Clinical utility of systemic venous sampling of FGF23 for identifying tumours responsible for tumour-induced osteomalacia


Dr. Seiji Fukumoto, Division of Nephrology & Endocrinology, Department of Medicine, University of Tokyo Hospital, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655 Japan.
(fax: +81 3 5800 9836; e-mail:

Dear Sir,Tumour-induced osteomalacia (TIO) is a rare paraneoplastic syndrome characterized by hypophosphataemia with impaired renal tubular phosphate reabsorption and low serum 1,25-dihydroxyvitamin D [1,25(OH)2D] for hypophosphataemia [1]. TIO is usually caused by a benign mesenchymal tumour known as phosphaturic mesenchymal tumour, mixed connective tissue variant (PMTMCT) [2]. Patients with TIO sometimes become completely bedridden because of severe muscle weakness and bone pain. However, the removal of the causative tumours completely cures these symptoms and biochemical abnormalities. Therefore, the identification of the tumours responsible for TIO is clinically important. It is not uncommon for tumours that cause TIO to be small and exist within bone, thus making them difficult to detect. Several techniques have been used to identify the tumours responsible for TIO, including computed tomography (CT), skeletal survey by magnetic resonance imaging (MRI), octreotide scintigraphy and positron emission tomography (PET), indicating that there is no standardized method for this purpose [3–5]. In addition, even if tumours are found by these methods, none is able to establish that the detected tumour is indeed the cause of TIO. Fibroblast growth factor 23 (FGF23) has been identified as a causative factor for TIO [6]. It has been shown that circulating levels of FGF23 are elevated in patients with TIO and this high FGF23 concentration rapidly decreases after the resection of tumours responsible for TIO [7, 8]. We have previously reported the case of a patient with TIO whose responsible tumour was confirmed by venous sampling for analysis of FGF23 [9]. Subsequent reports also suggested that systemic venous sampling for FGF23 analysis is useful for the identification and confirmation of tumours responsible for TIO [10–12]. However, the clinical utility of this method has not been evaluated in a series of patients with TIO. Therefore, we conducted a prospective study in 10 consecutive patients with suspected TIO.

Patients admitted to our hospital between May 2005 and December 2009 with adult-onset hypophosphataemic osteomalacia were evaluated in this study. The diagnosis of TIO was suspected based on clinical symptoms and laboratory data compatible with hypophosphataemic osteomalacia and high circulating levels of FGF23, as summarized in Table 1. FGF23 was measured in duplicate using an enzyme-linked immunosorbent assay (ELISA) that detects only full-length biologically active FGF23 (Kainos, Japan) [13]. None of these patients had a family history of hypophosphataemic rickets/osteomalacia. Written informed consent was obtained from all patients, and the study was approved by the institutional review board of the University of Tokyo. A total of 16–22 blood samples were collected through a catheter inserted in the right femoral vein in each patient. Approximately 2-mL blood was obtained at each sampling point. Samples were collected from the bilateral subclavian veins, bilateral internal jugular veins, bilateral brachiocephalic veins, superior vena cava proximal and distal to the junction of the azygous vein, inferior vena cava proximal and distal to the junction of the renal veins, bilateral common iliac veins, bilateral external iliac veins, bilateral internal iliac veins and bilateral femoral veins (proximal: between the lower limit of the femoral head and the minor trochanter, i.e. distal to the junction of the great saphenous vein; distal: distal to the junction of the deep femoral vein). In some patients, blood samples could not be obtained from several veins because of technical problems. In patients in whom venous sampling indicated the region of the responsible tumour, imaging studies (e.g. CT and/or MRI) were then targeted at these areas. All patients whose tumours were identified underwent surgery for resection of the tumours.

