Intellectual Developmental Disorder Autosomal Recessive 67

1. Disease Information

Overview

MRT67 is a rare, syndromic, autosomal recessive neurodevelopmental disorder caused by biallelic variants in EIF3F, which encodes the f subunit of eukaryotic translation initiation factor 3 (eIF3) — the largest and most structurally complex of the eukaryotic initiation factors. Its mechanistic logic is unusual among Mendelian neurodevelopmental disorders: the primary lesion is a partial, quantitative failure of general protein synthesis, not the disruption of a neuron-specific protein or a discrete signalling pathway.

The clinical core is global developmental delay with prominent speech delay in essentially every reported individual, followed by a variable syndromic tail: behavioural problems, sensorineural hearing loss, altered muscle tone, short stature, microcephaly, ophthalmological findings, reduced pain sensitivity, sleep disturbance, and occasional orofacial clefting.

Key identifiers

Gene-level OMIM number for EIF3F was not verified in this session — do not assert one without checking.

Synonyms

• MRT67
• Mental retardation, autosomal recessive 67 (legacy OMIM/MedGen title)
• EIF3F-related neurodevelopmental disorder
• EIF3F-related syndrome (Simons Searchlight usage)
• EIF3F deficiency

Nature of the evidence base

Aggregated disease-level, not EHR-derived. The entire published literature is small-n clinical genetics: one exome-wide burden analysis with case follow-up, one multi-centre GeneMatcher-assembled cohort, and scattered single-case reports. There is no registry, no natural-history study, and no EHR-phenotyped cohort. Simons Searchlight states plainly that "the information available about EIF3F is limited" and that "As of 2025, about 31 people with EIF3F-related syndrome have been described in medical research" (Simons Searchlight).

2. Etiology

Disease causal factors

Purely monogenic and genetic. Biallelic EIF3F variants are necessary and, on current evidence, sufficient. No environmental, infectious, or toxic contribution has been proposed or investigated.

The founding association came from a genome-wide recessive burden analysis of the Deciphering Developmental Disorders study (Martin et al., Science 2018; PMID:30409806):

"We identified two genes not previously associated with recessive developmental disorders, KDM5B and EIF3F, and functionally validated them with mouse and cellular models."

That study reported nine affected individuals from seven non-consanguineous families of European ancestry, all homozygous for the same missense allele (OMIM 618295).

Genetic risk factors

The disorder is driven by a single, dominant-in-frequency founder allele: NM_003754.3:c.694T>G, p.(Phe232Val). Hüffmeier 2021 (PMID:33736665):

"21 patients were homozygous and one compound heterozygous for c.694T>G/p.(Phe232Val) in EIF3F. Haplotype analyses in 15 families suggested that c.694T>G/p.(Phe232Val) was a founder variant."

Only one additional pathogenic allele has been published — the frameshift c.861dup, p.(Gln288AlafsTer14), seen in trans with the founder missense in a single compound heterozygote.

A key structural feature of this disease's genetics: almost all recessive NDDs are caused by many private, mostly truncating variants across a handful of families. MRT67 inverts that. Hüffmeier 2021:

"Therefore, an identical, rather frequent missense variant in EIF3F in almost all affected individuals is an uncommon finding in NDDs, particularly in a cohort with heterogeneous, ethnical and regional backgrounds."

The authors explicitly analogise to CFTR p.Phe508del, HFE, and GJB2 founder alleles.

Modifier genes

None established. One report (Huang 2022, PMID:35743796) describes a heterozygous EIF3F p.P4S variant as one of seven inherited rare variants in an oligogenic model of ID/ASD/psychosis in a Taiwanese proband. This is not MRT67 — the individual was heterozygous, the variant is not the founder allele, and the authors propose only cumulative moderate effects. Cite it, if at all, as a hypothesis about heterozygote burden, never as MRT67 evidence.

Environmental risk factors, protective factors, gene–environment interaction

None described. No CTD, GWAS, or exposure data exist for this entity. This is not a gap that could plausibly be filled by existing databases — it reflects the disorder's fully penetrant monogenic architecture and tiny cohort. Record explicitly as not applicable / not studied rather than leaving blank.

One indirect gene–environment observation exists in the mouse, and it is mechanistically interesting: Eif3f heterozygotes are normal at baseline but decompensate under a physiological stressor. Docquier 2019 (PMID:31026345):

"We found that eIF3f partial depletion amplifies muscle atrophy compared to wild-type mice."

That is a gene-by-disuse interaction in muscle, not a disease risk factor in humans, but it is the closest thing in the literature.

3. Phenotypes

The definitive frequency table is Hüffmeier 2021 Table 2 (PMID:33736665), which reports the new cohort (n=21 evaluable) alongside the original Martin cohort (n=9). Where the two disagree — most sharply on epilepsy — the disagreement itself is the finding.

Core, near-universal features

"All affected individuals had developmental delays including delayed speech development. About half of the affected individuals had behavioral problems, altered muscular tone, hearing loss, and short stature." — PMID:33736665 (abstract)

Motor milestone detail: delayed unassisted sitting (>10 months) in 4/12 (33%); delayed independent walking (≥18 months) in 14/20 (70%). Severity is genuinely variable: "one patient did not learn to walk independently, while most individuals learned walking late, and some individuals achieved motor milestones at a normal age."

Frequent features (~30–60%)

Behavioural detail: "obsessive compulsory disorder, social problems, anxiety, autism, hyperactivity, attention deficit, aggressivity or pica." Autism has been formally instrumented in at least one case (Lakatosova 2024, PMID:39723281): "The girl was diagnosed using Autism Diagnostic Observation Schedule-Second Edition (ADOS-2) and Autism Diagnostic Interview-Revised (ADI-R) instruments and met the criteria for autism spectrum disorder."

