Early-Onset Mitochondrial Encephalomyopathies

Primary disorders of oxidative phosphorylation with onset in the neonatal period, infancy and early childhood: class, classifications, types and treatable forms

Literature and regulatory information updated: 27 October 2026.

0. Initial Reference Table

This article covers a class of disorders, not a single entity: primary mitochondrial diseases (PMD) with onset in the neonatal period, infancy or early childhood that affect oxidative phosphorylation (OXPHOS), due to variants in mitochondrial DNA (mtDNA) or in nuclear genes (nDNA), with congenital lactic acidosis, encephalopathy, myopathy, cardiomyopathy, hepatopathy or combinations of these [1–3]. The first table summarises the data for the class; the second lists the types and genes.

The table of types brings together, for each entity, the Online Mendelian Inheritance in Man (OMIM) and Orphanet (ORPHA) phenotype numbers given by the sources consulted, the gene, the inheritance, the features that distinguish it within the class, the test or clue that discriminates it and the sections of this article in which it is developed. The abbreviations are: HCM, hypertrophic cardiomyopathy; SRNS, steroid-resistant nephrotic syndrome; COXPD, combined oxidative phosphorylation deficiency; CI to CV, complexes I to V; 3-MGA, 3-methylglutaconic aciduria; MMA, methylmalonic aciduria; MMDS, multiple mitochondrial dysfunctions syndrome.

1. Introduction

Definition and scope

OXPHOS disorders form a heterogeneous group of diseases with onset in infancy, childhood or adulthood, with variable involvement of the organs with the highest energy requirements: central and peripheral nervous system, skeletal and cardiac muscle, eyes, ears, kidneys and liver; any organ, or any combination of organs, may be affected at any age [1]. GeneReviews defines primary mitochondrial disorders as genetic disorders, known or presumed, caused by pathogenic variants in genes encoding the mitochondrial respiratory chain and related proteins, and distinguishes them from the many genetic and non-genetic disorders with secondary mitochondrial dysfunction, such as ethylmalonic aciduria due to ETHE1, Friedreich ataxia, Wilson disease or ageing, which "are not strictly mitochondrial disorders" [2]. The 2015 Mitochondrial Medicine Society (MMS) consensus describes PMD as disorders of mitochondrial structure or function caused by variants in nDNA or mtDNA and places them among the most frequent inborn errors of metabolism, with a prevalence conservatively estimated at about 1:5000 [21].

The paediatric diagnostic guideline of Wortmann and colleagues (Salzburg, Munich and Bern, 2017) uses a broad operational definition: disease with a primary defect at any point of the pyruvate oxidation pathway, including the pyruvate dehydrogenase complex (PDHc), the Krebs cycle and complexes I to V, and in the supporting machinery, which comprises mtDNA-dependent protein synthesis (mtDNA replication, mitochondrial RNA metabolism and translation), cofactor metabolism and mitochondrial homeostasis (protein import, lipid metabolism, fusion and fission, quality control). The same guideline estimates an incidence of at least 1:5000, counted 289 associated genes in 2017 and stresses that most affected children have at least one neurological sign, such as muscle weakness, peripheral neuropathy, ophthalmoplegia, movement disorder, epilepsy or migraine [22].

This article deals with forms with onset in the neonatal period, infancy and early childhood. PMD have two presentation peaks, in the first three years of life and in late adolescence or early adulthood, and early-onset forms are usually severe and multisystemic [18]. In most infantile forms progression is relentless and targeted treatments exist only for rare forms; the mainstay of treatment is supportive care [1]. Infantile-onset nuclear disease is almost always autosomal recessive [1], and only 15–30% of childhood OXPHOS defects are due to mtDNA mutations, a much higher proportion in adults [1]. The old idea that nDNA disorders began in childhood and mtDNA disorders in adulthood has been abandoned: many mtDNA disorders begin in childhood and many nuclear disorders in adulthood [2].

