Hidden PHEX Variants Beyond Exomes

Evidence

X-linked hypophosphatemia, or XLH, is an inherited disorder in which PHEX variants disrupt phosphate regulation and impair bone mineralization. Three primary studies examined cases with clinical and biochemical features consistent with XLH but no straightforward result from earlier genetic testing. Their findings do not estimate how often hidden variants occur. Instead, they show how different classes of PHEX variation can evade exon-focused analysis or routine variant calling.

The newest study, first published July 9, described one female patient whose short-read whole-exome sequencing had been unrevealing. Short-read whole-genome sequencing identified a previously unreported deep-intronic PHEX variant, c.2070+601C>T, and testing confirmed that it was de novo. RNA from peripheral blood was amplified by reverse-transcription PCR and examined with Nanopore long-read sequencing. The investigators reported two abnormal transcripts containing pseudoexons; each introduced a premature stop codon. This links the DNA change to altered splicing in the sampled tissue rather than relying only on a computational prediction. Because ingestion provided only the abstract, the precise read counts, transcript proportions, and comparison data are unavailable here.

An independent Finnish-family study reached a similar mechanism through a more extensive full-text workflow. A 29-year-old son and his 56-year-old mother had hypophosphatemia since childhood, while PHEX sequencing, hypophosphatemia gene panels, and whole-exome sequencing had not found a causal variant. Thirty-fold whole-genome sequencing identified c.2147+1197A>G deep within PHEX intron 21 in both affected relatives but not in the tested unaffected relatives. Sanger sequencing confirmed the genomic result.

Splice-prediction tools suggested that the Finnish variant created new splice sites. Whole-transcriptome data from patient-derived fibroblasts then showed reads at the predicted pseudoexon coordinates, although PHEX expression was low. Targeted reverse-transcription PCR produced a 389-base-pair fragment in affected samples versus a 306-base-pair fragment in controls. Sanger sequencing of the longer product confirmed an 84-nucleotide pseudoexon between exons 21 and 22. The inserted sequence contained premature stop codons. The male sample, which was hemizygous for the X-linked variant, mainly showed the longer product but also a faint normal band; the heterozygous mother showed both products.

The third study found a different kind of hidden PHEX change in a two-year-old Chinese boy. Routine analysis of whole-genome variant files found no relevant single-nucleotide variant or copy-number variant, but inspection of the alignment data showed a breakpoint in exon 22 in both the child and his mother. Long-range PCR targeted the region. Instead of the expected 614-base-pair product, sequencing produced a 1,013-base-pair fragment, of which 483 bases were an insertion. The insertion comprised a 62-base poly-T segment and 421 bases derived from LINE-1, a mobile genetic element. The researchers classified the insertion as pathogenic under their stated ACMG/AMP criteria. They did not perform RNA sequencing, so the study did not directly measure its effect on the PHEX transcript.

Analysis — Orthogonal Tests Expose Different Blind Spots

The cross-study pattern is an analysis, not a validated diagnostic pathway. The two deep-intronic studies independently connect genome-level candidates to abnormal RNA splicing, but they do so with different variants, tissues, and transcript assays. Their convergence is mechanistic: a variant outside the captured exons can create a pseudoexon and premature stop signal, and RNA provides evidence that the predicted splice event actually occurs. The LINE-1 study exposes a separate failure mode. Whole-genome reads contained a breakpoint, yet routine single-nucleotide and copy-number calls did not describe the 483-base insertion; targeted long-range PCR and Sanger sequencing resolved it. Together, the studies suggest that “negative exome” and “negative routine genome calls” are not equivalent endpoints. The informative next method depends on the suspected blind spot: deeper review of genome alignments for structural insertions, or transcript testing for splice consequences. That is a research inference from three case-based reports, not evidence that one fixed workflow will generalize across XLH or other rare disorders.

Limitations

These reports cover one patient, one two-person Finnish family, and one Chinese child with maternal carriage. They cannot establish prevalence, sensitivity, diagnostic yield, or comparative performance against a prespecified testing standard. The July study was available only as an abstract, limiting scrutiny of its sequencing depth, filtering, controls, and quantitative transcript evidence.

The Finnish study provides segregation and RNA evidence, but PHEX expression was low in fibroblasts, transcriptomic read support was limited, and the authors did not directly measure the resulting protein. The male sample retained a faint normal transcript, so the relationship between transcript proportions and phenotype remains unresolved. In the LINE-1 report, targeted DNA sequencing defined the insertion, but no RNA experiment tested transcript disruption, and the grandmother was unavailable for segregation analysis. The studies also used different tissues and assays, preventing direct comparison. Replication in additional unrelated families, standardized reanalysis of negative cases, quantitative RNA measurements, and functional protein studies would be needed to determine which combinations of genome and transcript testing are most reliable.