Occasional features (10–25%)

Rare / single-case features

Reduced pain sensitivity (HP:0007021 Pain insensitivity — 2 individuals, confirming a feature first seen by Martin); psychosis (HP:0000709, 1/21); encephalopathy (1/21, with concomitant B12 deficiency); nystagmus (HP:0000639, 1/21, 5%); coloboma (HP:0000589, 1/21, 5%); tetralogy of Fallot (HP:0001636, 1); congenital lobar emphysema; anal stenosis (HP:0002025); cryptorchidism (HP:0000028); nasal fistula; muscle atrophy (HP:0003202); drop attacks; arachnoid cyst (HP:0100702); prominent perivascular spaces; mild tonsillar ectopia (HP:0002707 Abnormal cerebellum morphology).

Overgrowth/macrocephaly — a phenotype-expanding outlier. An ACMG 2024 abstract (Shad, Mbonu, Aguiar, Hassan, Repnikova; Genetics in Medicine Open P319, doi:10.1016/j.gimo.2024.101213) reports two male siblings homozygous for c.694T>G, one with typical features and the older brother with additional overgrowth and macrocephaly — the opposite direction from the cohort's microcephaly/short stature. This is a conference abstract, not peer-reviewed full text; grade accordingly.

Neuroimaging in a single detailed case (Lakatosova 2024, PMID:39723281): "An MRI scan at 24 months of age showed randomly distributed cystic dilatations of the Virchow-Robin spaces bilaterally in the white matter, with a maximum concentration in the left parietal-temporal-occipital site, indicating atypical leukoencephalopathy. The findings also included an asymmetrically narrowed right internal auditory canal and cisterna magna." Suggested terms: HP:0002352 Leukoencephalopathy, HP:0002185 Neuronal loss in central nervous system (no — prefer HP:0012705 Abnormal cerebral white matter morphology), HP:0002700 Enlarged cisterna magna.

Phenotype characteristics

• Onset: congenital to infantile. Delays are apparent in the first two years. Pregnancy and delivery are largely uneventful (1/21 oligohydramnios; 1/19 suspected perinatal asphyxia).
• Severity: highly variable within an essentially uniform genotype — from near-normal motor milestones with usable speech to non-ambulatory and non-verbal.
• Progression: predominantly static/developmental. Regression is reported but confounded (see §11).
• Dysmorphology: subtle and not gestalt-recognisable. Hüffmeier 2021: "Dysmorphic features were generally subtle and non-specific and were not considered a recognizable facial gestalt." This matters clinically — you will not diagnose this on the face.

Quality-of-life impact

No EQ-5D, SF-36, or PROMIS data exist. The one instrumented case (PMID:39723281) provides the only structured functional data in the literature: Vineland-3 showed "a deeply below-average adaptive behavior profile in all subdomains of the test, including communication, daily living skills, socialization, and motor skills"; BRIEF-P showed clinically elevated Inhibit, Working Memory, and Plan/Organize scales; the Children's Sleep Habits Questionnaire score was 48 against a cutoff of 41, with bedtime resistance and night waking most pronounced. The child had urinary and stool incontinence at 5y8m and a 10-word vocabulary.

4. Genetic / Molecular Information

Causal gene

EIF3F (eukaryotic translation initiation factor 3 subunit F), 11p15.4. HGNC CURIE in dismech form: hgnc:3275. Formerly EIF3S5. Encodes eIF3f, one of 13 subunits (eIF3a–m) of the eIF3 complex.

Pathogenic variants

1. NM_003754.3:c.694T>G, p.(Phe232Val) — the founder allele

Population frequency, per Hüffmeier 2021:

"This variant represents the 7th most common EIF3F missense variant in gnomAD (0.07%) with highest frequencies in Ashkenazi Jewish (0.21%) and non-Finnish European individuals (0.12%)"

"In line with its pathogenicity, no individual in gnomAD was reported to be homozygous for this variant, in contrast to five of the six more frequent variants."

Independent ClinVar-submitter figures agree: allele frequency 0.0007 in 251,224 control chromosomes (gnomAD v2.1.1 total 0.000711), 0.0012 in non-Finnish Europeans, no homozygotes. The absence of homozygotes despite substantial heterozygote frequency is the population-genetic argument that carries most of the pathogenicity weight — five of the six more common EIF3F missense variants do have gnomAD homozygotes.

Functional characterisation (ClinVar interpretation summary): the variant "results in reduced protein expression, approximately 70% of wildtype, and was associated with reduced protein translation and cellular proliferation rates." Phe232 is evolutionarily conserved and predicted to stabilise the protein (PMID:39723281).

2. c.861dup, p.(Gln288AlafsTer14) — frameshift

Found in trans with the founder allele in one individual (P3 in Hüffmeier 2021). Absent from gnomAD. "was predicted to result in a frameshift and subsequently either in mRNA decay or a truncated protein (length reduced by 20%, exon 6 of 8) with altered structure."

Constraint and the null-lethality argument

"Of note, the overall number of truncating alleles in individuals in gnomAD v2.1.1 is extremely low (probability of being loss-of-function intolerant = 0.97, observed over expected variants = 0.07). Regarding lack of individuals with two truncating variants and of further affected individuals who are compound heterozygous for truncating variants, one might speculate that truncating variants on both alleles might not be compatible with life." — PMID:33736665

This is the single most important mechanistic inference in the disease, and it is triangulated three ways: (i) pLI 0.97 / o/e 0.07 constraint; (ii) no reported patient with two truncating alleles; (iii) mouse Eif3f-null embryonic lethality (§15). The founder allele is best read as a hypomorph that survives where a null would not.

