Wolfram Syndrome — Comprehensive Research Report

Prepared 2026-08-15 for dismech KB entry kb/disorders/Wolfram_Syndrome.yaml

Curation note on evidence. All PMIDs below were resolved against live PubMed (NCBI E-utilities) or ClinicalTrials.gov during preparation. Ontology CURIEs marked ✅ were verified against OLS/HPO APIs during this session; those marked ⚠️ are best-guess and must be checked with OAK before they go in a YAML file. I have deliberately not written any snippet:-ready quotes — fetch each reference and quote the abstract's own sentence.

1. Disease Information

The one-paragraph version

Wolfram syndrome is what happens when a single housekeeping protein in the endoplasmic reticulum — that folding-and-quality-control workshop inside every cell — stops showing up for work. The cells that suffer most are the ones with the biggest export orders: pancreatic beta cells cranking out insulin, and long-projection neurons. So you get childhood diabetes, then the optic nerves fade, then hearing, then the water-balance hormone system, then the brainstem. It's classically remembered by the acronym DIDMOAD: Diabetes Insipidus, Diabetes Mellitus, Optic Atrophy, Deafness. First described by Wolfram and Wagener in 1938 in four siblings with juvenile diabetes plus optic atrophy.

Think of it less as four separate diseases stapled together and more as one metabolic organelle failing, with the symptom list simply reading out which tissues were most dependent on it.

Identifiers (verified this session unless noted)

Two further MONDO terms in the same neighborhood, both verified: MONDO:0010800 "Wolfram syndrome, mitochondrial form" and MONDO:0100072 "neonatal diabetes, congenital sensorineural hearing loss and congenital cataracts" (the de novo dominant WFS1 presentation).

Genes: WFS1 — OMIM *606201 ✅ (appears in OMIM entry set), chromosome 4p16.1. CISD2 — OMIM *611507 ⚠️. HGNC IDs (hgnc:12762 for WFS1, hgnc:24212 for CISD2) are ⚠️ — verify with OAK, and remember this repo uses the lowercase prefix.

ICD: ICD-10 and ICD-11 codes were not retrievable from a reliable source in this session (Orphanet blocked). Flag for lookup rather than guessing — Wolfram is usually coded under the diabetes chapter with a specified-complication modifier, which is easy to get subtly wrong.

Synonyms

DIDMOAD; DIDMOAD syndrome; Wolfram-DIDMOAD syndrome; diabetes insipidus–diabetes mellitus–optic atrophy–deafness syndrome; WFS1 spectrum disorder (WFS1-SD) — this last one is the current GeneReviews nomenclature and a genuinely important recent shift (see §4).

Where the data comes from

A mix, and the mix matters for how much you trust each number: - Aggregated / curated: OMIM, Orphanet, GeneReviews (WFS1 Spectrum Disorder, NBK4144), HPO annotations. - Patient-level registries and prospective cohorts: the Washington University Wolfram Syndrome International Registry and Clinical Study (NCT02841553, still recruiting), the WashU Research Clinic longitudinal cohort (annual visits, under-30 at enrollment), the UK national specialist clinic in Birmingham, EURO-WABB, and the Italian SID/SIEDP consensus cohort (PMID:39527371). - The registry/clinic data are the reason frequency and onset-age numbers have shifted so much since the 1990s — the old literature was built on published case reports, which are systematically biased toward the severe end.

2. Etiology

Causal factors

Wolfram syndrome 1 is monogenic and autosomal recessive: biallelic loss-of-function variants in WFS1, accounting for ~90%+ of classic cases. Wolfram syndrome 2 is biallelic CISD2 (PMID:17846994 ✅ — the original ERIS/CISD2 report). No environmental cause. No infectious agent. There is no known way to acquire this.

Landmark gene-identification references: - PMID:9771706 ✅ — Inoue et al., Nature Genetics 1998, the WFS1/wolframin cloning paper. - PMID:7490992 ✅ — Barrett, Bundey & Macleod, Lancet 1995 — the 45-patient UK series that fixed prevalence, carrier frequency, and the natural-history skeleton still cited today. - PMID:17846994 ✅ — CISD2/ERIS in WFS2.

Risk factors

Genetic (the whole ballgame): - Biallelic pathogenic WFS1 variants — causative, not merely predisposing. - Consanguinity substantially raises risk; much of the reported case load comes from consanguineous families in the Middle East, North Africa, South Asia, and Sicily. - Heterozygous carriers are not silent bystanders. Carriers have been reported at markedly elevated risk of psychiatric hospitalization (Swift et al.; see also PMID:12707947 on WFS1 and suicidal/impulsive behavior), and common non-coding WFS1 variation (e.g. rs10010131) is an established type 2 diabetes susceptibility signal in GWAS. Curate the carrier phenotype with relationship_type: SUSCEPTIBILITY, not CAUSATIVE. - WFS1 heterozygosity also produces dominant disease in its own right (§4).

Environmental: none established as causal. Two things plausibly modulate severity but are not proven modifiers — chronic hyperglycemia (adds ER protein-folding load on top of an already-strained system) and ototoxic/nephrotoxic drug exposure in a population already losing hearing and bladder function. Treat both as notes:, not as evidence-bearing environmental: entries, unless you find a paper that actually measured it.

Protective factors

No genetic protective variant, modifier allele, or dietary/lifestyle factor is established. Reported milder courses track with variant type, not with anything the patient did — see genotype–phenotype below. Do not curate "good glycemic control is protective" as a protective factor; it's plausible and unproven.

Gene–environment interaction

Essentially unstudied in humans. The mechanistic hypothesis worth recording as a KNOWLEDGE_GAP discussion: because wolframin's job is buffering ER stress, any environmental stressor that raises the unfolded-protein load (glucotoxicity, inflammation, hypoxia) should hit harder in a WFS1-null cell. Preclinical support exists — cytokine-induced ER stress is amplified in WFS1-deficient beta cells (PMID:33693650) — but the human GxE data are absent.

3. Phenotypes

The HPO annotation set for OMIM:222300 was pulled live ✅. Note the HPO frequencies are small-denominator fractions from legacy OMIM sources; the GeneReviews frequencies (from specialist-clinic cohorts) are the better number for curation. I've given both.

Cardinal features

Neurologic / brainstem–cerebellar

Ataxia HP:0001251 ✅ (3/9), dysarthria HP:0001260 ✅, dysphagia HP:0002015 ✅, nystagmus HP:0000639 ✅ (2/10), tremor HP:0001337 ✅ (10/20), seizure HP:0001250 ✅ (1/20, uncommon), cerebral atrophy HP:0002059 ✅, stroke-like episode HP:0002401 ✅ (rare). Central sleep apnea and impaired central respiratory drive — the terminal feature — map to HP:0002870 ⚠️ (Central apnea; verify). Anosmia HP:0000458 ⚠️ is reported and often overlooked; it's a nice early-marker candidate.