The scope covers congenital lactic acidosis of OXPHOS origin; isolated deficiencies of complexes I to V due to structural subunits or assembly factors (ACAD9, NDUFS4, NDUFV1, NDUFS1, NDUFAF5, NDUFAF6, FOXRED1; SDHA; BCS1L, TTC19, UQCRQ; SURF1, SCO2, SCO1, COX10, COX15, PET100, LRPPRC; TMEM70 and ATP5F1 subunits); combined deficiencies due to defects of mitochondrial translation and gene expression (GFM1, TSFM, TUFM, MTO1, GTPBP3, TRMU, ELAC2, mitochondrial ribosomal proteins and aminoacyl-tRNA synthetases); cofactor defects (CoQ10, flavins, iron-sulphur clusters and lipoate, copper); membrane lipid defects (AGK, SERAC1); reversible infantile respiratory chain deficiency (MT-TE and TRMU); infantile mitochondrial cardiomyopathy; and PPA2 deficiency [1]. Leigh syndrome, the general approach to mitochondrial diseases, the m.3243A>G spectrum, which includes mitochondrial encephalomyopathy with lactic acidosis and stroke-like episodes (MELAS), myoclonic epilepsy with ragged-red fibres (MERRF), single large-scale mtDNA deletions, including Pearson syndrome, Leber hereditary optic neuropathy (LHON) and POLG and mtDNA depletion disorders, including TK2, SUCLA2 and SUCLG1, appear here only as differential diagnoses. Pyruvate dehydrogenase complex deficiency, pyruvate carboxylase deficiency, biotinidase deficiency, multiple acyl-CoA dehydrogenase deficiency (glutaric aciduria type II) and Barth syndrome likewise fall within the differential.

Historical overview

The first disease with a mitochondrial defect was described in 1962 by Luft, Ikkos, Palmieri, Ernster and Afzelius, in a correlated clinical, biochemical and morphological study of a case of severe hypermetabolism of non-thyroid origin with a defect in the maintenance of mitochondrial respiratory control [77]. In the same year, Hartman and colleagues described congenital lactic acidosis for the first time, according to the Atlas of Inherited Metabolic Diseases [78]. Leigh syndrome had been defined in 1951 on neuropathological grounds [16].

The molecular era began in 1988. Holt, Harding and Morgan-Hughes found, in the muscle of 9 of 25 patients with mitochondrial myopathy, two populations of mtDNA, one with deletions of up to 7 kb, the first demonstration of heteroplasmy of human mtDNA and of disease associated with defects of the mitochondrial genome [79]. In the same year, Wallace and colleagues described the first pathogenic point mutation of mtDNA, which converts arginine to histidine at codon 340 of the ND4 subunit in LHON [80]. In 1995, Bourgeron and colleagues identified the first nuclear gene with a causal variant for respiratory chain deficiency: the homozygous substitution Arg554Trp (NM_004168.4:c.1660C>T p.(Arg554Trp); ClinVar VCV000008742) [81] in the flavoprotein subunit of succinate dehydrogenase (SDHA), in two siblings with complex II deficiency and Leigh syndrome, children of first cousins [82]. Among the assembly factors, the descriptions of SURF1 in Leigh syndrome with COX deficiency followed (Zhu 1998; Tiranti 1998, according to the Atlas) [78]. Next-generation sequencing (NGS) accelerated discovery: Saudubray 2022 records 260 nuclear genes and known mutations in all 37 mtDNA genes, a number growing exponentially [1]; Frazier, Thorburn and Compton counted almost 290 genes in 2019 [12]; the 2022 North American Mitochondrial Disease Consortium (NAMDC) consensus refers to more than 350 nuclear and mitochondrial genes [13].

Current classifications

There is no single classification. The sources simultaneously use five axes, each with a different purpose, and a sixth syndromic axis harmonised by NAMDC.

The first axis is biochemical, by complex and by enzymatic pattern. It distinguishes isolated deficiencies of one complex (I, II, III, IV or V) from combined deficiencies, which are the most frequent, and recognises patterns of guiding value: combined reduction of complexes I, III, IV and V in mtDNA-dependent disorders (tRNA and rRNA mutations, rearrangements, nuclear defects of maintenance, expression and translation); reduction of I, II and possibly III with pyruvate dehydrogenase complex deficiency and raised glycine in defects of iron-sulphur cluster biosynthesis; and reduction of I+III and II+III with normal isolated activities of I, II and III in CoQ10 deficiency [1]. The biochemical classification guides investigation, but does not always distinguish primary disease from secondary dysfunction [21].

