AIFM1-Related Disorders

X-linked allelic spectrum of apoptosis-inducing factor: from severe infantile mitochondrial encephalomyopathy to isolated adolescent auditory neuropathy

Literature and regulatory information updated: 1 November 2026.

0. Initial Reference Table

The AIFM1 gene is a nuclear gene on the X chromosome whose protein works inside the mitochondrion, and the disorders resulting from it fall into four entities with their own OMIM number, plus a set of phenotypes without an independent nosological entry [1, 2]. The following table brings together, for each entity, the identification, diagnostic and management elements, and the prose of the following sections develops each of them.

Besides these four entities, the literature describes phenotypes that have no independent nosological entry and that fall within or lie adjacent to the existing entities: partially riboflavin-responsive childhood cerebellar ataxia, infantile motor neuron disease with a spinal muscular atrophy-like presentation, purely motor distal hereditary motor neuropathy, isolated late-onset axonal polyneuropathy and the X-linked Leigh spectrum [20, 33, 39–41]. The designation “AIFM1-related disorders” describes this reality better than any of the historical names, because assignment of a patient to a specific entity is made a posteriori, from the combination of manifestations and their evolution, and not from the variant found: a laboratory report naming AIFM1 does not, by itself, predict which of the phenotypes the patient will have [12, 24].

1. Introduction

1.1 The founding article and the clinical path before the gene

The clinical history of this group of disorders begins long before the gene was known. In 1985, Cowchock and colleagues described a large Italian-American family in which males of two generations, linked to each other through women, had onset in early childhood of a slowly progressive sensorimotor axonal neuropathy, accompanied by deafness and cognitive impairment, and entitled the work “X-linked motor-sensory neuropathy type II with deafness and mental retardation: a new disorder” [5, 42]. The same family had been reported the previous year by Sladky and Brown as infantile axonal polyneuropathy with X-linked inheritance, and was later included in a study of four families conducted by Fischbeck and colleagues [5]. Female carriers showed minor abnormalities in sensory nerve conduction, electromyography and hearing, but the authors considered them too inconsistent to serve for carrier identification, an observation that later molecular genetics came to put into context [5, 42].

Mapping of the locus occupied the following decade. Fischbeck and colleagues, in 1986, linked the gene to the centromeric region of the X chromosome, between the marker DXYS1, on the proximal long arm, and DXS14, on the proximal short arm, and in 1987 showed that the linkage relationships with phosphoglycerate kinase differed from those of X-linked Charcot-Marie-Tooth disease type 1, proving that it was a genetically distinct entity and not a clinical variant of the same disease [5]. In 1995, Priest and colleagues refined the interval with nineteen microsatellite markers from the long arm of the X in a three-generation family, obtaining as flanking markers DXS425, at Xq24, and the HPRT locus, at Xq26.1, and fixing the candidate interval at Xq24-q26 [5]. The clinical follow-up that accompanied this study gave the first systematic description of the natural history: seven affected males, three obligate female carriers and four unaffected males, with distal weakness, muscle atrophy, sensory loss, areflexia, pes cavus and hammer toes established in the first years of life, deafness in five of the seven males and mental retardation or delayed social development in three of those five [5].

In parallel, and without the link being suspected at the time, a line of research on deafness was developing. Wang and colleagues described, in 2003 and 2006, a five-generation Chinese family with a form of deafness they called familial auditory neuropathy, mapped the locus, designated AUNX1, to Xq23-q27.3, with maximum two-point and multipoint scores of 2.41, delimited it between the markers DXS1220 and DXS8084 and excluded the candidate gene then considered, SLCA14 [7]. In 2008, in a review of sex-linked deafness, Petersen, Wang and Willems noted that the AUNX1 locus overlapped those of CMTX4 and CMTX5, both with deafness and peripheral neuropathy, and proposed replacing the designation AUNX1 by DFNX5, because the disease was not, as the name suggested, a non-syndromic form of auditory neuropathy: patients also develop diffuse peripheral neuropathy [7]. The nomenclature proposal was accepted, but both acronyms still circulate in the literature, and the AUNX1 acronym persists in the names of animal models [7, 43].

1.2 The entry of AIFM1 into human pathology

The protein only reached medicine after it had been characterised in the biology of cell death. In 1999, Susin and colleagues identified and cloned a factor capable of inducing, on its own, apoptosis in isolated nuclei, which they named apoptosis-inducing factor; they cloned the murine homologue, with 92% amino acid identity to the human protein, and found that both full-length proteins contain two mitochondrial localisation sequences and two putative nuclear localisation signals, with the mature murine form weighing 57 kilodaltons and the precursor 67 kilodaltons, with homology to bacterial oxidoreductases [1]. For a decade, interest focused on this apoptotic function, and it took the chance of an Italian family to link it to a disease.

In 2010, Ghezzi and colleagues described two Italian first cousins, sons of monozygotic twin sisters and of unrelated fathers, with an early-onset neurodegenerative disease associated with mitochondrial respiratory chain dysfunction, and identified in them a hemizygous deletion in AIFM1 [3, 44]. This was the first human disease attributed to the gene, and the phenotype — severe X-linked mitochondrial encephalomyopathy — received the designation combined oxidative phosphorylation deficiency type 6, within the phenotypic series of combined oxidative phosphorylation deficiencies [3]. Two years later, in 2012, Rinaldi and colleagues sequenced the exome of a patient from the original Cowchock family and identified the p.Glu493Val variant, closing a twenty-seven-year cycle between the clinical description and its molecular cause [5, 45]. In 2015, Zong and colleagues resolved the AUNX1/DFNX5 locus, identifying five distinct missense variants in five unrelated Chinese families and, by gene screening, hemizygous variants in eleven of ninety-three males with sporadic auditory sensory disorder [7, 46].

The fourth entity followed its own path. In 1999, Bieganski, Dawydzik and Kozlowski described, on the basis of a single family, a spondyloepimetaphyseal dysplasia with mental retardation, probably of X-linked recessive inheritance and unknown molecular defect, which received OMIM number 300232 [19, 47]. In 2006, Neubauer and colleagues characterised, in a different family, a new type of leukoencephalopathy with metaphyseal chondrodysplasia mapping to Xq25-q27, without suspecting the relationship with the Polish family [11, 14]. The synthesis only came in 2017, when two independent groups arrived at the same gene: Mierzewska and colleagues sequenced the exome in Bieganski's original family and in a second Polish family, identified the p.Asp237Gly variant and confirmed segregation by linkage analysis, with a maximum logarithm of the odds score of 3.359 for the two families, while Miyake and colleagues brought together twelve patients from six families and showed that all the variants clustered around exon 7 [11, 19]. Mierzewska and colleagues stressed that spondyloepimetaphyseal dysplasia thereby became the third skeletal phenotype recognised as associated with mitochondrial disease, alongside EVEN-PLUS syndrome, due to HSPA9 variants, and CODAS syndrome, due to LONP1 variants [19].

1.3 Definition of each entity in scope

Combined oxidative phosphorylation deficiency type 6 is defined as a severe X-linked recessive encephalomyopathic disease, with intrauterine onset or onset in the first months of life, in which hypotonia and severe impairment of psychomotor development are associated with variably decreased enzymatic activity of several respiratory complexes in skeletal muscle or fibroblasts; sensorimotor neuropathy, seizures, severe muscle weakness, abnormal signals in the basal ganglia, hypertrophic cardiomyopathy, deafness, swallowing difficulties and respiratory failure are more variable manifestations, and death in childhood is frequent [3].

X-linked recessive Charcot-Marie-Tooth disease type 4, with or without cerebellar ataxia, is defined as a mitochondrial disease manifesting as progressive neurological dysfunction of highly variable expression, with age at onset between infancy and early adulthood; the presentation may be hearing loss, delayed motor development or difficulty walking, due to peripheral neuropathy, cerebellar ataxia or both, and most patients eventually develop all the features, including a progressive sensorimotor axonal neuropathy and an auditory neuropathy-type deafness [5]. Cognitive impairment, cerebellar atrophy on imaging, cerebellar signs — dysarthria, abnormal extraocular movements, tremor and dysmetria — and spasticity are variable, and intrafamilial variability is remarkable, as expected of a mitochondrial dysfunction [5, 25].

X-linked deafness type 5, with peripheral neuropathy, is defined as a neurological disease characterised by childhood onset of auditory neuropathy and later onset of distal sensory impairment of the peripheral nervous system [7]. Auditory neuropathy, in turn, is a type of hearing disorder defined by preserved outer hair cell function, with abnormal or absent auditory brainstem responses and mild to profound hearing loss [7, 48].

X-linked spondyloepimetaphyseal dysplasia with hypomyelinating leukodystrophy is defined by the unusual combination of a spondyloepimetaphyseal dysplasia with a hypomyelinating leukodystrophy, in a picture that sets in between twelve and thirty-six months of life after a normal first year and progresses slowly, through cerebellar, pyramidal and lower motor neuron degeneration, without seizures [11, 24].

The four definitions share the mode of transmission, and the pattern should be stated with the nuance that the literature now requires. The predominant pattern, and the only one recognised in the first descriptions of each entity, is X-linked recessive inheritance, with affected hemizygous males, usually asymptomatic transmitting carrier mothers and de novo variants documented in all entities [3, 5, 7, 11]. An X-linked dominant pattern has, however, been described: in auditory neuropathy, five women in a cohort of fifty were affected, all carriers of the p.(Leu344Phe) variant and with hearing thresholds indistinguishable from those of males, which led the authors to conclude that the gene also produces X-linked dominant inheritance; and in a Moroccan family with the p.(Ser349Gly) variant the pedigree analysis was compatible with that same pattern [17, 18]. Expression in heterozygous women depends on the X-chromosome inactivation pattern, and ranges from mild manifestations to the full encephalomyopathic phenotype [15, 16]. The two patterns are dealt with in detail in section 8, with the consequences for counselling, and taken up again in section 12.

1.4 Nomenclature, reclassifications and the problem of boundaries

The nosology of this group maintains, for historical reasons, a division that the clinical picture does not fully support. The OMIM entry for combined oxidative phosphorylation deficiency type 6 itself refers to CMTX4 as “a less severe disorder with some overlapping features”, and the CMTX4 entry refers to X-linked deafness type 5 as having “overlapping features” and to the combined deficiency as “a more severe disorder” [3, 5]. Clinical observations confirm the continuity: a patient with the p.Thr141Ile variant, classified in the CMTX4 spectrum because of ataxia and deafness, developed cardiomyopathy at ten and a half years, a finding that suggested combined oxidative phosphorylation deficiency type 6; and a patient with the p.Val243Leu variant, initially placed in the combined deficiency, had a protracted course closer to CMTX4 [5, 28]. The review of the neonatal spectrum stated the conclusion explicitly: the unifying feature of all AIFM1 disorders is the presence of axonal neuropathy, with the extent of involvement of the cerebellum, pyramidal tracts, extrapyramidal system and lower motor neurons varying across the spectrum [24].

There are, however, two boundaries of practical value. The first separates X-linked spondyloepimetaphyseal dysplasia with hypomyelinating leukodystrophy from the other entities, and rests on the genotype: all known variants of this entity lie in intron 6 or exon 7, in a narrow region at the end of the first flavin adenine dinucleotide-binding domain, while variants in the same region may also cause combined oxidative phosphorylation deficiency [11, 24]. The second separates X-linked deafness type 5 from the others: the variants associated with the isolated auditory phenotype studied in a Chinese cohort of thirty-six families did not overlap with the variants causing the other syndromes, which suggests that they affect oxidative phosphorylation less severely [14, 17]. Seizures are the only clinical sign with strong discriminating value within the spectrum: they have been reported only in patients with combined oxidative phosphorylation deficiency type 6 [24].

A final clarification concerns Orphanet, which maintains two entries for the same skeletal entity: ORPHA:83629, with OMIM code 300232, and ORPHA:168448, under the Bieganski eponym, with no associated OMIM code or gene [9, 10]. Since it was the family described by Bieganski and colleagues that Mierzewska and colleagues resolved to AIFM1, the two entries designate the same disease, and the second should be read as a synonym of the first [19, 47].

2. Molecular Genetics and Pathophysiology

2.1 The gene and the protein

AIFM1 occupies band Xq26.1, between positions 130,129,361 and 130,165,886 of the X chromosome in the GRCh38 assembly, and encodes apoptosis-inducing factor mitochondria-associated 1 [1, 12]. The MANE Select reference transcript is NM_004208.4, with protein NP_004199.1, confirmed in VariantValidator with the databases vvdb_2025_3 and vvta_2025_02 [13]. The older literature also uses the transcripts NM_004208.3, NM_001130846 and NM_001130847.3, and this diversity explains small numbering differences between publications, so all notation in this article refers to the MANE Select transcript [1, 12, 13]. The external identifiers of the gene are Entrez 9131, Ensembl ENSG00000156709, HGNC:8768 and UniProt O95831, and the aliases in use include AIF, PDCD8, COXPD6, COWCK, CMTX4, DFNX5, NAMSD and AUNX1 [2, 12].

The protein is a flavin adenine dinucleotide (FAD)-dependent oxidoreductase that uses reduced nicotinamide adenine dinucleotide (NADH) as substrate and resides in the mitochondrial intermembrane space [1, 12]. It has at least six functional domains: a mitochondrial targeting sequence at the amino terminus, a hydrophobic transmembrane sequence, three structural domains that coordinate FAD and NADH binding, and a carboxy-terminal motif [14]. The soluble form is known for its role in caspase-independent cell death, but the review dedicated to the gene establishes that human pathophysiology is more associated with impaired mitochondrial bioenergetics than with alteration of apoptotic pathways [14].

2.2 Molecular mechanism: the metabolic function and the apoptotic function

The function that explains the disease is the metabolic one. Apoptosis-inducing factor binds coiled-coil-helix-coiled-coil-helix domain-containing protein 4 (CHCHD4, also known as MIA40) and controls its import into the intermembrane space; CHCHD4, in turn, governs the import and oxidative folding of respiratory chain complex subunits, so a defective apoptosis-inducing factor destabilises assembly of the oxidative phosphorylation system [12, 14, 15]. The measurable consequence in the severe phenotype is a combined deficiency of several respiratory complexes, with predominance of complexes I and IV [12, 24]. The reference handbook of inborn errors of metabolism therefore classifies AIFM1 among the defects of mitochondrial protein import, alongside TIMM8A, with a typical deficiency of complexes I to V [30]. The overview chapter on the Leigh spectrum attributes to it, with a different emphasis, a mechanism of mitophagy dysregulation, grouping it with FBXL4 and VPS13D [20].

The apoptotic function is distinct and follows the cell death stimulus. After permeabilisation of the outer mitochondrial membrane — characteristic of almost all apoptotic pathways — apoptosis-inducing factor is released and translocated to the nucleus, where it mediates chromatin condensation and large-scale degradation of deoxyribonucleic acid (DNA) [1]. Joza and colleagues showed, in 2001, that genetic inactivation of the gene makes embryonic stem cells resistant to death from serum deprivation and that the factor is essential for programmed cell death in the cavitation of embryoid bodies, establishing the existence of a cell death pathway independent of apoptotic protease-activating factor 1 and caspase-9 [1]. Yu and colleagues showed, in 2002, that activation of poly(ADP-ribose) polymerase 1 is required for translocation of the factor to the nucleus and that the factor is required for cell death dependent on this polymerase, independently of caspases; work in 2006 identified the poly(ADP-ribose) polymer as the mediator of this death, now called parthanatos, and showed that poly(ADP-ribose) glycohydrolase prevents release of the factor [1]. In Caenorhabditis elegans, the homologue WAH-1 localises to the mitochondrion, is released through the action of the BH3-domain protein EGL1 in a CED3 caspase-dependent manner and cooperates with the endonuclease CPS6/EndoG in DNA degradation, defining a conserved pathway [1].

Several studies document both mechanisms in the same patient. In the original Cowchock family, muscle biopsy showed an increased number of apoptotic cells compared with controls, suggesting activation of the caspase-independent cell death pathway, while the recombinant p.Glu493Val protein altered the redox state without affecting respiratory complex activity [5, 45]. The dedicated review explicitly allows that a variable combination of oxidative phosphorylation failure and increased susceptibility to caspase-independent apoptosis may underlie the broad spectrum of presentations [14, 24].

2.3 Functional studies by variant

Deletion of arginine residue 201 was the first variant studied in detail. The residue is part of a hairpin that forms the FAD-binding pocket and confers conformational stability on the flavoprotein; later structural predictions show that it forms hydrogen bonds with glutamate 531 and that its loss may compromise folding of the regulatory peptide between alanine 509 and aspartate 559 [1, 14]. The mutant protein has a shorter lifetime of the FADH2-NAD complex, increased susceptibility to proteolytic cleavage, increased DNA-binding affinity and greater potential to cause DNA damage; about 75% of mutant cells, versus 23% of control cells, showed mitochondrial fragmentation under galactose treatment, associated with deficient oxidative phosphorylation [1, 44].

The p.Glu493Val variant has a lower Michaelis-Menten constant for NADH than the wild-type protein and is reduced by NADH four times faster; the charge-transfer complex has a shorter half-life, but the protein retains the ability to dimerise in the presence of NADH [1, 45]. The p.Gly262Ser, p.Gly338Glu and p.Val243Leu variants reduce the amount of protein, through instability [1]. The p.Pro169Leu variant, studied in fibroblasts from a family with three affected males, reduced steady-state levels of the protein and the abundance of complex I subunits, and increased sensitivity to staurosporine and nuclear condensation in galactose medium; in that case, the deficiency was not corrected by riboflavin or protease inhibitors [49].

The p.Glu336Lys variant, characterised in 2026, offers the most complete analysis of a single allele. The patient's fibroblasts showed reduced protein stability with preserved messenger ribonucleic acid (mRNA) expression, impaired growth under oxidative phosphorylation-dependent conditions, decreased basal respiration and altered assembly of respiratory supercomplexes, with reduced CHCHD4 levels and lower mitochondrial content; the purified protein showed impaired FAD retention, decreased thermal stability, impaired NADH affinity, destabilisation of the charge-transfer complex that sustains the interaction with CHCHD4 and a shift in coenzyme preference towards reduced nicotinamide adenine dinucleotide phosphate (NADPH) [50]. Substitution of glutamate 336 by lysine reshapes the electrostatic environment of the NADH-binding cleft, compromising redox function and weakening binding to CHCHD4; DNA binding, nuclease activity and binding to nuclear partners were maintained, although induction of parthanatos was attenuated in the patient's fibroblasts [50].

The mechanism of the auditory variants has been studied in neurons derived from induced pluripotent stem cells. The c.1265G>A variant generates an aberrant transcript, with loss of nucleotides c.1267 to c.1305, which produces the proteins p.Arg422Gln and p.423-435del and impairs dimerisation of apoptosis-inducing factor; impaired dimerisation weakens the interaction with CHCHD4, which on the one hand inhibits mitochondrial import of electron transport chain subunits, with an increased adenosine diphosphate to triphosphate ratio and increased reactive oxygen species levels, and on the other impairs heterodimerisation of the mitochondrial calcium uptake proteins MICU1 and MICU2, with mitochondrial calcium overload [51]. Calcium activates calpain, which cleaves the factor and allows its nuclear translocation, with caspase-independent apoptosis; correction of the variant by clustered regularly interspaced short palindromic repeats editing restored the structure and function of the protein and improved the physiological state of the neurons [51]. In cell lines with the gene inactivated and re-expression of mutant alleles, mean dimerisation was 34.5 to 49.7% of that obtained with the wild-type allele, and the mean percentage of apoptotic cells was 12.3 to 17.9% with the p.Thr260Ala, p.Arg422Trp and p.Arg451Gln variants, versus 6.9 to 7.4% in controls; NADH treatment reduced apoptosis by rescuing dimerisation [52]. A distinct variant, p.Phe210Leu, causes abnormal assembly of complexes I and III without altering the levels of apoptosis-inducing factor or MIA40, which indicates that the mechanism does not always involve quantitative deficiency of either protein [41].

The c.5T>C p.(Phe2Ser) variant, the only one described in the mitochondrial targeting sequence, illustrates a third mechanism. Studies showed no altered processing or mislocalisation of the protein, but the steady-state level was reduced, which suggests a negative impact on mitochondrial import with degradation of the cytosolic precursor; since the mature protein present in the mitochondrion, lacking the targeting sequence where the variant lies, is identical to the normal one, the reduced amount — and not an altered property — is responsible for the phenotype [24].

2.4 Variant spectrum

The variants described are, in the great majority, missense, and all pathogenic variants are assumed to be loss-of-function [14, 53]. A ClinVar query on 1 November 2026 returned 980 records for the gene, of which 778 single nucleotide variants; with the properties filter, 181 were classified as pathogenic and 259 as of uncertain significance [54]. The following table brings together the variants with a published phenotype, normalised in VariantValidator against the MANE Select transcript, with the GRCh38 genomic coordinate in NC_000023.11.

The list of variants associated with the auditory phenotype is longer than the table can accommodate and includes, besides those mentioned, the substitutions p.Ile304Met, p.Leu333Phe, p.Ser349Gly (c.1045A>G), p.Gly360Arg (c.1078G>C), p.Arg430Cys (c.1288C>T), p.Ala465Val (c.1394C>T), p.Ala472Val (c.1415C>T), p.Pro475Leu (c.1424C>T), p.Val498Met (c.1492G>A), p.Tyr560His (c.1678T>C) and p.Ile591Met (c.1773C>G) [13, 14, 18, 54]. Two notes limit the reading of this list. The first concerns nomenclature: the review gives only the protein designation, and for p.Ile304Met and p.Leu333Phe there is no ClinVar record from which the coding DNA notation can be inferred, so they remain as published protein-level designations and are not assigned a c. notation in this article; for the others, the c. notation was obtained from ClinVar and confirmed in VariantValidator [13, 54]. The second concerns classification: p.Arg430Cys and p.Val498Met appear in ClinVar as of uncertain significance, with no auditory phenotype assigned, so their grouping in the review should not be read as a demonstration of causality in the auditory phenotype [14, 54]. In a Chinese cohort of thirty-six families, eighteen variants were associated with the auditory neuropathy phenotype, nine located in the FAD domain, six in the NADH domain and four in the carboxy-terminal region, without overlap with the variants of the other syndromes [17]. In the series of Zong and colleagues, of the eleven missense variants identified, most occurred in the NADH domains and in the second FAD domain, essential for FAD-dependent oxidoreductase activity [7, 46].