The brainstem story deserves emphasis because it's the one that kills. Volume loss centers on the ventral pons and cerebellum, and — critically — it is already measurable at the very earliest clinical presentation, within about half a year of diabetes onset (PMID:22792385 ✅, Hershey et al., PLoS One 2012). Longitudinal morphometry over ~3.6 years in 29 patients vs 52 controls shows both failed developmental growth and frank degeneration: controls gained white-matter volume where the Wolfram group was flat (optic radiations) or shrinking (brainstem, ventral pons) (PMID:30979932 ✅, Lugar et al., Sci Rep 2019). So this is not purely neurodegenerative — there's a neurodevelopmental limb too, which changes when a disease-modifying therapy would have to start.

Psychiatric

Substantial and under-curated. Roughly 60% of WS1 patients have a history of severe psychiatric disorder — depression, psychosis, disorientation, memory deficits, irritability, impulsive aggression — with ~25% classed "very severe" and a similar figure for suicide-related behavior; first attempt/hospitalization typically between ages 15 and 32 (reviewed in PMID:39202345 ✅, Genes 2024, "Wolfram Syndrome 1: A Neuropsychiatric Perspective"; original signal PMID:12707947 ✅). Suggested terms: depressivity HP:0000716 ⚠️, psychosis HP:0000709 ⚠️, atypical behavior HP:0000708 ✅.

A 2025 sigma-1-receptor–based perspective piece (PMID:40955171 ✅) argues the WFS1/sigma-1/ER-stress axis may be mechanistically informative for depression and suicidality generally — interesting for a mechanistic_hypotheses block, but it's a hypothesis paper, tag accordingly.

Ophthalmic beyond optic atrophy

Cataract HP:0000518 ✅ (5/10), pigmentary retinopathy HP:0000580 ✅, ptosis HP:0000508 ✅, impaired color vision (dyschromatopsia) ⚠️. A 2026 retrospective specifically on visual outcomes and biomarker correlates: PMID:41870390 ✅ (J Neuroophthalmol).

Genitourinary / renal

Hydronephrosis HP:0000126 ✅ (8/10), hydroureter HP:0000072 ✅, recurrent UTI, and secondary renal failure. The bladder is a genuine mortality contributor — atonic bladder → obstructive uropathy → urosepsis/renal failure. There's a 2025 case report of neurogenic bladder presenting as acute kidney failure (PMC12141587).

Endocrine (beyond DM/DI)

Hypogonadism, especially primary hypogonadism in males; testicular atrophy HP:0000029 ✅; menstrual irregularity in females. Hypothyroidism HP:0000821 ✅ reported. Growth delay HP:0001510 ✅ (3/10).

Gastrointestinal / autonomic

Dysmotility, gastroparesis ⚠️, constipation ⚠️, and — notably — fecal incontinence, which has its own new mechanistic study (NCT07313085, not yet recruiting). Broad autonomic dysfunction is part of the picture.

Hematologic and cardiac (rarer)

Thrombocytopenia HP:0001873 ✅, sideroblastic anemia HP:0001924 ✅, megaloblastic anemia HP:0001889 ✅ — these are the features that overlap with thiamine-responsive megaloblastic anemia and mitochondrial disease, so they matter for differential diagnosis. Cardiomyopathy HP:0001638 ✅ is rare; a dedicated 2024 review of "cardiac wolframinopathies" including a myocarditis case: PMID:38542026 ✅.

Intellectual disability HP:0001249 ✅ appears at 2/20 in HPO annotations — worth flagging as not a core feature; cognition is typically preserved early, with later executive/memory decline tracking brain volume loss.

Quality of life

No Wolfram-specific validated QOL instrument is in wide use. The disease-specific severity instrument is the Wolfram Unified Rating Scale (WURS) (reliability/validity: PMID:23148655 ⚠️ — verify). Functional burden is dominated by the stacking of blindness + deafness + insulin dependence + incontinence in the same young adult, which is why the trial field has converged on visual acuity and C-peptide as endpoints rather than QOL scales. A methods paper on endpoint selection and analysis models for Wolfram neurodegeneration trials: PLoS One 2025, 10.1371/journal.pone.0321598.

4. Genetic / Molecular Information

The gene and its protein

WFS1, chromosome 4p16.1, 8 exons (exon 1 non-coding; exon 8 is large and carries the bulk of pathogenic variants). Encodes wolframin, an 890-amino-acid ER transmembrane glycoprotein. Topology as generally modeled: an N-terminal cytoplasmic region, 11 transmembrane segments, and a C-terminal ER-luminal domain, with an EF-hand-like element and a C-terminal OB-fold. It assembles into higher-order oligomers (tetramers/nonamers reported). No experimental cryo-EM structure of full-length human wolframin was found in this search — the structural work in the literature is homology modeling and molecular dynamics, which is a real limitation worth recording.

CISD2 (also ERIS, Miner1), 4q24 — a small iron-sulfur (2Fe-2S) protein of the mitochondria-associated ER membrane (MAM). Same neighborhood, different chair.

Variant landscape

• >200–300 distinct pathogenic variants reported; the majority loss-of-function — nonsense, frameshift, splice-site — with a substantial missense contingent clustered in the C-terminal luminal domain.
• Detection: sequence analysis catches >95%; deletion/duplication analysis adds ~3% (GeneReviews).
• Origin: germline; recessive alleles usually inherited, but de novo dominant variants are well documented (see below). Somatic mosaicism is not a feature.
• Functional consequence: predominantly loss of function. The dominant alleles behave as dominant-negative — a mutant subunit poisoning the wolframin oligomer — which is the standard explanation for why heterozygosity causes disease in some families and nothing in most.
• A striking recent mechanism paper: a WFS1 variant that disrupts an acceptor splice site and rewires alternative splicing to drive beta-cell apoptosis — PMID:39520565 ✅.

Genotype–phenotype correlation

GeneReviews states no established correlations, and that is the conservative position. But three lines of evidence say the picture is sharpening: - PMID:23429432 ✅ (de Heredia et al., Genet Med 2013) — the classic meta-analysis: two loss-of-function alleles associate with earlier onset and more complete phenotype than genotypes carrying a missense allele. - PMID:42524523 ✅ (Front Genet, July 2026) — a genotype-based severity scoring system correlating variant type with onset of cardinal symptoms. New and worth reading closely. - Zhang et al. 2025 (Pediatric Diabetes, PMC12331406) — variant topology (which part of the protein is hit) tracks with residual islet function and with urological symptom risk.

Record this honestly: GeneReviews says "none established"; the 2025–2026 literature says "emerging." That tension is a good mechanistic_hypotheses / discussions candidate rather than a flat assertion either way.