The second axis is genetic: genome of origin (mtDNA or nDNA) and mechanism. Table 10.4 of Saudubray groups 37 mitochondrial and 260 nuclear genes into OXPHOS subunits, assembly factors, mtDNA replication, transcription, translation, cofactors, mitochondrial homeostasis, inhibitors and unknown mechanism (reproduced in section 2) [1]. GeneReviews classifies representative nuclear disorders into five groups: structural subunits, assembly factors, translation factors, multiple mtDNA deletions or depletion, and mitochondrial membrane disorders [2]. The 2025 revision of the GeneReviews on the nuclear Leigh spectrum organises more than 120 genes by pathological mechanism (deficiencies of each complex, pyruvate metabolism, B vitamins, CoQ10, lipoic acid, mtDNA maintenance, gene expression, protein quality control, membranes, dynamics, mitophagy and toxicity) [16]. Frazier and colleagues proposed in 2019 a classification of the genes of mitochondrial energy generation disorders by mechanism, with almost 290 genes [12].

The third axis is nosological, with the numbered series of OMIM and Orphanet: COXPD 1 to 39, mitochondrial complex I deficiencies, nuclear type (MC1DN), complex III, IV and V deficiencies, nuclear type, mtDNA depletion syndromes (MTDPS) and MMDS 1 to 5 [4]. These series are useful for coding and for database searches, but the numbering reflects the order of discovery and not the biology. Orphanet classifies mitochondrial disease as a category (ORPHA:68380) and gives individual numbers to several types, for example isolated complex I deficiency (ORPHA:2609), isolated COX deficiency (ORPHA:254905) and mitochondrial myopathy with reversible COX deficiency (ORPHA:254864) [3, 44, 50, 52].

The OMIM entries consulted in October 2026 number these series beyond the GeneGuidance list of September 2026 (COXPD up to 39, MMDS up to 5): COXPD1 to COXPD60, MC1DN1 to MC1DN39, MC3DN1 to MC3DN10, MC4DN1 to MC4DN25, MMDS1 to MMDS9B and COQ10D1 to COQ10D9; the OMIM numbering is followed [4–10]. The same gene may belong to two series (MTFMT is COXPD15 and MC1DN27), COQ8B has no COQ10D number and is classified as nephrotic syndrome type 9 (NPHS9), and entry 256000 (nuclear Leigh syndrome) is a general entry that refers to the deficiencies of each complex, without its own gene [5, 6, 29, 45]. The table brings together the members of each series according to the reference entries.

The fourth axis is the International Classification of Inherited Metabolic Disorders (ICIMD), of 2021, which organises 1450 inherited metabolic disorders into a hierarchical classification by groups, endorsed by the main metabolic societies and continuously curated [83]. The fifth axis is the reclassification of the 3-methylglutaconic acidurias: the old numbering into types I to V has been replaced by a nomenclature by mechanism and protein, in which primary 3-MGA corresponds to 3-methylglutaconyl-CoA hydratase (AUH) deficiency and the secondary forms are grouped into phospholipid remodelling (TAZ, SERAC1, AGK), mitochondrial membrane-associated disorders (OPA3, DNAJC19, TMEM70), the MICOS complex (MIC13), mtDNA deletion, mitochondrial import (TIMM50) and unknown mechanism (CLPB) [84].

The syndromic axis corresponds to the recognised mitochondrial syndromes, about forty, of which the early-onset ones relevant to this article are congenital lactic acidosis, Leigh syndrome, mtDNA depletion syndromes, Pearson syndrome, reversible infantile respiratory chain deficiency, GRACILE, Sengers syndrome, MEGDEL, pontocerebellar hypoplasia type 6 and the aminoacyl-tRNA synthetase leucoencephalopathies [1]. NAMDC proposed in 2022 the harmonisation of this nomenclature, after finding that only 20.1% of the patients in its registry fitted "classic" syndromes; one of the changes directly concerns this article: "benign infantile mitochondrial myopathy due to reversible COX deficiency" is now termed "reversible infantile myopathy with respiratory chain deficiency", because there are defects in several complexes besides COX [13].