The historical notation deserves a technical note. The variant originally described as a “3 base pair deletion at 601AGA” is automapped by VariantValidator, by applying the 3'-most position rule of the sequence, to c.603_605del, with the same protein consequence p.(Arg201del); both notations are found in the literature, and the second is correct [1, 13].

2.5 Genotype–phenotype correlations, penetrance and expressivity

Genotype–phenotype correlations are poorly defined. Variants described in a specific clinical presentation may affect different domains, and variants in the same region of the protein — for example, in the FAD-binding domain — may cause different phenotypes, combined oxidative phosphorylation deficiency type 6 or spondyloepimetaphyseal dysplasia with hypomyelinating leukodystrophy [24]. The two exceptions already mentioned are the restriction of the skeletal entity's variants to intron 6 and exon 7 and the lack of overlap between the variants of the auditory phenotype and those of the others [11, 17]. A mechanistic hypothesis proposes that distinct substitutions in the same region exert a variable effect on the oxidised and reduced states of the protein, affecting not only redox properties but also stability and conformational transitions between the monomeric and dimeric forms; in addition, the degree of reduction of the protein is itself a possible modulator of the spectrum and severity [24].

Intrafamilial variability is a constant feature and, from the counselling point of view, the most important. In the Irish family with seven males carrying the same p.Met340Thr variant, six presented with hearing loss between fourteen and thirty-four years, followed by gait ataxia a few years later, whereas the seventh had onset of hearing loss at eighteen months and of gait abnormalities at four years, with marked cerebellar signs and frank cerebellar atrophy [25]. The same variant, in the patient described by Heimer and colleagues, produced a different phenotype, which the authors of the Irish series commented on as expected in diseases affecting mitochondrial function [5, 25]. In the two brothers with the p.Ile433Met variant, one walked with a cane at seventeen years, with normal brain magnetic resonance imaging, while the other, younger, needed a wheelchair and non-invasive ventilation, with cerebellar atrophy on imaging [59]. In the cumulative review of the twelve males published up to 2015, three died before the age of five while others were still walking in early adulthood, and the course could alternate rapid deterioration with long plateaux [28].

Expressivity in heterozygous women depends on X-chromosome inactivation, and is dealt with in detail in section 8. The molecular principle should be established here: the women described with a phenotype had strongly skewed X inactivation, 98 to 2 in the case of the girl with the full phenotype, and highly skewed in the mother with mild cerebellar ataxia and intellectual disability [15, 16].

2.6 Allelic disorders, modifiers and confounding factors

All the entities in this article are allelic to one another, by definition: they result from the same gene, and the OMIM entry for each explicitly refers to the others [3, 5, 7]. For AIFM1, there are no identified genetic modifiers, nor any published study of skewed X inactivation as a modifier in carriers, although the cases described imply it [12, 15]. The family with the p.Gly399Ser variant documented, on the other hand, the confounding effect of non-genetic factors on the phenotype: alcohol and drug use and medication side effects contributed to the neurological picture observed, which calls for caution in attributing all signs to the variant [16]. An additional example is instructive as to superimposed pharmacological risk: the carrier mother of a patient with p.Met340Thr developed severe axonal neuropathy at fifty years, after vincristine treatment for breast cancer [5, 33].

2.7 Animal models

Mouse models support the loss-of-function hypothesis. The constitutive knockout is embryonic lethal, so the initial model was the hypomorphic Harlequin mutant, in which a proviral insertion reduces expression by up to about 80%, variably between tissues; it was proposed as a model of complex I deficiency [1, 14]. These animals develop skeletal muscle atrophy and progressive degeneration of the retina and cerebellum, phenotypically evident between three and four months of age, with glial fibrillary acidic protein-positive astrogliosis in the thalamus and cerebellum as early as two months and extensive microglial activation at three to four months, concomitant with neurodegeneration of Purkinje cells and granular layer neurons [1, 14]. Mutant cerebellar granule cells are susceptible to peroxide-mediated apoptosis and can be rescued by expression of the gene, and overexpression in wild-type cells further reduces that death, which led to the proposal that the factor functions as a free radical scavenger [1].

Inactivation in the early embryo does not affect cavitation or the apoptosis associated with neural tube closure, but from embryonic day 9 causes abnormal cell death, presumably through reduced complex I activity; null embryos do not increase significantly in size after that day, while still maintaining the somite number of their wild-type siblings, which showed that pattern formation can occur independently of embryo size and cell number [1].

Two models targeting human variants are particularly informative. The knock-in of the murine arginine 200 deletion, corresponding to human arginine 201, develops, within a few months, severe myopathy with a reduced number of cytochrome c oxidase-positive fibres, through post-translational loss of the mutant protein, and manifests peripheral neuropathy later, without the cerebellar neurodegenerative processes observed in Harlequin mutants of equivalent age; quantitative proteomics showed molecular signatures common to mitochondrial diseases, including aberrant folate-dependent one-carbon metabolism and sustained signalling through the protein kinase B and mammalian target of rapamycin pathway [14]. The murine p.Arg450Gln knock-in, corresponding to human p.Arg451Gln, reproduces X-linked auditory neuropathy: hemizygous males develop progressive hearing loss from postnatal day 30, more severe at day 60 and stable until day 210, with muscle atrophy by that date, and histology shows a gradual reduction in the number of spiral ganglion neurons at day 30 and of ribbon synapses at day 60, coinciding with nuclear translocation of the factor from days 21 and 30 [43]. At day 210, spiral ganglion neurons lose cell membrane integrity, with abnormal nuclear morphology and dendritic and axonal demyelination, and the mitochondria of the inner hair cells and the myelin sheath are morphologically abnormal [43]. Muscle-specific loss of the gene, in another model, causes severe dilated cardiomyopathy, heart failure and skeletal muscle atrophy with lactic acidaemia, which provides the preclinical basis for the human cardiac phenotype; in the Harlequin mutant, decreased expression in cardiac tissue makes the myocardium more susceptible to oxidative stress and pressure overload [15].

3. Clinical Spectrum and Manifestations by System

3.1 Overview of the spectrum

AIFM1-related disorders are distributed along a continuum whose severity decreases with age at onset. At the most severe end are fetuses and newborns with mitochondrial encephalomyopathy, in whom ventriculomegaly may be detected at thirteen or fourteen weeks of gestation and death occurs in the first years of life; at the mildest end are adolescents and adults with isolated auditory neuropathy or late-onset axonal polyneuropathy, with normal cognition and unreduced survival [7, 27, 41]. The element common to the whole spectrum is axonal neuropathy, while the extent of involvement of the cerebellum, pyramidal tracts, extrapyramidal system and lower motor neurons varies [24]. Respiratory dysfunction is predominant, with variable age at onset, in both combined oxidative phosphorylation deficiency type 6 and spondyloepimetaphyseal dysplasia with hypomyelinating leukodystrophy, and seizures have been reported only in the former [24].

The 2015 review, which brought together the twelve males then published with variants in this gene, identified as the unifying picture of the spectrum the association of progressive myopathy, ataxia and neuropathy, with highly variable severity even within the same family: three of these twelve died before the age of five, while others were still walking in early adulthood, and the course could alternate rapid deterioration with long plateaux [28]. This proportion is not a mortality rate for the spectrum and should not be used as such: the twelve males are the cumulative total of the literature as of 2015, assembled by review of published cases, and not a cohort with defined follow-up; the denominator has, moreover, been superseded by later descriptions, which extended the spectrum to the mild auditory and neuropathic forms, in which survival is not compromised [17, 28, 41]. The proportion therefore describes the severity of the cases published up to 2015, with the ascertainment bias that favours severe forms, and not the risk of a patient diagnosed today. Hearing loss, seizures and psychomotor regression were, in that review, less frequent than the triad of myopathy, ataxia and neuropathy [12].

3.2 Combined oxidative phosphorylation deficiency type 6

3.2.1 Prenatal and neonatal presentation

The earliest documented presentation is ultrasonographic. In the two infant brothers of unrelated Palestinian parents described by Berger and colleagues, ultrasound at thirteen to fourteen weeks of gestation showed enlarged ventricles, enlarged cisterna magna and bilateral septated choroid plexus cysts; at birth, both had marked hypotonia with few spontaneous movements and swallowing difficulties, required tube feeding and occasional mechanical ventilation, and developed controllable neonatal seizures [3, 27]. In one of them, brain imaging at two months showed T2-hyperintense lesions in the basal ganglia; this patient made no developmental progress and died at four years of recurrent respiratory failure, while his brother died at three months of aspiration pneumonia and hypertrophic cardiomyopathy [3, 27]. Ultrasound in a subsequent pregnancy showed similar abnormal findings in a male fetus [3].

Neonatal presentation with seizures is a recognised picture and, in 2023, its extreme was described. A newborn with the c.5T>C p.(Phe2Ser) variant had drug-resistant multifocal electroclinical seizures six hours after birth, and magnetic resonance imaging showed prominent cerebral oedema of both hemispheres and generalised signal change in much of the cortex and in the thalami, with sparing of the basal nuclei [24]. This was the first case associated with neonatal seizures and diffuse white matter involvement with relative sparing of the basal ganglia, in the absence of signs suggestive of myopathy or motor neuron disease; fibroblasts showed a clear reduction of the protein and deficient activities of complexes I, III and IV [24]. At six months, after supplementation with riboflavin, coenzyme Q10 and thiamine, the infant had microcephaly but had not suffered further deterioration, was still fed orally and had no muscle weakness or atrophy, a course that contrasts with the rapid progression usual in cases presenting in the first year of life [24].

3.2.2 Presentation in the first year of life and regression

In the two Italian cousins described by Ghezzi and colleagues, development was normal in the first months, and psychomotor delay was noted at five and eleven months respectively [3, 44]. The first showed decreased spontaneous movement of the right lower limb and continuous involuntary movements of the hands and feet; magnetic resonance imaging at one year showed abnormal signals in the basal ganglia and, at fifteen months, he could no longer stand, with tongue fasciculations, hypotonia and areflexia [3, 44]. Electromyography and nerve conduction studies documented sensory and motor axonal neuropathy, lactate and pyruvate were moderately increased in plasma and cerebrospinal fluid, muscle biopsy showed mitochondrial deoxyribonucleic acid copy number reduced to 20% of control and fibroblasts had complexes I, II, III and IV at between 51 and 73% of normal values [44]. Psychomotor regression continued, with seizures and loss of contact with the environment, and at five years the patient was tetraplegic, in a wheelchair, unable to communicate and in need of permanent artificial ventilation [3, 44]. The second cousin followed a similar course, with muscle biopsy showing ragged-red fibres, mitochondrial deoxyribonucleic acid depletion to 35% of controls and complex I to V activities between 11 and 63% of normal; he died suddenly during sleep at sixteen months [3, 44].

A 2017 report added the pathological characterisation of this presentation. An infant with a new variant had mitochondrial disease of unusually early onset, rapid deterioration and death, and autopsy at four months revealed mitochondrial encephalopathy, myopathy and peripheral nerve involvement with axonal degeneration, together with hepatic microvesicular steatosis, absence of thymic involution, follicular bronchiolitis and medial hypertrophy of the pulmonary arteries, findings that broaden the pathological spectrum attributable to the gene [61].

3.2.3 Infantile motor neuron disease phenotype

Diodato and colleagues described two Italian first cousins, sons of sisters, whose picture began shortly after birth with marked hypotonia, proximal muscle weakness and rapidly progressive limb involvement with absent reflexes [3, 39]. Progressive swallowing difficulties and respiratory failure requiring tube feeding and tracheostomy followed, with severely impaired global development with poor eye contact, abnormal visual and brainstem evoked potentials, continuous involuntary movements of the hands and feet, multifocal epileptic discharges on the electroencephalogram and signs of motor axonal neuropathy [39]. Brain imaging showed frontotemporal atrophy with enlarged ventricles and a lactate peak on spectroscopy, and muscle biopsy revealed neurogenic changes, with groups of atrophic and hypertrophic fibres, cytochrome c oxidase deficiency and decreased enzyme activity of that enzyme, with the other activities within the control range [39]. The authors stressed that the phenotype closely resembled spinal muscular atrophy, which suggests simultaneous neuronal injury in the cerebral cortex and spinal cord [3, 39].

3.2.4 Protracted form

The case described by Kettwig and colleagues delimits the other end of the entity. An eleven-year-old boy, with normal early psychomotor development, presented at two and a half years with rapidly progressive hearing loss which, within three weeks, resulted in deafness and absence of speech; severe rapidly progressive ataxia followed, with inability to walk or stand [3, 28]. Muscle weakness and atrophy, hyporeflexia, a single seizure, ptosis, external ophthalmoplegia and swallowing and respiratory difficulties requiring tube feeding and tracheostomy were added; the patient communicated by sign language and his mother reported no cognitive impairment [3, 28]. Laboratory studies were normal, except for nerve conduction, compatible with sensorimotor axonal neuropathy; imaging showed mild atrophy of the cerebellar vermis and non-specific periventricular and occipital white matter changes, and muscle biopsy showed non-specific myopathic changes with type II fibre hypotrophy, without evident ragged-red or cytochrome c oxidase-negative fibres, although biochemical and western blot analysis indicated mild decreases in complexes I, II and III [28]. The authors noted the episodic rapid deterioration, common in mitochondrial diseases [3, 28].

3.2.5 Cardiac involvement

Cardiac involvement is now a recognised manifestation of the spectrum, although it has been described in few patients. Before 2026, it was documented in four male children: an infant with the p.Gly308Glu variant who died at three months of hypertrophic cardiomyopathy and aspiration pneumonia; a ten-and-a-half-year-old boy with the p.Thr141Ile variant and mild left ventricular dilatation with borderline systolic function; and two related patients with the p.Pro169Leu variant, one with biventricular myocardial hypertrophy detected on fetal echocardiography and the other who died at thirty-nine days of life, with biventricular cardiac hypertrophy at autopsy [15]. All four had neurological manifestations and three of the four had severe multisystem disease with onset in infancy [15]. The case published in 2026 added longitudinal follow-up: left ventricular hypertrophy at eight months, with z-scores of +3.5 for the interventricular septum and +3.6 for the posterior wall and preserved systolic function, ventricular dilatation at around eighteen months with an ejection fraction of 51.2%, and criteria for paediatric dilated cardiomyopathy at two years and six months, with a diastolic diameter z-score of +4.4, fractional shortening of 16.9% and ejection fraction of 35.3% [15]. The hypertrophy–dilatation–heart failure sequence is thus the expected evolution of the cardiac phenotype, and is partially reversible with heart failure therapy [15].

3.3 X-linked recessive Charcot-Marie-Tooth disease type 4

3.3.1 The original family and the classical phenotype

In the original family, described in 1985 and followed up in 1995 and 2012, the seven affected males were severely affected in the first years of life, with distal weakness, muscle atrophy, sensory loss, areflexia, pes cavus and hammer toes [5, 42]. Deafness accompanied five of the seven, and three of those five also had mental retardation or delayed social development; motor conduction velocities were normal to mildly slowed and sensory conduction markedly abnormal, a pattern that confirms the axonal nature of the neuropathy [5]. The 2012 follow-up added that distal muscle weakness and atrophy and sensory loss affected the lower limbs more than the upper, that serum transaminases and serum creatine kinase were increased, that muscle biopsy showed neurogenic atrophy and that electron microscopy revealed an increased number of abnormally shaped mitochondria in subsarcolemmal areas; brain magnetic resonance imaging in two patients showed T2 hyperintensities in the supratentorial white matter [5, 45].

3.3.2 The cerebellar phenotype

From 2015, the entity was redefined to include cerebellar ataxia explicitly, and the OMIM title came to contain the expression “with or without cerebellar ataxia” [5, 56]. The patient who prompted this redefinition, a thirty-nine-year-old man, had normal development until about one year, when walking difficulties appeared; in childhood he developed limb and gait ataxia, hearing loss and cognitive impairment, and in the second decade visual deficits, distal muscle weakness and atrophy and cerebellar dysfunction with dysarthria, head and trunk titubation and intention tremor were added, with wheelchair dependence at twenty years [5, 56]. Imaging showed progressive cerebellar atrophy, mild cortical and thalamic atrophy and, later, brainstem and spinal cord involvement; nerve conduction documented a mainly sensory axonal neuropathy, and there was optic nerve atrophy and retinopathy with worsening visual loss [56]. Muscle biopsy showed chronic denervation, ragged-red fibres, decreased immunostaining for cytochrome c oxidase and a marked reduction in complex IV activity, with smaller reductions in the other respiratory enzymes; the authors concluded that the ataxia resulted from both cerebellar atrophy and sensory neuropathy, a distinction with consequences for rehabilitation [5, 56].

The 2019 Irish series established the natural history of this form in a single family. Of the seven affected males, six presented with hearing loss between fourteen and thirty-four years, followed by walking difficulties and gait ataxia a few years later, although two had abnormal gait since childhood; the seventh had a more severe course, with hearing loss at eighteen months and gait abnormalities at four years, significant cerebellar signs — slow saccades, broken smooth pursuit, cerebellar dysarthria, dysdiadochokinesia and incoordination — and frank cerebellar atrophy on imaging [25]. Of the others, only one had dysarthria and another slow broken pursuit, associated with subtle cerebellar atrophy, which confirms that vermian atrophy may develop over time and not be evident at the first assessment [24, 25]. Other common, though variable, features were tinnitus, pes cavus, hammer toes, distal muscle weakness, atrophy and sensory impairment mainly of the lower limbs, and hyporeflexia of all four limbs; four patients had spasticity, generally with extensor plantar responses, and electrophysiological studies showed length-dependent sensorimotor axonal neuropathy with predominant radial nerve involvement [25]. Three patients had colour blindness in the absence of variants in the genes classically associated with that trait, a finding the subsequent literature has retained as part of the spectrum [24, 25].

3.3.3 The movement disorder phenotype

Heimer and colleagues described, in 2018, two patients whose phenotype shifted the clinical emphasis to movement disorder. The first, a seventeen-year-old boy, had hearing loss at five years, with auditory brainstem responses compatible with auditory nerve neuropathy; at eight years he developed gait ataxia and hyporeflexia, progressing to wide-based gait, dysmetria, dysarthria, tremor, bradykinesia and pes cavus, with cerebellar vermian atrophy on imaging and sensorimotor axonal neuropathy on nerve conduction [5, 33]. The second, an eleven-year-old boy, had mild hypotonia with gross motor delay and poor speech in early childhood and, at four years, progressive and severe myoclonus of the limbs, face, larynx and palate, unresponsive to antiepileptic drugs and without abnormalities on the electroencephalogram, so the authors classified it as a movement disorder of possible brainstem origin; cerebellar ataxia and ophthalmoplegia with cerebellar atrophy on imaging, hearing loss with decreased evoked potentials and decreased tendon reflexes followed, and at ten and a half years cardiomyopathy appeared [5, 33]. Both had normal intellect, which shows that cognitive impairment, present in the historical name of the disease, is not obligatory [33]. Tremor may reach disabling severity and drug resistance, as in the two brothers with the p.His457Tyr variant who prompted the use of deep brain stimulation [34].

3.3.4 Purely neuropathic forms

At the mildest end, AIFM1 causes neuropathy without central involvement. A family with the p.Phe210Leu variant developed isolated late-onset axonal polyneuropathy, sparing the central nervous system and other organs [41]. Two brothers with the p.Phe210Ser variant had axonal polyneuropathy with onset in early childhood with exclusive involvement of motor fibres, that is, a distal hereditary motor neuropathy, so the authors recommended including AIFM1 in the diagnosis of these neuropathies [40]. A Chinese family with the p.Met171Ile variant had isolated motor and sensory axonal neuropathy, of childhood onset and slowly progressive, in which leg magnetic resonance imaging revealed fatty infiltration and atrophy with severe involvement of the peroneal compartment muscles, a pattern the authors proposed as an early feature of the entity; pathology showed abnormal mitochondrial morphology and accumulation of mitochondria in the axoplasm of the nerve fibre and in the subsarcolemmal area of the muscle [55].

3.4 X-linked deafness type 5 and the auditory phenotype

3.4.1 Audiological characterisation

The five-generation Chinese family that gave rise to the locus had type I auditory neuropathy, with primary degeneration of auditory nerve axons, accompanied by late-onset peripheral neuropathy [7]. Neurological examination showed diffuse peripheral sensory neuropathy and, on nerve conduction assessment, the sensory potential could not be elicited in the median, ulnar, peroneal, tibial and sural nerves bilaterally in one patient, in the left median and ulnar in a second and in the right median and bilateral sural nerves in a third, with normal motor conduction in all three tested [7].

The series of Zong and colleagues described fourteen patients from four Chinese families with childhood-onset auditory neuropathy and late peripheral sensory neuropathy, manifesting as numbness of the extremities, unsteadiness and areflexia [7, 46]. Electrophysiological studies showed reduced or absent sensory conduction velocities and sensory action potential amplitudes in the sural, median and ulnar nerves, with normal motor results; hearing loss and sensory neuropathy were slowly progressive, imaging of the internal auditory canals showed bilateral cochlear nerve hypoplasia, all patients had normal cognitive function and none had muscle weakness or atrophy [7, 46]. Eleven Chinese patients with sporadic occurrence and a similar phenotype were also described, although not all had symptoms of peripheral neuropathy [7].