The dominant WFS1 allelic series — do not collapse this into Wolfram syndrome

This is the single easiest place to get a dismech entry wrong. Heterozygous WFS1 variants cause a distinct, dominantly inherited spectrum: - DFNA6/14/38 — autosomal dominant low-frequency (<2000 Hz) sensorineural hearing loss, congenital, slowly progressive, rarely severe-to-profound, speech perception preserved, often with tinnitus. >50 distinct heterozygous variants, mostly in the ER-luminal domain (PMID:37041640 ✅). Note the frequency inversion versus recessive Wolfram, where hearing loss is high-frequency. - Wolfram-like syndrome (MONDO:0013673, OMIM 614296) — adult-onset DM + progressive hearing loss + juvenile optic atrophy, dominantly transmitted. - p.Ala684Val is a documented mutational hotspot producing a severe hearing-loss phenotype (PMC11764508). - Neonatal diabetes + profound congenital deafness + congenital cataracts from de novo heterozygous variants (MONDO:0100072). - Fresh natural-history data on 15 patients with AD WFS1 variants (SNHL + optic atrophy): PMID:42001184 ✅ (Orphanet J Rare Dis, April 2026).

GeneReviews formalizes this as classic vs nonclassic WFS1-SD, and reports that nonclassic accounts for ~15% of molecularly confirmed cases in UK specialist clinics. If dismech is going to model this, "Wolfram syndrome" (recessive) and "Wolfram-like syndrome / DFNA6-14-38" (dominant) should be separate Disease entries under a Grouping, not one blended entry — the inheritance, the audiogram shape, and the molecular mechanism (null vs dominant-negative) all differ.

Modifier genes, epigenetics, chromosomal abnormalities

• Modifier genes: none established. Variation in residual beta-cell function between siblings with identical genotypes implies modifiers exist; nobody has found them.
• Epigenetics: no disease-specific methylation or chromatin signature reported. Nothing in ENCODE/Roadmap specific to Wolfram. Record as absent, not as unexamined-therefore-negative.
• Chromosomal abnormalities: not a feature. Large WFS1 deletions exist (~3% of alleles) but there is no recurrent CNV syndrome. CMA/karyotype/FISH have no diagnostic role here.
• Interesting adjacency: WFS1 has been implicated beyond Wolfram syndrome in Alzheimer disease and sleep disorders — review: PMID:39595565 ✅ (Biomolecules, Oct 2024). Good candidate for a comorbidities/mechanism cross-link, tagged as hypothesis-grade.

Allele frequency: carrier frequency in the UK derived from the Barrett series is ~1 in 354. gnomAD-based carrier estimates were not directly retrieved this session — pull them fresh if you want to curate a number.

5. Environmental Information

Short section, and it should be short.

• Environmental factors: none causal. Nothing in CTD/TOXNET links a specific exposure to Wolfram syndrome onset.
• Lifestyle factors: none causal. Glycemic management modifies diabetes complications exactly as it does in type 1 diabetes, which is downstream care, not etiology.
• Infectious agents: none. No NCBITaxon entity belongs in this entry.

If you populate an environmental: block at all, the honest content is a notes: line recording that ECTO was searched and no exposure term applies — which per the dismech environmental-term audit guidance is a legitimate UNBOUND outcome, not a gap.

6. Mechanism / Pathophysiology

This is the interesting part, and it has a clean causal spine you can build a pathograph on.

The chain, top to bottom

Step 1 — Loss of wolframin at the ER membrane. Biallelic LOF removes an ER transmembrane glycoprotein that does at least three jobs: negative regulation of the unfolded protein response, ER calcium handling, and maintenance of ER–mitochondrial contact sites.

Step 2 — Unrestrained ER stress signaling. The mechanistic keystone: wolframin stabilizes the E3 ubiquitin ligase HRD1 and thereby drives ubiquitination and proteasomal degradation of ATF6α. Without wolframin, ATF6α accumulates and ATF6-branch UPR signaling runs hot (PMID:20160352 ✅, Fonseca et al.). The IRE1α/XBP1 and PERK/ATF4/CHOP branches are likewise chronically engaged. Chronic UPR is the difference between a fire alarm and a fire alarm that never stops — eventually the building evacuates permanently, i.e. apoptosis.

Step 3 — ER calcium depletion and MAM failure. Wolframin physically interacts with SERCA2b (PMID:25274773 ✅) and with a complex of NCS1 (neuronal calcium sensor 1) and the IP3 receptor at mitochondria-associated ER membranes. In WFS1-null patient fibroblasts, NCS1 abundance falls (wolframin normally protects NCS1 from proteasomal degradation), ER–mitochondria contacts are reduced, and Ca²⁺ exchange between the two organelles drops (PMID:30352948 ✅, Angebault et al., Sci Signal 2018). CISD2 sits in the same MAM compartment, which is why WFS2 phenocopies WFS1 — the CISD2 p.Asn72Ser variant pushes Ca²⁺ the other way (enhanced ER→mito flux, increased contacts, swollen ER lumen, hyperfused mitochondria), so the two genes converge on "MAM Ca²⁺ handling is broken" from opposite directions.

Step 4 — Mitochondrial consequence. Reduced ER→mitochondrial Ca²⁺ delivery starves the Ca²⁺-dependent dehydrogenases of the TCA cycle → lower ATP output. Mitochondrial dynamics and axonal trafficking are disturbed, and in Wfs1-deficient neurons this impairs dendrite/neurite growth (PMID:27434582 ✅, Cagalinec et al., PLoS Biol 2016) — this is the molecular correlate of the neurodevelopmental limb seen on MRI. hiPSC-derived neuronal models confirm compromised mitochondrial function on WFS1 depletion (Stem Cell Reports 2023).

Step 5 — Cytosolic Ca²⁺ dysregulation → calpain activation. Elevated cytosolic Ca²⁺ hyperactivates calpain-2, and calpain inhibition (or ibudilast) rescues beta-cell function in cellular models (PMID:32632005 ✅, PNAS 2020). This is the node the whole "ER calcium stabilizer" therapeutic strategy — including dantrolene, a ryanodine-receptor blocker — was aimed at.

Step 6 — Cell death in the two most vulnerable populations. CHOP-driven intrinsic apoptosis in pancreatic beta cells (→ insulin-dependent diabetes) and in retinal ganglion cells and long-projection CNS neurons (→ optic atrophy, brainstem degeneration).

A parallel amplifier — inflammation. WFS1 deficiency upregulates pro-inflammatory cytokines and chemokines, producing cytokine-induced ER stress and death in beta cells, and patients show a systemic inflammatory signature (PMID:33693650 ✅, Hum Mol Genet 2021). Preclinically, liraglutide's benefit runs partly through reduced neuroinflammation.

