I've gathered the main sources. Now I'm writing the report.

Glycosylphosphatidylinositol Biosynthesis Defect 15 (GPIBD15; GPAA1 deficiency): research report

Source basis. Most claims below come from ten references cached in this worktree (references_cache/PMID_*.md), with quotes copied verbatim from those files. Identifiers and the structural literature were added from web searches. Where a claim rests only on a search result, an unread abstract, or my own inference, it is marked [unverified] or [inference].

1. Disease information

Overview. GPIBD15 is an autosomal recessive inherited GPI deficiency (IGD) caused by biallelic variants in GPAA1. GPAA1 is a non-catalytic subunit of the ER-resident GPI transamidase (GPI-T). GPI-T attaches the GPI anchor to the C-terminus of precursor proteins. The core picture is: - global developmental delay or intellectual disability - hypotonia - early-onset epilepsy, often fever-sensitive - cerebellar atrophy, frequently acquired or progressive - osteopenia in some patients

"we report bi-allelic mutations in GPAA1 in ten individuals from five families… Most individuals presented with global developmental delay, hypotonia, early-onset seizures, cerebellar atrophy, and osteopenia." — PMID:29100095 (Nguyen et al., Am J Hum Genet 2017)

"Glycosylphosphatidylinositol biosynthesis defect 15 is a rare autosomal recessive disorder due to biallelic loss of function of GPAA1." — PMID:37510348 (Fontana et al., Genes 2023)

Identifiers

Synonyms: GPIBD15; GPAA1 deficiency; GPAA1-related GPI deficiency; GPAA1-CDG; "GPI anchoring deficiency due to biallelic variants in GPAA1" (PMID:34703884).

Data source type. Everything is aggregated case-level literature: about 30 patients in case series and case reports. There are no registries or EHR cohorts.

2. Etiology

• Cause. Biallelic (homozygous or compound heterozygous) germline hypomorphic variants in GPAA1. These reduce GPI-T activity, which lowers the amount of GPI-anchored protein (GPI-AP) on the cell surface.
• Genetic risk factors.
• Consanguinity and genetic isolates explain the homozygous cases (PMID:37510348; PMID:39152716 in consanguineous Pakistani families).
• No susceptibility loci or modifier genes are known.
• Environmental risk factors. None cause the disease. Fever is a seizure trigger: febrile sensitivity was present in 66.7% of the pooled cohort (PMID:42001530), and in 4 of 7 patients in PMID:34703884.
• Protective factors and gene–environment interactions. None reported.
• Disputed heterozygous phenotype. One study (PMID:32533362) proposed that a heterozygous GPAA1 c.968A>G variant causes familial vascular anomalies. Castle et al. (PMID:34703884) found no vascular anomalies in biallelic patients or in carrier parents. Fontana et al. state the association "has not been confirmed by other authors." This should be treated as unconfirmed and kept out of the GPIBD15 entity.

3. Phenotypes

Pooled frequencies are from Ouyang et al. 2026 (PMID:42001530), covering 24 patients: 5 new plus 19 published.

"The clinical manifestations included global developmental delay (100%), hypotonia (95.8%), and epilepsy (83.3%). 60% of patients experienced seizure onset before the age of one. The main seizure types were generalized tonic-clonic seizures (GTCS) (82.4%) and myoclonic seizures (70.6%). Febrile sensitivity was presented in 66.7% of patients. 60.9% of patients had cerebellar atrophy."

Course and impact on daily life. - Most patients are non-ambulant or walk late. One patient walked at 6 years. - Many are nonverbal or have severely limited speech. - The oldest reported patient (38 years) cannot stand, read or write, but eats and undresses independently (PMID:34703884). - No formal quality-of-life (QoL) instruments have been used.

4. Genetic and molecular information

• Gene. GPAA1 (glycosylphosphatidylinositol anchor attachment 1), RefSeq NM_003801.3, 8q24.3.
• Variant spectrum. Ouyang 2026 counts 28 variants: 19 missense, 5 frameshift, 3 intronic splicing and 1 nonsense.