3.4.2 The 2020 Chinese cohort and the detailed clinical profile

The most complete cohort of the auditory phenotype brought together thirty-six families and fifty patients with variants in the gene, thirty previously studied and twenty new [17]. Forty-five were male and five female; none had a history of hyperbilirubinaemia or hypoxia, nine (18%) had a history of ototoxic drug use and one of noise exposure [17]. Thirty-seven patients (74%) reported tinnitus at the first visit, numbness of the extremities was the second most common symptom, in twelve patients, ten had visual disturbance and eight vertigo [17]. In the fifteen patients assessed with cervical vestibular evoked myogenic potentials, nineteen of thirty ears (63.3%) showed ipsilateral saccular dysfunction; in nineteen patients who underwent vestibular testing, fifteen of thirty-eight sides (39.5%) had abnormalities [17]. Computed tomography of the temporal bone was normal in all, and magnetic resonance imaging showed bilateral cochlear nerve hypoplasia in three of seven patients studied (42.9%) [17].

Age at onset ranged from five to twenty years, with a mean of 13.4 ± 3.9 years, and the earlier the onset the more severe the hearing impairment: the mean pure-tone threshold between 250 and 1000 hertz was 55.7 decibels hearing level in the group under twelve years, 51.2 in the twelve to sixteen years group and 47.7 in the group over sixteen years, although the differences did not reach statistical significance [17]. Disease duration at the first visit ranged from zero to thirty-three years, with a mean of 9.7 ± 7.7 years, and thresholds did not differ significantly between the groups with duration under five years, five to fifteen years and over fifteen years [17]. Of the twenty new cases, eighteen were sporadic, only one did not report onset in childhood and, with the exception of one woman with unilateral auditory neuropathy, all had bilateral disease, with a rising audiogram configuration in thirty-one of thirty-seven ears (83.8%) [17].

3.4.3 Auditory progression over decades

Follow-up of this cohort, with sixteen patients (53.3%) undergoing repeated pure-tone audiometry over one to twenty-three years, with a mean of 9.75 ± 9.89 years, showed a pattern of slow and uneven progression [17]. With short follow-up, of one to ten years, deterioration was not apparent at any frequency; with follow-up of more than ten years, thresholds at the low frequencies, 250 and 500 hertz, and at the high frequencies, 4 and 8 kilohertz, increased significantly [17]. The finding with the greatest practical consequence concerns speech discrimination: half of the patients with mild hearing loss had a speech discrimination score of zero, a proportion much higher than in the groups with moderate and severe loss, so the decline in discrimination is not proportional to the pure-tone threshold and is, paradoxically, more severe in patients with mild pure-tone loss [17]. Disease duration is a determining factor: none of the nine patients with a duration under five years had a discrimination score of zero, whereas above five years that proportion rose to half of the ears assessed [17].

Distortion product otoacoustic emissions remained obtainable in twenty-nine of the thirty patients assessed, with at least five frequencies, and disappeared only in one patient with thirty-three years of disease duration, which indicates that auditory neuropathy may progress to sensorineural loss with outer hair cell damage over decades [17]. On electrocochleography in twenty-two patients, nine of forty-four ears (20.5%) had only the negative summating potential wave, without an evident compound action potential wave, and the degree of hearing loss in these patients was significantly greater [17].

The largest retrospective series of auditory neuropathy, with six hundred and forty-five patients diagnosed at a single Chinese centre between January 1997 and December 2023, confirmed the comparative profile by gene: carriers of AIFM1 variants had later onset and better hearing thresholds than carriers of OTOF and ATP1A3 variants [32]. In that series, auditory brainstem responses remained stably absent or abnormal on follow-up, distortion product otoacoustic emissions gradually disappeared, the cochlear microphonic remained stable, obtained in 96.36 to 100% of patients at all ages, and poor speech discrimination persisted [32]. A relatively large proportion of patients had hearing loss predominantly at the low frequencies [32].

3.5 Spondyloepimetaphyseal dysplasia with hypomyelinating leukodystrophy

The reference series brought together twelve patients from six families, who underwent full assessment and exome sequencing [11]. Presentation typically occurred between twelve and thirty-six months, after a normal first year of life, and rested on two defining features: spondylometaphyseal dysplasia and hypomyelination on magnetic resonance imaging [11]. Common signs and symptoms included motor deterioration, spasticity, tremor, ataxia, dysarthria, cognitive deficits, pulmonary hypertension, nystagmus and visual loss due to retinopathy, and the course was slowly progressive [11]. The complementary description adds that cerebellar signs — nystagmus and truncal ataxia — predominate at presentation, that cognitive impairment is variable but significant disability results mainly from cerebellar, pyramidal and lower motor neuron degeneration, with severe dysarthria, spasticity, muscle weakness and respiratory failure, and that progressive skeletal deformities, such as scoliosis, probably contribute to the severe respiratory phenotype [24]. Visual loss may be associated with macular degeneration, achromatopsia and optic nerve atrophy, no seizures have been reported in these patients and early death may occur in the second decade of life [24]. The skeletal phenotype includes short stature and large, deformed joints, with motor deficit and visual defects appearing after the normal first year, followed by regression [12].

In Bieganski's original family and in the second Polish family, all patients had slowly progressive neurodegeneration with central and peripheral involvement and identical skeletal dysplasia [19]. Two additional patients, described in 2021, documented the natural history in more radiological detail, one with an intronic variant associated with exon 7 skipping and the other with a synonymous variant in that exon [26]. One of the patients in the reference series, previously published as an isolated case, died at twenty years [1, 11].

3.6 Manifestations by system

3.7 Natural history and treated disease

The distinction between natural history and treated disease is difficult in this group, because there is no disease-modifying treatment and the reports combine multiple interventions. All the descriptions in the previous sections correspond to natural history, plus supportive care. Three exceptions should be made explicit. The first is the effect of high-dose riboflavin on ataxia, with a reduction of 39% and 20% in the International Cooperative Ataxia Rating Scale score at twelve months in two patients, which does not correspond to the expected natural course, generally progressive [33]. The second is the cardiac course under heart failure therapy, in which ejection fraction rose from 35.3% to 44.4% between two and a half and four years of age, with reduced ventricular dilatation, which does not correspond to the natural history of untreated mitochondrial dilated cardiomyopathy [15]. The third is hearing function after cochlear implantation, which has no natural equivalent [35]. In all other respects, including auditory progression over decades, the figures from the series describe the disease with supportive care, and not a course modified by targeted therapy [17, 32].

4. Clinical Recognition and Red Flags

4.1 When to suspect, by age

In the prenatal period, suspicion arises in the face of early-detected ventriculomegaly in a male fetus, at around thirteen to fourteen weeks of gestation, especially when accompanied by an enlarged cisterna magna and bilateral septated choroid plexus cysts, and when there is a family history compatible with X-linked inheritance [3, 27]. This association was the finding that led to the diagnosis in the Palestinian family and was repeated in a subsequent pregnancy of the same mother [27].

In the newborn, the combination of drug-resistant multifocal electroclinical seizures in the first hours of life with diffuse signal change in the white matter and thalami on magnetic resonance imaging and relative sparing of the basal nuclei should raise suspicion, especially when the picture is initially interpreted as hypoxic-ischaemic encephalopathy but the clinical course, the persistence of the suppression-burst pattern on the electroencephalogram and the imaging and spectroscopy findings do not confirm it [24]. The practical rule formulated by the authors of that description is that, when there is a discrepancy between the clinical features and the neuroradiological or neurophysiological findings, the mimics of hypoxic-ischaemic encephalopathy should be considered, among them combined oxidative phosphorylation deficiency, molybdenum cofactor deficiency and sulphite oxidase deficiency [24].

In infancy and the first year of life, the grounds for suspicion are severe hypotonia with few spontaneous movements, areflexia, arrest or regression of psychomotor development after a few months of normal progress, tongue fasciculations, continuous involuntary movements of the hands and feet, swallowing difficulties and T2 hyperintensities of the basal ganglia [3, 44]. When hypotonia, proximal weakness and areflexia dominate the picture, the phenotype may be indistinguishable from spinal muscular atrophy, and the absence of a deletion or copy number change of the survival motor neuron 1 gene requires the investigation to continue [3, 39]. In any boy with encephalomyopathy and hypotonia, the presence of left ventricular hypertrophy on echocardiography strengthens the suspicion of mitochondrial disease [15].

At preschool and school age, four presentations should bring this gene to mind. The first is rapidly progressive hearing loss, with loss of speech within weeks, followed by ataxia [28]. The second is the appearance of gait ataxia and hyporeflexia in a child with previous auditory neuropathy-type hearing loss, progressing to wide-based gait, dysmetria, dysarthria and tremor [33]. The third is motor deterioration with spasticity, tremor, truncal ataxia and nystagmus setting in between twelve and thirty-six months after a normal first year, in a child with skeletal dysplasia or with short stature and large, deformed joints [11, 24]. The fourth is severe and progressive myoclonus of the limbs, face, larynx and palate, without an electroencephalographic correlate and without response to antiepileptic drugs [33].

In adolescence and young adulthood, the typical presentation is hearing loss with an auditory neuropathy pattern, appearing between five and twenty years, accompanied by tinnitus and speech discrimination disproportionately poor for the pure-tone threshold, to which numbness of the extremities, unsteadiness and areflexia are added years later [17, 46]. The finding that late-onset auditory neuropathy has this gene as its cause in about one in five cases justifies including it in the first-line genetic study of that presentation [17]. In adulthood, it may present as isolated axonal polyneuropathy, sometimes only in the fourth or fifth decade, sparing the central nervous system and other organs [41].

4.2 Combinations of signs with guiding value

Three associations have particular value in diagnostic orientation. The first is auditory neuropathy with peripheral neuropathy: the combination of absent auditory brainstem responses with preserved otoacoustic emissions and signs of distal sensory neuropathy is the signature of this gene's auditory entity and distinguishes it from forms of non-syndromic deafness due to variants in hair cell genes [7, 22]. The second is deafness with cerebellar ataxia and axonal neuropathy, which shifts the hypothesis towards this gene and a small set of alternatives covered in section 6 [25, 59]. The third is skeletal dysplasia with hypomyelinating leukodystrophy, a combination unusual enough practically to identify the entity [11, 23].

It should be noted, on the other hand, that findings usually considered obligatory in mitochondrial disease are often missing. In one of the families with severe neonatal presentation, extensive laboratory investigations, including plasma and cerebrospinal fluid lactate and urinary organic acids, were normal [3, 27]. In the reference series, clinical chemistry, brain imaging and muscle biopsy were described as non-specific and inconsistent, so a normal metabolic study does not exclude this gene [12, 28]. Likewise, normal respiratory chain enzyme activities in fibroblasts do not exclude the disease, because the defects may be tissue-specific: in one female patient, all activities measured in fibroblasts were normal, whereas in a male patient with the same variant there was reduced complex IV activity in muscle [15, 49].

4.3 Red flags and emergency situations

There are five situations that require immediate intervention. The first is respiratory failure, which may set in acutely, in the context of intercurrent infection, or progressively, through respiratory muscle weakness, and is the most frequent cause of death in the severe forms; advanced forms include restrictive lung disease due to scoliosis and obstructive sleep apnoea [24, 27, 62]. The second is dysphagia with risk of aspiration, which caused the death of an infant from aspiration pneumonia and requires a low threshold for clinical feeding assessment and radiographic swallowing study [20, 27]. The third is episodic rapid deterioration, characteristic of mitochondrial diseases, in which periods of stability alternate with sudden functional losses, often triggered by infection or other metabolic stress [12, 28]. The fourth is the onset or worsening of heart failure, since the sequence from hypertrophy to ventricular dilatation with systolic dysfunction can be rapid and benefits from early therapy [15]. The fifth is metabolic acidosis with raised lactate, which requires correction and avoidance of lactate-containing solutions [20, 49].

Two iatrogenic exposures are, in themselves, red flags. The first is the prescription of chemotherapy with vincristine to a patient or a female carrier: the mother of a patient developed severe axonal neuropathy at fifty years after vincristine treatment for breast cancer, and vincristine toxicity on already compromised peripheral nerves is documented in hereditary neuropathy in general [5, 21, 33]. The second is general anaesthesia with volatile agents or propofol, for the reasons set out in section 7 [62].

4.4 Newborn screening

There is no newborn screening for AIFM1-related disorders in any of the jurisdictions covered by the sources consulted, and the biochemical features of the disease would not allow it with current methodologies, given that the metabolic markers are inconsistent and non-specific [12, 30]. Universal newborn hearing screening, on the other hand, has an indirect relationship with this group: because it uses otoacoustic emissions, it may be normal in a newborn who cannot hear, because the outer hair cells are intact and the defect is in the nerve [48]. This limitation is relevant to the recognition of auditory neuropathy in general, although this gene's entity has post-neonatal onset and is not detectable at birth: AIFM1 was identified exclusively in the non-infant group of a cohort of three hundred and eleven patients with auditory neuropathy, whereas OTOF was identified almost exclusively in the infant group [29, 48].

5. Diagnosis: Criteria, Tests and Algorithms

5.1 Applicable formal criteria

There are no specific diagnostic criteria for AIFM1-related disorders. The diagnosis is molecular, and the formal criteria available belong to the syndromic entities within which the gene falls.

The criteria for the nuclear-encoded Leigh spectrum, applicable to patients with combined oxidative phosphorylation deficiency type 6 who fit the picture, require the combination of three elements [20]. The first is typical neuroradiological findings: bilateral, typically symmetrical, hyperintensities on T2-weighted magnetic resonance images, or hypodensities on computed tomography, in the brainstem and basal ganglia, with or without bilateral hyperintensities in the thalamus, cerebellum, subcortical white matter or spinal cord [20]. The second is at least one neurological feature, among developmental regression, developmental delay and neurobehavioural or psychiatric features [20]. The third is at least one biochemical or mitochondrial abnormality, among raised lactate in blood or cerebrospinal fluid, raised glycine when the gene involved is required for lipoic acid biosynthesis, a lactate peak on magnetic resonance spectroscopy in the absence of acute seizures, a respiratory chain enzyme activity deficiency below 30% in affected tissues — muscle, liver or fibroblasts —, a pyruvate dehydrogenase complex deficiency in fibroblasts more than two standard deviations below the mean, a mitochondrial fission or fusion defect detected by electron microscopy of muscle or by immunofluorescence of cultured fibroblasts, or decreased respiratory activity measured by microscale oxygraphy [20]. The original 1951 neuropathological definition, with symmetrical focal necrotic lesions and a spongiform appearance in the basal ganglia, thalamus, brainstem, dentate nuclei and optic nerves, has been replaced in practice by imaging diagnosis, and post-mortem examination is now rarely performed outside a research context [20].

For auditory neuropathy, the operational criterion is audiological and consists of the combination of absent or abnormal auditory brainstem responses with preserved outer hair cell function, documented by present otoacoustic emissions, and mild to profound hearing loss [7, 48]. The series of six hundred and forty-five patients applied this criterion with the additional requirement of binaural onset and a normal or slightly altered cochlear microphonic [32].

For Charcot-Marie-Tooth disease, the diagnosis is clinical, with a common or classical phenotype, followed by an appropriate neurophysiological study, according to the 2025 Spanish guideline, which sets specific recommendations for the parameters to include in that study [31].

5.2 Diagnostic algorithm

The sequence of steps depends on the presentation.

In an infant or child with encephalomyopathy, hypotonia and regression, the investigation follows that of any suspected mitochondrial disease: full blood count, pH, bicarbonate and lactate as first-line tests, with an interpretation that takes into account that lactate is the product of anaerobic adenosine triphosphate production by glycolysis and that raised values may be appropriate in a crying child or after an epileptic seizure [30]. These are followed by plasma amino acids, in which ratios between concentrations are more informative than isolated values — the alanine/lysine ratio is normally below 3:1 —, urinary organic acids, in which elevation of lactate, pyruvate and Krebs cycle intermediates indicates an oxidative phosphorylation defect, acylcarnitines, useful mainly in defects of flavin metabolism, and, in myopathic patients, fibroblast growth factor 21 and growth differentiation factor 15, whose raised values correlate with mitochondrial dysfunction [30]. Brain magnetic resonance imaging, with spectroscopy, looks for the Leigh pattern, hypomyelination, cerebellar atrophy and the lactate peak [20, 30]. Screening for multisystem involvement includes echocardiography, electrocardiography, assessment of renal tubular and endocrine function — pituitary, thyroid, parathyroids, pancreas and adrenal — and ophthalmological and audiological assessment, and should be repeated over time, because evolution with involvement of new systems is typical [30]. The definitive step is genetic, and identification of variants in a known gene by next-generation sequencing is now the most frequent scenario, even in patients in whom clinical suspicion of mitochondrial disease was low before testing [30].

In a child or adult with neuropathy, the Spanish guideline proposes clinical diagnosis followed by a neurophysiological study and then by sequential genetic diagnosis: once duplication of the peripheral myelin protein 22 gene has been excluded, a next-generation sequencing study is considered, bearing in mind the limitations of the available techniques [31]. The clinical genetics handbook details the reference neurophysiological parameters: motor conduction velocities in hereditary motor and sensory neuropathy type 1 are usually between 10 and 38 metres per second, whereas in healthy adults they are above 40 to 45 metres per second, and in term newborns velocities are about half those of the adult, approaching adult values at around three years of age [63]. In AIFM1 disorders, the pattern is axonal: normal or mildly slowed motor conduction with markedly abnormal sensory conduction [5]. In sporadic cases in early childhood, the handbook recommends considering nerve conduction studies in both parents to detect asymptomatic carriers, testing for Friedreich ataxia, ophthalmological assessment for pigmentary retinopathy or optic atrophy, phytanic acid for Refsum disease, magnetic resonance imaging if there is evidence of regression, seizures or another central nervous system abnormality, and sural nerve biopsy if molecular studies are not informative [63].

In a child or adolescent with hearing loss, audiological assessment precedes and guides genetic testing. The pure-tone audiogram assesses the auditory pathway from the tympanic membrane to the auditory cortex; otoacoustic emissions assess outer hair cell function and may be normal in a newborn who cannot hear, due to an eighth nerve defect; auditory brainstem evoked responses assess the nerve and brainstem pathways; and magnetic resonance imaging or computed tomography look for inner ear malformations [48]. For this gene, magnetic resonance imaging of the internal auditory canals looks for bilateral cochlear nerve hypoplasia, whereas temporal bone computed tomography was normal in all patients in a cohort of fifty [17]. Speech audiometry is indispensable, because the decline in speech discrimination is disproportionate to the pure-tone threshold [17]. Electrocochleography and cervical vestibular evoked myogenic potentials add information on the degree of damage [17].

In a child with skeletal dysplasia and neurological deterioration, skeletal radiographs document the spondyloepimetaphyseal dysplasia and brain magnetic resonance imaging the hypomyelination; molecular diagnosis requires particular attention, because the variants of this entity include synonymous and intronic changes that escape conventional filters [11, 26].

5.3 Molecular testing: strategy and report content

The testing strategy is that of an X-linked disease with great phenotypic heterogeneity, so a broad approach is preferable to a targeted one. Exome or genome sequencing identified most published variants, and genome sequencing was the technique that resolved the Irish family and the two brothers with the p.Ile433Met variant [25, 59]. Targeted panels have documented yield in specific contexts: a panel of four hundred and six hearing loss genes identified the causal variant in a Chinese family, and a mitochondrial disease gene panel resolved the Italian patient with cerebellar ataxia [17, 56]. The overview chapter on Charcot-Marie-Tooth neuropathy proposes a three-step strategy, from the most frequent variant to the panel or exome, with the note that, when X-linked inheritance is suspected without male-to-male transmission, the first gene to test is that for connexin 32 [21].

The report should state the gene, the transcript with its version, Human Genome Variation Society nomenclature at the coding DNA and protein level, zygosity and the pathogenicity class according to the criteria of the American College of Medical Genetics and Genomics and the Association for Molecular Pathology. The current reference transcript is NM_004208.4, with protein NP_004199.1 [13]. This requirement is not a formality: the literature uses at least four different transcripts, and the resulting small numbering differences hamper comparison between publications and database queries [1, 13]. Normalisation should be done in VariantValidator or ClinVar, and never by manual calculation [13]. It is also important to record that the expected zygosity is hemizygosity in males and heterozygosity in females, and that identification of a heterozygous variant in a symptomatic woman should prompt study of X-chromosome inactivation [15, 16].

5.4 Limits of the tests

No negative result excludes the diagnosis, and there are five material limitations to consider.

The first concerns intronic and synonymous variants. The entire skeletal entity is caused by variants located in intron 6 and exon 7, in a region of about seventy base pairs, and includes synonymous changes, such as c.720C>T p.(Asp240=), and intronic ones, such as c.697-44T>G, which act through a splicing defect [11, 13]. A panel that covers only canonical coding regions and a filter that eliminates synonymous variants will miss these alleles. A case of fatal infantile encephalomyopathy was caused by the intronic variant c.1164+5G>A, in intron 11, with exon 11 skipping and a predicted truncated protein of three hundred and sixty-one amino acids [58]. The authors of that description stated the conclusion directly: analysis of intronic sequences and integrative functional studies are needed in variant interpretation [58].

The second concerns variants of uncertain significance. A ClinVar query on 1 November 2026 returned 980 records for the gene, of which 259 were classified as of uncertain significance by the properties filter, versus 181 pathogenic [54]. The disproportion reflects the rarity of the disease and the difficulty of demonstrating pathogenicity: in a large proportion of published cases, no functional study of the variant was performed [1]. Family segregation and functional studies are therefore decisive. The body of evidence used in one of the cohorts to classify seven new variants as pathogenic illustrates what is required: the presence of variants in 18.6% of cases of late-onset auditory neuropathy and their absence in the sensorineural loss group, absence or extremely low frequency in population databases, co-segregation with the phenotype in families with an unaffected carrier mother, concordant deleterious prediction by several algorithms and phenotypic concordance [17].

