Validation of a next-generation sequencing (NGS) panel to improve the diagnosis of X-linked hypophosphataemia (XLH) and other genetic disorders of renal phosphate wasting

Background Hypophosphataemic rickets (HR) comprise a clinically and genetically heterogeneous group of conditions, defined by renal-tubular phosphate wasting and consecutive loss of bone mineralisation. X-linked hypophosphataemia (XLH) is the most common form, caused by inactivating dominant mutations in PHEX, a gene encompassing 22 exons located at Xp22.1. XLH is treatable by anti-Fibroblast Growth Factor 23 antibody, while for other forms of HR such as therapy may not be indicated. Therefore, a genetic differentiation of HR is recommended. Objective To develop and validate a next-generation sequencing panel for HR with special focus on PHEX. Design and methods We designed an AmpliSeq gene panel for the IonTorrent PGM next-generation platform for PHEX and ten other HR-related genes. For validation of PHEX sequencing 50 DNA-samples from XLH-patients, in whom 42 different mutations in PHEX and 1 structural variation have been proven before, were blinded, anonymised and investigated with the NGS panel. In addition, we analyzed one known homozygous DMP1 mutation and two samples of HR-patients, where no pathogenic PHEX mutation had been detected by conventional sequencing. Results The panel detected all 42 pathogenic missense/nonsense/splice-site/indel PHEX-mutations and in one the known homozygous DMP1 mutation. In the remaining two patients, we revealed a somatic mosaicism of a PHEX mutation in one; as well as two variations in DMP1 and a very rare compound heterozygous variation in ENPP1 in the second patient. Conclusions This developed NGS panel is a reliable tool with high sensitivity and specificity for the diagnosis of XLH and related forms of HR.


Introduction
X-linked hypophosphataemia (XLH) (OMIM#307800) is the most common genetic disorder of phosphate homeostasis characterized by renal phosphate wasting and hypophosphataemia. It affects about one in 20.000 individuals (1) and follows an Xchromosomal dominant inheritance pattern.
Children affected by XLH present with a broad phenotypic spectrum ranging from isolated hypophosphatemia with few clinical signs up to severe symptoms, such as rickets with extreme lower limb deformities, distinct tooth problems (such as dental abscesses), and a disproportionate short stature (reviewed in (2)). In adulthood further symptoms may occur, such as osteomalacia, arthrosis, pseudo fractures, and diminished final height (3). Further clinical signs are hearing difficulties, enthesopathy, and muscular dysfunction. As the disorder is more under focus during the last few years, more clinical signs have been associated with XLH, such as Arnold-Chiari malformation and other craniofacial abnormalities. Furthermore, bone pain is a very pronounced sign in XLH leading to an impairment of the quality of life in affected children and adults (reviewed in (2)).
XLH is caused by mutations in the PHEX gene encoding the cleavage enzyme phosphateregulating neutral endopeptidase (PHEX) located on the X chromosome. Today, more than 588 mutations have been reported, spread all over the 22 exons of the PHEX gene (http://www.hgmd.cf.ac.uk/). This includes point mutations, deletions, insertions, as well as intronic variations, presumably altering PHEX function. Although the pathophysiology of XLH is not fully understood, the inactivation of the PHEX protein (expressed predominantly in osteoblasts) leads to an increase of fibroblasts growth factor 23 (FGF23) levels. High FGF23 levels cause urinary phosphate wasting by down-regulating the renal sodium phosphate transporters 2a and 2c (NaPi2a and NaPi2c, respectively) and reducing transformation of 25-OH-Vitamin D3 to the active vitamin D form 1,25-OH 2 -Vitamin D 3 (4)(5)(6) leading to abnormal low levels of 1,25-OH 2 -Vitamin D 3 despite of hypophosphatemia.
In addition, elevated alkaline phosphatase is seen as a marker of higher bone turnover related to rickets.
There is a high overlap between XLH and other forms of HR both in clinical as well as in laboratory findings. Some of those components that have been associated with elevated FGF23 expression or decreased degradation of FGF23 include (amongst others) FAM20C (family with sequence similarity 20 member C) (7), furthermore, ENPP1 (ectonucleotide pyrophosphatase/ phosphodiesterase) encoded by the ENPP1 gene (8), and DMP1 (dentin matrix acidic phosphoprotein 1) encoded by the DMP1 gene (9, 10).
Consequently, inactivating mutations in these genes also lead to an elevation of serum FGF23 levels and to disorders with a similar phenotype to XLH. A further known pathomechanism is caused by increased FGF23 levels due mutations in FGF23 itself, which affect the cleavage site for degradation. This condition follows an autosomaldominant (AD) inheritance and the phenotype seems to be milder (11). In contrast to XLH, ADHR shows incomplete penetrance, variable age at onset, and vanishing of the phosphate-wasting defect in rare cases (11)(12)(13). Table 1 summarizes different forms of HR with their biochemical characteristics in comparison to nutritional rickets.
Conventional therapy of HR includes oral phosphate supplementation and, in forms with FGF23-mediated hypophosphatemia, calcitriol; however, this therapy further stimulates FGF23 excretion, enforcing the renal phosphate wasting (6). Recently, a novel therapy with an anti-FGF23 antibody has been approved and current results demonstrate an enormous impact on medical outcome for patients with XLH in children (14)(15)(16) and adults (17). The novel therapy with Burosumab has only been approved for XLH, and it is currently unknown if patients with other forms of HR with FGF23 elevation might profit or not from this therapy. Some forms of HR, e.g. caused by ENPP1 mutation, may even have unfavourable effects from Burosumab such as hypercalcemia and calcification, although this is also currently unknown.
In XLH, early diagnosis followed by an immediate treatment has a strong impact on the patient's long-term outcome (18). However, the diagnosis of this rare condition is often delayed. The first clinical signs are often mild and occur when a toddler starts standing alone and walking, leading to bowing of the lower limbs, which can also be the first sign of rickets due to vitamin D deficiency. Even the biochemical signs are not always straight forward and may altogether not differentiate between XLH and other forms of HR (19). Therefore, the molecular genetic confirmation of the clinically and biochemically based diagnosis of XLH and differentiation from other forms of HR has been recommended by many specialists of this disorder (2). Until today, the gold standard for the search of mutations in the PHEX gene has been Sanger sequencing of all twenty-two exons including the exon/intron boundaries for detecting splice site mutations, followed by multiplex ligation-mediated probe amplification analysis (MLPA). By this approach, the diagnosis is relatively expensive and time consuming and cannot detect other forms of HR. If several patients were to be analyzed in one approach and several genes of one panel were to be examined for some of the patients, we were able to determine a total time and cost saving from 10 samples onward by NGS analysis.

