FGF23 expression (4,18). Cross-sectionally, medical treatment does not alter FGF23 concentrations in tumor-induced osteomalacia, but treated XLH subjects have higher FGF23 concentrations than untreated subjects (6). After washout in eight XLH subjects, FGF23 concentrations increased 6 wk after restarting treatment in another study that did not include a titration phase and used higher phosphate doses (38–62 mg/kg/d) than our study (19).

Although previous XLH studies documented that calcitriol with phosphate improves osteomalacia (7), simultaneously increasing FGF23 concentrations might limit therapeutic effectiveness or contribute to complications of therapy. If so, our findings could have potential clinical importance in the treatment of XLH, and perhaps of other FGF23-driven disorders. In autosomal dominant hypophosphatemic rickets, a disorder of FGF23 excess due to FGF23 mutations, patients clinically wax and wane in terms of the clinical phenotype and hypophosphatemia, at times with complete resolution of the biochemical phenotype. We have shown that the biochemical phenotype does correlate with FGF23 concentrations in autosomal dominant hypophosphatemic rickets, and resolution of the phenotype is associated with normalization of FGF23 (20). Although this does not prove adverse outcomes from further increasing FGF23 in XLH, it suggests that alterations in FGF23 do have meaningful clinical effects in disorders of underlying FGF23 excess. Treatments that do not directly address the underlying production of FGF23 might further increase bone FGF23 expression, counteracting the goal of raising the phosphate concentration.

In conclusion, our data suggest that current medical therapy of XLH may worsen the underlying phosphaturia by increasing FGF23 production. The consequences of worsening FGF23 concentrations are not yet clear. Because it is not known whether the degree of elevation in FGF23 level correlates with disease severity in XLH, further study is needed to determine whether adjusting therapy to minimize effects on FGF23 concentration is warranted.