• Genotype. No patient carries two null alleles:

  "Similar to the previous cohort, no homozygous loss-of-function variants were identified… suggestive of the importance of having residual functional gene product" (PMID:34703884)

The functional class is hypomorphic, i.e. partial loss of function. Complete GPAA1 knockout is presumed embryonic lethal [inference, by analogy with null GPI-pathway genes]. - Variant origin. All germline. Missense variants cluster in the luminal domain (residues 66–348) (PMID:34703884, 37510348). - Variant interpretation. Flow cytometry is recommended to support variant classification (PMID:34703884). - Gene curation. Genomics England PanelApp rates GPAA1 Green for early-onset or syndromic epilepsy. ClinGen gene-disease validity: not found in the local cache (no CGGV file). - Not reported: modifier genes, epigenetic findings, chromosomal abnormalities. A SNP microarray was normal in the Fontana case.

5. Environmental information

There are no environmental, lifestyle or infectious causes. Fever, including infection-driven fever, precipitates seizures. One patient had a recurrent sterile periodic-fever syndrome (PMID:38112147).

6. Mechanism and pathophysiology

Ordered causal chain

1. Biallelic hypomorphic GPAA1 variants (missense in the luminal domain, frameshift, splice) lead to reduced GPAA1 mRNA or protein, or a destabilized subunit. Splice c.1164+5C>T lowers mRNA (demonstrated, PMID:29100095). Missense variants are predicted to disrupt the luminal domain (PMID:37510348, computational).
2. A deficient GPAA1 subunit impairs assembly and activity of the heteropentameric GPI transamidase (PIGK–GPAA1–PIGT–PIGS–PIGU) in the ER membrane.

3. "GPI-TA consists of five subunits: PIGK, GPAA1, PIGT, PIGS, and PIGU, and the absence of any subunit leads to the loss of activity." (PMID:34576938, in vitro)

4. Cryo-EM structures (Zhang 2022, PMID:35165458; Xu 2022, Nat Commun, PMID:35551457 [PMID from search, not cached]) place catalysis in PIGK (C206-H164-N58). GPAA1 is structural, not catalytic: "they suggest that unlike previously proposed, GPAA1 is not a catalytic subunit… most of the disease mutations occur near the active site or the subunit interfaces" (PMID:38963495, review). GPAA1 Asp250 is required for activity (PMID:34576938).
5. Reduced GPI-T activity results in failed replacement of the C-terminal GPI-attachment signal with a GPI anchor on proproteins. This reduces the delivery of GPI-APs to the cell surface:

   "several GPI-anchored proteins had decreased cell-surface abundance in leukocytes (FLAER, CD16, and CD59) or fibroblasts (CD73 and CD109). Transduction of fibroblasts with a lentivirus encoding the wild-type protein partially rescued the deficiency" (PMID:29100095; human ex vivo / in vitro)

Because the defect sits at the attachment step, the GPI precursor and the unanchored protein are not released as soluble protein, which is why ALP is not raised. Instead, ALP is normal or low, unlike defects of the post-attachment / remodeling steps that cause hyperphosphatasia (PMID:34703884). 4. Branch A: nervous system. Loss of neural GPI-APs, which act in axon growth, synapse formation and synaptic plasticity (PMID:34703884), leads to: - impaired neurodevelopment, giving developmental delay and intellectual disability - neuronal hyperexcitability, giving epilepsy with fever sensitivity. The specific GPI-APs responsible are not demonstrated. - cerebellar micro-organization and Purkinje-cell dendritic defects, giving progressive cerebellar atrophy and ataxia. Fontana cites a transamidase model showing "it impairs cerebellum micro-organization and Purkinje cell dendritic arborization." That is model-organism evidence from a related subunit; GPAA1 itself has no mouse model. A Purkinje-specific Pigk knockout mouse shows cerebellar atrophy, ataxia and Purkinje-cell loss with apoptosis and neuroinflammation [web search, PMID not retrieved]. - Atrophy is often acquired: the cerebellum was normal at 7 months and atrophic at 2 years (PMID:34703884), and normal at 15 months and atrophic at 33 months (PMID:37510348). 5. Branch B: bone. Reduced osteoblast-surface tissue-nonspecific ALP is hypothesized to reduce bone formation, giving osteopenia (PMID:34703884; stated as a hypothesis). Immobility and antiseizure drugs may contribute. 6. Branch C: immune system [speculative]. Reduced complement regulators (CD59, CD55) and loss of CD16 on granulocytes could contribute to the inflammatory or periodic-fever phenotype in one patient (PMID:38112147). This parallels autoinflammation in PIGT-PNH (PMID:34576938 background) but is not demonstrated for GPAA1.