Table 1. Clinical characteristics of 10 suspected patients with tumour-induced osteomalacia
Patient no. 12345678910Reference range
  1. Conversion factors for SI unit are as follows: Albumin 10 (g L−1), Ca 0.25 (mmol L−1), Pi 0.323 (mmol L−1), Cr 88.4 (μmol L−1), 1,25(OH)2D 2.6 (pmol L−1l)

Age 53 614152673361664051 
Sex MaleMaleMaleFemaleMaleFemaleMaleFemaleFemaleMale 
Alb (gdL−1)–4.9
Ca (mgdL−1)–9.7
Pi (mgdL−1)–4.5
Cr (mgdL−1)–1.20 (Male)
            0.40–0.90 (Female)
1,25(OH)2D(pg mL−1) 25.940.527.7<235.962.529.435.86.792.920–60
Intact PTH (pgmL−1) 34020864237284375362210–65
BAP (U L−1) 71.554.8131.0306.0106.045.4203.0138.0220.0340.013.0–33.9
TmP/GFR (mgdL−1)–4.3
FGF23 (pg mL−1) 264481082412907026822516116248210–50
Neutral phosphate ++--++---+ 
Active vitamin D3 ++-+++++-+ 

Nine patients (cases 2–10) had been referred to our hospital. In eight of these patients (cases 2–8 and 10), some imaging studies had been performed previously in other hospitals. However, tumours responsible for TIO had not been detected in any of these patients. Systemic venous sampling suggested the region of responsible tumours in eight patients (Figs 1 and 2).

Figure 1.

Venous sampling for fibroblast growth factor 23 (FGF23) in 10 patients with suspected tumour-induced osteomalacia. Blood samples were collected (16–22 per patient), and FGF23 levels were measured using an ELISA that detects full-length biologically active FGF23. Approximate positions of responsible tumours identified by subsequent imaging studies are shown as black circles.

Figure 2.

Imaging studies of tumours responsible for tumour-induced osteomalacia (TIO). In eight patients, the areas suspected of tumour involvement following venous sampling of fibroblast growth factor 23 were examined by computed tomography (case 3) and/or magnetic resonance imaging. The responsible tumours were found in all of these patients. Arrows indicate the tumours responsible for TIO. Gd, gadolinium enhancement.

Case 1: the FGF23 level was highest in the right femoral vein and gradually decreased in the right external and common iliac veins. Subsequent imaging identified a tumour in the distal epiphysis of the right femoral bone. Case 2: there was laterality of FGF23 levels in several veins. FGF23 levels were somewhat variable but highest in the proximal inferior vena cava. Re-examination of previously obtained CT scans after systemic sampling and subsequent MRI indicated the presence of a tumour in the right diapophysis of the third lumbar vertebra. Case 3: the FGF23 level was highest in the left internal jugular vein, and CT identified a tumour in the left maxilla. Case 4: the FGF23 level was highest in the distal superior vena cava. In addition, the second highest level was observed in the right internal jugular vein. Again, re-examination of a previously obtained CT scan and subsequent MRI identified a tumour in the right temporal bone. Further measurement of FGF23 in the right external jugular vein showed clearly elevated levels (1272 pg mL−1). Case 5: the highest level of FGF23 was obtained in the distal inferior vena cava and the second highest in the right common iliac vein. Because of these results, subsequent MRI was focused on the right lower body and a tumour was detected in the cortex of the right proximal femur. Case 7: the highest FGF23 level was observed in the distal superior vena cava. The level in the right brachiocephalic vein was almost as high. During venous sampling, the patient’s left brachiocephalic vein was found to be thin and rudimentary. Therefore, it was considered that most blood from his head drained into the right jugular veins. Subsequent imaging studies identified a tumour in the skull base. The tumour was found by MRI and was present on the left side but just adjacent to the midline. Case 9: the FGF23 level was highest in the right femoral veins, and imaging studies indicated the presence of a tumour in the distal epiphysis of the right femoral bone. Case 10: the highest FGF23 level was observed in the left external iliac vein. After sampling, the patient noticed a tumour in his left sole. Subsequent MRI indicated the presence of a tumour in his sole, and the FGF23 concentration in the left great saphenous vein was found to be relatively high (6500 pg mL−1). The location of the responsible tumours was not detected in cases 6 and 8. In these patients, FGF23 levels were almost the same throughout the body. In addition, PET did not find a tumour in either case. The systemic venous sampling was not associated with any significant adverse events in any case.