The systematic comparison of published variants with ClinVar, carried out for this article on 1 November 2026, reveals three situations that the laboratory report and communication with the patient should distinguish [13, 54]. The first is the absence of a record. Nine of the published variants have no ClinVar record at all, among them c.5T>C p.(Phe2Ser) from the neonatal presentation, c.470C>T p.(Ala157Val), the two at codon 210, c.931C>G p.(Leu311Val), c.1357G>C p.(Glu453Gln) and c.1369C>T p.(His457Tyr) from the brothers treated with deep brain stimulation; for these, the only existing classification is that of the authors of the original article, and it should be presented as such, without the apparent reinforcement that mention of a public database confers [54]. The second is conflicting classification despite functional evidence. c.506C>T p.(Pro169Leu), c.1006G>A p.(Glu336Lys), c.1030C>T p.(Leu344Phe), c.1045A>G p.(Ser349Gly) and c.1478A>T p.(Glu493Val) — the latter the variant of the original Cowchock family, with robust linkage and segregation evidence — appear in ClinVar with conflicting classifications, in several cases without criteria declared by the submitters; for these, ClinVar is not the reference classification, and the primary evidence prevails [45, 49, 50, 54]. The third is the reverse discordance: c.513G>A p.(Met171Ile) and c.1299C>G p.(Ile433Met) were described as causal by the authors, with maternal inheritance and American College of Medical Genetics and Genomics criteria applied, but remain of uncertain significance in ClinVar through a single submission, and it is the ClinVar classification that should be stated alongside that of the article [54, 55, 59]. A final caveat concerns submission noise. The ClinVar record for p.(Leu344Phe) associates it with seven distinct conditions, among them Bieganski-type spondyloepimetaphyseal dysplasia, an association the primary literature does not support, since the variants of that entity are restricted to the intron 6 and exon 7 cluster [11, 19, 54]. The phenotype list of an aggregated record does not, therefore, replace reading the articles that described the variant.

The third concerns biochemical studies and biopsy. In the reference series, clinical chemistry, brain imaging and muscle biopsy were non-specific and inconsistent, so a normal metabolic study does not exclude the gene [12, 28]. Normal respiratory chain enzyme activities in fibroblasts do not exclude mitochondrial disease, because of tissue-specific heterogeneity: in one female patient, all activities measured in fibroblasts — complexes I, I+III, II, II+III, III, V and pyruvate dehydrogenase — were normal, relative to protein content and the reference enzymes citrate synthase and cytochrome c oxidase, whereas a male patient with the same variant had reduced complex IV activity in muscle [15, 49]. A definitive diagnosis of mitochondrial disease typically requires impaired respiratory complex activity in samples from the affected organ, abnormal findings on electron microscopy and a pathogenic variant, but invasive biopsy is often not performed in infancy, and in cases with new variants or atypical phenotypes the diagnosis rests on the integration of clinical manifestations, metabolic findings, neuroimaging, genetic analysis, segregation studies and functional validation [15]. Functional assays such as analysis of oxygen consumption rate in extracellular flux analysers have emerged as useful complementary tools [15].

The fourth is the distinction between the proportion of detectable variants and clinical sensitivity. The available series describe diagnostic yield in selected populations — 18.6% of cases of late-onset auditory neuropathy, 11 of 93 males with sporadic auditory neuropathy, a single patient in a Czech cohort of prelingual hearing loss of unknown cause — and these figures do not transfer to the probability of finding a variant in this gene in a different clinical population [7, 17, 64].

The fifth is the tissue-specific nature of the defects. The phenotype may develop independently of measurable defects in mitochondrial enzyme activity, possibly through mechanisms of the protein that are not directly related to respiratory chain metabolism [15]. In the published neonatal cases, by contrast, reduced respiratory complex activity was demonstrated in cultured fibroblasts in all, so punch skin biopsy is recommended to support the diagnosis in that age group [24].

5.5 Prenatal and preimplantation diagnosis

Once the familial variant is known, prenatal diagnosis and preimplantation genetic diagnosis are technically straightforward and follow the principles of any Mendelian disease, since this is a nuclear gene: counselling, prenatal diagnosis and preimplantation diagnosis are relatively simple when the variant has been identified, unlike in mitochondrial DNA diseases, in which heteroplasmy and the genetic bottleneck prevent precise risk estimates [30]. The authors of one of the descriptions explicitly stressed this usefulness for genetic counselling, embryo screening in in vitro fertilisation and prenatal diagnosis [57].

Prenatal ultrasound has limited but documented value: in the Palestinian family, ultrasound in a subsequent pregnancy showed ventriculomegaly and the other abnormal findings in a male fetus, which confirmed the risk before any molecular test [3, 27]. This observation does not make ultrasound a diagnostic method, since most phenotypes have no prenatal expression, but it justifies targeted ultrasound surveillance in at-risk pregnancies.

6. Differential Diagnosis

6.1 Comparative table

6.2 Within the group of X-linked neuropathies

X-linked Charcot-Marie-Tooth disease has six numbered forms, and the distinction between them rests on accessible clinical and electrophysiological data [65]. The most frequent form, type 1, results from variants in the connexin 32 gene and produces a demyelinating or intermediate neuropathy, with the particularity that female carriers are frequently symptomatic, although later than males, which can make the family history appear autosomal dominant until the absence of male-to-male transmission is established [21, 63]. Disproportionate weakness of thumb abduction relative to the first dorsal interosseous is a guiding clinical sign of this form [63]. Types 2 and 3 remain, in the sources consulted, defined by locus — Xp22.2 and Xq26 respectively — without an identified gene, and are characterised, the former, by association with intellectual disability and, the latter, by spasticity with pyramidal signs [65]. Type 5 results from variants in the phosphoribosylpyrophosphate synthetase 1 gene and should be suspected when a boy with early-onset sensorineural hearing loss develops gait disturbance and visual disturbance [65]. Type 6 results from variants in the pyruvate dehydrogenase kinase 3 gene and produces a largely axonal motor and sensory neuropathy of childhood onset, with steppage gait and absent reflexes, without deafness or cerebellar ataxia [65].

The practical distinction from the form caused by AIFM1 rests on three elements. The first is the axonal nature of the neuropathy, with preserved or slightly altered motor conduction, which excludes type 1 in most cases [5, 21]. The second is the association with auditory neuropathy-type deafness, that is, with preserved otoacoustic emissions, which distinguishes it from types 2, 3 and 6 [7, 65]. The third is the presence of cerebellar signs with vermian atrophy, which is not part of the other X-linked forms [25, 65]. Type 5, because it shares early deafness with neuropathy, is the one that requires the most careful distinction, and the element that guides it is optic atrophy with early visual disturbance, although visual deficits may also occur in AIFM1 forms, due to optic atrophy and retinopathy [56, 65].

6.3 Within the group of X-linked deafness and auditory neuropathies

Six genes are associated with X-linked prelingual or postlingual non-syndromic hearing loss: AIFM1, COL4A6, GPRASP2, POU3F4, PRPS1 and SMPX [22]. The distinctive feature of AIFM1, within that set, is its association with a phenotypic spectrum that includes X-linked non-syndromic auditory neuropathy and X-linked syndromic auditory neuropathy with other associated neuropathies, such as ataxia and Cowchock syndrome [22]. In a Chinese cohort of three thousand six hundred and forty-six patients with hearing loss, the aggregate contribution of X-chromosome genes was about 1.14% of diagnosed cases, and POU3F4 accounted for about 59% of these; all X-linked hearing loss was congenital or of childhood onset, with representative audiological profiles or typical cochlear malformations in specific genes, and no phenotypic differences were identified between carriers of POU3F4 and COL4A6 variants [53].

Within the narrower group of auditory neuropathy, the distribution of genes differs from that of other sensorineural losses, and the two most common genes are OTOF and AIFM1 [29]. The practical discriminator is age at onset: in a cohort of three hundred and eleven patients, OTOF was identified almost exclusively in the infant group, with twenty-eight of the twenty-nine cases, whereas AIFM1 was identified only in the non-infant group; ATP1A3 and OPA1 occurred in both groups [29]. The comparison of profiles in the series of six hundred and forty-five patients adds severity: carriers of OTOF and ATP1A3 variants had early onset and worse hearing thresholds, whereas those with AIFM1 had later onset and better thresholds [32]. The prevalence of variants in that cohort was also highly dependent on study design: 54.4% in trios and 56.2% in families, versus 22.9% in cases with only the proband studied, and 33.7% in bilateral cases versus none in unilateral cases [29]. Finally, it is worth recalling that variants in the otoferlin and pejvakin genes cause autosomal recessive non-syndromic auditory neuropathy, an alternative to consider whenever the family history is not compatible with X-linked inheritance [48].

Childhood deafness has, overall, environmental causes in about 35% of cases — congenital rubella or cytomegalovirus infection, perinatal asphyxia, hyperbilirubinaemia with serum bilirubin above 250 micromol per litre for a prolonged period, meningitis and aminoglycosides — and genetic causes in the remainder, of which about 35% are syndromic and 65% non-syndromic, with more than sixty autosomal dominant and more than ninety autosomal recessive loci and about 80 to 85% autosomal recessive inheritance [48]. Identification of a syndromic cause is important for assessing associated problems and for genetic counselling [48].

6.4 Within the group of mitochondrial diseases

The handbook of inborn errors of metabolism classifies the X-chromosome oxidative phosphorylation defects into eight genes — NDUFA1, NDUFB11, COX7B, HSD17B10, HCCS, TAZ, TIMM8A and AIFM1 — and notes that they fall into three patterns of manifestation: in the first, only heterozygous females are affected, embryonic lethality in males being presumed, as in NDUFB11, COX7B and HCCS; in HSD17B10 both sexes are affected; and in NDUFA1, TAZ, TIMM8A and AIFM1 the disease is described only in males or with a very mild phenotype in females [30]. Knowledge published since then requires this third group to be qualified, as discussed in section 8, but the classification retains guiding value [15, 17].

Among mitochondrial diseases with deafness and neurological signs, Mohr-Tranebjærg syndrome is the closest differential, because it shares with AIFM1 the mechanism of defective mitochondrial protein import and X-linked inheritance; it is distinguished by prominent dystonia, optic neuronopathy, cognitive decline and psychiatric symptoms [30]. Cowchock syndrome is, moreover, listed among the recognised mitochondrial syndromes, with the genes AIFM1 and COX7B given as causes, so COX7B should be considered in a child with the Cowchock phenotype and no AIFM1 variant, bearing in mind that its clinical expression follows the first pattern described, with heterozygous females affected [30].

The question with the greatest practical consequence in the differential of mitochondrial diseases is the identification of entities with targeted therapy, which AIFM1 lacks. Among the defects that have one are biotin-thiamine-responsive basal ganglia disease, due to SLC19A3 variants, in which manifestations typically resolve within days with biotin and thiamine; biotinidase deficiency; primary coenzyme Q10 deficiency, with response to high-dose supplementation; primary pyruvate dehydrogenase complex deficiency, with ketogenic diet and thiamine; and disorders of thiamine metabolism due to SLC25A19 and TPK1 variants, with lifelong thiamine [20]. ACAD9 deficiency responds to oral riboflavin, at doses of 100 to 400 milligrams per day, and is the situation closest, from the therapeutic point of view, to AIFM1 disorders [30]. Exclusion of these entities is therefore a priority in the investigation of a mitochondrial encephalomyopathy, before accepting a diagnosis without targeted therapy.

6.5 Within the group of hypomyelinating leukodystrophies and skeletal dysplasias

Spondyloepimetaphyseal dysplasia with hypomyelinating leukodystrophy appears in the differential diagnosis table of the Pelizaeus-Merzbacher-like diseases, with which it shares hypomyelination and from which it is distinguished by the skeletal dysplasia, which that table calls spondyloepiphyseal, although OMIM (300232), Orphanet and the 2017 description classify it as spondyloepimetaphyseal [1, 9, 19, 23]. This combination is rare enough to be practically identifying: before molecular resolution, the entity had been observed in only four families, with linkage to Xq25-q27 [11]. On the skeletal side, spondyloepimetaphyseal dysplasia associated with mitochondrial disease was, at the time of description, the third known example, alongside EVEN-PLUS syndrome, due to HSPA9 variants, and CODAS syndrome, due to LONP1 variants [19]. The common forms of spondyloepiphyseal and spondylometaphyseal dysplasia, including spondyloepiphyseal dysplasia congenita, the Kozlowski type, the Strudwick type and X-linked spondyloepiphyseal dysplasia tarda, do not present with leukodystrophy, and the absence of hypomyelination on magnetic resonance imaging rules them out as the entity of this article [11, 23]. The atlas of bone dysplasias adds, on the skeletal side, a differential that requires molecular analysis for the distinction: NANS deficiency, or spondyloepimetaphyseal dysplasia with intellectual disability of the Camera-Geneviève type, whose association of skeletal dysplasia with developmental impairment overlaps with that of the entity of this article [68]. The distinction is made on three kinds of data. The first is the course: in NANS deficiency the intellectual disability is intermediate to severe but non-progressive, and the adult patients observed remained stable until the end of the fourth decade, whereas the AIFM1 entity progresses through cerebellar, pyramidal and lower motor neuron degeneration [11, 68]. The second is brain imaging: NANS deficiency does not present with hypomyelinating leukodystrophy, and hypomyelination is obligatory in the AIFM1 entity [11, 68]. The third is the skeletal and biochemical pattern: NANS deficiency has advanced carpal ossification in the newborn and child, longitudinal striations in the knee metaphyses that disappear in the adult and accumulation of the enzyme's substrate, N-acetylmannosamine, in urine and cerebrospinal fluid, findings absent from the AIFM1 entity [68]. Inheritance is also discriminating, autosomal recessive without manifestations in heterozygotes, versus X-linked transmission [68].

6.6 Within the group of motor neuropathies and motor neuron disease

The infantile motor neuron disease phenotype due to AIFM1 variants has been described as very similar to spinal muscular atrophy, so the latter is the first differential in a child with hypotonia, proximal weakness, areflexia and tongue fasciculations [3, 39]. What rules out spinal muscular atrophy is cerebral involvement — abnormal visual and brainstem evoked potentials, multifocal epileptic discharges, frontotemporal atrophy and a lactate peak on spectroscopy — which indicates simultaneous injury to the cerebral cortex and spinal cord [39]. In the purely motor forms of adolescents and adults, the differential is that of the distal hereditary motor neuropathies, in which more than thirty genes are implicated, and the authors who described two brothers with the p.Phe210Ser variant recommended including AIFM1 in that panel [40].

7. Treatments and Risk Management Strategies

7.1 Framework: what exists and what does not

There is no disease-modifying treatment for any of the entities in this article. The position is the same as for mitochondrial diseases in general, for which the reference handbook states that there are no curative treatments that correct the underlying process in the great majority of cases, and the same as for the hereditary neuropathies, for which the 2025 Spanish guideline states that, to date, no disease-modifying drug treatment is available [30, 31]. The only intervention with a targeted molecular rationale is supplementation with riboflavin, a cofactor of the flavin adenine dinucleotide that the protein uses, and its efficacy is partial, inconsistent and the subject of explicit divergence in the literature [14, 33]. The rest of the treatment is symptomatic, supportive and aimed at preventing complications, delivered by a multidisciplinary team [20, 21, 31].

A search of the ClinicalTrials.gov registry, carried out on 1 November 2026 with the terms “AIFM1”, “Cowchock” and “Charcot-Marie-Tooth disease type 4”, returned no clinical trial targeting these disorders; the only two studies returned by the gene term — one on apoptosis markers in aggressive periodontitis and the other on skeletal muscle apoptosis in ageing — are unrelated to the pathology [69]. Likewise, there is no gene therapy in clinical development, nor any specific regulatory application, opinion or authorisation in the databases consulted [69, 70]. A negative search does not demonstrate the absence of research activity, but documents that, to date, nothing has reached the registration stage.

7.2 Riboflavin: the targeted intervention

7.2.1 Rationale and evidence in the condition itself

Riboflavin, or vitamin B2, forms the functional group of the cofactors flavin mononucleotide and flavin adenine dinucleotide and is required for numerous reactions catalysed by flavoproteins; complexes I and II of the respiratory chain contain flavoprotein reductases and electron-transfer flavoproteins, and the vitamin is also required for myelin synthesis, its deficiency being able to disorganise the myelin lamellae [71]. Since the protein encoded by AIFM1 is a flavin adenine dinucleotide-dependent oxidoreductase, increased availability of the cofactor may stabilise the mutant protein, and that is the rationale for supplementation [33, 72].

Experimental data support the rationale. In the study that identified the first human variant, treatment of cells with the p.(Arg201del) variant with riboflavin restored the filamentous mitochondrial network and improved cell viability, a result that corresponded to the clinical improvement observed in one of the patients [3, 44]. In a 2023 study of a seven-year-old boy with ataxia, sensorimotor neuropathy and muscle weakness due to the p.(Met340Thr) variant, riboflavin deprivation in fibroblasts reduced protein levels, whereas supplementation at high concentrations partially increased them and partly improved mitochondrial respiratory function; the authors concluded that mutant fibroblasts are intolerant of riboflavin deprivation and that supplementation helps maintain protein levels, proposing early treatment as a worthwhile attempt [72].

The targeted clinical evidence consists of a single prospective study with two patients. Heimer and colleagues treated two unrelated males, with the p.(Met340Thr) and p.(Thr141Ile) variants, with up to 200 milligrams of riboflavin per day for twelve months, and assessed ataxia using the International Cooperative Ataxia Rating Scale at baseline and at six and twelve months of treatment; the ataxia score decreased by 39% in the first patient and 20% in the second at twelve months [33]. The authors' conclusion — that variants in the gene cause a childhood cerebellar ataxia that is partially treatable in some patients with high-dose riboflavin — came to be included in the synopsis of the entity's OMIM entry, which states that prolonged riboflavin treatment may produce some mild improvement in ataxia [5, 33]. The exposure denominator should be stated precisely, because it is what delimits the strength of this claim: among all published patients with variants in this gene, five were exposed to riboflavin and described, and only two have quantified dose, duration and outcome [15, 24, 33, 44]. In the remaining three, riboflavin was given in combination with other cofactors and the dose, formulation or duration were not published, so none of them can be counted as a treated case with a measured outcome [15, 24, 44]. There is, moreover, a counter-example with a functional basis: in the fibroblasts of the patient with the p.(Pro169Leu) variant, the reduced abundance of the protein was not stabilised by riboflavin, and the males of that family had severe multisystem disease with early death [49]. The response cannot, therefore, be generalised to the spectrum, nor presented as expected in a specific patient before a therapeutic trial with an objectively measured outcome.

7.2.2 The divergence in the literature

The review dedicated to the gene, published in 2022, takes the opposite position, stating that currently available treatments, such as riboflavin supplementation, do not produce substantial improvements, and citing in support the cases of Ghezzi, Moss, Heimer and Kawarai [14]. There is a factual basis for this reading: in the fibroblasts of the patient with the p.(Pro169Leu) variant, the reduced abundance of the protein was not stabilised by riboflavin or protease inhibitors, and the three males of that family had severe multisystem disease with early death [49]. The review of the neonatal spectrum states the intermediate position: better understanding of disease mechanisms would facilitate the targeted use of supplements such as riboflavin or coenzyme Q10, whose efficacy in affected patients remains variable and unclear; in that specific neonatal case, the limited duration of follow-up did not allow it to be determined whether the absence of deterioration resulted from vitamin supplementation [24]. The authors of the 2023 study explicitly acknowledge that efficacy remains controversial and that few studies have explored the reasons for that controversy [72].

The clinically defensible reading is as follows: there is prospective evidence of partial improvement in ataxia in two patients, cellular evidence of protein stabilisation with some variants and cellular evidence of no effect with another variant; the response is therefore variant-dependent and not predictable before a therapeutic trial. The decision to treat rests on the absence of an alternative, the favourable safety profile and the possibility of objectively assessing the response by ataxia scale, neurophysiological studies or hearing thresholds.

7.2.3 Dose, dosing regimen and monitoring

Riboflavin is a water-soluble vitamin. There is no single-component riboflavin medicinal product with marketing authorisation in Portugal: a search of Infomed, of INFARMED, on 1 November 2026 returned only two records containing riboflavin, both with revoked authorisation, corresponding to Stimulex® coated tablet and Stimulex® powder and solvent for oral solution, a combination of cobamamide, cocarboxylase, pyridoxine and riboflavin [73]. In the United States, a search of Food and Drug Administration drug labels on the same date, by the generic name of the substance, returned fifty records, all of multivitamin combinations — multivitamins with fluoride, prenatal multivitamins and similar — with a dosage of one tablet per day, without any single-component prescription drug label with a therapeutic dose for mitochondrial disease [70]. In Brazil, a search of the open medicines data of the Agência Nacional de Vigilância Sanitária, in the file of 13 October 2026 consulted on 1 November 2026, identified one thousand and twenty records whose active ingredient includes riboflavin, of which fifty-nine with active status, all combinations — multivitamins, vitamin B complexes, appetite stimulants and iron-containing antianaemics — and none single-component, nor with an indication in mitochondrial disease [74]. Riboflavin has no centralised authorisation from the European Medicines Agency. High-dose use is therefore outside the terms of any label, from a food supplement or compounded preparation, which should be explained to families.

The dose sourced from the condition itself is up to 200 milligrams per day, orally, for twelve months, used in the prospective study of two patients [33]. This dose and the associated ataxia outcomes were obtained from the structured abstract of the publication, since the full text was not accessible; the abstract does not specify the division of doses, the weight or age of the patients at the time of treatment, the formulation used, the maximum tolerated dose or the maximum duration of the regimen, and these elements therefore cannot be stated [33].

In the absence of a guideline-established dose for this condition, the additional reference comes from an analogous condition treated with the same drug, with the extrapolation declared. The closest analogous condition is riboflavin transporter deficiency, caused by biallelic variants in the SLC52A2 and SLC52A3 genes, in which high-dose oral supplementation is effective and possibly life-saving and for which the reference chapter sets out an explicit regimen [67]. The following table brings together the available doses and their origin.