Ontology suggestions (all unverified; look up before binding)

No transcriptomic, proteomic, metabolomic, single-cell or CRISPR-screen data specific to this disease exist. HEK293 GPAA1-KO cells exist as a complementation system (PMID:34576938).

7. Anatomical structures

8. Temporal development

• Onset. Hypotonia from birth to 4 months. Delay is evident by about 3 months. Seizures start at 3 months to 3 years, before 1 year in 60%.
• Course.
• Chronic and lifelong.
• Cerebellar atrophy can progress in early childhood and then remain stable into adulthood: stable at 21, 25 and 29 years in the oldest patient (PMID:34703884).
• Developmental skills are acquired slowly, not lost.
• Formal staging and remission. None. Seizures may be controlled or drug-resistant.
• Critical window [inference]. Neuroimaging should be repeated after infancy, because a normal early MRI does not exclude later atrophy.

9. Inheritance and population

• Inheritance. Autosomal recessive (HP:0000007).
• Penetrance and expressivity. Penetrance appears complete for biallelic pathogenic variants. Expressivity is markedly variable: one 22-month-old with p.Leu350Arg had only mild delay and no seizures (PMID:34703884).
• Carriers. Heterozygous parents are unaffected and have normal flow cytometry (PMID:34703884).
• Prevalence. Unknown. Count of reported patients over time:

Plus further cases in Pakistani (PMID:39152716) and Chinese (PMID:38112147) families. Suggested prevalence_class: CASES_IN_LITERATURE / ULTRA_RARE. - Populations. Reported from European, North American, Egyptian, Finnish, Israeli, Kurdish-Turkish, Italian (genetic isolate in Benevento), Pakistani and Chinese families. - Founder effects and anticipation. No founder variant is established. p.Ala316Val recurred in two unrelated families. No anticipation (not a repeat disorder). Sex ratio is about 1:1.

10. Diagnostics

• Genetic testing. Exome or genome sequencing is the main route, used in essentially all cases. GPAA1 is also on CDG and epilepsy panels (PanelApp Green).
• Functional confirmation.
• Flow cytometry of granulocytes for FLAER, CD16, CD59 and CD55. CD16 is the most sensitive marker, with reductions of 12–47% (PMID:34703884).
• Fibroblast CD73/CD109 and lentiviral rescue (PMID:29100095).
• GPAA1 has no characteristic flow profile, unlike PIGT and PIGU.
• Biochemistry.
• Serum ALP is normal or low, never high. This separates the transamidase disorders from the HPMRS (hyperphosphatasia) group.
• Inflammatory markers (CRP, ESR) rise only during febrile episodes in the periodic-fever case.
• Suggested LOINC for ALP: 6768-6 [unverified; seed with just loinc-seed-cache].
• Imaging. Brain MRI shows cerebellar (vermian) atrophy, delayed myelination and a thin corpus callosum. DXA can detect osteopenia.
• EEG. No consistent pattern: findings include normal, multifocal spike-and-wave, polyspike and slow discharges.
• Differential diagnosis.
• Other transamidase IGDs: PIGT (intractable seizures, nephrocalcinosis, dental and skeletal findings); PIGS (coarse face, gingival hypertrophy, pontine hypoplasia); PIGK (microcephaly); PIGU (thin corpus callosum).
• Other IGDs: PIGA, PIGN, PIGG, PGAP1.
• Hyperphosphatasia–intellectual disability syndromes (HPMRS: PIGV, PIGO, PGAP2/3), which have high ALP.
• Other CDGs, e.g. PMM2-CDG with cerebellar atrophy.
• Pontocerebellar hypoplasias and early-onset spinocerebellar ataxias. GPIBD15 spares the brainstem (PMID:37510348).
• Screening. No newborn screening. Carrier and prenatal or preimplantation testing are possible once familial variants are known.