A downstream axonal branch. In zebrafish, wfs1b mutation suppresses Mauthner-cell axon regeneration via the ER stress pathway (PMID:36527091 ✅) — so it's not just cell death, it's failed repair. And a 2026 mouse study finds synaptic alterations precede axonal loss in the optic atrophy (PMID:42255937 ✅, Front Neurosci) — meaning the therapeutic window may open earlier than "axons are dying" implies.

Suggested GO terms (all ⚠️ — verify with OAK before curating)

Biological process: response to endoplasmic reticulum stress GO:0034976; ATF6-mediated unfolded protein response GO:0036500; IRE1-mediated unfolded protein response GO:0036498; PERK-mediated unfolded protein response GO:0036499; ERAD pathway GO:0036503; regulation of ER calcium ion concentration GO:0032469; calcium ion transmembrane transport GO:0070588; intrinsic apoptotic signaling pathway in response to ER stress GO:0070059; protein ubiquitination GO:0016567; mitochondrion organization GO:0007005; neuron apoptotic process GO:0051402; insulin secretion GO:0030073.

Cellular component: endoplasmic reticulum membrane GO:0005789; mitochondria-associated endoplasmic reticulum membrane GO:0044233 (label may have been revised to a "membrane contact site" form — check); endoplasmic reticulum lumen GO:0005788.

Molecular function: calmodulin binding GO:0005516 (wolframin was independently characterized as a calmodulin-binding protein), ubiquitin protein ligase binding GO:0031625.

Suggested CL terms (⚠️)

Pancreatic beta cell CL:0000169; retinal ganglion cell CL:0000740; neuron CL:0000540; oligodendrocyte CL:0000128; cochlear inner hair cell CL:0000589; magnocellular vasopressin-secreting neuron (supraoptic/paraventricular) — likely needs a broader term.

Molecular profiling available

• Transcriptomics: single-cell RNA-seq of CRISPR-corrected vs uncorrected patient SC-β cells showed increased insulin and decreased ER-stress gene expression after correction (PMID:32321868 ✅).
• Multi-omics: a human cell model multi-omic study reporting mitochondrial morphology and function changes (Cell Commun Signal 2021).
• Proteomics/metabolomics/lipidomics: thin. No canonical dataset. Real gap.
• Biomarker: serum neurofilament light chain (NfL) is elevated, while GFAP is not — PMID:41929703 ✅ (Front Neurosci, March 2026). That's a clean, recent, curatable biochemical readout.

7. Anatomical Structures Affected

Organ level (UBERON ⚠️ — verify)

• Primary: pancreatic islet UBERON:0000006 (beta cells); optic nerve UBERON:0000941 and retina UBERON:0000966; brainstem UBERON:0002298, especially pons UBERON:0000988; cerebellum UBERON:0002037; hypothalamo-neurohypophyseal axis — hypothalamus UBERON:0001898, posterior pituitary/neurohypophysis UBERON:0002196; cochlea UBERON:0001844.
• Secondary: urinary bladder UBERON:0001255 (neurogenic/atonic) → ureter UBERON:0000056 and kidney UBERON:0002113 (hydroureter, hydronephrosis, renal failure); gonad UBERON:0000991 / testis UBERON:0000473; gastrointestinal tract (dysmotility); heart UBERON:0000948 (rare).
• Systems: endocrine, nervous (central + autonomic), special sense (visual, auditory), renal/urinary, gastrointestinal, and — secondarily — cardiovascular.

Tissue and cell level

Neuroectodermal and endocrine tissues dominate. Retinal ganglion cell layer and the optic nerve/chiasm; ventral pontine white matter and cerebellar structures; the organ of Corti and stria vascularis; islet beta cells specifically (alpha cells are relatively spared, which is a nice specificity argument for the "high secretory load = high ER load" model).

Subcellular level

This is a subcellular disease in the truest sense — the lesion is at the ER membrane and the ER–mitochondrial contact site (MAM), with the mitochondrion as the injured downstream party. If dismech has a node granularity for organelle-level pathology, this entry should use it.

Localization and laterality

Bilateral and broadly symmetric throughout — optic atrophy, hearing loss, and brainstem volume loss are all bilateral. Asymmetry should prompt reconsideration of the diagnosis.

8. Temporal Development

Onset

Childhood, insidious, sequential. The classic order — diabetes → optic atrophy → diabetes insipidus/deafness → neurologic/urologic — is reliable enough that a child with antibody-negative insulin-dependent diabetes who develops optic atrophy before 16 meets the clinical diagnostic bar without anything else.

Median ages of onset (GeneReviews, classic WFS1-SD):

Brainstem and cerebellar volume abnormalities, though, are already present at the earliest clinical presentation (PMID:22792385 ✅). The clock starts before the symptoms do.

Progression

• Course: chronic, progressive, lifelong. No relapsing-remitting pattern, no spontaneous remission, no plateau.
• Rate: slow but relentless, with substantial inter-individual variability — the source of that variability is largely unexplained (modifiers, §4).
• Stages (informal, no consensus staging system):
• Early — diabetes alone, imaging abnormalities already detectable.
• Intermediate — optic atrophy with progressive visual loss; hearing loss; DI.
• Advanced — neurogenic bladder, upper-tract renal complications, ataxia/dysarthria/dysphagia.
• End-stage — bulbar dysfunction, central apnea, respiratory failure.
• Remission: none spontaneous. Treatment-induced stabilization has now been reported in a single-arm open-label Phase 2 (§12) but never in a controlled trial.

Critical periods

The neurodevelopmental finding is the practically important one: because part of the brain deficit reflects growth that never happened rather than tissue that degenerated, disease-modifying intervention plausibly has to begin before or at diabetes diagnosis — i.e. in early childhood — to capture the full benefit. The 2026 mouse work showing synaptic changes preceding axonal loss (PMID:42255937 ✅) points the same direction. Note the gene-therapy proof of concept deliberately dosed mice at one month of age, stated as corresponding to roughly 10 human years, i.e. when vision loss typically begins (PMID:41998758 ✅).

9. Inheritance and Population

Epidemiology — prevalence estimates vary by more than 25-fold, and that's a finding, not noise

The Sicily-vs-Italy spread (~25×) is the signature of founder effects plus consanguinity in a geographically constrained population, not measurement error. For a dismech prevalence: record, curate the Orphanet band as BELOW_1_IN_1000000 or BAND_1_9_PER_1000000 depending on which source you anchor to, always with population: naming the country and measure_type: POINT_PREVALENCE, and put the verbatim phrasing in notes:.

Incidence: not separately established; the disease is too rare for reliable incidence figures outside registries.