The liquid form exists but is difficult to obtain, so capsules are often opened and mixed with food, such as yoghurt [67]. Parenteral administration may, theoretically, increase uptake and effect, but there is no clinical evidence to support it [67]. The safety precedent for the high dose comes from outside mitochondrial disease: in the randomised trial of migraine prophylaxis, four hundred milligrams per day for three months in fifty-five patients produced two minor adverse effects in the riboflavin arm, diarrhoea and polyuria, and one in the placebo arm, none serious [75]. Side effects of the high dose, when present, include diarrhoea, increased diuresis and urine discolouration, and allergic reactions are rare; any adverse effect, problem or deterioration at any time requires discussion with the responsible clinician [67]. Two management principles from the analogous condition transfer usefully: a positive response may appear with latency, so treatment should continue even without apparent initial improvement, and the response is better when treatment is given earlier in life or in the course of the disease, late initiation often being associated with absence of improvement, probably reflecting neuronal damage already established [67]. Transferring these principles is an extrapolation, and should be presented as such, since the mechanism of the two diseases is distinct: in transporter deficiency, the problem is the intracellular availability of the vitamin, whereas in AIFM1 disorders the problem is the stability of a flavoprotein that uses it as a cofactor [67, 72].

7.2.4 Other cofactors

The two consensus statements of the Mitochondrial Medicine Society, obtained by a Delphi process with an expert panel, apply to any genetically confirmed primary mitochondrial disease and, therefore, to the four entities of this article. The 2015 statement formulates four supplementation recommendations with a defined target: coenzyme Q10 should be offered to most patients with a diagnosis of mitochondrial disease, and not only to those with primary coenzyme Q10 deficiency; alpha-lipoic acid and riboflavin should be offered to patients with mitochondrial disease; folinic acid should be considered in patients with central nervous system manifestations, and is in routine use in those with documented cerebral folate deficiency; and L-carnitine should be given only when there is documented deficiency, with monitoring of levels during treatment [76]. To these is added a rule of method with practical consequences: when starting supplementation, one supplement at a time should be introduced, whenever possible, rather than a simultaneous set [76]. This rule is precisely the one that the reports of patients with variants in this gene did not follow — riboflavin was given in combination with coenzyme Q10, thiamine and pyridoxine and, in one case, with a ketogenic diet —, and is the reason why the contribution of each component cannot be isolated in these patients [15, 24, 76]. The 2017 standards of care statement adds two targeted indications, riboflavin in myopathy due to ACAD9 deficiency and riboflavin with coenzyme Q10 in myopathy due to ETFDH variants, and states that, in critical illness, supplements and cofactor therapy should be continued whenever possible [77].

The evidence base for these recommendations is, however, explicitly weak, and the report of the December 2014 United States National Institutes of Health workshop sets out its limits: anecdotal reports do not allow evidence-based conclusions; existing controlled trials were judged weak, often with five patients or fewer, or with inadequate doses; many supplements have never been studied; and the target levels of most of them are not known [78]. The same report documents the cost of supplementation: in the patient survey of the North American mitochondrial disease consortium, about 28% of respondents reported side effects, most often gastric discomfort, nausea, strong body odour and diarrhoea [78]. Restriction of L-carnitine to documented deficiency has its own basis: carnitine is now considered less benign than was thought, because of gut microbiome-dependent production of trimethylamine N-oxide and the possible associated cardiovascular risk [78]. There is no regulator-approved treatment for genetic mitochondrial disease [78].

Coenzyme Q10, thiamine and pyridoxine have been used in specific patients, in combination with riboflavin, without the contribution of each being isolable [15, 24]. Coenzyme Q10 biosynthesis defects respond to high-dose oral coenzyme Q10, of at least 30 milligrams per kilogram per day in childhood and up to 3 grams per day in adults, but that indication is specific to those defects and does not transfer to AIFM1 [30]. The reference handbook warns that, although some of the vitamin and cofactor “cocktails” advocated by certain centres have a logical rationale — such as trials of riboflavin in complex I deficiency or thiamine in pyruvate dehydrogenase complex deficiency —, many components are non-specific antioxidants and there is no evidence base to support or refute their use [30]. Reduced nicotinamide adenine dinucleotide has been proposed as a therapeutic candidate for auditory neuropathy, as it rescues protein dimerisation and reduces apoptosis in cell lines with the p.(Thr260Ala), p.(Arg422Trp) and p.(Arg451Gln) variants; this is a preclinical cellular result, without any human study [52].

7.3 Treatment of the neuropathy and the neuromuscular component

Treatment of the neuropathy is symptomatic and delivered by a multidisciplinary team including neurology, physical and rehabilitation medicine, orthopaedics, physiotherapy and occupational therapy [21]. The international paediatric guideline, developed by a consortium with systematic reviews of ten clinical questions, a modified Delphi methodology and application of the Grading of Recommendations, Assessment, Development and Evaluation approach, includes three evidence-based and thirty-one consensus-based recommendations, covering the management of muscle weakness, impaired balance and mobility, sensory symptoms, muscle cramps, impaired upper limb function, respiratory impairment, maintenance of joint range of motion and non-surgical treatment of joint deformity; in some areas there was no consensus, which the authors attribute to differences in practice between clinicians and care settings, and the guideline advocates better access to multidisciplinary care [79]. The full text of this guideline was not accessible, so the individual numbered recommendations are not transcribed here; the management described in the following paragraphs comes from the reference chapter on Charcot-Marie-Tooth neuropathy, revised on 20 November 2025, which incorporates it, and from the 2025 Spanish guideline [21, 31, 79].

The established measures are as follows. Footwear should have good ankle support, and ankle-foot orthoses correct foot drop and aid walking; night splints did not improve ankle range of motion in dedicated studies [21, 63]. Some patients need crutches or canes for gait stability, and fewer than 5% need a wheelchair in Charcot-Marie-Tooth neuropathy in general, a proportion that does not apply to the severe forms of this gene, in which wheelchair dependence is frequent [21, 56, 63]. Daily Achilles tendon stretching prevents shortening, and grip exercises address hand weakness; exercise is encouraged within individual capacity, and many patients remain physically active [21]. Orthopaedic surgery may be needed to correct severe pes cavus, and a study of forty-five patients showed that surgical correction of pes cavus is very beneficial; the decision to refer and to intervene should involve neurology, physiotherapy and orthopaedics, and hip dysplasia surgery is sometimes needed [21]. Foot care prevents ulceration in patients with sensory deficit, and scoliosis should be monitored in adolescents, being more common in those with early onset and rarely severe in neuropathy in general — an observation that does not transfer to the forms with progressive neuromuscular kyphoscoliosis of this gene [62, 63]. The Spanish guideline adds that orthopaedic and rehabilitation management should start early, to identify and improve functional deficits, and include individualised exercise guidance, orthotic adaptation and assessment of conservative surgery, such as tendon transfer [31].

Pain requires characterisation before treatment. Musculoskeletal pain may respond to paracetamol or non-steroidal anti-inflammatory drugs, and neuropathic pain to tricyclic antidepressants or drugs such as carbamazepine or gabapentin [21]. Modafinil has been used to treat fatigue [21].

Exercise is the subject of a specific recommendation in the mitochondrial medicine consensus, with a precondition that must be met in these patients. The combination of progressive exercise with resistance exercise is considered optimal for patients with mitochondrial disease, and the statement establishes that they should undergo cardiac screening before starting an exercise programme [76]. Since hypertrophic and dilated cardiomyopathy are part of this gene's spectrum, that screening is not a formality: it is the echocardiogram and electrocardiogram described in section 11 [15, 76]. The standards of care statement specifies that exercise may benefit patients with mitochondrial myopathy, that exercise testing is not recommended for all, because of the risk of metabolic decompensation, and that activity intensity should be set by cardiology according to cardiac function [77]. Assessment by physical and rehabilitation medicine aims to maintain safety and mobility, with inpatient rehabilitation when needed, and, in movement disorder manifestations, to maximise mobility, prevent contractures and relieve pain [76, 77].

Two orthopaedic complications documented in patients with variants in this gene define concrete surveillance targets. The first is bilateral neurogenic hip subluxation, observed at two years and five months in the female patient with encephalomyopathy, which justifies hip surveillance in the non-ambulant child [15]. The second is progressive neuromuscular kyphoscoliosis, which in the patient with the encephalomyopathic form required posterior spinal instrumentation from T3 to L3 at eleven years, which justifies spinal surveillance [62]. Patients at increased risk of vocal cord paralysis merit an otorhinolaryngology consultation at diagnosis, and hoarseness or stridor at any time justifies periodic monitoring, because vocal cord hypomotility and airway obstruction are a potentially lethal complication; this observation belongs to Charcot-Marie-Tooth neuropathy in general, and the sources consulted do not document vocal cord paralysis in AIFM1 disorders [21].

7.4 Treatment of the auditory component

Audiological assessment with auditory brainstem evoked responses and otoacoustic emissions guides hearing aid fitting [12]. In the series of six hundred and forty-five patients with auditory neuropathy, interventions included hearing aids, mostly bilateral, and cochlear implants, mostly unilateral, and implant recipients had lower ages at diagnosis and onset and better aided hearing thresholds than hearing aid users [32].

Cochlear implantation has specific, though limited, evidence. In a study of five people with Charcot-Marie-Tooth neuropathy associated with sensorineural hearing loss and auditory neuropathy spectrum disorder, the implants were safe and reliable and improved hearing and speech, with the caveat that four of those described belonged to a family with a peripheral myelin protein 22 gene variant and not to this gene [21]. A series of four cases published in 2026 included a twelve-year-old boy with auditory neuropathy spectrum disorder due to AIFM1 and mitochondrial dysfunction, who achieved moderate speech recognition — 56% on the consonant-nucleus-consonant test and 75% on the AzBio test — but remained limited in noisy environments [35]. In the other cases in the same series, a sixty-year-old woman with bilateral auditory neuropathy and associated multiple sclerosis went from 8% to 68% on the consonant-nucleus-consonant test one year after implantation; a twenty-month-old boy with the X-linked form due to a connexin 32 variant achieved rapid auditory gains; and a thirty-two-year-old man with type 5 due to a phosphoribosylpyrophosphate synthetase 1 variant and profound prelingual loss improved sound awareness to 20 to 30 decibels hearing level but obtained no gain in speech perception, because of prolonged auditory deprivation [35]. The authors conclude that outcomes are variable but significantly beneficial, with the best results with early intervention, and that a personalised approach integrating the genetic profile, tailored rehabilitation and the setting of realistic expectations is essential [35]. The reference chapter on the Leigh spectrum recommends, in general, hearing aids or cochlear implants for sensorineural loss, referral to speech therapy and hearing support services, and advice to avoid exposure to excessive noise, which may worsen the loss [20].

Two caveats condition the decision to implant and should be made explicit, because they follow from the post-synaptic nature of the lesion: the defect lies in the auditory nerve and spiral ganglion neurons, with preserved outer hair cells, and the implant stimulates precisely the nerve downstream of these cells. The first caveat is anatomical. The work that identified the gene as the cause of X-linked auditory neuropathy concluded that the finding of cochlear nerve hypoplasia in some patients implies that magnetic resonance imaging may be useful to localise the lesion and that cochlear implantation in these patients may have limited success [46]. The series of fifty patients quantified the problem and identified the examination that detects it: temporal bone computed tomography was normal in all patients, whereas magnetic resonance imaging showed bilateral cochlear nerve hypoplasia in three of the seven patients studied by that method [17]. The practical consequence is that pre-implant assessment should include magnetic resonance imaging targeted at the cochlear nerves, and not only computed tomography, and that the expectations communicated to the family should be adjusted to the state of the nerve [17, 46].

The second caveat is functional and concerns the examination on which the decision is based. In the same series, the speech discrimination score deteriorated disproportionately to the pure-tone threshold: half of the patients with only mild hearing loss had a discrimination score of zero, a proportion significantly higher than in the groups with moderate and severe loss, and none of the nine patients with less than five years of disease had a score of zero, versus half of the ears with more than five years' duration [17]. The decision to implant cannot, therefore, rest on the pure-tone audiogram, which underestimates functional impairment, and should be based on speech audiometry [17]. The marker to follow to document outer hair cell function is the cochlear microphonic, which remained obtainable in 96.4 to 100% of patients in all age bands of the cohort of six hundred and forty-five, whereas distortion product otoacoustic emissions gradually disappear [32]. The temporal corollary of the three observations converges with the result of the implant series itself: intervention should be early, because prolonged auditory deprivation cancelled the gain in speech perception in the patient with profound prelingual loss [17, 35]. As for the auditory brainstem implant, which would be the theoretical option in severe nerve hypoplasia, no published case was identified in patients with variants in this gene, and its efficacy in this disease cannot be asserted [46].

7.5 Treatment of the movement disorder

Tremor may reach disabling severity with a poor response to drugs. In two brothers with the p.(His457Tyr) variant, identified by next-generation sequencing, and a phenotype compatible with Cowchock syndrome with a complex and progressive movement disorder, deep brain stimulation of the ventral intermediate nucleus of the thalamus improved contralateral tremor and quality of life, which the authors interpreted as suggestive of a beneficial role for this intervention in treatment-resistant tremor in disorders of this gene [34]. This is off-label use of an established device, supported by a series of two patients, and should be presented to families in those terms.

Dystonia, when present, is treated according to the Leigh spectrum recommendations: benzhexol, baclofen, tetrabenazine and gabapentin may be useful, alone or in combination, with a low starting dose and gradual increase until the manifestations are controlled or intolerable side effects appear; botulinum toxin has been used in patients with severe and intractable dystonia, with the caution that it may worsen muscle weakness and with a contraindication to use in the head and neck, because of the risk of compromising swallowing [20]. In one specific patient, clonidine was used to reduce dystonia and motor restlessness [15]. Clonidine, clonazepam, trihexyphenidyl, gabapentin, chlorpromazine, melatonin and propranolol made up the chronic regimen of an eleven-year-old patient with the encephalomyopathic form, and the episode of rhabdomyolysis that this patient suffered postoperatively was attributed to dystonia precipitated by benzodiazepine withdrawal, which requires that these drugs not be stopped abruptly [62].

7.6 Supportive treatment of the encephalomyopathy

The reference chapter on the Leigh spectrum sets out supportive management by manifestation, applicable to patients with the encephalomyopathic form of this gene. The following table reproduces its recommendations, with the specific additions documented in patients with AIFM1 variants.

Other supportive measures applicable to mitochondrial diseases in general include ptosis surgery, transfusions for anaemia, pacemakers for conduction defects and, occasionally, organ transplantation in an appropriate clinical context, such as isolated end-stage involvement of one organ in an otherwise healthy child; L-arginine may attenuate stroke-like episodes in mitochondrial encephalomyopathy with lactic acidosis, and folinic acid has been reported to improve seizures and other neurological problems in patients with cerebral folate deficiency [30].

Respiratory care deserves a statement of its own, because the standards of care statement sets out a series of assessments that does not follow from the reference chapter on the Leigh spectrum. Polysomnography should be considered at baseline in all patients at diagnosis, to screen for apnoea and sleep disturbances [77]. Spirometry should be considered in the sitting and supine positions, because the difference between the two positions reveals diaphragmatic weakness, and respiratory tests should be repeated during acute decompensations [77]. Preoperative pulmonary assessment should include, at a minimum, pulse oximetry, with blood gas analysis if peripheral oxygen saturation is below 95% [77]. Non-invasive ventilation should be considered preoperatively in patients with neuromuscular weakness, with extubation to non-invasive ventilation postoperatively, and postoperative atelectasis requires aggressive airway clearance with cough assistance and chest physiotherapy [77]. Opioids should be limited, because they suppress coughing and delay recovery [77]. The actual course documented in the patient with the encephalomyopathic form confirms the relevance of each of these points: he was on nocturnal continuous positive airway pressure at six centimetres of water for obstructive sleep apnoea and as-needed oxygen therapy before surgery and, postoperatively, had respiratory failure requiring oxygen, desaturation with activation of the rapid response team and readmission to intensive care, atelectasis and left lower lobe pneumonia, resolved with chest physiotherapy and resumption of continuous positive airway pressure [62].

For swallowing and nutrition, the same statement sets out swallowing studies when aspiration or bulbar dysfunction is suspected, with speech therapy guidance on feeding strategies, and the decision on gastrostomy or jejunostomy in close consultation with gastroenterology, with early gastrostomy placement if oral supplements are not tolerated [77]. Nutritional support should be started as early as is clinically safe, ideally within the first forty-eight hours of admission, weight and nutritional status should be monitored at least every six months, and pseudo-obstructive episodes require bowel rest with intravenous hydration with dextrose and electrolytes [77]. The female patient with encephalomyopathy followed this course, with nasogastric tube feeding for dysphagia and then percutaneous endoscopic gastrostomy [15].

In acute care, the principle is to avoid entering catabolism: acute decompensation should be treated with dextrose-containing intravenous fluids, outpatient therapies should be maintained whenever possible, and each patient should have a written emergency plan and medical alert identification [76]. The critically ill patient requires careful triage and referral to a centre capable of managing multisystem dysfunction, with monitoring of lactate and pyruvate when feeding is introduced and when respiratory support is weaned [77]. Neuroimaging is useful for following neurological progression and should be repeated in any patient with an acute change in neurological status [76]. Plasma or leucocyte coenzyme Q10 levels are useful for assessing absorption and adherence, and carnitine levels should be monitored during treatment [76].

7.7 Agents and circumstances to avoid

The following table brings together the avoidances with their source.

The convergence between sources deserves explicit recording, because it reinforces two of the avoidances. Vigabatrin and propofol appear on the caution list of the mitochondrial medicine consensus, and the two patients with variants in this gene whose pharmacological course has been published confirm the relevance of that list: in one, vigabatrin was needed for the treatment of West syndrome and the decision had to be weighed; in the other, propofol and halogenated agents were deliberately avoided in anaesthesia, because of their action on complexes I, IV and V and the risk of propofol infusion syndrome [15, 62, 77]. It is worth noting that the international Delphi consensus stresses that there is no evidence to exclude the other drugs in most patients with primary mitochondrial disease and that prescribing decisions should be tailored to each patient's needs and risks; this recommendation is represented here by indirect attribution, from the secondary sources that cite it [20, 30, 80].

7.8 Perioperative care

Anaesthesia may worsen respiratory manifestations and precipitate respiratory failure, so its use and monitoring before, during and after require careful consideration [20]. The most detailed perioperative report available describes an eleven-year-old boy, weighing 41 kilograms and American Society of Anesthesiologists class 4, who underwent posterior spinal instrumentation of segments T3 to L3 for progressive neuromuscular kyphoscoliosis [62].

Preoperative assessment found a feeding gastrostomy in place, a chronic regimen with clonidine, clonazepam, trihexyphenidyl, gabapentin, chlorpromazine, melatonin, propranolol, vitamin supplements and inhaled bronchodilators, nocturnal continuous positive airway pressure at 6 centimetres of water for obstructive sleep apnoea and as-needed home oxygen therapy; coagulation, blood count, serum electrolytes and renal function were normal [62]. Anaesthetic management consisted of six hours' fasting, premedication with oral midazolam and aprepitant, ketamine 5 milligrams per kilogram intramuscularly for preoperative agitation, induction with 50% nitrous oxide in oxygen for peripheral access placement and then lidocaine 1 milligram per kilogram, fentanyl 2 micrograms per kilogram and remimazolam in two doses of 0.25 milligrams per kilogram, a total of 0.5 milligrams per kilogram, with rocuronium 0.5 milligrams per kilogram; maintenance was with methadone 0.1 milligrams per kilogram, remifentanil 0.1 to 0.5 micrograms per kilogram per minute, remimazolam 5 to 25 micrograms per kilogram per minute and lidocaine 1 milligram per kilogram per hour, titrated to a bispectral index of 50 to 60 and a mean arterial pressure of 50 to 65 millimetres of mercury [62]. Surgical site infection prophylaxis used cefazolin 50 milligrams per kilogram every three hours and gentamicin 5 milligrams per kilogram as a single dose, tranexamic acid was given at 50 milligrams per kilogram followed by 5 milligrams per kilogram per hour, with intraoperative cell salvage, and mean arterial pressure was controlled with clevidipine 0.3 to 0.6 micrograms per kilogram per minute or phenylephrine 0.4 to 0.7 micrograms per kilogram per minute; the theatre was warmed and a forced-air warming device was used to maintain normothermia [62]. Total anaesthesia time was five hours, fluids totalled 1300 millilitres, estimated blood loss was 100 millilitres and urine output 670 millilitres; residual neuromuscular blockade was reversed with sugammadex 2 milligrams per kilogram and extubation was performed with the patient awake, with additional boluses of fentanyl 1.2 micrograms per kilogram for pain and dexmedetomidine 1 microgram per kilogram for emergence agitation [62].

The choice of total intravenous anaesthesia with remimazolam and remifentanil rested on the theoretical concerns about the effect of volatile anaesthetics and propofol on mitochondrial function, set out in the previous section; remimazolam was approved by the US regulatory agency for use in adults in 2020, so its use in this child was off-label [62]. The postoperative course illustrates the vulnerability of these patients: multimodal analgesia with hydromorphone by patient-controlled analgesia, scheduled intravenous paracetamol and ketorolac, lidocaine infusion and as-needed diazepam, chest physiotherapy and resumption of nocturnal continuous positive airway pressure; and, as complications, leakage around the gastrostomy, ileus requiring feeding to be stopped, respiratory failure requiring supplemental oxygen, benzodiazepine withdrawal with dystonia and rhabdomyolysis, atelectasis and left lower lobe pneumonia, intercurrent viral respiratory infection, deep and superficial vein thrombosis of the right upper limb with superimposed cellulitis, tenosynovitis and osteomyelitis, treated with enoxaparin, and persistent ileus requiring jejunostomy; inpatient rehabilitation began on day forty-two and discharge took place on day eighty-two [62]. This report does not define a protocol, but documents the risks to anticipate and justifies multidisciplinary planning and postoperative intensive care admission after major surgery.

7.9 Pregnancy and lactation

The sources consulted do not describe pregnancies in women affected by AIFM1-related disorders, which is explained by the predominance of male patients and by the severity of the forms affecting women. Two sets of data are applicable by analogy and should be presented as such.

The first concerns Charcot-Marie-Tooth neuropathy in general, which is an independent risk factor for maternal complications in pregnancy and childbirth: in a study of one hundred and fifty-seven births in one hundred and ninety-three pregnancies, in 9.3% of pregnancies new manifestations appeared or existing ones worsened — reduced strength and sensation, cramps and pain — which may persist after pregnancy, and placenta praevia, abnormal non-cephalic presentation and preterm birth occurred more frequently, in 1.6%, 8.4% and 20.3% of pregnancies respectively [21]. These figures describe the neuropathy as a whole, and not the form caused by this gene.