11. Outcome and prognosis

• Survival. No survival data. Patients have been reported up to 38 years, so survival into adulthood is documented.
• Morbidity.
• Severe neurodevelopmental disability: most are non-ambulant and many nonverbal.
• Drug-resistant epilepsy in a subset.
• Fracture risk from osteopenia is inferred, not documented for GPAA1, though fractures occur in PIGT deficiency (PMID:34703884).
• Prognostic factors. None validated. Milder cases may carry higher residual activity, but no genotype–phenotype correlation is established.

12. Treatment

No disease-modifying therapy exists. Management is symptomatic. Suggested NCIT terms below are unverified; look them up before binding.

• Experimental therapy. No trials. Lentiviral GPAA1 gene transfer rescued GPI-APs in patient fibroblasts in vitro (PMID:29100095). That is a proof of concept only.
• Butyrate. Effective only in PIGM promoter disease, so not applicable here.
• Pharmacogenomics. None known.

13. Prevention

• Primary: genetic counseling, which carries a 25% recurrence risk; carrier testing; prenatal or preimplantation diagnosis.
• Tertiary:
• Fever control or avoidance to reduce fever-triggered seizures [inference from the febrile-sensitivity data].
• Bone-density surveillance.
• Serial MRI.
• Not applicable: vaccines, public-health measures.

14. Other species

No naturally occurring GPAA1 disease is known in animals. A query of OMIA was not performed. GPAA1 is conserved across eukaryotes; the yeast ortholog is GAA1 (S. cerevisiae). Zoonotic potential: not applicable.

15. Model organisms

Curation notes for the dismech entry

1. Disease-specific evidence anchors are PMIDs 29100095, 34703884, 37510348, 42001530 and 38112147. Their abstracts are cached and the quotes above are verbatim.
2. Mechanism and structure are covered by 34576938 (full text), 35165458 and 38963495. Xu 2022 (PMID:35551457) still needs just fetch-reference.
3. Keep the vascular-anomaly paper (32533362) out of the GPIBD15 phenotypes. It concerns a heterozygous, disputed claim. At most, record it as a REFUTE/discussion item quoting Castle's rebuttal.
4. Pathophysiology node to conform to. The current node should conform to an existing GPI-anchor-deficiency module if there is one; check with just list-modules gpi.
5. Do not use PMID:20301507. The retired GeneReviews N-linked CDG chapter is not GPIBD15-specific.
6. GeneReviews baseline. Run just check-genereviews to confirm that no GPIBD15 chapter exists.

Sources

Cached abstracts and full texts (all in references_cache/): - PMID:29100095 (Nguyen 2017) - PMID:34703884 (Castle 2021) - PMID:37510348 (Fontana 2023) - PMID:42001530 (Ouyang 2026) - PMID:38112147 (Chen 2023) - PMID:39152716 (Bibi 2024) - PMID:32533362 (Li 2020) - PMID:34576938 (Liu 2021) - PMID:35165458 (Zhang 2022) - PMID:38963495 (Li 2024)

Web: - MedGen C4540520 - LOVD GPIBD15 - PanelApp GPAA1 (Genomics England) - PanelApp Australia CDG - Xu et al. 2022 Nat Commun (PMC) - RCSB 7WLD - Zhang et al. 2022 NSMB - Liu et al. 2021 Molecules - PMC7206848 (GPI mouse model review) - bioRxiv Pigk Purkinje model

Reference Validation

Checked with linkml-reference-validator 0.3.0.