Genotype–phenotype correlation

Exactly one signal, from the single compound heterozygote:

"Symptoms in the compound heterozygous individual with an additional truncating variant were at the severe end of the spectrum in regard to motor milestones, speech delay, organic problems and pre- and postnatal growth of body and head, suggesting some genotype-phenotype correlation." — PMID:33736665

That individual did not walk independently at age 5 (crawled), had no speech, tetralogy of Fallot, GI problems, IUGR, and was both short and microcephalic. This is a dosage disease: missense/missense < missense/null < null/null (lethal).

Other missense variants are of unknown consequence. Hüffmeier 2021 is explicit: "As only the identical missense variant in EIF3F has been functionally characterized … it remains speculative whether other missense variants might have comparable effects." Do not extrapolate pathogenicity from p.Phe232Val to novel EIF3F missense variants without functional work.

Modifier genes, epigenetics, chromosomal abnormalities

• Modifier genes: none identified.
• Epigenetics: no methylation episignature has been reported for EIF3F. Nothing in ENCODE/Roadmap/DiseaseMeth is disease-specific. Record as not studied.
• Chromosomal abnormalities: not a mechanism. CMA is characteristically negative in these patients — both the Lakatosova case ("Microarray analysis of CNVs (aCGH) showed negative findings") and routine workups. This is diagnostically important (§10).
• Somatic variation: MRT67 is germline. Somatic EIF3F variants were surveyed in COSMIC for an unrelated reason (the meningioma question, §11): "somatic EIF3F variants have been detected in 0.9% of 38,579 samples (n = 353) and did not include 130 meningioma samples."

5. Environmental Information

Not applicable. No environmental factors, lifestyle factors, or infectious agents are implicated in MRT67. Searching CTD/TOXNET/NHANES for this entity is not a productive use of curation effort. Any dismech environmental: block should either be absent or carry a notes: line recording that the search was made and returned nothing, so a later curator does not repeat it.

6. Mechanism / Pathophysiology

The causal chain

EIF3F c.694T>G p.(Phe232Val) homozygous  [MOLECULAR]
↓  reduced eIF3f protein (~70% of wild-type)
Reduced eIF3f abundance in the eIF3 complex  [MOLECULAR]
↓
Reduced global translation initiation rate  [CELLULAR]
   (↓ polysome content, ↓ protein synthesis rate, ↓ mTOR pathway activation)
↓
Reduced cell proliferation and cell growth  [CELLULAR]
↓
Impaired brain development and somatic growth  [ORGANISM]
   → GDD, ID, speech delay, microcephaly, short stature, altered muscle tone

Upstream is the variant → protein-abundance step (measured). Midstream is the translation → proliferation step (measured in iPSCs and mouse muscle). Downstream is the step to clinical phenotype, which is inferred, not measured — no neural model of the disease allele has been published.

Molecular pathway: translation initiation

eIF3 stimulates multiple steps of initiation — 43S pre-initiation complex assembly, mRNA recruitment, prevention of premature 60S joining, AUG scanning (Moretti 2010, PMID:21124883). Crucially, eIF3 is not a uniform amplifier:

"EIF3F encodes an essential subunit of the largest eukaryotic translation initiation factor eIF3 which binds to a highly specific group of mRNAs involved in cell proliferation and growth, including cell cycle control, differentiation and apoptosis" — PMID:33736665

That selectivity is the best available answer to the obvious objection — why does a housekeeping translation factor produce a selective neurodevelopmental phenotype rather than uniform cellular failure?

GO terms: - GO:0003743 translation initiation factor activity (modifier: DECREASED) - GO:0006413 translational initiation (DECREASED) - GO:0005852 eukaryotic translation initiation factor 3 complex (cellular component) - GO:0002181 cytoplasmic translation (DECREASED) - GO:0008283 cell population proliferation (DECREASED) - GO:0031929 TOR signaling (DECREASED) - GO:0007420 brain development (ABNORMAL)

Reactome: R-HSA-72613 (Eukaryotic Translation Initiation), R-HSA-72649 (Translation initiation complex formation). KEGG: hsa03013 (Nucleocytoplasmic transport/RNA transport).

The mTORC1 coupling — how a translation defect becomes a growth defect

This is the best-characterised non-canonical arm and it is where the mouse data land. eIF3f is a scaffold connecting mTORC1 and S6K1:

• Under nutrient-rich conditions, the mTOR/raptor complex binds the TOS motif of eIF3f and phosphorylates/activates S6K1; active S6K1 is released from eIF3, increasing protein synthesis (Csibi et al., PLoS ONE 2010, doi:10.1371/journal.pone.0008994).
• Conversely, the E3 ligase MAFbx/atrogin-1 physically interacts with the Mov34 domain of eIF3f via its leucine-charged domain, polyubiquitinating it for proteasomal degradation during muscle wasting. An eIF3f mutant insensitive to MAFbx polyubiquitination maintains S6K1/rpS6 phosphorylation.
• eIF3f thus sits at the atrophy/hypertrophy antagonism node in skeletal muscle (reviewed PMID:23769948).

In vivo confirmation, Docquier 2019 (PMID:31026345):

"Skeletal muscles of heterozygous mice show a deficiency in polysome content, a decrease in protein synthesis rate and an inhibition of the mechanistic target of rapamycin (MTOR) pathway."

Hüffmeier 2021 draws the link explicitly: "partial depletion of murine eIF3f amplified muscle atrophy compared to wild-type mice and reduced the MTOR pathway activation."

This is mechanistically load-bearing for the human phenotype because it supplies a route from reduced eIF3f → reduced mTORC1 signalling → reduced growth, which is otherwise a hand-wave. GO: GO:0031929 TOR signaling; GO:0038202 TORC1 signaling.