The second concerns riboflavin supplementation in the analogous condition: in women with riboflavin transporter deficiency, the vitamin should be taken before and during pregnancy and during breastfeeding, to avoid inducing deficiency in the fetus and infant, at a dose of 10 milligrams per kilogram per day [67]. This recommendation does not transfer directly, because the mechanism is distinct, but it establishes that doses of this order were considered acceptable in pregnancy in that condition.

7.10 Transition to adult care

The sources consulted contain no transition programme specific to these disorders. The elements to ensure follow from the disease profile: continuity of neurological, audiological, cardiological, respiratory and orthopaedic follow-up; maintenance of the rehabilitation team; repeated genetic counselling in adulthood, with information on reproductive options; and anticipatory guidance on career and employment, since persistent weakness of the hands and feet has important implications in these areas [21]. In Charcot-Marie-Tooth disease, adults who drive should inform the competent driving licensing authority [63].

7.11 Regulatory information

The following table summarises the regulatory status of the interventions described.

7.12 Clinical trials and developments in the last twelve months

There are no registered clinical trials targeting AIFM1-related disorders, nor any trial results, preliminary results announcements, accepted applications, committee opinions or authorisations [69, 70, 82]. A search of the US registry on 1 November 2026 returned only two studies mentioning the gene, and in both the protein is used as an apoptosis marker, unrelated to the disease: a study of apoptosis markers in aggressive periodontitis and a study of skeletal muscle apoptosis in ageing, both completed [82]. The search for “Cowchock” returned no study [82]. This search has a methodological limitation that should be declared: the free-text search of the registry performs diffuse synonym expansion and returns high totals with poorly relevant ordering — the query for “Charcot-Marie-Tooth” returned nine hundred and eighty-six records, mostly on carpal tunnel syndrome —, so the studies identified are those relevant in the listings read and not the whole registry, and a negative search does not prove absence [82]. For this reason, it is not possible to present a table of trials with population, number of participants, duration, primary outcome and result with confidence interval: the only published prospective intervention study is the open-label treatment of two patients with riboflavin, described in section 7.2, which is not a clinical trial and whose reading is subject to all the limitations of an open-label series of two cases without a comparison group, blinding or pre-registration [33]. The percentage reduction in the ataxia score in that study should not be read without the absolute value of the scale, which the publication presents only in the detailed results, nor compared with reductions obtained in other diseases with other interventions.

The publications of the last twelve months, classified by level of evidence, are as follows. In November 2025 the report of perioperative care with total intravenous anaesthesia and remimazolam in a patient with the encephalomyopathic form was published, which constitutes a published result in the form of a case report [62]. In March and April 2026 two retrospective series were published: the natural history of six hundred and forty-five patients with auditory neuropathy over twenty-seven years, with comparative profiles by gene, and a series of four cases of cochlear implantation in Charcot-Marie-Tooth neuropathy including one patient with a variant in this gene [32, 35]. In April 2026 the complete clinical and molecular characterisation of the p.(Glu336Lys) variant was published, a functional study in fibroblasts and purified protein [50]. In June 2026 the report of the first clinically affected woman was published, with a literature review on the cardiac phenotype, likewise a published result in the form of a case report with review [15]. None of these publications constitutes a clinical trial, sponsor announcement, regulatory opinion or authorisation. The routes of participation that exist for these patients therefore lie in Leigh spectrum and primary mitochondrial disease trials that accept nuclear gene defects, and should be presented as such. Two phase 2 studies have criteria that, in principle, admit a patient with Leigh spectrum due to a variant in this gene: the study of sirolimus in Leigh syndrome, which requires a genetically confirmed diagnosis with neurodevelopmental manifestations, from six months to fifty-five years of age and weighing five kilograms or more, without gene restriction, recruiting by invitation with fifteen estimated participants and expected completion in December 2027; and the study of cysteamine-pantetheine disulphide in the Leigh syndrome spectrum, which requires a confirmed diagnosis of Leigh syndrome or Leigh-like disease with neurodevelopmental manifestations, from five to fifty-five years of age, not yet recruiting, with eighteen estimated participants [82]. Two negative observations complete the picture and are clinically more informative than the previous ones. The paediatric trial of sonlicromanol, which expressly accepted any genetically confirmed mitochondrial disease whose defect reduces one or more enzymes of the oxidative phosphorylation system, from zero to seventeen years of age, and for which a patient with a variant in this gene would be eligible, is suspended by strategic decision of the sponsor, declared to be unrelated to safety concerns [82]. The phase 3 trial of the same compound, recruiting since April 2026 with two hundred and twenty estimated participants, excludes these patients, because it requires the mitochondrial deoxyribonucleic acid variant m.3243A>G with age-adjusted heteroplasmy of at least 20% and adulthood [82]. The pivotal trial of elamipretide in primary mitochondrial myopathy, with two hundred and eighteen participants, one hundred and nine per arm, and forty milligrams per day subcutaneously for twenty-four weeks, did not meet its primary endpoints, the six-minute walk test and total fatigue on the mitochondrial myopathy symptom assessment scale; three quarters of the participants had a mitochondrial deoxyribonucleic acid alteration and the rest nuclear gene defects, so patients with nuclear genes were included and the result applies to them [83]. The vatiquinone, REN001 and peroxisome proliferator-activated receptor delta agonist programmes were terminated by decision of the sponsors [82].

Two routes remain open and should be offered to families, as they do not require intervention: the patient registry and biorepository of the North American mitochondrial disease consortium, which accepts patients with diagnosed or suspected mitochondrial disease, including nuclear gene variants affecting mitochondrial function, recruiting until December 2026 with one thousand estimated participants; and the international Leigh syndrome registry, open to any participant with that diagnosis, recruiting until June 2030 with two hundred estimated participants [82].

As regards orphan medicinal product designations, there is no designation for AIFM1-related disorders. In the European Union there are six designations for Leigh syndrome, verified on the European Medicines Agency pages: sonlicromanol, designated in October 2014; vatiquinone, in December 2011; an adeno-associated viral vector of serotype 9 carrying the human SURF1 gene, in November 2021; cannabidiol, in July 2023; sildenafil citrate, in October 2023; and a butanoic acid methyl ester from Pharming, in December 2023 [84]. Reading these data requires two cautions. The first is that orphan designation does not mean that the medicine is available or authorised: it is a status supporting development. The second is that three of the six designations were withdrawn at the sponsors' request between December 2025 and April 2026 — sonlicromanol, vatiquinone and the SURF1 gene vector —, which describes a field contracting rather than expanding [84]. For the sildenafil citrate designation, the agency's own page is internally inconsistent, with the overview section indicating designation in May 2023 for myotonic disorders and the title and key facts indicating October 2023 for Leigh syndrome; the key facts date is adopted here and the divergence is recorded [84]. As for the United States, the orphan designation database of the regulatory agency's office of orphan products development returned no results in the queries of 1 November 2026, so it was not possible to list the US designations; this is an access limitation of this data collection, and does not allow their existence to be either affirmed or denied [82].

8. Reproductive Options and Family Counselling

8.1 The first point of counselling: a nuclear gene on a sex chromosome

The starting point of counselling is to clear up a frequent confusion. Although the protein acts inside the mitochondrion and the disease is, in the pathophysiological sense, a mitochondrial disease, AIFM1 is a nuclear gene located on the X chromosome; inheritance is X-linked and not maternal or mitochondrial, so recurrence risks and testing strategy are those of an X-linked disease [12]. The practical consequence is important: there is no heteroplasmy, there is no genetic bottleneck, the difficulties of risk estimation in mitochondrial DNA diseases do not apply, and counselling, prenatal diagnosis and preimplantation genetic diagnosis are relatively simple when the familial variant has been identified, following the principles of any Mendelian disease [12, 30]. Likewise, mitochondrial donation techniques, designed to prevent transmission of mitochondrial DNA variants, do not apply [30].

8.2 Recurrence risks by scenario

In a family in which the mother is a known heterozygote, transmission of the allele occurs in 50% of pregnancies, so each pregnancy has a 25% probability of resulting in an affected son and 25% of resulting in a heterozygous daughter [12]. An affected male who survives and reproduces transmits the variant to all his daughters, who are obligate heterozygotes, and to none of his sons [12].

As for the parents of an affected male, the father will not have the disease or be hemizygous for the variant, so he does not require further assessment [20]. In a family with more than one affected individual, the mother is an obligate heterozygote; if a woman has more than one affected son and no other affected relative, and the familial variant is not detectable in leucocyte DNA, she most probably has gonadal mosaicism [20]. If the male is the only affected member of the family, that is, in a simplex case, the mother may be heterozygous or have somatic or gonadal mosaicism, but may also not be a carrier, in which case the variant occurred de novo [20].

De novo variants are not rare in this gene and should be considered in the risk calculation. They are documented for the p.(Thr141Ile) variant in a patient with cerebellar ataxia, the synonymous p.(Asp240=) variant in a patient with the skeletal entity, the p.(Asp237Gly) variant in another patient with the same entity and the p.(Pro169Leu) variant in the girl with the full phenotype [1, 11, 15]. In the series of twelve patients with the skeletal entity, the variants occurred de novo or were inherited from unaffected mothers [11].

8.3 Heterozygous women: the necessary revision

The traditional position is that heterozygous women are unaffected carriers. The gene synopsis states that all published affected individuals are hemizygous males, that variants are typically maternally inherited, that the mothers in the reported families were unaffected transmitting carriers and that, therefore, heterozygotes should be counselled as expected to be unaffected; it adds, prudently, that there is no published study of skewed X-chromosome inactivation as a modifier in carriers of this gene and that, therefore, a carrier cannot be promised a total absence of risk, only that none has been described [12]. The handbook of inborn errors of metabolism likewise classifies AIFM1, with NDUFA1, TAZ and TIMM8A, in the group of X-linked oxidative phosphorylation defects in which the disease is described only in males or with a very mild phenotype in females [30]. The original family followed this pattern: the heterozygous women were asymptomatic, although they showed minor abnormalities in sensory nerve conduction, electromyography and hearing, judged too inconsistent for carrier identification [5, 42].

Four published sets of observations require this position to be revised, and should be included in counselling.

The first is the case of the girl with the full phenotype, published in June 2026, carrying a de novo heterozygous c.506C>T p.(Pro169Leu) variant with extremely skewed X-chromosome inactivation, in a 98 to 2 pattern, who developed infantile-onset mitochondrial encephalomyopathy with West syndrome and cardiomyopathy [15]. The authors concluded that a heterozygous variant can result in disease manifestation in females and that careful cardiac surveillance should include female carriers of variants [15]. It is noteworthy that the mother of the affected males with the same variant, described earlier, had no cardiac or cerebral symptoms despite being a germline carrier, which shows that expression in women depends on the inactivation pattern and not only on the variant [15, 49].

The second is the Chinese auditory neuropathy cohort, in which five of the fifty patients were female, all with the same c.1030C>T p.(Leu344Phe) variant, the most common in the gene among the variants associated with the auditory phenotype [17]. There was no difference in hearing threshold between male and female patients with this variant, although the audiograms of the affected women varied from normal to profound over time; the authors concluded that, besides the X-linked recessive pattern, an X-linked dominant pattern is also observed in auditory neuropathy associated with this gene, with affected women [17]. With one exception, the affected women had no family history [17].

The third is the mother of the proband of the family with the c.1195G>A p.(Gly399Ser) variant, who had mild cerebellar ataxia, intellectual disability and mood and behavioural disorder, but no neuropathy or hearing loss, with highly skewed X-chromosome inactivation [16]. The fourth is the Moroccan family with auditory neuropathy due to the c.1045A>G p.(Ser349Gly) variant, in which pedigree analysis was compatible with X-linked dominant inheritance [18].

The appropriate formulation for counselling is therefore as follows: the predominant pattern is X-linked recessive inheritance, with carriers usually asymptomatic, but skewed X-chromosome inactivation can make a carrier symptomatic, with severity ranging from mild manifestations to a full phenotype, and there are descriptions of an X-linked dominant pattern, particularly in auditory neuropathy due to p.(Leu344Phe). Both patterns should be recorded in the clinical record and explained to the family. There is, in the sources consulted, no male lethality in the sense of loss of male pregnancies, although the severe infantile phenotype is frequently fatal in early childhood [12].

8.4 Cascade testing and carrier detection

Once the variant has been identified in the proband, cascade testing begins with the mother and extends to the proband's sisters, the mother's sisters and their female descendants, following the maternal line. Testing is molecular and targeted at the familial variant; carrier detection by indirect means is unreliable, as the original family showed, with minor abnormalities in sensory conduction, electromyography and hearing in the women that the authors considered insufficiently consistent for that purpose [5, 42].

Two particularities should be anticipated. The first is that an identified carrier may develop manifestations, for the reasons set out in the previous section, so the result should not be communicated as absence of personal risk; study of the X-chromosome inactivation pattern is informative when there are symptoms [15, 16]. The second is that an identified carrier has, in herself, an increased risk of neurological toxicity from neurotoxic drugs, as the observation of the mother who developed severe axonal neuropathy after vincristine documents, so the information should be included in her clinical record and communicated in any future oncological decision [5, 21, 33].

8.5 Prenatal diagnosis and preimplantation genetic diagnosis

With the familial variant known, prenatal diagnosis is technically simple, by analysis of fetal DNA obtained by chorionic villus sampling or amniocentesis, with prior or simultaneous determination of fetal sex. Preimplantation genetic diagnosis is equally applicable, and the authors of one of the descriptions explicitly stressed the usefulness of identifying the variant for genetic counselling, embryo screening in in vitro fertilisation and prenatal diagnosis [30, 57].

Targeted ultrasound surveillance has complementary value in the severe forms. In the Palestinian family, ultrasound in a subsequent pregnancy showed ventriculomegaly and the other abnormal findings in a male fetus, which identified the recurrence; the parents opted for termination of pregnancy [3, 27]. This observation does not turn ultrasound into a diagnostic method, because most phenotypes have no prenatal expression and even the severe form may not have it, but it justifies an early anomaly scan in pregnancies of known risk.

8.6 Counselling of minors and related issues

Genetic testing in children is justified when they are symptomatic, as is the rule in hereditary neuropathy: testing is usually done in children if they have symptoms [63]. Predictive testing in asymptomatic at-risk minors, in the absence of an intervention that alters prognosis, has no established indication in the sources consulted; the practical exception follows from the profile of this gene, since early identification of an at-risk male would allow targeted auditory and cardiac surveillance and would avoid exposure to neurotoxic drugs, arguments that should be weighed case by case with the family.

Anticipatory guidance on career and employment is appropriate, given that persistent weakness of the hands and feet has implications in these areas [21]. Quality of life and the definition of disability have been measured and compared between groups of patients with Charcot-Marie-Tooth neuropathy, and these data support the usefulness of formal functional assessment in follow-up [21]. Finally, the documented intrafamilial variability — brothers with the same variant and very different courses, males of the same family with onset of hearing loss between eighteen months and thirty-four years — should be communicated explicitly, because it precludes individual prognosis based on the course of an affected relative [25, 59]. One concrete piece of information should be included in counselling, because it is the only research participation currently within reach of these families and does not require therapeutic intervention: there are two registries open to patients with variants in this gene, the registry and biorepository of the North American mitochondrial disease consortium, which accepts patients with diagnosed or suspected mitochondrial disease, including nuclear gene variants affecting mitochondrial function, and the international Leigh syndrome registry, open to any participant with that diagnosis [82]. Given that there is no clinical trial targeting these disorders and that the routes of participation in Leigh spectrum trials are limited and, in part, closed, enrolment in a registry is the way to help ensure that future cohorts have real denominators, which is precisely what every prognostic figure in this article lacks today [82].

9. Special Considerations: Population Diversity and Epidemiology

9.1 Declared prevalence and what each figure measures

The four entities have, in Orphanet, the same prevalence class: less than one case per million people [4, 6, 8, 9]. This classification indicates extreme rarity and is not a population estimate obtained by an epidemiological study; it corresponds to the lowest category of the scale used by Orphanet when the number of published cases is very small. There is, in the sources consulted, no prevalence or incidence study dedicated to any of these entities in any population.

The number of published patients allows an order of magnitude. For combined oxidative phosphorylation deficiency type 6, the descriptions bring together about two dozen patients [3, 24]. For Charcot-Marie-Tooth disease type 4, there are the original family with seven affected males, the Irish family with seven males, isolated cases in Italy, Israel and China and the subsequent series [5, 25, 57]. For the skeletal entity, the reference series brought together twelve patients from six families, plus later isolated cases [11, 26]. For the auditory entity, the numbers are substantially higher, because the phenotype is milder and there are audiological screening programmes that identify these patients: a single Chinese cohort brought together fifty patients from thirty-six families [17].

The contrast with the prevalence of the broader categories is instructive. Charcot-Marie-Tooth disease, as a whole, is the most common hereditary neuromuscular disease, with an estimated prevalence of one in two thousand five hundred, and the form caused by this gene represents a minimal fraction of that total [63]. Mitochondrial disease, as a whole, has an estimated prevalence of one in five thousand births, and AIFM1 accounts for less than one per cent of cases of the nuclear-encoded Leigh spectrum [15, 20].

9.2 Diagnostic yield in the populations studied

The most robust figures for this group are not prevalence figures, but diagnostic yield in selected clinical populations, and should be read as such.

In late-onset auditory neuropathy, cases positive for this gene accounted for 18.6% in a Chinese cohort of thirty-six families and fifty patients, and the variants were absent in the sensorineural hearing loss group of the same institution [17]. In an earlier series, gene screening identified hemizygous variants in eleven of ninety-three males (10%) with sporadic auditory sensory disorder [7, 46]. In a retrospective cohort of three hundred and eleven patients with auditory neuropathy, pathogenic and likely pathogenic variants in twenty-three genes were identified in ninety-eight patients, that is, in 31.5% of the total, and AIFM1 was, with OTOF, among the two most common genes; the prevalence of variants varied with design, being 54.4% in trios and 56.2% in families, versus 22.9% in cases studied in the proband only, and 45.7% in the infant group versus 25.6% in the non-infant group, with AIFM1 identified exclusively in the latter [29]. In the series of six hundred and forty-five patients with auditory neuropathy followed over twenty-seven years at the same Chinese centre, the diagnosis rate ranged from one to forty-four cases per year, reflecting the evolution of knowledge of the disease, hearing screening and diagnostic methods [32].

In hereditary hearing loss in general, the molecular-epidemiological investigation of three thousand six hundred and forty-six unrelated Chinese patients obtained an overall diagnostic rate of 52.72%, with the aggregate contribution of X-chromosome genes accounting for about 1.14% of diagnosed cases, that is, twenty-two of one thousand nine hundred and twenty-two; POU3F4 accounted for about 59% of these, and AIFM1 for part of the remainder [53]. In a Czech cohort of prelingual non-syndromic hearing loss of unknown cause after testing of the connexin 26 gene, AIFM1 explained the disease in a single patient, out of a total of fifty-four resolved cases [64].

The difference between the Chinese figures and the Czech figure should not be interpreted as a difference in the population frequency of the gene: the cohorts select different phenotypes. The Chinese series select auditory neuropathy, that is, exactly this gene's phenotype, whereas the Czech cohort selected prelingual hearing loss, a presentation in which this gene practically does not occur, because its onset is postlingual [17, 29, 64].

9.3 Geographical distribution of the descriptions and absence of a founder effect

The published families are distributed across several regions. In the United States, the original Italian-American family and the two brothers with the p.(Ile433Met) variant [42, 59]. In Italy, the two cousins with the first variant identified, the adult patient with cerebellar ataxia, the two cousins with the motor neuron disease phenotype and the neonatal case with the variant in the mitochondrial targeting sequence [24, 39, 44, 56]. In the Palestinian territories, the family with prenatal ventriculomegaly [27]. In Israel, the two patients with ataxia treated with riboflavin [33]. In Ireland, the multigenerational family with seven affected males [25]. In Germany, the patient with the protracted form [28]. In Poland, the two families with the skeletal entity [19]. In Japan, part of the series of twelve patients with that entity and a previously published case [11]. In Morocco, the family with auditory neuropathy and an X-linked dominant pattern [18]. In Belgium, the family with the p.(Gly399Ser) variant [16]. In Spain, the two brothers with distal motor neuropathy [40]. In China, the great majority of patients with the auditory phenotype, from twenty-nine provinces and eleven different ethnic groups in the largest series [17, 32, 55, 57, 60]. In the Czech Republic, one isolated patient [64].

No variant has been described as a founder variant. The most important recurrent variant is p.(Leu344Phe), present in 36.1% of positive cases in the Chinese auditory neuropathy cohort, followed by p.(Arg422Trp), in 13.9%; other recurrent variants in that cohort were p.(Ile304Met) and p.(Tyr560His) [17]. The p.(Met340Thr) variant occurred in two unrelated families, one Israeli and one Irish, and p.(Asp237Gly) in three families, two Polish and one Japanese, one of them through a de novo mutation [11, 19, 25, 33]. The recurrence of these variants in distinct populations suggests points of greater susceptibility in the gene sequence rather than common origin, although the sources consulted present no haplotype analysis to demonstrate it. In the original Polish family, the p.(Asp237Gly) variant was absent from an in-house database of one thousand three hundred and forty-three individuals from the Polish population, which excludes its being a local polymorphism [1, 19].

The over-representation of China in the auditory phenotype data and of Europe in the severe phenotype data should be noted. This asymmetry reflects the organisation of services — the existence of auditory neuropathy reference centres with large cohorts in China and of mitochondrial disease centres with access to enzymology and sequencing in Western Europe — and not a difference in the distribution of the disease. Any inference about the relative frequency of the phenotypes from these figures would be unwarranted.