Genetics of transmission

• Inheritance: autosomal recessive (HP:0000007 ✅) for classic Wolfram syndrome 1 and 2. Autosomal dominant (HP:0000006 ⚠️) for Wolfram-like syndrome / DFNA6-14-38 / the de novo neonatal-diabetes-deafness-cataract presentation.
• Penetrance: high, effectively complete for diabetes mellitus + optic atrophy in biallelic LOF genotypes; the later features (DI, deafness, bladder) are age-dependent and incompletely penetrant. Dominant WFS1 alleles show more variable penetrance.
• Expressivity: variable, including between siblings sharing a genotype.
• Anticipation: none — not a repeat-expansion disorder.
• Germline mosaicism: not reported as a recurrent issue; de novo dominant variants are documented.
• Founder effects: yes — Sicily is the clearest example; also reported in Lebanese, Ashkenazi, and various Middle Eastern/South Asian consanguineous populations.
• Consanguinity: a major driver of case load in high-consanguinity regions.
• Carrier frequency: ~1/354 (UK). Note that with a carrier frequency that high, heterozygotes are ~0.3% of the population — which is precisely why the carrier psychiatric-risk and type-2-diabetes-susceptibility signals are epidemiologically interesting.

Demographics

• Sex ratio: ~1:1, as expected for autosomal recessive. No consistent sex difference in severity reported.
• Geographic distribution: worldwide; enriched where consanguinity is common.
• Age distribution of affected individuals: heavily weighted to children and young adults, because life expectancy truncates the distribution in the 4th decade.

10. Diagnostics

Clinical criteria

The operative rule is refreshingly simple: insulin-requiring diabetes mellitus plus optic atrophy, both with onset before age 16, in the absence of another explanation = clinical Wolfram syndrome. GeneReviews formalizes classic WFS1-SD as biallelic pathogenic WFS1 variants + DM and optic atrophy before 16; nonclassic WFS1-SD as a single heterozygous pathogenic variant with a milder/partial phenotype.

The SID/SIEDP expert consensus on early detection and management (Italian diabetes societies) is the current best practice document: PMID:39527371 ✅ (J Endocrinol Invest 2025).

Laboratory

• Diabetes workup that distinguishes it from type 1: islet autoantibodies (GAD, IA-2, ZnT8, IAA) negative; no HLA-DR3/DR4 risk association; C-peptide preserved longer than in autoimmune T1D; often lower insulin requirement and less ketoacidosis at presentation. Antibody-negative "type 1 diabetes" in a child is the single highest-yield screening trigger — see NCT03988764, "Monogenic Diabetes Misdiagnosed as Type 1."
• Diabetes insipidus: water deprivation test with desmopressin challenge; plasma copeptin (with or without arginine/hypertonic saline stimulation) is the modern replacement.
• Emerging biomarker: serum NfL elevated, GFAP not (PMID:41929703 ✅).
• LOINC binding for glucose, HbA1c, C-peptide, plasma/urine osmolality, and copeptin — pull IDs from LOINC directly; I did not verify any this session.

Imaging

• Brain MRI: absent posterior pituitary "bright spot" on T1 (the hallmark of central DI); brainstem, ventral pontine, and cerebellar atrophy; thinning of optic nerves, chiasm, and tracts; reduced white matter volume. Longitudinal neuroradiologic features characterized in AJNR 2020 (ajnr.org/content/41/12/2364).
• OCT: retinal nerve fiber layer thinning — the most sensitive, quantitative, repeatable measure of the optic neuropathy, and the reason visual acuity/OCT became the trial endpoint of choice.
• Renal/bladder ultrasound: hydronephrosis, hydroureter, post-void residual.

Functional and electrophysiologic

Pure-tone audiometry (recessive WS1: high-frequency loss; dominant DFNA6/14/38: low-frequency loss — the inversion is diagnostically useful); visual evoked potentials; urodynamic studies; sleep study / overnight oximetry for central apnea in advanced disease.

Genetic testing — the actual diagnostic gold standard

• Single-gene WFS1 sequencing when the clinical picture is classic; >95% detection by sequencing, +3% by deletion/duplication analysis.
• Multigene panels (monogenic diabetes / inherited optic neuropathy / syndromic hearing loss panels) when the presentation is partial.
• WES/WGS for atypical presentations or when panels are negative; also how CISD2 cases get found.
• Not indicated: chromosomal microarray, karyotype, FISH, repeat-expansion testing. mtDNA testing is indicated only to exclude mitochondrial mimics.
• Two open studies are specifically interrogating the boundary: NCT07485413 ("Looking for VUS to Confirm Dominant Wolfram-like Syndrome Instead of Recessive Wolfram Syndrome") and NCT07336966 ("Does Recessive Optic Atrophy Due to WFS1 Exist?"). Both are 2026-vintage and signal that the allelic-series boundaries are actively contested.

Differential diagnosis

That last row is the practical WFS1-vs-WFS2 discriminator worth curating as a distinguishing_features entry.

Screening

• Newborn screening: not performed anywhere; no biochemical marker exists at birth.
• Carrier screening: not population-based; offered in consanguineous families and after a proband is identified.
• Cascade screening: yes — test siblings; the recessive siblings of a proband are the population where early diagnosis is actually achievable.
• Opportunistic case-finding: the highest-yield strategy is genetic testing of antibody-negative, non-HLA-associated childhood diabetes, plus fundoscopy/OCT surveillance in every child with monogenic-suspect diabetes.

11. Outcome / Prognosis

Mortality

• Median age at death: historically ~30 years (range 25–49), commonly quoted as ~35 and, in the widely cited life-expectancy figure, 39 years. GeneReviews now explicitly revises this upward: median 37 years in specialist-clinic populations, with a maximum of 65 — because modern cohorts include milder and nonclassic cases that historical case-report literature never captured.
• Leading cause of death: central respiratory failure from brainstem atrophy (central apnea, bulbar dysfunction). This is the endpoint the whole neurodegeneration trial program is trying to move.
• Other causes: complications of urinary tract atony (obstructive uropathy, urosepsis, renal failure); hypoglycemic coma; suicide — not a footnote, given a ~25% rate of suicide-related behavior.

Morbidity and function

By the third decade a typical patient carries insulin-dependent diabetes, legal blindness, significant hearing loss, incontinence, and progressive gait/speech impairment simultaneously. Disability is multi-domain and cumulative — the phenotypes don't just add, they compound (losing vision and hearing removes both compensatory channels at once). No Wolfram-specific validated QOL instrument; the WURS is the disease-severity instrument.

Complications

Diabetic complications (retinopathy is confounded by the optic atrophy; nephropathy), obstructive uropathy and CKD, recurrent UTI/urosepsis, aspiration pneumonia from dysphagia, falls from ataxia and blindness, depression/suicidality, central sleep apnea.

Prognostic factors

• Variant type — two null alleles → earlier onset, more complete phenotype (PMID:23429432 ✅); newer genotype severity score (PMID:42524523 ✅); variant topology → residual islet function and urological risk (PMC12331406).
• Age at diabetes onset — earlier onset generally tracks with a more aggressive course.
• Rate of brainstem/ventral pons volume loss on serial MRI — the best imaging prognostic marker.
• Serum NfL — emerging, plausibly a progression marker (PMID:41929703 ✅).
• Residual C-peptide — used as both a prognostic and a therapeutic-response measure (it was the primary endpoint that moved in HELIOS).