Moonlighting function 1: deubiquitinase activity on Notch

eIF3f carries an MPN/JAMM domain and has intrinsic DUB activity (Moretti 2010, PMID:21124883):

"An immunofluorescence-based screening of an shRNA library allowed us to identify eIF3f, previously known as one of the subunits of the translation initiation factor eIF3, as a DUB targeting the activated Notch receptor. We show that eIF3f has an intrinsic DUB activity."

"we also show that eIF3f is recruited to activated Notch on endocytic vesicles by the putative E3 ubiquitin ligase Deltex1, which serves as a bridging factor … eIF3f is a new positive regulator of the Notch pathway."

Curation caution: this is elegant in vitro cell-biology (U2OS, HEK293T, HeLa; shRNA knockdown; coculture reporter assays) and it has never been connected to MRT67 patient biology. Notch signalling is central to neurodevelopment, so the temptation to draw a causal edge from p.Phe232Val → impaired Notch → neurodevelopmental phenotype is strong — and unsupported. Phe232 has not been shown to affect DUB activity. If curated at all, this belongs in a discussions: block as a KNOWLEDGE_GAP with a proposed experiment, not as a pathophysiology edge. GO: GO:0004843 cysteine-type deubiquitinase activity; GO:0007219 Notch signaling pathway.

Moonlighting function 2: white matter and RAN translation

eIF3f regulates repeat-associated non-ATG (RAN) translation in spinocerebellar ataxia type 8 (Ayhan et al., EMBO J 2018, PMID:30206144): knockdown of eIF3F reduces steady-state levels of SCA8 polySer and other RAN proteins, and the polySer aggregates accumulate preferentially in white matter regions showing demyelination and axonal degeneration.

Lakatosova 2024 (PMID:39723281) uses this to interpret their patient's leukoencephalopathy:

"EIF3F was shown to be overexpressed in white matter compared to gray matter in various brain regions [19,20], suggesting its importance in myelination and white matter homeostasis, including regulation of oligodendrocyte functions."

Grade this carefully. It is a single-case imaging finding plus an inference from a different disease's biology. Note also that the Human Protein Atlas reports EIF3F as low tissue specificity (tau 0.21, "detected in all"), clustering with ribosomal/translation genes and showing no white-matter or neuron-enriched pattern (HPA ENSG00000175390). The white-matter-enrichment claim and the HPA data are in tension. Curate as a hypothesis with the tension recorded, not as an established mechanism.

Cell types, tissues, subcellular compartments

• Subcellular: cytoplasm (GO:0005737), eIF3 complex (GO:0005852), cytosolic ribosome-associated (GO:0022626), and — for the Notch DUB function — early endosome (GO:0005769).
• Cell types: no disease-specific cell type is established. Speculative/inferred candidates only: CL:0000540 neuron, CL:0000128 oligodendrocyte (from the white-matter hypothesis), CL:0000187 muscle cell (from mouse), CL:0002322 embryonic stem cell (the model system actually used).
• Do not manufacture a cell-type binding. The measured cellular work was done in undifferentiated iPSCs and mouse skeletal muscle. Neither is the affected tissue in patients.

Molecular profiling

• Transcriptomics: no patient RNA-seq published. No GEO series specific to MRT67 was identified.
• Proteomics / metabolomics / lipidomics: none.
• Single-cell / spatial: none.
• Functional genomics: EIF3F is a broadly essential gene in DepMap-class screens (consistent with pLI 0.97 and mouse null lethality), but no MRT67-specific screen exists.

This is a real and citable gap. No omics dataset exists for this disorder in any repository. Any datasets: block would have to be gene-level or model-level, and per dismech policy would need just verify-datasets plus manual relevance triage — searching EIF3F in GEO will return translation-biology and cancer datasets that are not about this disease (the classic Named Entity Confusion trap).

7. Anatomical Structures Affected

Organ level

Body systems: nervous, sensory (auditory, visual), musculoskeletal, digestive; growth as a cross-cutting axis.

Lateralization

Predominantly bilateral and symmetric where laterality is reported (bilateral mild SNHL; bilateral Virchow-Robin space dilatation). Two asymmetric findings in the single detailed imaging case: right internal auditory canal stenosis, and left parieto-temporo-occipital predominance of the perivascular changes (PMID:39723281). One unilateral single palmar crease was noted in four individuals.

The negative finding is important

"In concordance with the previous study, brain imaging did not reveal specific findings and were therefore not considered diagnostically indicative in EIF3F related NDD." — PMID:33736665

There is no structural lesion to point at. Nonspecific findings in ~38–43% across both cohorts. This is consistent with a diffuse quantitative translation deficit rather than a focal developmental malformation, and it is why the disease is diagnosed by sequencing, not imaging.

8. Temporal Development

Onset

• Congenital to early infantile. Developmental delay is the presenting feature. Microcephaly and short stature may be present at birth (4/10 and 3/15 respectively) or acquired later.
• Onset pattern: insidious, developmental — not acute.
• Pregnancy/delivery largely uneventful; 2/20 premature.

Progression

• Predominantly static. No defined disease stages. Most affected individuals show delay rather than deterioration.
• Growth trajectory is informative: head circumference tracks its centiles more steadily than height does. Hüffmeier 2021: "longitudinal data indicated that head growth had a more constant course along the centiles than height." Four of seven individuals with short stature at last exam had normal length at birth — so the stature phenotype is at least partly postnatally acquired, not purely prenatal.
• Duration: chronic, lifelong.
• Cohort maturity caveat: mean age at last exam 12.1 ± 9.6 years (median 8.5); only four individuals were over 18. The natural history beyond young adulthood is essentially unknown.