All eight patients in whom the suspicious tumours had been identified by systemic sampling underwent surgery. All the tumours were pathologically diagnosed as PMTMCT. FGF23 levels became undetectable (<3 pg mL−1) in cases 3, 4, 5, 7 and 9 within 2 days after surgery. In addition, FGF23 levels decreased to below the lower limit of the reference range in case 10 within 2 days. This suggests that the production of FGF23 from normal bone, which produces FGF23 under physiological conditions, is suppressed in patients with TIO and therefore the tumours responsible for TIO are the only source of elevated FGF23 in these patients. Hypophosphataemia and symptoms of TIO disappeared in cases 3, 4, 5, 7, 9 and 10. In patients 1 and 2, as much tumoural tissue as possible was removed. However, the FGF23 levels did not decrease to within the reference range in these patients. Hypophosphataemia persisted indicating that tissues producing FGF23 remained and TIO was not cured in these two patients.

In this series, venous sampling suggested the region of tumours responsible for TIO in eight of 10 patients. Practically, it would be difficult to determine the smallest significant increase in FGF23 concentration because the distances between the tumours and the sampling points and draining routes vary between patients. Therefore, we speculated that the responsible tumours should be present in the region draining into the vein with the highest FGF23 level. Several points should be noted when evaluating the results of systemic venous sampling. First, blood samples could not be obtained from the peripheral or external jugular veins because of technical difficulties. Therefore, it would be difficult to detect tumours draining into these veins. Second, as shown in cases 5, 9 and 10, the veins with the highest FGF23 levels do not necessarily indicate the exact location of the responsible tumours. One of the reasons for this discrepancy is the site of junctions of veins. In case 9, the highest FGF23 level was observed in the proximal femoral vein, whereas the tumour was in fact found in the patient’s knee. Because the blood sample obtained from the proximal femoral vein contains blood draining into the deep femoral vein, the high FGF23 level in the proximal femoral vein might have indicated the further higher level in the deep femoral vein that receives blood from the knee. Similarly, the highest FGF23 level was observed in the external iliac vein in case 10. It is likely that this FGF23 level reflected the even higher FGF23 concentration in the great saphenous vein because this vein joins the femoral vein between the sampling points of the external iliac vein and the proximal femoral vein. It is necessary to specify the regions of tumours considering possible routes of draining veins. It is noteworthy that previous imaging studies had failed to detect the responsible tumours in eight cases and the re-evaluation of previously obtained CT images identified the tumours after systemic sampling in two cases (cases 2 and 4) in our study. This result demonstrates the difficulty in identifying tumours responsible for TIO without prior knowledge of the suspected regions to investigate.

Venous sampling did not indicate the regions of responsible tumours in two patients and the diagnosis of TIO could not be established in these patients. They experienced mild symptoms and no bone pain after the initiation of therapy with neutral phosphate and active vitamin D3. Because there have been several reports of a genetic cause of adult-onset hypophosphataemic osteomalacia [14], we sequenced all the exons and exon–intron junctions of phosphate-regulating gene with homologies to endopeptidases on the X chromosome (PHEX), FGF23 and dentin matrix protein 1 (DMP1) genes by direct sequencing but found no mutations (data not shown). The results suggest that these two patients indeed have TIO; however, other unknown causes of FGF23-related hypophosphataemia remain a possibility as evidenced by a recent report that intravenous administration of iron polymaltose increases FGF23 levels and induces hypophosphataemia [15].

In conclusion, we have demonstrated that systemic venous sampling for FGF23 is useful for the identification of the responsible tumours in most but not all patients with TIO.

Disclosure summary

No conflict of interest to declare.