Definition of the entities covered

Primary CoQ10 deficiency is the reduction of CoQ10 in tissues or cultured cells associated with biallelic variants in one of the ten biosynthesis genes (COQ2, COQ4, COQ5, COQ6, COQ7, COQ8A, COQ8B, COQ9, PDSS1, PDSS2), classified as a respiratory chain disorder [14]. ACAD9 deficiency is a complex I deficiency due to loss of function of an assembly factor that also has acyl-CoA dehydrogenase activity [33, 75]. TMEM70 deficiency is ATP synthase deficiency nuclear type 2, also called TMEM70-related mitochondrial encephalocardiomyopathy [36, 37]. GRACILE (growth restriction, aminoaciduria, cholestasis, iron overload, lactic acidosis, early death) is the most severe phenotype of BCS1L deficiency, alongside complex III deficiency nuclear type 1 and Björnstad syndrome [40]. COXPD are combined OXPHOS deficiencies due to defects of mitochondrial protein synthesis [1, 4]. Reversible infantile respiratory chain deficiency (RIRCD) encompasses reversible infantile myopathy due to homoplasmic MT-TE variants and transient infantile liver failure due to biallelic TRMU variants [1]. MMDS are the disorders of iron-sulphur cluster and lipoate biosynthesis that simultaneously affect complexes I and II, the pyruvate dehydrogenase complex and, in several, the glycine cleavage system [74].

2. Molecular Genetics and Pathophysiology

The OXPHOS system and the two genomes

The respiratory chain is formed by complexes I to IV; the OXPHOS system also includes complex V, all in the inner mitochondrial membrane [1]. Complex I (NADH-ubiquinone oxidoreductase) has 44 subunits, 7 encoded by mtDNA; complex II (succinate-ubiquinone oxidoreductase), 4 subunits, all nuclear, and includes succinate dehydrogenase (SDH), shared with the Krebs cycle; complex III (ubiquinol-cytochrome c oxidoreductase), 11 subunits, one from mtDNA (cytochrome b); complex IV (cytochrome c oxidase, COX), 14 polypeptides, 3 from mtDNA; and complex V (ATP synthase), 16 subunits, 2 from mtDNA (ATP6 and ATP8) [1]. CoQ10 (ubiquinone) and cytochrome c are mobile electron carriers; flavins, iron-sulphur clusters, quinones and haem groups are the internal carriers. Complexes I, III and IV pump protons and create a proton-motive force of about 150 mV, which complex V uses to synthesise ATP [1]. Complexes I, II and III depend on iron-sulphur clusters, and these are also required for lipoate synthesis, which explains the raised glycine in some defects of iron-sulphur cluster biosynthesis [1, 74]. The Atlas gives subunit numbers different from those of Saudubray (for example, 44 nuclear subunits for complex II); those of Saudubray 2022 are followed [1, 78]. The 2017 paediatric guideline attributes 45 subunits to complex I, encoded by 37 nuclear and 7 mitochondrial genes, and adds that its formation depends on more than 10 assembly factors, 7 iron-sulphur clusters, flavin mononucleotide as a cofactor and supporting processes (mtDNA-dependent protein synthesis, protein import, lipid composition); it then counted more than 150 genes associated with isolated complex I deficiency or with defects involving it [22]. The difference between 44 and 45 subunits does not alter the number of mtDNA-encoded subunits (7), and this article keeps Saudubray's count.

mtDNA is circular, 16,569 bp long, and encodes 13 polypeptides (7 of complex I, cytochrome b of complex III, 3 of complex IV and 2 of complex V), 2 rRNAs and 22 tRNAs; the 1.1 kb D-loop is the only region not directly involved in polypeptide synthesis [2, 11]. About 1500 proteins are required for normal mitochondrial function; only about 100 participate directly in OXPHOS and about 1% are encoded by mtDNA [18]. The mtDNA replication, transcription and translation machinery involves more than 200 enzymes, rRNAs and tRNAs, and OXPHOS depends on cofactors (CoQ10, iron-sulphur clusters, haem, copper), cardiolipin, assembly factors, protein import, and organelle turnover and dynamics [1]. There is no complete loss of OXPHOS in systemic human disease: residual function always persists, variable between tissues, which complicates diagnosis [1].

Metabolic consequences

OXPHOS oxidises the NADH and FADH2 produced mainly from pyruvate, fatty acids and amino acids. When it fails, anaerobic glycolysis, about 20 times less efficient, generates lactate, and the accumulated pyruvate is transaminated to alanine; raised lactate, pyruvate and alanine are typical but not specific [1]. Lactate depends on pyruvate and on the NAD+/NADH ratio; the normal lactate/pyruvate (L/P) ratio does not exceed 15, and a cytosolic ratio above 25 indicates OXPHOS disease, whereas in pyruvate dehydrogenase deficiency the ratio is low or normal. The 3-hydroxybutyrate/acetoacetate ratio better reflects the mitochondrial redox state [78]. In a newborn with fatal neonatal lactic acidosis, the L/P ratio was 136:1 and the 3-hydroxybutyrate/acetoacetate ratio 42:1 (Robinson 1985, according to the Atlas) [78].