Regression — an unresolved question

"Developmental regression or neurodegeneration at various ages (2.5 to ~ 30 years) was observed in three of the 20 affected individuals which might be relevant for prognosis. However, two of the three individuals had additional diagnoses that are not necessarily related to this syndromic disorder" — PMID:33736665

The confounds were vitamin B12-deficiency encephalopathy in one and meningioma-associated psychosis in another. The authors decline to resolve it: "The cohort size and the relatively young ages of the majority of individuals did not allow conclusions as to whether those symptoms are part of the disease spectrum or might have an independent cause." Curate as supports: PARTIAL with the confound stated.

Critical periods

Not defined in humans. In mouse, the critical window for complete loss is peri-implantation: Eif3f^-/- embryos "die at an early stage of development but after the pre-implantation stage" (PMID:31026345). No intervention window has been identified in patients.

9. Inheritance and Population

Inheritance

Autosomal recessive (HP:0000007). All tested parents were heterozygous carriers and essentially unaffected:

"All tested parents were heterozygous carriers of the variant (parents of pedigree 16 (P16) were not available for testing)." — PMID:33736665

"Heterozygous carriers were mainly asymptomatic. One father was reported to have epilepsy, and one mother migraines. Those overlapping symptoms are likely unrelated and of different etiology." — PMID:33736665

Recurrence risk for siblings: 25%, as counselled in the Slovak case (PMID:39723281).

• Penetrance: appears complete for the biallelic genotype in ascertained families. No unaffected homozygote has been reported — though ascertainment bias is total here, and the absence of gnomAD homozygotes is consistent with either full penetrance or embryonic/early attrition.
• Expressivity: markedly variable, with an essentially uniform genotype. This is the disease's defining epidemiological puzzle and no modifier has been found.
• Anticipation: not applicable (not a repeat expansion).
• Germline mosaicism: not reported.

Consanguinity — and why it matters here

Parental consanguinity in only 1/17 (6%) of families in Hüffmeier 2021 and 0/7 in Martin 2018. For a recessive NDD, that is remarkable, and it is a direct consequence of the founder allele being common enough that two unrelated carriers meet by chance. Practical implication: absence of consanguinity is not a reason to deprioritise a recessive hypothesis in a European-ancestry proband.

Founder effect

"This revealed that the EIF3F variant was on an identical haplotype (minimal 7.8 kb) in 15 affected individuals of all tested families, suggesting a founder variant."

The founding haplotype (C-C-A-C-C-G-C across rs79714374, rs12421289, rs12278319, rs7941782, rs4758267, rs12420464, rs56392532) had a frequency of 3.5% in 1,818 internal German control exomes. The variant arose once; "the nascence of the missense variant cannot be determined to a more localized region."

Carrier frequency and derived prevalence

Reported carrier/allele frequencies (gnomAD v2.1.1):

Derived birth prevalence — my own Hardy-Weinberg arithmetic, not a published figure: q² for NFE ≈ (0.0012)² ≈ 1.4 × 10⁻⁶ ≈ 1 in ~690,000; for Ashkenazi Jewish ≈ (0.0021)² ≈ 4.4 × 10⁻⁶ ≈ 1 in ~230,000. Carrier frequency ≈ 1 in 420 (NFE) and 1 in 240 (AJ). Label these clearly as derived; they are almost certainly higher than observed diagnosed prevalence, and the gap between ~31 published patients and this expectation is the size of the under-diagnosis problem.

Epidemiology

• Prevalence: not determined. PMID:39723281 states plainly: "the population frequency of this syndromic disorder has not been determined since then."
• Cases in literature: ~31 as of 2025 (Simons Searchlight), consistent with 9 (Martin) + 22 (Hüffmeier) + scattered singles.
• For a dismech prevalence: block: measure_type: CASES_IN_LITERATURE, prevalence_class: NOT_YET_DOCUMENTED or ULTRA_RARE, with the derived HWE estimate in notes — not in rate_per_100000, since it is an inference rather than an observation.

Demographics

• Ancestry: European and West Asian. Hüffmeier 2021 lists "French, English, Irish, Scottish, German, Bulgarian, Ukranian, Russian, Ashkenazi Jewish, and Iraqi." Plus Slovak (PMID:39723281). No African, East Asian, or South Asian cases reported — consistent with a European founder allele, though ascertainment in European genomics cohorts inflates this.
• Sex ratio: 15 F : 7 M (Hüffmeier) and 5 F : 4 M (Martin) — no evidence of sex bias; the female excess in the larger cohort is unremarkable at this sample size and there is no biological reason to expect one for an autosomal recessive disorder. Report as ~1:1.
• Age distribution: ascertained pediatric (median 8.5 y); adults under-represented.

10. Diagnostics

The diagnostic route is sequencing, full stop

Everything else is negative or nonspecific:

Genetic testing

Recommended approach: WES or WGS, or an ID/ASD gene panel that includes EIF3F. Hüffmeier 2021 is unambiguous: "Thus, genome-wide sequencing approaches (genome or exome sequencing) represent an essential component of the diagnostic work-up." Lakatosova 2024 makes the same argument as a health-systems point for Slovakia.

• EIF3F is on Genomics England PanelApp as a GREEN (high-evidence) gene on the "Early onset or syndromic epilepsy" panel (panel 402), mode of inheritance BIALLELIC, phenotypes: MRT67 (OMIM:618295), intellectual disability, seizures, behavioral abnormality, sensorineural hearing impairment (PanelApp).
• SFARI Gene score: 3 (Suggestive Evidence) for autism, with 8 supporting reports (SFARI).
• Targeted single-variant testing is genuinely viable here — unusually for a recessive NDD. Because one founder allele explains ~95% of reported cases, Sanger or targeted genotyping of c.694T>G is a reasonable low-cost first-line test in a European-ancestry proband with GDD + speech delay + SNHL, and is the natural cascade/carrier test in families.