Recovery potential

None spontaneous. Lost retinal ganglion cells and beta cells do not come back. This is why every credible therapeutic strategy is either preventive (stop further loss) or replacement (gene therapy, cell therapy) — and why intervention timing is the field's central question.

12. Treatment

Standard of care — entirely supportive, multidisciplinary, no approved disease-modifying therapy anywhere in the world

Surveillance is annual and comprehensive per GeneReviews and the SID/SIEDP consensus (PMID:39527371 ✅).

Disease-modifying attempts — the trial ledger, honestly reported

1. Dantrolene sodium (ryanodine-receptor Ca²⁺ blocker; ER-calcium-stabilizer rationale) - NCT02829268, Phase 1b/2a, open-label, WashU (Urano). Completed. Published JCI Insight 2021, PMID:34185708 ✅. - Result: safe and well tolerated; efficacy essentially negative. Beta-cell function not significantly improved overall (there was a correlation between baseline beta-cell function and change in responsiveness, which is a subgroup signal, not an efficacy result); visual acuity and neurologic function not improved at 6 months. - Historically important as the first-ever Wolfram syndrome clinical trial.

2. Sodium valproate — TREATWOLFRAM - NCT03717909, Phase 2, randomized double-blind placebo-controlled, 36 months, up to 40 mg/kg/day, 63 participants across Birmingham (UK), Paris and Montpellier (France), Almería (Spain), Łódź (Poland). Sponsor: University of Birmingham. Protocol paper: PMID:40010822 ✅. A separate Italian VPA study: NCT04940572. - Result: negative on the primary endpoint. No statistically significant reduction in the rate of visual acuity loss. First MRI brainstem-volume data showed gradual decline in all participants, marginally more in the valproate arm. No benefit, no harm, no unexpected safety signals. Secondary outcome analysis was still ongoing at last public update, and investigators flagged plans to reanalyze against a better-matched placebo group. - This is the most rigorous trial the field has run, and it was negative. Curate it that way — supports: REFUTE or NO_EVIDENCE on any valproate-neuroprotection claim, not a hedge.

3. AMX0035 — sodium phenylbutyrate + taurursodiol (PB&TURSO) — the current bright spot, with caveats - NCT05676034, HELIOS, Phase 2, single-site, single-arm, open-label, 12 adults, up to 208 weeks. Amylyx. - Peer-reviewed publication: PMID:42138079 ✅ — Journal of Clinical Investigation, 15 May 2026, "Phase II trial of sodium phenylbutyrate and taurursodiol in Wolfram syndrome." - Results: significant improvement in the primary endpoint of C-peptide response on mixed-meal tolerance testing at Week 24; at Week 48, sustained stabilization or improvement in pancreatic function, glycemic control (HbA1c and CGM time-in-range), visual acuity, and overall symptom burden. Most participants reported improvement in ≥1 symptom domain — vision, bladder control, insulin-requiring diabetes, fatigue, swallowing, headache/migraine. Well tolerated; adverse events mild-to-moderate, predominantly diarrhea; no serious AEs causing discontinuation. - Mechanistic fit: phenylbutyrate is a chemical chaperone, taurursodiol (TUDCA) is an ER-stress/apoptosis modulator — both act directly on the pathway wolframin normally regulates. The mechanism and the result point the same way, which is reassuring. - The caveat, which must be curated alongside the result: single-center, single-arm, open-label, n=12, no placebo. Amylyx themselves state you cannot draw long-term disease-modification conclusions from this. Given that a well-powered randomized trial (TREATWOLFRAM) in the same disease was negative, discipline here matters — this is supports: PARTIAL territory with an explicit HUMAN_MODEL_MISMATCH-adjacent note about design limitations, not SUPPORT for disease modification.

4. GLP-1 receptor agonists — strongest preclinical package in the field, real off-label human use, no controlled trial - Preclinical: liraglutide in Wfs1 KO rats prevents/delays glucose intolerance and diabetes (PMID:29976929 ✅, PMID:31673100 ✅), reduces islet ER stress, inflammation, and proliferation, and provides extra-pancreatic protection — less neuroinflammation, better learning, prevention of optic nerve degeneration, effects on sensorineural hearing loss. Exenatide restores glucose-stimulated insulin secretion and relieves beta-cell ER stress in Wfs1 KO mice. Dulaglutide prevents and reverses glucose intolerance. In human preclinical models: PMID:36995380 ✅ (Diabetologia 2023). - Human data: an observational evaluation of GLP-1 RA use in Wolfram patients — PMID:42597412 ✅ (Front Endocrinol, July 2026; preprint PMID:41959758). NCT01302327 (exenatide) was withdrawn. NCT05659368 (tirzepatide monotherapy in WS1, Phase 2) has status Unknown. - Verdict: the mechanism is coherent, the rodent data are the best in the field, and human evidence is observational. Curate as EMERGING.

5. Gene therapy — AAV-mediated WFS1 replacement - PMID:41998758 ✅ — "WFS1 gene delivery rescues visual function in a mouse model of Wolfram syndrome," Acta Neuropathol Commun. Vector AAV2/2-CMV-WFS1, intravitreal delivery, overexpression in retinal ganglion cells, protection against optic nerve damage and preservation of visual function. Dosed at one month of age in mice, stated as corresponding to ~10 human years — the age vision loss typically begins. - Additional strategy under exploration: scAAV9-mediated NCS1 overexpression (rationale from PMID:30352948 ✅; zebrafish rescue of mitochondrial activity and behavior, PMC9594121). - Status: preclinical. No human gene therapy trial for Wolfram syndrome is open.

6. Cell therapy / regenerative - PMID:32321868 ✅ — Maxwell et al., Science Translational Medicine 2020. CRISPR-Cas9 correction of a pathogenic WFS1 variant in patient-derived iPSCs, differentiation to stem-cell-derived beta cells, robust dynamic insulin secretion in vitro, and reversal of pre-existing diabetes after transplantation into mice. Single-cell transcriptomics showed increased insulin and decreased ER-stress gene expression in corrected cells. This was the first demonstration of CRISPR correcting a patient's diabetes-causing defect and reversing diabetes. - Multidimensional patient-iPSC disease modeling and therapeutic development: PMID:36134655 ✅. - Status: preclinical, but this is the most concrete path to actually replacing the beta cells rather than protecting them.