Inheritance, heteroplasmy and threshold

Because two genomes are involved, mitochondrial disease may have maternal, AR, AD, X-linked or sporadic inheritance, and there are somatic mutations in both genomes [1]. The oocyte contains about 100,000 copies of mtDNA and the sperm head about 100 [18]. mtDNA variants may be homoplasmic or heteroplasmic; the threshold effect lies at 80–95% for many common point mutations and at 50–70% for large deletions, and some tRNA anticodon mutations have a dominant-negative effect at loads below 20% [1]. tRNA and rRNA mutations usually reduce several complexes, unlike mutations of a single subunit [1]. The mtDNA mutation rate is 10 to 20 times that of nDNA, the load varies between tissues and over time, and an early bottleneck in embryonic development explains the large shifts in load between mother and children [11].

In nuclear disease, the phenotype is usually more homogeneous, with variable expressivity [18]. The eight X-linked OXPHOS genes follow three patterns: only heterozygous females affected, presumably with embryonic lethality in males (NDUFB11, COX7B, HCCS); both sexes (HSD17B10); only males, or a very mild phenotype in females (NDUFA1, TAZ, TIMM8A, AIFM1) [1]. This third pattern reflects a tendency and not a limit: in a girl with mitochondrial encephalomyopathy and cardiomyopathy, a heterozygous de novo AIFM1 variant with skewed X inactivation produced the full phenotype; COXPD6 and the other disorders of the gene (CMTX4 or Cowchock syndrome, X-linked deafness type 5 and spondyloepimetaphyseal dysplasia with hypomyelinating leucodystrophy) form an allelic spectrum [85].

Genetic classification by mechanism

Variants in structural subunits or in assembly factors usually cause isolated deficiency of one complex; the structural subunits of complex I and the assembly factors of complex IV appear to be relatively frequent causes of isolated CI or CIV deficiency [1]. mtDNA depletion and multiple deletions usually reduce CI, CIII, CIV and CV, although in early stages there may be isolated CI or CIV deficiency; defects of gene expression (RNA processing, mRNA and tRNA modification) and of translation (ribosomal proteins, aminoacyl-tRNA synthetases, initiation, elongation and regulation factors) also reduce CI, CIII, CIV and CV [1]. Cofactor defects form a growing subgroup, and defects of organelle maintenance (import, dynamics, quality control) mainly affect the central nervous system [1]. The table reproduces the Saudubray 2022 classification (single asterisk: depletion and multiple deletions; double: multiple deletions; triple: depletion).

Source: Rahman and Mayr, Saudubray 2022, Table 10.4 [1].

Gene-by-gene mechanisms in the main types

ACAD9. The ACAD9 gene (3q21.3; transcript NM_014049) encodes a protein that, together with NDUFAF1 and ECSIT, forms the mitochondrial complex I assembly complex (MCIA); binding to ECSIT converts it from a β-oxidation enzyme into an assembly factor, and the assembly function is independent of the dehydrogenase activity [33]. ACAD9 is also homologous to very-long-chain acyl-CoA dehydrogenase (VLCAD), with dehydrogenase activity towards long-chain acyl-CoA [75]. Riboflavin is the precursor of FAD and FMN, and more mitochondrial FAD improves FAD binding, folding and stability of ACAD9 and thus complex I assembly [33]. In the cohort of 70 patients there was no complete biallelic loss of function, the variants are distributed along the gene without an evident genotype-phenotype correlation, and severity correlates with residual ACAD9 activity, not with complex I activity [23, 33].

TMEM70. TMEM70 (transcript NM_017866) is classified as a complex V assembly factor; its deficiency causes mitochondrial complex V deficiency nuclear type 2, and 3-methylglutaconic acid accumulates as in other defects of inner membrane integrity [1, 36]. The founder splicing allele c.317-2A>G is homozygous in all 27 Roma patients of the series of 48 [36, 86].

BCS1L. BCS1L is a complex III assembly factor [1]. The Björnstad syndrome variants alter protein-protein interaction residues and those of complex III deficiency alter ATP-binding residues; all disturb respirasome assembly and increase reactive oxygen species, and biallelic variants do not by themselves predict a severe outcome [40]. The Finnish founder variant of GRACILE is c.232A>G p.(Ser78Gly) [87, 88].