A diagnostic pitfall worth flagging explicitly

The founder allele sits at 0.12% in non-Finnish Europeans. Many recessive-model variant-filtering pipelines apply a minor-allele-frequency cutoff at or near 0.1% and would discard this variant before a human ever sees it. Combined with the absence of a recognisable facial gestalt and uninformative MRI, this is a plausible mechanism for systematic under-diagnosis. (This inference is mine, but it follows directly from the reported AF and from Hüffmeier's own observation that carrier frequency here "is higher than for most other autosomal recessive NDDs.")

Adjunct clinical assessments (not diagnostic, but indicated once the diagnosis is made)

• Audiology — SNHL in ~57%, and undetected hearing loss in a child with speech delay will be mis-attributed to the developmental delay. This is the highest-yield adjunct test.
• Ophthalmology — refractive error/strabismus/nystagmus/coloboma in up to 38%. Note Hüffmeier's own caveat that hyper-/myopia "might not necessarily be related to EIF3F deficiency, as they are common in the general population."
• Growth monitoring with head circumference plotted longitudinally.
• Sleep assessment (CSHQ used in the Slovak case).
• Developmental/behavioural assessment: ADOS-2, ADI-R, Vineland-3, BRIEF-P, SON-R all used in published assessment.
• EEG if seizures are suspected.

LOINC: no disease-specific analyte. No biomarker, enzyme assay, or metabolite is known. Do not curate one.

Differential diagnosis

Screening

• Newborn screening: not included anywhere; no biochemical marker exists to screen for.
• Carrier screening: EIF3F is not on standard expanded carrier screening panels as far as could be determined. Given the AJ frequency of 0.21%, inclusion on Ashkenazi-focused panels would be a defensible proposal — this is my inference, not a published recommendation.
• Cascade screening: straightforward and appropriate once a familial variant is known.

11. Outcome / Prognosis

Be honest about how thin this is. There is no natural-history study, no survival analysis, and no validated prognostic model.

Survival and mortality

• No deaths attributable to MRT67 have been reported. Individuals have been described up to 48 years (Martin cohort) and ~30 years (Hüffmeier cohort).
• Life expectancy: unknown, but not obviously shortened on current evidence. Do not assert a figure.
• The one hard mortality statement in the literature is about the null genotype, which appears incompatible with life (no human reported; mouse embryonic lethal).

Morbidity and function

Lifelong intellectual disability of variable degree with major communication impairment (24% never develop speech). The single instrumented case showed "deeply below-average" adaptive functioning across all Vineland domains. Most affected individuals will require lifelong support; some achieve simplified but usable language and independent ambulation.

No EQ-5D/SF-36/PROMIS/ICF data exist.

Complications

• Sequelae of SNHL on language acquisition.
• Injury risk from reduced pain sensitivity (a practical safety issue, though no incident data exist).
• Feeding/GI problems in ~24%.
• Surgical burden where clefting or congenital heart disease is present.

The meningioma question — genuinely unresolved

Two of 21 individuals (10%) had meningioma, and a third had psychosis co-occurring with meningioma. In a cohort of median age 8.5 years that is a striking number. Hüffmeier 2021 checked and reported a negative result:

"In large collections of malignancies (COSMIC; accessed on 3rd of December 2020 at cancer.sanger.ac.uk), somatic EIF3F variants have been detected in 0.9% of 38,579 samples (n = 353) and did not include 130 meningioma samples, providing no further evidence for a role of EIF3F in tumorigenesis of meningioma"

Two events in 21 people is not a tumour-predisposition claim, and the somatic data do not support one. But it is also not nothing, and it is the single most consequential open question for surveillance recommendations. Curate as a discussions: KNOWLEDGE_GAP with the COSMIC negative recorded — not as a phenotype with an asserted frequency, and certainly not as a basis for recommending imaging surveillance.

Prognostic factors

The only signal is genotype: the compound heterozygote (missense + frameshift) sat at the severe end for motor milestones, speech, organ involvement, and pre-/postnatal growth. No biomarkers, no imaging predictors.

12. Treatment

There is no disease-specific or disease-modifying therapy.

No approved drug, no gene therapy, no ASO, no cell therapy, no clinical trial. A search of ClinicalTrials.gov concepts for EIF3F/MRT67 surfaces nothing interventional. Say this plainly rather than assembling a speculative pipeline.

Management is supportive, symptom-directed, and multidisciplinary — as practised in the reported cases, where the Slovak child was "under the care of several clinical specialists, including neurologists, otorhinolaryngologists, psychologists, and a speech therapist."

Supportive and rehabilitative care (with NCIT terms)

therapeutic_modality for the therapy/counselling entries: BEHAVIORAL; for surgery: SURGERY; for antiseizure drugs: SMALL_MOLECULE.

Hearing amplification is clinically indicated given ~57% SNHL, but note the dismech-specific constraint: there is no reliable NCIT clinical-action term for device usage (the former "hearing aid usage" term had no NCIT equivalent and was dropped in the MAXO removal), so a hearing-aid treatment entry should carry therapeutic_modality: DEVICE with a free-text preferred_term and no term: binding.

Pharmacogenomics

None. No EIF3F entry in PharmGKB/CPIC relevant to this disorder.

A speculative therapeutic direction — label it as such

The mTORC1 coupling (§6) raises the theoretical question of whether pathway modulation could compensate for reduced eIF3f-dependent S6K1 activation. Nothing supports this in patients. Direction of effect is also non-obvious: the defect is reduced mTOR activation, so an mTOR inhibitor would be expected to worsen it, and there is no validated agonist strategy. Record as a hypothesis in discussions: if at all; do not curate as a treatment.

Antipsychotics — a cautionary observation

In the EIF3F-heterozygous oligogenic case (PMID:35743796, which is not MRT67), both siblings "responded poorly to pharmacotherapy" and one failed ECT. In the one MRT67 individual with psychosis, the psychosis co-occurred with a meningioma. Neither supports a treatment-response claim for MRT67. Mentioned only to forestall over-reading.