Quotes not found in the cited source

Searched the abstract, any retrieved full text, and the title. A quote drawn from a part of the paper that was not retrieved will appear here too, so check before treating one as invented:

1 of these was searched against an abstract alone, with no full text retrieved - marked abstract only below. Where full text can be fetched, re-running with it will settle them; where the source publishes only a summary to PubMed, as GeneReviews chapters do, it will not, and the quote has to be checked by hand against the chapter itself.

• PMID:34703884: "GPI anchoring deficiency due to biallelic variants in GPAA1"
• closest text in source: "There is a case report of an individual with GPI deficiency due to variants in PIGM whose severe seizures were treated successfully with the histone deacetylase inhibitor sodium butyrate.30,31 However, this is likely not a feasible treatment for all GPI-related disorders, as the patient's specific pathogenic variant was in the PIGM promotor and led to decreased acetylation and therefore decreased gene expression"
• PMID:34703884: "Similar to the previous cohort, no homozygous loss-of-function variants were identified… suggestive of the importance of having residual functional gene product"
• closest text in source: "Similar to the previous cohort, no homozygous loss-of-function variants were identified in this series of patients"
• PMID:38963495 (abstract only): "they suggest that unlike previously proposed, GPAA1 is not a catalytic subunit… most of the disease mutations occur near the active site or the subunit interfaces"
• closest text in source: "Consistent with the assembly mechanisms and the active site architecture, most of the disease mutations occur near the active site or the subunit interfaces"

References that may not be about this subject

These identifiers resolve, so they are not fabrications, but the records they resolve to share almost none of this report's vocabulary. That is a clue and not a verdict - a paper can be relevant in ways its title and abstract do not spell out - so read them before deciding:

• PMID:35551457 (2 mentions) - Molecular insights into biogenesis of glycosylphosphatidylinositol anchor proteins.
• shared terms: gpi-ap, anchor
• PMID:20301507 (1 mention) - Congenital Disorders of N-Linked Glycosylation and Multiple Pathway Overview – RETIRED CHAPTER, FOR HISTORICAL REFERENCE ONLY.
• shared terms: anchor

Weighed against this report's own most characteristic terms: gpaa1, patient, seizure, cerebellar, variant, gpibd15, phenotype, atrophy, fever, epilepsy, gpi-ap, fibroblast, fontana, unverified, disease, year, delay, developmental, normal, anchor.

Term Validation

Checked with linkml-term-validator 0.4.5, through the ols: adapter.

Terms the report names something else

These identifiers resolve, so nothing about them looks wrong, and the ontology calls them something unrelated to what the report calls them. That usually means the identifier is not the one the sentence needs:

• HP:0032794 (1 mention) - the report calls it "unverified"; HP calls it Myoclonic seizure
• HP:0033052 (1 mention) - the report calls it "unverified"; HP calls it Non-epileptic seizure

Terms whose name is worth a second look

The report's name for these is recognisably related to the term's own name without being one of them. A loose paraphrase reads the same way as a citation of the wrong sibling term - and so does a related synonym, which the ontology records precisely because it names something adjacent rather than the same thing - so these are listed rather than judged:

• HP:0000384 (1 mention) - the report calls it "Preauricular and cheek skin tags"; HP calls it Preauricular skin tag, and lists "Preauricular skin tags" among its other names
• CHEBI:50858 (1 mention) - the report calls it "Pharmacotherapy plus a corticosteroid agent"; CHEBI calls it corticosteroid, and lists "corticosteroides" among its other names

Prefixes with no resolver

Terms carrying these prefixes were not checked either way, because no configured ontology covers them. An unrecognised prefix may name an ontology this run could not reach as easily as one that does not exist, so nothing here is evidence of fabrication: ORPHA.