Copper and complex IV (SCO2, SCO1, COX10, COX15, SURF1). SCO1, SCO2, COX10, COX15 and COA6 are classified as copper cofactor defects, with complex IV deficiency; in COX10 deficiency, the anaemia reflects a defect of haem A biosynthesis [1, 16]. SURF1 is the most studied complex IV assembly factor [78]. In SCO2, the combination of the common E140K allele (p.Glu140Lys in three-letter nomenclature) or of D135G (NM_005138.3:c.404A>G p.(Asp135Gly)) with a second missense variant near the CxxxC motif gives the mild end of axonal motor neuropathy, whereas the homozygous E140K genotype and compound heterozygosity with severe alleles give infantile cardioencephalomyopathy [43].

TRMU and MT-TE. TRMU uses cysteine to thiolate the wobble-position uridine of the mitochondrial tRNAs for lysine, glutamine and glutamate, a modification required for the synthesis of the 13 mtDNA-encoded proteins [1, 15]. Cystathionase, the endogenous source of cysteine, has a physiological nadir in the first months of life, which makes cysteine an essential amino acid in that period; this combination explains the early hepatic decompensation and the reversibility [15]. In reversible deficiency due to MT-TE, the m.14674T>C and m.14674T>G variants are homoplasmic and the mechanism of spontaneous remission has not been elucidated [1, 66].

CoQ10 biosynthesis. The ten genes of the pathway (COQ2 to COQ9, PDSS1, PDSS2) cause primary deficiency; genes not directly linked to biosynthesis, such as ETFDH (multiple acyl-CoA dehydrogenase deficiency) or APTX (ataxia with oculomotor apraxia type 1), and respiratory chain defects themselves cause secondary deficiency, which only molecular testing distinguishes from primary deficiency [14]. In COQ4 deficiency, patients without a response to CoQ10 had a null variant and responders two missense variants [14].

Iron-sulphur clusters and lipoate. GLRX5, BOLA3, NFU1, ISCA2 and IBA57 build the iron-sulphur clusters of complexes I and II, aconitase and lipoyl synthase (LIAS); LIPT2 and LIAS synthesise lipoate, and LIPT1 transfers it to the E2 subunits of the pyruvate dehydrogenase complex, 2-oxoglutarate dehydrogenase, the branched-chain keto acid dehydrogenase complex and 2-oxoadipate dehydrogenase. LIPT2, LIAS and GCSH add a glycine cleavage system deficiency; LIPT1 does not [74].

Membrane lipids. AGK is an acylglycerol kinase of the intermembrane space and a subunit of the TIM22 complex, which explains the deficiency of the ANT1 carrier in Sengers syndrome; SERAC1, at the interface between the endoplasmic reticulum and the mitochondrion, remodels phosphatidylglycerol [67].

Mitochondrial translation. Mitochondrial aminoacyl-tRNA synthetases are all nuclear-encoded and inherited in an AR manner; except for GARS1 and KARS1, which serve both compartments, they are encoded by genes distinct from those of the cytoplasmic enzymes, which do not replace them [89]. GFM1 and TSFM are elongation factors; the same TSFM p.Arg333Trp homozygosity gave encephalomyopathy in one patient and HCM in another, which points to modifiers of translation [54]. This variant, designated R333W in the original description, corresponds in ClinVar to NM_005726.6:c.934C>T p.(Arg312Trp) (VCV000005379), and was also found in homozygosity in a Chilean infant with lethal COXPD3 [53, 81]. ELAC2 is the mitochondrial RNase Z: in its deficiency, mitochondrial tRNA intermediates unprocessed at the 3′ end and mRNA and rRNA precursors accumulate, with reduced mitochondrial translation [60]. MRPL44 is a protein of the large subunit of the mitochondrial ribosome, and the p.(Leu156Arg) variant causes defective assembly of that subunit and complex IV deficiency in fibroblasts [18].

Recurrent, founder and clinically relevant variants

The table uses Human Genome Variation Society (HGVS) nomenclature on the Matched Annotation from NCBI and EMBL-EBI (MANE) Select transcript, normalised in VariantValidator, and indicates the historical nomenclature.