13. Prevention

Primary prevention

There is no way to prevent the disease in a conceptus who inherits two pathogenic alleles. Prevention here means reproductive risk management:

• Genetic counseling (NCIT:C15240) with a 25% sibling recurrence risk, as delivered in the Slovak case: "The identification of the causal variant c.694T>G, p.(Phe232Val) in the EIF3F gene indicates a 25% recurrence risk in potential siblings of the patient, due to its autosomal recessive mode of inheritance, as explained by the clinical geneticist to the parents of the proband." (PMID:39723281)
• Carrier testing of at-risk relatives (cascade testing on the familial variant).
• Prenatal diagnosis and preimplantation genetic testing (PGT-M) are technically straightforward once the familial variant is known. No published MRT67-specific experience.
• Population carrier screening: not currently offered. The AJ allele frequency of 0.21% makes a case for consideration on Ashkenazi panels — again, my inference, not a guideline position.

Secondary prevention (early detection and intervention)

The realistic wins are all here, and they follow from the phenotype frequencies:

1. Early audiological assessment — the single highest-value intervention, because untreated SNHL compounds an already-universal speech delay and is present in over half of patients.
2. Early developmental intervention — speech, physical, and occupational therapy.
3. Ophthalmological review — correctable refractive error and strabismus.
4. Growth surveillance with plotted centiles.

Tertiary prevention

• Injury surveillance in individuals with reduced pain sensitivity — a child who does not report pain normally is at risk of unrecognised fracture, burn, or intra-abdominal pathology. This follows from the phenotype and is standard practice in congenital insensitivity to pain; it is not written down anywhere for MRT67.
• Nutritional monitoring where GI symptoms or feeding difficulty are present. Note the individual whose "regression" turned out to involve vitamin B12 deficiency — a reminder that treatable causes should be excluded before attributing deterioration to the underlying disorder.
• Seizure management where epilepsy is present.

Immunization, public health, environmental interventions, prophylaxis

Not applicable. Standard childhood immunisation as for any child; nothing disease-specific.

14. Other Species / Natural Disease

• Taxonomy: Homo sapiens NCBITaxon:9606; Mus musculus NCBITaxon:10090 (experimental only).
• Breed: not applicable — no breed-associated disease. No VBO term.
• Orthologous gene: Eif3f, mouse, MGI:1913335. eIF3 is deeply conserved across eukaryotes; eIF3f orthologues exist throughout metazoa and in yeast the eIF3 core is conserved though subunit composition differs (mammalian eIF3 has 13 subunits including five non-PCI/MPN subunits — Moretti 2010).
• Naturally occurring disease in other species: none known. No OMIA entry was identified for an EIF3F orthologue phenotype. Record as not applicable.
• Veterinary relevance: none.
• Comparative pathology: the mouse does not reproduce the human disease (§15). The comparative story is one of divergence, not convergence — complete loss is embryonic-lethal in mouse and (inferentially) in human, while the human hypomorphic missense allele has no mouse counterpart.
• Evolutionary conservation: Phe232 is "evolutionarily conserved and is predicted to stabilize the protein" (PMID:39723281), which is part of the pathogenicity argument.
• Zoonotic potential / cross-species transmission: not applicable.

15. Model Organisms

Mouse — the only in vivo model, and it does not model the disease

Model 1: Eif3f-null homozygote (Eif3f^-/-) — Docquier et al. 2019, PMID:31026345

"We showed that homozygous eIF3f knockout fail to develop and that eIF3f-/- embryos die at an early stage of development but after the pre-implantation stage."

• Relationship to disease: PERTURBS, fidelity LOW. This genotype exists in no patient and produces a qualitatively different outcome (embryonic lethality, not a neurodevelopmental disorder).
• What it is good for: bounding the mechanism from the severe end. It is the experimental leg of the tripod supporting "the human founder allele must be hypomorphic."
• MGI/IMPC classify this line as lethal (no live homozygous null pups) (MGI:1913335).

Model 2: Eif3f heterozygote (Eif3f^+/-) — same paper

"Although heterozygous mice are phenotypically indistinguishable from wild-type mice, they present a diminished body weight and a lean mass reduction associated with normal body size."

"Skeletal muscles of heterozygous mice show a deficiency in polysome content, a decrease in protein synthesis rate and an inhibition of the mechanistic target of rapamycin (MTOR) pathway."

"Interestingly, skeletal muscles are mainly affected and display an altered cell size without modification of fibre number."

"We found that eIF3f partial depletion amplifies muscle atrophy compared to wild-type mice."

• Relationship to "Reduced Global Translation Initiation": RECAPITULATES, fidelity MODERATE. It reproduces the translational deficit itself and identifies the mTOR coupling.
• Relationship to "Reduced Cell Proliferation and Growth": PARTIALLY_RECAPITULATES, fidelity MODERATE — a cell-size rather than cell-number deficit, with normal body size.
• Limitations (required for any dismech model link): measurements were made in skeletal muscle, not brain; the animal carries a null on one allele, not the human missense on both; heterozygous mice are behaviourally and developmentally normal, so the model captures the biochemistry of partial eIF3f depletion without reproducing the neurodevelopmental phenotype at all. MGI additionally notes partial prenatal lethality in heterozygotes.

Readouts available for curation (ExperimentalReadout): polysome content (DECREASED), protein synthesis rate (DECREASED), mTOR pathway activation (DECREASED), lean mass (DECREASED), muscle fibre cross-sectional area (DECREASED), fibre number (UNCHANGED — a real negative result worth recording).