7. Other pharmacologic strategies - Ibudilast / calpain-2 inhibition — rescues beta-cell function in cellular models, PMID:32632005 ✅ (PNAS 2020). Ibudilast is already an approved drug in Japan, so repurposing is plausible. - Deferiprone (iron chelation) + incretin therapy for Wolfram syndrome 2 — NCT02882477, Phase 2/3, interventions deferiprone, acetylcysteine, sitagliptin and metformin. Status Unknown. The rationale is CISD2's iron-sulfur cluster chemistry, which is WFS2-specific — do not generalize it to WFS1. - Historically explored and not established: valproate (now negative), chemical chaperones generally, thiamine (only relevant for the TRMA mimic).

Pharmacogenomics

Nothing Wolfram-specific. Standard considerations apply — valproate hepatotoxicity/POLG interaction being the one relevant safety note, given valproate was trialed here.

Treatment algorithm

There is no disease-modifying algorithm because there is no approved disease-modifying therapy. Practically: confirm genetically → establish the multidisciplinary surveillance schedule → manage each manifestation as it appears → enroll in the registry (NCT02841553) and, where available, a trial. The Urano group's stated framing is a layered, individualized strategy — small-molecule ER-stress modulation now, gene editing and regenerative therapy layered on later.

13. Prevention

Primary prevention

None possible for the recessive disease. You cannot prevent a genotype you're born with. The only genuine primary prevention is reproductive: genetic counseling for at-risk couples, carrier testing in consanguineous families, and where families choose it, prenatal diagnosis or preimplantation genetic testing (PGT-M) — technically straightforward once the familial variants are known.

No vaccination, no environmental intervention, no behavioral modification, no prophylactic medication prevents Wolfram syndrome. Say so plainly rather than padding the section.

Secondary prevention (early detection)

This is where real gains are available: - Genetic testing of every child with autoantibody-negative, non-HLA-associated insulin-dependent diabetes. This is the highest-yield case-finding intervention that exists and is currently under-done — hence NCT03988764, "Monogenic Diabetes Misdiagnosed as Type 1." - Fundoscopy and OCT in children with monogenic-suspect diabetes. - Cascade testing of siblings once a proband is identified. - The SID/SIEDP consensus (PMID:39527371 ✅) exists precisely to standardize early detection. - No newborn screening exists or is proposed — there's no biochemical marker at birth.

Tertiary prevention (preventing complications in diagnosed patients)

This is where the annual surveillance schedule earns its keep, and where preventable deaths hide: - Bladder and upper urinary tract: post-void residuals, renal ultrasound, urodynamics. Timely clean intermittent catheterization prevents hydronephrosis → renal failure → urosepsis. This is a genuinely preventable cause of death. - Respiratory: monitor for central apnea and bulbar dysfunction as neurologic disease advances; ventilatory support where appropriate. - Psychiatric: proactive depression screening and suicide-risk assessment, given ~25% suicide-related behavior with a documented 15–32 age window. Arguably the most under-implemented preventive measure in this disease. - Aspiration: swallow assessment once dysphagia appears. - Glycemic: standard diabetes complication prevention.

Counseling

Standard autosomal recessive counseling — 25% recurrence risk per pregnancy for carrier couples, sibling carrier risk 2/3 among unaffected sibs. Extra care required for the dominant allelic series: if the family's variant is one of the dominant WFS1 alleles, the counseling arithmetic is completely different (50% transmission, variable penetrance), and misclassifying which series a family is in is a real counseling error. Note also the carrier psychiatric-risk literature — whether and how to disclose that to heterozygotes is an open ethical question, not a settled one.

Public health

Not applicable in the usual sense. The population-level lever that would matter most is consanguinity-aware genetic services in high-prevalence regions, which is a health-systems intervention rather than a sanitation/vector-control one.

14. Other Species / Natural Disease

• Naturally occurring Wolfram syndrome in non-human animals: none reported. I found no OMIA entry for a spontaneous WFS1 disorder in companion animals or livestock. Every animal model in the literature is engineered. Record this as an explicit negative, not as an unexamined gap — it's a meaningful contrast with, say, canine SOD1 degenerative myelopathy.
• Zoonotic potential / cross-species transmission: not applicable. Monogenic, non-transmissible.

Orthologs and taxonomy (NCBITaxon ⚠️):

Evolutionary conservation: wolframin is conserved across vertebrates and present in Drosophila, and — importantly for mechanism — its functional partnership is conserved too: the fly homolog synergizes with the IP3 receptor to affect mitochondrial morphology and function, recapitulating the human WFS1–IP3R–MAM axis in an organism separated from us by ~600 million years. That's about as good a conservation argument as a mechanism gets.

Comparative pathology: the informative cross-species observation is a dissociation. Mice largely fail to reproduce the optic atrophy and show only mild diabetes; rats reproduce both the diabetes and the brainstem/optic nerve neurodegeneration; zebrafish reveal an axon-regeneration phenotype invisible in mammals. The species differences are themselves data about which parts of the human disease depend on which conserved functions.

15. Model Organisms

A comprehensive review of every available Wolfram model — read this first: PMID:38351344 ✅, "Comprehensive overview of disease models for Wolfram syndrome: toward effective treatments," Mammalian Genome, March 2024.

Mammalian in vivo

Mouse — Wfs1 knockout (several independent lines: exon 8 deletion, exon 2 deletion, and beta-cell-specific conditional KO). - Recapitulates: progressive glucose intolerance and beta-cell loss, elevated ER stress markers, impaired glucose-stimulated insulin secretion, behavioral/anxiety phenotypes. - Limitations (important, curate as PARTIALLY_RECAPITULATES or FAILS_TO_RECAPITULATE with explicit limitations:): diabetes is mild relative to human disease; optic atrophy is weak or absent in most lines; diabetes insipidus is not well reproduced. Background strain strongly modifies severity. - A 2026 optic-nerve study in a mouse model nonetheless found synaptic alterations preceding axonal loss — PMID:42255937 ✅ — so the mouse retains value for early-stage mechanism even where it undershoots the endpoint. - Resources: MGI, IMPC, KOMP, IMSR.

Rat — Wfs1 KO (University of Tartu). PMID:28860598 ✅ (Plaas et al., Sci Rep 2017). - Recapitulates: a more prominent diabetic phenotype than mouse models, plus neurodegeneration of the brainstem and optic nerve — i.e. it captures the two features that actually kill and blind patients. - This is the workhorse for pharmacology. The entire liraglutide package (PMID:29976929 ✅, PMID:31673100 ✅) was built in it, including the lifelong-treatment study covering visual neurodegeneration, sensorineural hearing loss, and the diabetic phenotype simultaneously. - Limitations: rodent lifespan and brain scale still don't map onto a 30-year human course; the human neurodevelopmental component (§8) is hard to model.

Non-mammalian in vivo

Zebrafish — wfs1b mutants. - Distinctive finding: wfs1b mutation suppresses Mauthner-cell axon regeneration via ER stress signaling (PMID:36527091 ✅) — a repair-failure phenotype rather than a death phenotype. - NCS1 overexpression restored mitochondrial activity and behavioral alterations in a zebrafish Wolfram model (PMC9594121) — direct in vivo validation of the MAM/NCS1 therapeutic hypothesis. - Resource: ZFIN.