7
For these reasons we developed a next generation sequencing (NGS) panel comprising not only all exons and the intron boundaries of the PHEX gene but also ten other genes, in which mutations are known to cause renal phosphate wasting disorders to ensure distinct differential diagnosis of XLH.

Material and Methods
The ethical committee of the University of Lübeck approved this part of the study in January 2004 (04-020) and confirmed the ethical correctness for developing of a NGS panel with the same samples in October 2018 .  Table 2  Sequence analysis was carried out with the software module SeqNext (SeqPilot™, JSI, Ettenheim, Germany). Small gaps in the designed panel, mainly due to large homopolymer stretches, were complemented by Sanger sequencing (Table 2 and Supplementary Tables 2   and 3). Hamburg, Germany). The iPLEX reaction was performed according to the standard protocol recommended by the system supplier (21). The homogeneous MassEXTEND (hME) and iPLEX process relies on a small volume PCR amplifying the target regions including the SNP position in a multiplex fashion. The basic principle of hME and iPLEX reaction is identical. Both methods use a third, so called MassEXTEND primer, which anneals directly adjacent to the SNP position. In an enzymatic primer extension reaction, this primer will be elongated.

DNA samples with known PHEX-Mutations
During that process the allele specific analytical products are generated. The products differ by mass according to the incorporated bases. Primers were designed:

Results
The technical data of the established NGS panel for the molecular confirmation of the diagnosis of XLH are shown in Table 2. Supplementary Sanger sequencing has been established to completely cover sequences with homopolymers in particular. All in all, close to 100% coverage of all amplicons of the 11 genes in the NGS panel was achieved.
The method proved to be sufficiently robust to process 20 patients in parallel in one reaction approach without loss of quality.
In DNA samples from 49 patients (patients No. 1 to 50, except for patient 17) with confirmed XLH, the NGS assay correctly re-identified the PHEX mutations that were already known and classified as pathogenic. Using the NGS method, 15 nonsense, 12 missense, 4 splice, 7 deletions and 4 insertions were found (Supplementary Table 4). For validation, the sequence In the last two samples no variation was detected in the PHEX sequence, nor was MLPA suspicious. Therefore, the sequence data of the other genes of the NGS panel were analyzed. The already known nonsense mutation c.31delT (p.Trp11Glyfs*9) in the DMP-1 gene was discovered in a homozygous fashion in the first sample( P2; Table 3). In the second sample, two heterozygous missense mutations, c.475C>A (p.Gln159Lys) and c.205A>T  Table 3).