Cellular models

Gene-edited iPSCs homozygous for c.694T>G — the isogenic model on which the loss-of-function interpretation rests (Martin 2018, described in Hüffmeier 2021):

"In vitro studies of induced pluripotent stem cells (iPSC), gene-edited to be homozygous for the c.694T>G/ p.(Phe232Val) variant, demonstrated lower EIF3F protein levels and reduced proliferation rates"

"Furthermore, both heterozygous and homozygous variants reduced translation rates in iPSC cells"

Quantified via ClinVar submitter summary as ~70% of wild-type protein expression.

• Fidelity: this is the right allele in human cells, which is its major strength — and it is undifferentiated stem cells, which is its major limitation. No neuronal, cerebral-organoid, oligodendrocyte, or cochlear model of the disease allele has been published.

Other cell systems (mechanism, not disease): U2OS, HEK293T, HeLa, and MEF lines with shRNA knockdown for the Notch/DUB work (PMID:21124883); C2C12 myotubes for the mTORC1/MAFbx work; cell lines for SCA8 RAN-translation modulation (PMID:30206144). None carries the disease allele.

The largest model gap

There is no neural model of MRT67. The disease is a brain disorder; the cellular work is in undifferentiated iPSCs and the in vivo work is in mouse muscle. This is precisely the situation the dismech schema's HUMAN_MODEL_MISMATCH discussion kind exists for — evidence exists in models, but its translational validity to the human neurodevelopmental phenotype is the open question. Obvious proposed experiments: differentiate the existing isogenic iPSC lines to cortical neurons and cerebral organoids and measure translation rate, proliferation, and neurite/network phenotypes; generate a Eif3f p.Phe232Val knock-in mouse (which does not currently exist).

Model resources

• MGI: MGI:1913335
• IMPC: mousephenotype.org (note: the search result found no publications using IMPC mice/data for the human-disease question)
• MMRRC / IMSR for line availability
• Simons Searchlight is actively recruiting an EIF3F patient cohort (link) — the most likely source of future natural-history data

Summary of evidence quality

Sources

• OMIM #618295 — Intellectual developmental disorder, autosomal recessive 67; MRT67
• Monarch Initiative — MONDO:0032662
• MedGen — MRT67 (C4749019)
• ClinVar RCV000754608 — NM_003754.3(EIF3F):c.694T>G (p.Phe232Val)
• dbSNP rs141976414
• Martin HC et al., Science 2018 — Quantifying the contribution of recessive coding variation to developmental disorders (PMID:30409806)
• Hüffmeier U et al., Orphanet J Rare Dis 2021 — EIF3F-related neurodevelopmental disorder (PMID:33736665)
• Lakatosova S et al., Cureus 2024 — Genetic Diagnostics and Phenotypic Profiling of a Girl With AR IDD and Autism (PMID:39723281)
• Docquier A et al., J Physiol 2019 — eIF3f depletion impedes mouse embryonic development (PMID:31026345)
• Moretti J et al., PLoS Biol 2010 — eIF3f exhibits a deubiquitinase activity regulating Notch activation (PMID:21124883)
• Csibi A et al., PLoS ONE 2010 — eIF3f controls kinase-dependent mTOR signaling in muscle
• eIF3f: a central regulator of the antagonism atrophy/hypertrophy in skeletal muscle (PMID:23769948)
• Ayhan F et al., EMBO J 2018 — SCA8 RAN polySer accumulates in white matter and is regulated by eIF3F (PMID:30206144)
• Huang YS et al., J Pers Med 2022 — Two Genetic Mechanisms in Two Siblings (PMID:35743796)
• Shad Z et al., Genet Med Open 2024, P319 — Expanding the phenotype of EIF3F-related neurodevelopmental disorder
• Simons Searchlight — EIF3F
• Genomics England PanelApp — EIF3F (Early onset or syndromic epilepsy)
• SFARI Gene — EIF3F
• MGI:1913335 — Eif3f
• Human Protein Atlas — EIF3F tissue expression
• GeneReviews — PPP1R21-Related El-Hattab-Schmidts Syndrome (for disambiguation only)

Recap for the curator

MRT67 is the EIF3F disorder (MONDO:0032662, OMIM #618295) — not PPP1R21/FERRY, despite four PPP1R21 reference files sitting in this worktree's cache; two of those have no retrievable body text and could not yield a verifiable snippet in any case. The disease is driven almost entirely by one European founder missense allele, c.694T>G p.(Phe232Val), on a single 7.8 kb haplotype, present at 0.12% in non-Finnish Europeans and 0.21% in Ashkenazi Jewish individuals with zero gnomAD homozygotes. Mechanism is a ~30% reduction in eIF3f protein → reduced translation initiation → reduced mTOR signalling and proliferation → impaired brain development and somatic growth; the last arrow is inferred rather than measured, since no neural model of the disease allele exists. Global developmental delay and speech delay are universal; SNHL, behavioural problems, altered tone and short stature run near 50%; epilepsy frequency is contradictory between the two cohorts (86% vs 15%) and should not be assigned a frequency band. There is no treatment, no biomarker, no prevalence figure, and no omics dataset. Three items are genuinely open and belong in discussions: rather than as asserted facts: the epilepsy discrepancy, the 2/21 meningioma observation (with its negative COSMIC counter-evidence), and the human-model mismatch between undifferentiated iPSCs / mouse muscle and a brain disease.

Reference Validation

Checked with linkml-reference-validator 0.2.1.

Unresolved references

These identifiers did not resolve to a record and may be fabricated. A lookup that failed for transport reasons is indistinguishable from one that failed because the record does not exist, so spot-check before acting on them:

• DOI:10.1016/j.gimo.2024.101213](https://www.gimopen.org/article/S2949-7744(24)00359-5/fulltext (1 mention) - Identifier did not resolve to a record
• DOI:10.1371/journal.pone.0008994](https://journals.plos.org/plosone/article (1 mention) - Identifier did not resolve to a record