9.4 Data from Portugal and Brazil

The sources consulted do not describe patients with AIFM1-related disorders in Portugal or Brazil, nor do they contain prevalence estimates for these countries. This absence does not demonstrate that there are no diagnosed patients or publications not covered by the sources consulted; it is noted only that, in the series, cohorts and cases assembled, none is identified as Portuguese or Brazilian.

Two access observations apply to Portugal. The first is the absence of a single-component riboflavin medicinal product with marketing authorisation, verified by a search of Infomed on 1 November 2026, with the consequences described in section 7 [73]. The second is the existence of three relevant European Reference Networks, with centres in Portugal, described in the following section.

9.5 Special populations and situations

The paediatric patient is the most frequent situation in all entities except the auditory one, whose onset lies between late childhood and young adulthood [17]. The specific aspects of paediatric management — nutrition, development, education, orthopaedics during growth — are described in sections 7 and 11.

The adult patient with a mild form is an under-recognised population: isolated late-onset axonal polyneuropathy, sparing the central nervous system and other organs, has been described in one family and may be diagnosed in an adult neuropathy clinic without any suspicion of mitochondrial disease [41]. The same applies to late-onset ataxic sensory neuropathy with hearing impairment [14].

The patient with oncological exposure is a situation of specific risk. Vincristine toxicity on compromised peripheral nerves is documented in this gene, in a female carrier, and the need to avoid or replace vincristine-containing regimens should be recorded in the clinical record of all identified patients and carriers [5, 21, 33].

The patient undergoing major surgery requires multidisciplinary planning, for the reasons and with the details described in section 7 [62].

9.6 European Reference Networks and patient associations

Three European Reference Networks cover the manifestations of this group. The network for rare neuromuscular diseases is relevant for the neuropathy and the myopathic component [38]. The network for hereditary metabolic disorders is relevant for the encephalomyopathic form and for the framing as a mitochondrial disease [36]. The network for rare neurological diseases is relevant for the ataxia, the leukodystrophy and the movement disorder [37]. The skeletal entity benefits, in addition, from referral to the rare bone diseases network. Referral through a reference network is the mechanism that in practice allows access to respiratory chain enzymology, functional variant studies and second opinions in diseases with such a small number of patients.

As for patient associations, the Charcot-Marie-Tooth Association, in the United States, maintains the updated list of medicines with neurotoxic risk in the hereditary neuropathies that the reference chapter on the disease recommends consulting, and is therefore a resource of direct clinical usefulness [21, 81]. The United Mitochondrial Disease Foundation, in the United States, and Mitocon, in Italy, support patients with mitochondrial diseases [85, 86]. In Portugal, the Associação Portuguesa de Doentes Neuromusculares supports patients with hereditary neuropathies and myopathies [87]. In the United Kingdom, the support group cited by the clinical genetics handbook for Charcot-Marie-Tooth disease is CMT UK [63]. The Orphanet pages for each entity bring together the reference information in several languages, including Portuguese [4, 6, 8, 9].

10. Illustrative Clinical Case

The case described by Sandmann and colleagues in June 2026 is the first of a clinically affected woman with a variant in this gene and, because it has six years of longitudinal follow-up with serial cardiac documentation, it is the one that best illustrates the progression of the encephalomyopathic and cardiac phenotype [15].

The patient was born at thirty-nine weeks by planned caesarean section, weighing 2440 grams, measuring 47 centimetres and with Apgar scores of 9, 10 and 10. Early pregnancy ultrasound had shown increased nuchal translucency, and chorionic villus sampling excluded chromosomal abnormalities. The mother took levothyroxine 50 micrograms per day throughout pregnancy, and uterine artery Doppler showed slightly raised vascular resistance; the pregnancy was otherwise uneventful, and there were no perinatal complications. The patient has a healthy older brother and there was no family history of neurological, cardiac or metabolic diseases [15].

The first hospital admission was at seven months, for assessment of suspected West syndrome, developmental regression and loss of social interaction. According to the parents, the symptoms had begun at around two months, with poor feeding, frequent crying episodes and decreased responsiveness; a period of increased irritability of one to two months followed, and then decreased vocalisation, loss of social smiling and stagnation of weight gain. The mother described seizure-like episodes since early infancy, about six times a day, with sudden forward flexion of the head and trunk, adduction of the arms, absence of crying or cyanosis and subsequent sleep. The parents also noticed a musty odour in the urine and sticky stools with an abnormal odour [15].

Physical examination showed generalised muscular hypotonia, microcephaly with a head circumference of 39.8 centimetres, below the first centile and with a standard deviation score of −4.07, and growth retardation, with a length of 62.5 centimetres, below the first centile and with a standard deviation score of −2.92, and a weight of 5.71 kilograms, at the first centile and with a standard deviation score of −2.27. The electroencephalogram showed hypsarrhythmia without focal abnormalities, confirming the diagnosis of West syndrome. Subsequent investigation showed raised lactate in blood, at 31.0 milligrams per decilitre for a normal value below 16, and in cerebrospinal fluid, at 4.00 millimoles per litre for a normal range of 1.1 to 1.8, slightly increased cerebrospinal fluid protein and raised alanine; urine showed raised lactate/creatinine ratios and mitochondrial metabolites, specifically 2-methyl-2,3-dihydroxybutyrate, 3-hydroxy-methylglutaconate and 3-hydroxyglutaric acid. Serial brain magnetic resonance imaging, at seven and twenty months, documented progressive structural changes characteristic of mitochondrial encephalopathy, with abnormal myelination, rounded T2-hyperintense pallidal lesions, deep grey matter involvement and cortical atrophy, shown in figure 1 of the publication [15].

Genetic testing at eight months identified the de novo variant c.506C>T, p.(Pro169Leu), in transcript NM_004208.4, previously described in affected males of another family. Given the expected X-linked recessive inheritance, mitochondrial DNA variation was excluded. To assess the functional impact, enzyme activities were measured in the patient's fibroblasts: complexes I, I+III, II, II+III, III and V and the pyruvate dehydrogenase complex were all normal relative to protein content and the reference enzymes citrate synthase and cytochrome c oxidase. X-chromosome inactivation study showed extremely skewed inactivation, in a 98 to 2 pattern, which, together with the characteristic multisystem phenotype and the typical magnetic resonance imaging changes, supported causality of the variant despite the normal enzyme activities [15].

Initial antiepileptic treatment with prednisolone and vigabatrin produced temporary improvement in seizures and electroencephalogram, but epileptic spasms recurred during prednisolone reduction, so levetiracetam, gabapentin, lacosamide and phenobarbital were added and a ketogenic diet was started. With mitochondrial disease suspected, pyridoxine, thiamine, riboflavin and coenzyme Q were supplemented. Feeding, initially by nasogastric tube because of dysphagia, was transferred to percutaneous endoscopic gastrostomy. At thirteen months, spasms, gelastic seizures and myoclonus persisted, and dystonic movements and urinary retention appeared, treated with clonidine for the dystonia and motor restlessness and with chloral hydrate and midazolam for the sleep disturbance. At two years and five months bilateral neurogenic hip subluxation was diagnosed [15].

The cardiac course was the axis of follow-up. At seven and a half months, given recurrent daytime and night-time bradycardias, the resting twelve-lead electrocardiogram and twenty-four-hour recording showed regular sinus rhythm without pathological intervals or arrhythmia. At eight months, echocardiography revealed left ventricular hypertrophy, with a diastolic interventricular septum z-score of +3.5 and a diastolic posterior wall z-score of +3.6, without outflow tract obstruction and with preserved systolic function, with fractional shortening of 42.4% and ejection fraction of 76.2%, a finding interpreted as a possible hypertrophic cardiomyopathy phenotype; these images appear in figure 2 of the publication. At around eighteen months, myocardial hypertrophy remained stable and ventricular dilatation appeared, with a diastolic diameter z-score of +3.9 and ejection fraction falling to 51.2%, with fractional shortening of 25.6%. At two years and six months, the diagnostic criteria for paediatric dilated cardiomyopathy were met, with a diastolic diameter z-score of +4.4, fractional shortening of 16.9% and ejection fraction of 35.3%, despite persistent hypertrophy, and the electrocardiogram showed left ventricular hypertrophy with terminal repolarisation changes in the inferolateral leads. At that point heart failure therapy was started, with good tolerance of low-dose propranolol and ramipril, to which spironolactone and furosemide were later added. Follow-up imaging at four years showed partial stabilisation, with improved systolic function — fractional shortening of 21.4% and ejection fraction of 44.4% — and reduced ventricular dilatation, with hypertrophy persisting; the longitudinal progression of echocardiographic parameters is shown in figure 3 of the publication [15].

At six years, the patient remained non-ambulant, tetraplegic, non-verbal, with dysphagia and tube feeding, dystonia and occasional seizures, alive and clinically stabilised [15].

The figures actually consulted are as follows: figure 1, with brain magnetic resonance imaging at seven and twenty months, available at https://pmc.ncbi.nlm.nih.gov/articles/PMC13287274/figure/Fig1/ ; figure 2, with the echocardiographic images and electrocardiograms at eight and thirty months, at https://pmc.ncbi.nlm.nih.gov/articles/PMC13287274/figure/Fig2/ ; and figure 3, with the longitudinal progression of echocardiographic parameters, at https://pmc.ncbi.nlm.nih.gov/articles/PMC13287274/figure/Fig3/ [15].

Three aspects of this report should not be read as routine management. The first is the ketogenic diet, which was used here in the context of drug-resistant epilepsy, an indication in which it has a place, but which has no evidence of efficacy in forms of the Leigh spectrum not resulting from pyruvate dehydrogenase complex deficiency [15, 20]. The second is simultaneous supplementation with four cofactors, whose individual contribution cannot be assessed and for which there is no evidence base to support or refute it; the authors themselves attribute the clinical stabilisation to the combination of polypharmacy, cofactor replacement and ketogenic diet, without isolating any component [15, 30]. The third is the interpretation of the normal enzyme activities in fibroblasts: in this case, normality did not exclude the diagnosis, and the previously described patient with the same variant had reduced complex IV activity in muscle, which illustrates the tissue heterogeneity discussed in section 5 [15, 49].

Two of the authors' conclusions have the value of recommendations and converge with the general mitochondrial disease guidelines: early cardiac screening, with echocardiography and biomarkers such as lactate and N-terminal pro-B-type natriuretic peptide, should be part of the diagnostic work-up of suspected cases; and timely initiation of heart failure therapy, in addition to cofactor replacement, may lead to partial functional recovery [15, 20].

11. Prognosis and Long-Term Follow-up

11.1 Prognosis by entity

The prognosis of early-onset mitochondrial diseases is, in general, poor, but the natural history is extremely variable even between patients with the same genetic defect, so any individual prediction should be made with reservation [30].

In combined oxidative phosphorylation deficiency type 6, the documented deaths occurred at three months, from aspiration pneumonia and hypertrophic cardiomyopathy, at sixteen months, from sudden death during sleep, and at four years, from recurrent respiratory failure; one patient was tetraplegic, in a wheelchair, unable to communicate and on permanent artificial ventilation at five years [3, 27, 44]. In a review of twelve males in the spectrum, three died before the age of five, while others were still walking in early adulthood, and the course could alternate rapid deterioration with long plateaux [12, 28]. The protracted form and the neonatal case with stabilisation at six months show that early presentation does not necessarily determine rapid progression [24, 28]. The female patient with the full phenotype and cardiomyopathy was alive and stabilised at six years, on antiepileptic therapy, ketogenic diet, cofactor replacement and heart failure therapy [15].

In Charcot-Marie-Tooth disease type 4, the best-documented functional milestone is loss of walking: one patient became wheelchair-dependent at twenty years [56]. In the Irish family, six of the seven males had hearing loss between fourteen and thirty-four years and walking difficulties a few years later, with survival preserved into adulthood; the seventh, with onset at eighteen months, had a more severe course [25]. Of the two brothers with the p.(Ile433Met) variant, one walked with a cane at seventeen years and the other needed a wheelchair and non-invasive ventilation at thirteen [59]. In the family with the p.(Gly399Ser) variant, progression of the neurological abnormalities was limited after the first decade, which is the most favourable prognosis described [16]. The proportion of patients needing a wheelchair in Charcot-Marie-Tooth neuropathy in general, below 5%, does not apply to this form [21, 63].

In X-linked deafness type 5, survival is not compromised and the prognosis is functional, auditory and communicative. Progression is slow and uneven: with follow-up of one to ten years there was no apparent deterioration of pure-tone thresholds, whereas with follow-up of more than ten years thresholds at the low and high frequencies increased significantly [17]. The communicative prognosis is worse than the audiometric one: half of the patients with mild hearing loss had a speech discrimination score of zero, and the proportion of ears with zero discrimination rose from none, in patients with less than five years of disease, to half in patients with longer duration [17]. Over decades, auditory neuropathy may progress to sensorineural loss with outer hair cell damage, as shown by the disappearance of otoacoustic emissions in a patient with thirty-three years of disease [17]. In the series of six hundred and forty-five patients, poor speech discrimination persisted on follow-up, despite stability of the cochlear microphonic [32].

In spondyloepimetaphyseal dysplasia with hypomyelinating leukodystrophy, the course is slowly progressive, significant disability results from cerebellar, pyramidal and lower motor neuron degeneration, and early death may occur in the second decade of life; one patient in the reference series died at twenty years [1, 11, 24]. Progressive skeletal deformities, such as scoliosis, contribute to the severe respiratory phenotype, so the prognosis depends in part on orthopaedic and respiratory management [24].

11.2 Prognostic factors

Four factors emerge from the series. The first is age at onset: in the encephalomyopathic forms, intrauterine onset or onset in the first months is associated with death in early childhood, albeit with exceptions [24, 27]. In the auditory entity, the earlier the onset, the more severe the hearing impairment, with pure-tone averages of 55.7, 51.2 and 47.7 decibels hearing level in the groups with onset before twelve years, between twelve and sixteen and after sixteen, although the differences do not reach statistical significance [17]. The second is disease duration, a determinant of the decline in speech discrimination [17]. The third is the presence of cardiac involvement, which adversely affects prognosis in mitochondrial diseases in general and whose progression to dilatation and failure can be rapid [15]. The fourth is the presence of respiratory failure and dysphagia, which determine the documented causes of death [24, 27].

The variant, by itself, does not allow individual prognosis. The same p.(Met340Thr) variant produced different phenotypes in patients from two families, and in the Irish family produced, among seven males, courses that differed by more than three decades in age at onset [25, 33]. This observation should be made explicit when communicating prognosis.

11.3 Surveillance table

The following table brings together the applicable surveillance recommendations. The basis is the surveillance programme for the nuclear-encoded Leigh spectrum, adapted with the surveillance recommendations of the standards of care statement for primary mitochondrial disease, with the specific recommendations documented in AIFM1 disorders and with those for follow-up of hereditary neuropathy [15, 20, 21, 77]. The intervals indicated are those of the sources, and coincide with those referred to in section 7. For two examinations the sources diverge, and the divergence is flagged in the table itself rather than silently resolved: the mitochondrial medicine consensus allows electrocardiography and echocardiography to be spaced to every two to three years after three years of stability and accepts audiological assessment every one to two years, whereas this article adopts the annual interval for both, because of the cardiac risk documented in this gene, with the express recommendation of early cardiac screening and careful monitoring including female carriers, and because of the centrality of the auditory phenotype, whose speech discrimination deteriorates disproportionately to the pure-tone threshold [15, 17, 77].

The principle that organises this surveillance is that involvement of multiple organs is typical of oxidative phosphorylation diseases, that the pattern of organs affected can guide the diagnosis and that it is imperative to look for involvement for which supportive therapy exists; reassessment over time is essential, because clinical evolution with progression and involvement of new systems is the rule [30]. Surveillance can be guided by knowledge of the phenotypes associated with specific gene defects, which in this case means giving priority to hearing, the peripheral nerve, the cerebellum, the heart and respiratory function [20].

11.4 Follow-up of female carriers

Identified carriers should have cardiac surveillance, following the explicit recommendation of the authors of the case of the affected woman, and the need to avoid neurotoxic drugs should be recorded in their clinical record, following the observation of the carrier who developed severe axonal neuropathy after vincristine [5, 15, 33]. Baseline audiological and neurological assessment is defensible in carriers of variants associated with the auditory phenotype, given that five women with the p.(Leu344Phe) variant had hearing loss with thresholds no different from those of males [17]. The sources consulted do not set out a formal carrier surveillance programme, and this approach follows from interpretation of the published data.

12. Summary and Final Considerations

AIFM1 is a nuclear gene on the X chromosome, at Xq26.1, whose protein resides in the mitochondrial intermembrane space and functions as a flavin adenine dinucleotide-dependent oxidoreductase. The function that explains human disease is not the apoptotic one, which gave it its name, but the metabolic one: the protein controls the import of CHCHD4 and, through this route, the assembly and maintenance of the oxidative phosphorylation complexes. This results, in the severe forms, in a combined deficiency of several respiratory complexes, with predominance of complexes I and IV, and it also results in the practical consequence most relevant to counselling: although this is a mitochondrial disease, inheritance is X-linked and not maternal, so recurrence risks, testing strategy and reproductive options are those of a Mendelian disease [1, 12, 14].

Current nosology divides these disorders into four entities. Combined oxidative phosphorylation deficiency type 6 is the severe encephalomyopathic form, with intrauterine onset or onset in the first months, hypotonia, psychomotor regression, areflexia, seizures, basal ganglia lesions, dysphagia, respiratory failure and, in some patients, cardiomyopathy progressing from hypertrophic to dilated; death in early childhood is frequent. X-linked recessive Charcot-Marie-Tooth disease type 4, described in 1985 by Cowchock and resolved in 2012, combines a progressive sensorimotor axonal neuropathy with auditory neuropathy-type deafness and, in many patients, cerebellar ataxia with vermian atrophy, tremor, dysarthria and spasticity, with age at onset between infancy and young adulthood. X-linked deafness type 5 is the mildest form, with auditory neuropathy of onset between five and twenty years, speech discrimination disproportionately poor for the pure-tone threshold and late distal sensory neuropathy, without compromise of survival or cognition. Spondyloepimetaphyseal dysplasia with hypomyelinating leukodystrophy is the only one with a skeletal phenotype and the only one defined by genotype, since all known variants lie in intron 6 or exon 7 [3, 5, 7, 11].

The nosological division does not correspond to clear clinical boundaries. The OMIM entries refer to one another as overlapping entities, a patient classified in one may acquire features of another over time, and the unifying feature of the whole spectrum is the presence of axonal neuropathy, with varying extent of involvement of the cerebellum, pyramidal tracts, extrapyramidal system and lower motor neurons. Genotype–phenotype correlations are poorly defined, variants in the same domain may cause distinct phenotypes, and intrafamilial variability is remarkable: in the Irish family, seven males with the same variant had ages at onset of hearing loss that differed by more than three decades. Seizures are the only sign with strong discriminating value within the spectrum, having been reported only in the encephalomyopathic form [5, 24, 25].

The diagnosis is molecular and requires exome or genome sequencing, or panels covering the gene, interpreted with attention to three pitfalls. The first is that synonymous and intronic variants are pathogenic in this gene and escape conventional filters. The second is that clinical chemistry, brain imaging and muscle biopsy are non-specific and inconsistent, and a normal metabolic study does not exclude the gene. The third is that normal respiratory chain enzyme activities in fibroblasts do not exclude the disease, because of tissue-specific heterogeneity. Notation should refer to the MANE Select transcript NM_004208.4, with protein NP_004199.1, and be normalised in VariantValidator or ClinVar, since the literature uses at least four different transcripts [11–13, 15].

There is no disease-modifying treatment. The only intervention with a targeted rationale is high-dose riboflavin, a cofactor of the defective flavoprotein, and its efficacy is the subject of explicit divergence in the literature: an open-label prospective study in two patients obtained a reduction of 39% and 20% in the ataxia score at twelve months with up to 200 milligrams per day, cellular studies showed protein stabilisation with some variants, and the review dedicated to the gene concludes that available treatments do not produce substantial improvements, relying on variants that did not respond. The response is variant-dependent and not predictable before a therapeutic trial. Riboflavin has no authorised single-component medicinal product in Portugal nor a single-component prescription label in the United States, so high-dose use is outside the terms of any label, and the additional reference doses come from analogous conditions treated with the same drug, by declared extrapolation: 10 to 50 milligrams per kilogram per day in three doses, with monthly titration, in riboflavin transporter deficiency, and 100 to 400 milligrams per day in ACAD9 deficiency [14, 30, 33, 67, 70, 72, 73].

The remaining treatment is symptomatic and organised along four axes. That of the neuropathy follows the Charcot-Marie-Tooth disease guidelines, with ankle-foot orthoses, daily stretching, exercise within capacity, surgery for severe pes cavus and differentiated treatment of musculoskeletal and neuropathic pain. That of hearing rests on hearing aids and cochlear implantation, with a useful but limited result in noisy environments and dependent on the duration of auditory deprivation. That of the movement disorder includes deep brain stimulation of the ventral intermediate nucleus of the thalamus for refractory tremor, supported by a series of two patients and used off-label. That of the encephalomyopathy follows the Leigh spectrum supportive programme, by manifestation, with particular attention to acidosis, epilepsy, nutrition, respiratory function and cardiomyopathy [20, 21, 31, 34, 35].

Avoidances are of clinical importance equivalent to that of interventions. Sodium valproate should be avoided because of its inhibitory effect on the respiratory chain, lactate-containing solutions should be avoided in patients at risk of acidosis, volatile anaesthetics and propofol as a prolonged infusion are a concern because they depress oxidative phosphorylation, and vincristine chemotherapy can severely worsen the neuropathy — including in carriers, as documented by the mother who developed severe axonal neuropathy at fifty years after treatment for breast cancer. Multidisciplinary perioperative planning is mandatory for major surgery, and the available report of total intravenous anaesthesia with remimazolam in a child with the encephalomyopathic form documents both the strategy and the complication profile to anticipate [20, 21, 62].