Discussion
Recently, first international clinical practice recommendations for the diagnosis and management of XLH have been published, recommending that XLH should be diagnosed not only on the basis of clinical signs of rickets and/or osteomalacia in association with renal phosphate wasting, but also on the basis of molecular analysis, confirming the clinical diagnosis on a genetic level (2). Improvements in gene sequencing technologies in combination with rapidly declining costs have led to the development of a large amount of targeted NGS panels. These panels allow investigating multiple known disease-causing genes in one assay. Therefore, in this study, we developed an NGS panel for the diagnosis of XLH and related disorders. We validated the panel for PHEX using 50 DNA-samples with a known PHEX mutation. Since the NGS tool gave a 100% agreement in 49 patient samples, the coverage and the sensitivity must be rated very high, proving that we was very low (reference base C about 91,5%, but mutated base T in only 8,5% of all reads), but the mosaic mutation has been confirmed by iPLEX and MALDI-TOF MS (Supplementary Figure 2) ). Since mosaic mutations are difficult to detect by Sanger sequencing, their description in PHEX is rare in the literature (22)(23)(24). Therefore, the identification of this mosaic mutation demonstrated the high sensitivity of the developed NGS panel.
However, the advantage of using the panel is not only the molecular diagnosis of XLH, but also of related disorders of renal phosphate wasting in one investigation. For this goal we included ten other candidate genes in the panel. These genes were decided to be included into our tool, since all these genes are encoding proteins, which are involved in renal phosphate reabsorption and most of them are known to cause a type of renal phosphate wasting disorder in case of a mutation in one of these genes (for details see Table 1). A validation of variations in these genes except for PHEX was not possible because of the rarity of these conditions and the unavailability of samples with known mutations.
In one patient we proved a known homozygous DMP1 mutation, which has been detected before by Sanger sequencing. DMP1 encodes for dentin matrix protein type 1 and is produced by osteoblasts and osteocytes, regulating cell attachment and cell differentiation (25). Homozygous DMP1-mutations are the cause for ARHR type 1 (ARHR1) (9, 10, 26); a rare autosomal recessive disorder with biochemical and skeletal signs similar to those observed in XLH. Although there are similarities in the pathophysiology between XLH and AHRH1 especially with elevated or inappropriate normal FGF23-levels, the patient can be treated only by the conventional therapy since Burosumab is exclusively licenced for XLH.
In the third sample without proven PHEX mutation we detected several molecular genetic changes, which could be responsible for the phenotype of the patient. While the DMP-1 variants may be common variants, the most probable reason for the phenotype are the compound heterozygous mutations mentioned above in the ENPP1 gene, affecting different positions of the same amino acid codon. Both variants are very rare (allele frequency:<0.00003 at gnomAD) (27) and are considered as probably damaging by Polyphen 2 (28) and deleterious by SIFT (Sorting Intolerant From Tolerant) (29).
Homozygous or compound heterozygous ENPP1 mutations have been previously described to lead not only to generalized arterial calcification of infancy (GACI1) (30)(31)(32), but also to autosomal-recessive hypophosphatemic rickets type 2 (ARHR2) in rare cases (8). Although in ARHR2 patients inappropriate high FGF23 levels are seen (8) (33). And lastly, all molecular methods applied will be profitable only in light of very informative patients with respect to clinical and biochemical phenotyping.
In conclusion, the use of NGS technology has major advantages for exact diagnosis of the different forms of HR. In contrast to commercially available NGS panels, the panel was validated with known mutated samples and therefore the application of the panel developed in this study seems to be a sensitive and specific tool which can not only detect mutations in PHEX, but also in other genes associated with HR. This differentiation is favourable for the patients as it readily leads to very specific treatment options.

Declaration of interests
OH and ST received honoraria from Kyowa Kirin.

Funding
This work was in supported by a grant obtained from Kyowa Kirin.

Legends for supplementary tables and figures:
Supplementary Hereditary hypophosphataemic rickets with hypercalciuria (HHRH)