Drosophila — wfs1 homolog. Synergizes with IP3R to affect mitochondrial morphology and function; a fast genetic-interaction screening platform for the Ca²⁺/MAM arm. Resource: FlyBase.

Cellular / in vitro (the non-animal models — experimental_models: in dismech terms)

Patient fibroblasts. The substrate for the NCS1/ER–mitochondria discovery: reduced NCS1 abundance, reduced ER–mitochondria interactions, impaired Ca²⁺ exchange (PMID:30352948 ✅).

Patient iPSC-derived neurons. WFS1 depletion compromises mitochondrial function (Stem Cell Reports 2023). Bridges the human-genetics-to-neurodegeneration gap the rodent models can't fully close.

Patient iPSC-derived beta cells ± CRISPR correction. The flagship. PMID:32321868 ✅ — isogenic corrected vs uncorrected SC-β cells; corrected cells indistinguishable from healthy-donor-derived cells, reversed pre-existing diabetes on transplantation into mice, and showed reduced ER-stress gene expression by scRNA-seq. Isogenic pairs are the cleanest causal design available in human cells. Further multidimensional iPSC modeling: PMID:36134655 ✅.

Immortalized cell models. Used for the ATF6α/HRD1 mechanism (PMID:20160352 ✅), calpain-2/ibudilast rescue (PMID:32632005 ✅), SERCA interaction (PMID:25274773 ✅), and the human-preclinical GLP-1 RA work (PMID:36995380 ✅).

Cross-model limitations worth recording explicitly

1. No model reproduces the full DIDMOAD tetrad — DI in particular is poorly modeled anywhere.
2. The neurodevelopmental limb (failed brain growth, visible on human MRI at earliest presentation) has essentially no dedicated model.
3. Psychiatric phenotypes — depression, suicidality, ~60% severe psychiatric burden — are the least modeled and arguably most under-served dimension of the disease.
4. Timescale mismatch: a 30-year human course compressed into rodent months means every "prevention" result is really a "prevention in a rapidly-progressing animal" result.

These are good candidates for HUMAN_MODEL_MISMATCH discussions rather than generic KNOWLEDGE_GAP entries — evidence exists, it's the translational validity that's open.

Curation notes for the dismech entry

A few things I'd flag before you build kb/disorders/Wolfram_Syndrome.yaml:

1. NEC risk is real here. "Wolfram syndrome" resolves to at least six adjacent MONDO entities (verified above), and the numbered-series pattern (WFS1/WFS2) plus the recessive-vs-dominant split is exactly the risk class the preflight tool exists for. Run just preflight-dr <report> MONDO:0009101 on any deep-research output. The canonical causal gene for MONDO:0009101 is WFS1 — a report dominated by CISD2 mentions is a wrong-entity report.
2. disease_term should be MONDO:0009101 (Wolfram syndrome 1) if the entry is the recessive WFS1 disease, with MONDO:0018105 available as a broader grouping mapping. Remember the two enum caches (DiseaseTerm and DiseaseOrSubtypeTerm) both need seeding, and mirror into the primary checkout.
3. Module conformance candidates: loss_of_proteostasis (ER stress/UPR — check whether the existing module's scope covers UPR hyperactivation rather than aggregate formation), photoreceptor_degeneration is not the right one (this is retinal ganglion cell loss, not photoreceptor), sensorineural_hair_cell_loss for the deafness arm, peripheral_axonal_degeneration probably not (this is central). glaucoma_optic_neuropathy shares the RGC-apoptosis node but has a pressure-driven trigger — conformance would be node-qualified at best, so check the module's criteria before wiring it.
4. The valproate result is negative and the AMX0035 result is uncontrolled-positive. Both belong in the entry, tagged honestly. That contrast is genuinely informative content, not a blemish.
5. Do not curate a single "life expectancy 30 years" number. The historical figure and the modern specialist-clinic figure differ by ~7 years for a reason that is itself a finding (ascertainment bias). Two prevalence/prognosis records with different population: values, or one with the caveat in notes:.

Sources

• WFS1 Spectrum Disorder — GeneReviews (NBK4144)
• OMIM #222300 — Wolfram Syndrome 1
• MONDO via EBI OLS4 API
• HPO annotations for OMIM:222300 (ontology.jax.org)
• ClinicalTrials.gov — Wolfram syndrome studies
• Wolfram syndrome 1 gene negatively regulates ER stress signaling (PMID:20160352)
• ER-mitochondria cross-talk regulated by NCS1 is impaired in Wolfram syndrome — Science Signaling
• Novel CISD2 mutation alters Ca2+ homeostasis and ER-mitochondria interactions
• Novel mutations in WFS1 associated with Wolfram syndrome and systemic inflammation (PMID:33693650)
• A phase Ib/IIa clinical trial of dantrolene sodium in Wolfram syndrome — JCI Insight
• TREATWOLFRAM trial protocol (PMC11865774)
• TREATWOLFRAM interim results — The Snow Foundation
• Amylyx — Positive long-term Phase 2 HELIOS results for AMX0035
• Amylyx — Peer-reviewed HELIOS publication in JCI
• WFS1 gene delivery rescues visual function in a mouse model (PMID:41998758)
• Gene-edited human stem cell–derived β cells reverse preexisting diabetes — Sci Transl Med
• GLP-1R agonists in human preclinical models of Wolfram syndrome — Diabetologia
• Liraglutide neuroprotection in an aged rat model of Wolfram syndrome — Sci Rep
• Evaluating the use of GLP-1 receptor agonists in Wolfram syndrome patients — Front Endocrinol
• Evidence for altered neurodevelopment and neurodegeneration using longitudinal morphometry — Sci Rep
• Early Brain Vulnerability in Wolfram Syndrome — PLoS One
• Longitudinal Assessment of Neuroradiologic Features in Wolfram Syndrome — AJNR
• Clinical trials for Wolfram syndrome neurodegeneration: design, endpoints, analysis — PLoS One
• Wolfram Syndrome 1: A Neuropsychiatric Perspective (PMC11353439)
• WFS1 autosomal dominant variants linked with hearing loss (PMID:37041640)
• Topology of WFS1 variants linked with islet function and urological symptoms (PMC12331406)
• Multidimensional analysis using patient iPSC-derived models of Wolfram syndrome (PMID:36134655)
• Wolfram syndrome: MAMs' connection? — Cell Death & Disease
• Wolfram syndrome, from pathogenesis to future treatment perspectives — J Transl Med
• Genomics of Wolfram Syndrome 1 — Biomolecules
• Wolfram Syndrome International Registry (WashU)

Reference Validation

Checked with linkml-reference-validator 0.2.1.

All extracted references resolved successfully.