Two revisions of established doctrine should be incorporated into practice. The first is that heterozygous women are not necessarily asymptomatic: a girl with a de novo variant and X-chromosome inactivation skewed in a 98 to 2 pattern developed the full encephalomyopathic phenotype with cardiomyopathy, five women in a cohort of fifty had auditory neuropathy with thresholds indistinguishable from those of males, a carrier mother had mild cerebellar ataxia and intellectual disability with skewed inactivation, and a Moroccan family has a pedigree compatible with X-linked dominant inheritance. The counterproof that completes this revision is decisive for counselling and does not contradict it: the same c.506C>T p.(Pro169Leu) variant that produced the full phenotype in the girl with skewed inactivation was asymptomatic in the germline carrier described earlier, mother of the affected males, who had no cardiac or cerebral manifestations. The variant, by itself, does not predict the carrier's phenotype, and what modulates it is the X-chromosome inactivation pattern; a carrier cannot, therefore, be told that she has no personal risk, nor that she will necessarily have it. The predominant pattern remains X-linked recessive, but both patterns should be recorded in the clinical record and explained to the family, and carriers should have cardiac surveillance, baseline audiological and neurological assessment when the familial variant is one of those associated with the auditory phenotype, and a record of avoidance of neurotoxic drugs [49]. The second is that cardiac involvement is part of the spectrum and follows a predictable sequence from hypertrophy to dilatation with systolic failure, partially reversible with timely therapy, so echocardiography should be part of the diagnostic work-up and annual follow-up [15–18].

The prognosis is highly heterogeneous. In the encephalomyopathic forms, three of the twelve males brought together by the cumulative review of the literature published up to 2015 died before the age of five, while others were still walking in early adulthood, a proportion that describes the published cases of that period and not a current mortality rate; in the neuropathic and ataxic forms, loss of walking may occur in the third decade or progression may practically cease after the first; in the auditory forms, survival is not compromised and the problem is communicative, with a decline in speech discrimination disproportionate to the pure-tone threshold and aggravated by disease duration. Surveillance should be multisystem and repeated, because progression with involvement of new organs is the rule, and communication of the prognosis should make explicit that the variant identified does not allow individual prediction [12, 16, 17, 25, 30].

Finally, the gaps. There is no registered clinical trial, gene therapy in clinical development or any regulatory process under way for these disorders. There is no prevalence study in any population, and the figures available are diagnostic yield in selected clinical cohorts, not transferable to other populations. There are no published data on Portuguese or Brazilian patients in the sources consulted. There is no actionability statement or gene–disease validity classification by the ClinGen consortium, nor any formal carrier surveillance programme. The central mechanistic question — whether the disease results from bioenergetic failure, from increased susceptibility to caspase-independent cell death or from a variable combination of the two — remains open, and its resolution is the condition for cofactor supplementation to be replaced by targeted therapy [12, 14, 24, 69].

13. References

1. Online Mendelian Inheritance in Man. APOPTOSIS-INDUCING FACTOR, MITOCHONDRIA-ASSOCIATED, 1; AIFM1. MIM 300169. Baltimore: Johns Hopkins University. Available from: https://www.omim.org/entry/300169
2. OMIM Gene Map / Morbid Map (files genemap2.txt, morbidmap.txt, mimTitles.txt). Baltimore: Johns Hopkins University
3. Online Mendelian Inheritance in Man. COMBINED OXIDATIVE PHOSPHORYLATION DEFICIENCY 6; COXPD6. MIM 300816. Baltimore: Johns Hopkins University. Available from: https://www.omim.org/entry/300816
4. Orphanet. Severe X-linked mitochondrial encephalomyopathy. ORPHA:238329. https://www.orpha.net/en/disease/detail/238329
5. Online Mendelian Inheritance in Man. CHARCOT-MARIE-TOOTH DISEASE, X-LINKED RECESSIVE, 4, WITH OR WITHOUT CEREBELLAR ATAXIA; CMTX4. MIM 310490. Baltimore: Johns Hopkins University. Available from: https://www.omim.org/entry/310490
6. Orphanet. X-linked Charcot-Marie-Tooth disease type 4. ORPHA:101078. https://www.orpha.net/en/disease/detail/101078
7. Online Mendelian Inheritance in Man. DEAFNESS, X-LINKED 5, WITH PERIPHERAL NEUROPATHY; DFNX5. MIM 300614. Baltimore: Johns Hopkins University. Available from: https://www.omim.org/entry/300614
8. Orphanet. X-linked hereditary sensory and autonomic neuropathy with deafness. ORPHA:139583. https://www.orpha.net/en/disease/detail/139583
9. Orphanet. Leukoencephalopathy-spondyloepimetaphyseal dysplasia syndrome. ORPHA:83629. https://www.orpha.net/en/disease/detail/83629
10. Orphanet. Spondyloepimetaphyseal dysplasia, Bieganski type. ORPHA:168448. https://www.orpha.net/en/disease/detail/168448
11. Miyake N, Wolf NI, Cayami FK, et al. X-linked hypomyelination with spondylometaphyseal dysplasia (H-SMD) associated with mutations in AIFM1. Neurogenetics. 2017;18(4):185-94. doi:10.1007/s10048-017-0520-x
12. GeneGuidance. AIFM1 gene synopsis, version of 26 September 2026
13. VariantValidator (vvdb_2025_3, vvta_2025_02). Normalisation of 44 AIFM1 variants on the MANE Select transcript NM_004208.4, with genomic mapping to GRCh38. Accessed 1 November 2026. https://variantvalidator.org/
14. Wischhof L, Scifo E, Ehninger D, Bano D. AIFM1 beyond cell death: an overview of this OXPHOS-inducing factor in mitochondrial diseases. EBioMedicine. 2022;83:104231. doi:10.1016/j.ebiom.2022.104231
15. Sandmann C, Gudapati S, Mayr JA, et al. Cardiomyopathy and mitochondrial encephalomyopathy in a female child associated with a heterozygous X-linked AIFM1 variant. Mol Cell Pediatr. 2026;13(1). doi:10.1186/s40348-026-00246-z
16. Pandolfo M, Rai M, Remiche G, Desmyter L, Vandernoot I. Cerebellar ataxia, neuropathy, hearing loss, and intellectual disability due to AIFM1 mutation. Neurol Genet. 2020;6(3):e420. doi:10.1212/NXG.0000000000000420
17. Wang H, Bing D, Li J, et al. High Frequency of AIFM1 Variants and Phenotype Progression of Auditory Neuropathy in a Chinese Population. Neural Plast. 2020;2020:5625768. doi:10.1155/2020/5625768
18. Elrharchi S, Riahi Z, Salime S, et al. Novel Mutation in AIFM1 Gene Associated with X-Linked Deafness in a Moroccan Family. Hum Hered. 2020;85(1):35-9. doi:10.1159/000512712
19. Mierzewska H, Rydzanicz M, Bieganski T, et al. Spondyloepimetaphyseal dysplasia with neurodegeneration associated with AIFM1 mutation - a novel phenotype of the mitochondrial disease. Clin Genet. 2017;91(1):30-7. doi:10.1111/cge.12792
20. Rahman S, Thorburn DR. Nuclear Gene-Encoded Leigh Syndrome Spectrum Overview. In: GeneReviews. Seattle: University of Washington
21. Bird TD. Charcot-Marie-Tooth Hereditary Neuropathy Overview. In: GeneReviews. Seattle: University of Washington; created 1998-09-28, updated 2018-06-28, revised 2025-11-20. NBK1358
22. Shearer AE, Hildebrand MS, Schaefer AM, Smith RJH. Genetic Hearing Loss Overview. In: GeneReviews. Seattle: University of Washington
23. Hobson GM, Garbern JY. Pelizaeus-Merzbacher Disease, Pelizaeus-Merzbacher-Like Disease 1, and Related Hypomyelinating Disorders. In: GeneReviews. Seattle: University of Washington
24. Zambon AA, Ghezzi D, Baldoli C, et al. Expanding the spectrum of neonatal-onset AIFM1-associated disorders. Ann Clin Transl Neurol. 2023;10(10):1844-53. doi:10.1002/acn3.51876
25. Bogdanova-Mihaylova P, Alexander MD, Murphy RP, et al. Clinical spectrum of AIFM1-associated disease in an Irish family, from mild neuropathy to severe cerebellar ataxia with colour blindness. J Peripher Nerv Syst. 2019;24(4):348-53. doi:10.1111/jns.12348
26. Edgerley K, Barnicoat A, Offiah AC, et al. AIFM1-associated X-linked spondylometaphyseal dysplasia with cerebral hypomyelination. Am J Med Genet A. 2021;185(4):1228-35. doi:10.1002/ajmg.a.62072
27. Berger I, Ben-Neriah Z, Dor-Wolman T, et al. Early prenatal ventriculomegaly due to an AIFM1 mutation identified by linkage analysis and whole exome sequencing. Mol Genet Metab. 2011;104(4):517-20. doi:10.1016/j.ymgme.2011.09.020
28. Kettwig M, Schubach M, Zimmermann FA, et al. From ventriculomegaly to severe muscular atrophy: expansion of the clinical spectrum related to mutations in AIFM1. Mitochondrion. 2015;21:12-8. doi:10.1016/j.mito.2015.01.001
29. Wang H, Guan L, Wu X, et al. Clinical and genetic architecture of a large cohort with auditory neuropathy. Hum Genet. 2024;143(3):293-309. doi:10.1007/s00439-024-02652-7
30. Rahman S, Mayr JA. Disorders of Oxidative Phosphorylation. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 10, sections 10.1-10.5.6
31. Sivera Mascaró R, García Sobrino T, Horga Hernández A, et al. Clinical practice guidelines for the diagnosis and management of Charcot-Marie-Tooth disease. Neurologia (Engl Ed). 2025;40(3):290-305. doi:10.1016/j.nrleng.2024.02.008
32. Zhou X, Wu X, Li D, et al. Auditory Function and Natural History of 645 Auditory Neuropathy Patients Over a 27-Year Span in China. Ear Hear. 2026;47(2):517-31. doi:10.1097/AUD.0000000000001746
33. Heimer G, Eyal E, Zhu X, et al. Mutations in AIFM1 cause an X-linked childhood cerebellar ataxia partially responsive to riboflavin. Eur J Paediatr Neurol. 2018;22(1):93-101. doi:10.1016/j.ejpn.2017.09.004
34. Tunyi J, Abreu NJ, Tripathi R, et al. Deep Brain Stimulation for the Management of AIFM1-Related Disabling Tremor: A Case Series. Pediatr Neurol. 2023;142:47-50. doi:10.1016/j.pediatrneurol.2023.02.007
35. Hughes SE, Mehta AH, Stucken EZ, et al. Evaluating Cochlear Implantation Outcomes in Charcot-Marie-Tooth Disease: A Case Series Analysis of Genetic Profiles and Intervention Timing. Otol Neurotol. 2026;47(3):e525-8. doi:10.1097/MAO.0000000000004815
36. European Reference Network for Hereditary Metabolic Disorders (MetabERN). https://metab.ern-net.eu/
37. European Reference Network on Rare Neurological Diseases (ERN-RND). https://www.ern-rnd.eu/
38. European Reference Network on Rare Neuromuscular Diseases (EURO-NMD). https://ern-euro-nmd.eu/
39. Diodato D, Tasca G, Verrigni D, et al. A novel AIFM1 mutation expands the phenotype to an infantile motor neuron disease. Eur J Hum Genet. 2016;24(3):463-6. doi:10.1038/ejhg.2015.141
40. Sancho P, Sánchez-Monteagudo A, Collado A, et al. A newly distal hereditary motor neuropathy caused by a rare AIFM1 mutation. Neurogenetics. 2017;18(4):245-50. doi:10.1007/s10048-017-0524-6
41. Hu B, Wang M, Castoro R, et al. A novel missense mutation in AIFM1 results in axonal polyneuropathy and misassembly of OXPHOS complexes. Eur J Neurol. 2017;24(12):1499-506. doi:10.1111/ene.13452
42. Cowchock FS, Duckett SW, Streletz LJ, Graziani LJ, Jackson LG. X-linked motor-sensory neuropathy type-II with deafness and mental retardation: a new disorder. Am J Med Genet. 1985;20(2):307-15. doi:10.1002/ajmg.1320200214
43. Shi T, Chen Z, Li J, et al. AIF translocation into nucleus caused by Aifm1 R450Q mutation: generation and characterization of a mouse model for AUNX1. Hum Mol Genet. 2024;33(10):905-18. doi:10.1093/hmg/ddae010
44. Ghezzi D, Sevrioukova I, Invernizzi F, et al. Severe X-linked mitochondrial encephalomyopathy associated with a mutation in apoptosis-inducing factor. Am J Hum Genet. 2010;86(4):639-49. doi:10.1016/j.ajhg.2010.03.002
45. Rinaldi C, Grunseich C, Sevrioukova IF, et al. Cowchock syndrome is associated with a mutation in apoptosis-inducing factor. Am J Hum Genet. 2012;91(6):1095-102. doi:10.1016/j.ajhg.2012.10.008
46. Zong L, Guan J, Ealy M, et al. Mutations in apoptosis-inducing factor cause X-linked recessive auditory neuropathy spectrum disorder. J Med Genet. 2015;52(8):523-31. doi:10.1136/jmedgenet-2014-102961
47. Bieganski T, Dawydzik B, Kozlowski K. Spondylo-epimetaphyseal dysplasia: a new X-linked variant with mental retardation. Eur J Pediatr. 1999;158(10):809-14. doi:10.1007/s004310051211
48. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 121-3
49. Moss T, May M, Flanagan-Steet H, et al. Severe multisystem pathology, metabolic acidosis, mitochondrial dysfunction, and early death associated with an X-linked AIFM1 variant. Cold Spring Harb Mol Case Stud. 2021;7(3):a006081. doi:10.1101/mcs.a006081
50. Ferrer M, Pedron M, Soriano O, et al. Clinical and molecular characterization of a novel pathogenic AIFM1 E336K mutation connecting mitochondrial dysfunction and neurodegeneration. Cell Commun Signal. 2026;24(1). doi:10.1186/s12964-026-02862-8
51. Qiu Y, Wang H, Fan M, et al. Impaired AIF-CHCHD4 interaction and mitochondrial calcium overload contribute to auditory neuropathy spectrum disorder in patient-iPSC-derived neurons with AIFM1 variant. Cell Death Dis. 2023;14(6):375. doi:10.1038/s41419-023-05899-6
52. Qiu Y, Wang H, Pan H, et al. AIFM1 variants associated with auditory neuropathy spectrum disorder cause apoptosis due to impaired apoptosis-inducing factor dimerization. J Zhejiang Univ Sci B. 2023;24(2):172-84. doi:10.1631/jzus.B2200081
53. Feng H, Huang S, Ma Y, et al. Genomic and phenotypic landscapes of X-linked hereditary hearing loss in the Chinese population. Orphanet J Rare Dis. 2024;19(1):342. doi:10.1186/s13023-024-03338-z
54. ClinVar. Query of the AIFM1 gene via the NCBI E-utilities, 1 November 2026. https://www.ncbi.nlm.nih.gov/clinvar/?term=AIFM1%5Bgene%5D
55. Wang B, Li X, Wang J, et al. A novel AIFM1 mutation in a Chinese family with X-linked Charcot-Marie-Tooth disease type 4. Neuromuscul Disord. 2018;28(8):652-9. doi:10.1016/j.nmd.2018.05.008
56. Ardissone A, Piscosquito G, Legati A, et al. A slowly progressive mitochondrial encephalomyopathy widens the spectrum of AIFM1 disorders. Neurology. 2015;84(21):2193-5. doi:10.1212/WNL.0000000000001613
57. Wang C, Lin Z, Yuan Z, et al. Whole-exome sequencing detected a novel AIFM1 variant in a Han-Chinese family with Cowchock syndrome. Hereditas. 2023;160(1):22. doi:10.1186/s41065-023-00282-z
58. Peng Q, Ma K, Wang L, et al. Case Report: A Novel Intronic Mutation in AIFM1 Associated With Fatal Encephalomyopathy and Mitochondrial Disease in Infant. Front Pediatr. 2022;10:889089. doi:10.3389/fped.2022.889089
59. Vasquez A, Schimmenti LA, Demirel N, et al. Novel AIFM1 Variant in 2 Siblings With Sensorineural Hearing Loss and Cerebellar Ataxia. Neurol Genet. 2024;10(6):e200216. doi:10.1212/NXG.0000000000200216
60. Wang Q, Xingxing L, Ding Z, Qi Y, Liu Y. Whole exome sequencing identifies a novel variant in an apoptosis-inducing factor gene associated with X-linked recessive hearing loss in a Chinese family. Genet Mol Biol. 2019;42(3):543-8. doi:10.1590/1678-4685-GMB-2018-0051
61. Morton SU, Prabhu SP, Lidov HGW, et al. AIFM1 mutation presenting with fatal encephalomyopathy and mitochondrial disease in an infant. Cold Spring Harb Mol Case Stud. 2017;3(2):a001560. doi:10.1101/mcs.a001560
62. Olakunle IC, Tobias JD. Perioperative Care of a Child With Combined Oxidative Phosphorylation Deficiency 6: Total Intravenous Anesthesia With Remimazolam. J Med Cases. 2025;16(11):434-9. doi:10.14740/jmc5179
63. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 362-5
64. Safka Brozkova D, et al. Spectrum and frequencies of non GJB2 gene mutations in Czech patients with early non-syndromic hearing loss detected by gene panel NGS and whole-exome sequencing. Clin Genet. 2020;98(6):548-54. doi:10.1111/cge.13839
65. Kim HJ, Sohn KM, Shy ME, et al. Charcot-Marie-Tooth Neuropathy X Type 5. In: GeneReviews. Seattle: University of Washington
66. Olpin S, Morris A. Disorders of Mitochondrial Fatty Acid Oxidation and Riboflavin Metabolism. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 12, section 12.2
67. Cali E, Dominik N, Manole A, Houlden H. Riboflavin Transporter Deficiency. In: GeneReviews. Seattle: University of Washington; created 2015-06-11, updated 2021-04-08. NBK299312
68. Spranger JW, Brill PW, Hall C, Nishimura G, Superti-Furga A, Unger S. NANS deficiency (MIM 610442). In: Bone Dysplasias: An Atlas of Genetic Disorders of Skeletal Development. 4th ed. Oxford: Oxford University Press; 2018. p. 236-239.
69. ClinicalTrials.gov. Queries for AIFM1, Cowchock and Charcot-Marie-Tooth disease type 4, 1 November 2026. https://clinicaltrials.gov/
70. Food and Drug Administration. Drug labels containing riboflavin. Query of 1 November 2026. https://open.fda.gov/apis/drug/label/
71. Plantone D, Pardini M, Rinaldi G. Riboflavin in Neurological Diseases: A Narrative Review. Clin Drug Investig. 2021;41(6):513-27. doi:10.1007/s40261-021-01038-1
72. Zhao Y, Lin Y, Wang B, et al. A Missense Variant in AIFM1 Caused Mitochondrial Dysfunction and Intolerance to Riboflavin Deficiency. Neuromolecular Med. 2023;25(4):489-500. doi:10.1007/s12017-023-08750-5
73. INFARMED. Infomed — Database of medicinal products for human use. Query for riboflavin, 1 November 2026. https://extranet.infarmed.pt/INFOMED-fo/
74. Agência Nacional de Vigilância Sanitária (Brazil). Open medicines data, file of 13 October 2026. Query for riboflavin, 1 November 2026. https://dados.anvisa.gov.br/dados/
75. Schoenen J, Jacquy J, Lenaerts M. Effectiveness of high-dose riboflavin in migraine prophylaxis. A randomized controlled trial. Neurology. 1998;50(2):466-70. doi:10.1212/wnl.50.2.466
76. Parikh S, Goldstein A, Koenig MK, et al. Diagnosis and management of mitochondrial disease: a consensus statement from the Mitochondrial Medicine Society. Genet Med. 2015;17(9):689-701. doi:10.1038/gim.2014.177
77. Parikh S, Goldstein A, Karaa A, Koenig MK, et al. Patient care standards for primary mitochondrial disease: a consensus statement from the Mitochondrial Medicine Society. Genet Med. 2017;19(12):1380-97. doi:10.1038/gim.2017.107
78. Camp KM, Krotoski D, Parisi MA, Gwinn KA, et al. Nutritional interventions in primary mitochondrial disorders: developing an evidence base. Mol Genet Metab. 2016;119(3):187-206. doi:10.1016/j.ymgme.2016.09.002
79. Yiu EM, Bray P, Baets J, et al. Clinical practice guideline for the management of paediatric Charcot-Marie-Tooth disease. J Neurol Neurosurg Psychiatry. 2022;93(5):530-8. doi:10.1136/jnnp-2021-328483
80. De Vries MC, Brown DA, Allen ME, et al. Safety of drug use in patients with a primary mitochondrial disease: an international Delphi-based consensus. J Inherit Metab Dis. 2020;43(4):800-18. doi:10.1002/jimd.12196
81. Charcot-Marie-Tooth Association. https://www.cmtausa.org/
82. ClinicalTrials.gov. Queries for AIFM1, Cowchock, auditory neuropathy, mitochondrial disease, primary mitochondrial myopathy, Leigh syndrome, riboflavin and Charcot-Marie-Tooth, 1 November 2026. https://clinicaltrials.gov/
83. Karaa A, Bertini E, Carelli V, Cohen BH, et al. Efficacy and safety of elamipretide in individuals with primary mitochondrial myopathy: the MMPOWER-3 randomized clinical trial. Neurology. 2023;101(3):e238-52. doi:10.1212/WNL.0000000000207402
84. European Medicines Agency. Orphan designations for the treatment of Leigh syndrome: EU/3/14/1336, EU/3/11/937, EU/3/21/2531, EU/3/23/2800, EU/3/23/2831, EU/3/23/2867. Query of 1 November 2026. https://www.ema.europa.eu/en/medicines/human/orphan-designations/
85. United Mitochondrial Disease Foundation. https://www.umdf.org/
86. Mitocon — Insieme per lo studio e la cura delle malattie mitocondriali. https://www.mitocon.it/
87. Associação Portuguesa de Doentes Neuromusculares. https://www.apn.pt/