The TRMU, SCO2, COQ6, COQ4 and MRPL44 variants without a genomic coordinate were not normalised in VariantValidator in this data collection; they are kept with the notation of the sources. The rows with a VCV number were confirmed in ClinVar from the OMIM allelic variant identifier, with the transcript indicated by ClinVar; TRMU p.(Tyr77His) corresponds in ClinVar to NM_018006.5:c.229T>C (VCV000001291), TMEM70 c.317-2A>G to VCV000000540, BCS1L p.(Ser78Gly) to VCV000006167 and SURF1 c.312_321delinsAT to VCV000215237, where the protein consequence is described as p.Pro104_Leu105insTer [81].

Genotype-phenotype correlations, penetrance and allelic disorders

In TRMU deficiency, loss of function is associated with lower survival (10/20 versus 10/42 with other variants; p=0.022) [15]. In Sengers syndrome, homozygous nonsense variants are associated with the infantile form and splicing or start codon variants with the mild forms [67]. In COQ2 deficiency, the response to CoQ10 correlates with the severity of the variants [14]. MT-TE variants have incomplete penetrance [1]. The following are allelic disorders: Björnstad syndrome and complex III deficiency nuclear type 1 (BCS1L) [40]; early-onset axonal motor neuropathy (SCO2) [43]; isolated neuropathy CMT4K (SURF1) [46]; cerebellar ataxia SCAR9 (COQ8A) [14]; spastic ataxia type 10 (COQ4) [30]; and, in AARS2, leucoencephalopathy with ovarian failure, alongside infantile cardiomyopathy [57]. In COQ8A, in a cohort of 64 patients from 51 families (Traschütz 2020), biallelic loss-of-function variants were associated more with isolated ataxia and missense variants with multisystem involvement with epilepsy and myoclonus [27]. In ACAD9, the dehydrogenase activity of 16 variants from 24 patients did not correlate with complex I deficiency, but correlated inversely with clinical severity (Schiff 2015) [32]. In MTO1, compound heterozygous siblings with a truncating variant had fatal HCM and a patient homozygous for the missense A428T (OMIM nomenclature, = NM_012123.4:c.1282G>A p.(Ala428Thr)) had longer survival [58]. In TMEM70, the same c.317-2A>G homozygosity is associated with severe and mild forms, attributed to differences in the efficiency of nonsense-mediated mRNA decay [35]. In complex I deficiency, OMIM does not recognise evident genotype-phenotype correlations that would allow the gene to be inferred from the clinical or biochemical picture [6]. Isolated steroid-resistant nephrotic syndrome NPHS9 and CoQ10 deficiency due to COQ8B are also allelic [29]. The TRMU p.(Ala10Ser) variant modifies the risk of aminoglycoside-induced deafness associated with m.1555A>G [15].

3. Clinical Spectrum and Manifestations by System

Common syndromic description

The presentation ranges from the prenatal period, with intrauterine growth restriction (IUGR) and congenital anomalies, to isolated myopathy in older people [1]. Saudubray describes three classic scenarios of suspicion: a constellation of signs within a recognised mitochondrial syndrome; a multisystem presentation with two or more unrelated systems, explicable by an energy deficit; and lactic acidosis or another suggestive result, such as characteristic neuroimaging, 3-MGA or myopathy with ragged-red fibres. A fourth scenario is increasingly common: pathogenic variants in a mitochondrial gene revealed by genomic sequencing in a patient with low prior suspicion [1]. A high incidence of second- and third-trimester pregnancy losses is a common feature [2].

Early-onset PMD manifests as hypotonia, feeding difficulties, faltering growth, exercise intolerance, seizures and encephalopathy [18]. Neurological involvement includes developmental delay or regression, hypotonia, seizures, irritability, microcephaly, abnormal reflexes, weakness, visual and hearing abnormalities, spasticity, dystonia or dyskinesia and ataxia [18]. Cardiac involvement is present at first presentation in 40% of children (Scaglia 2004, 113 paediatric patients, according to Clinical Cardiogenetics), in the form of cardiomyopathy, arrhythmias and conduction defects [18]. The table summarises the recognised manifestations by organ.

Source: Saudubray 2022, Table 10.1 [1].

Presentations by age

Neonatal period. Congenital lactic acidosis typically presents with tachypnoea and non-specific sepsis-like illness, and may be associated with features of Leigh syndrome and with multisystem involvement, such as cardiomyopathy and renal tubulopathy [1]. The Atlas describes, whatever the cause, recurrent episodes of acidosis with hyperventilation, any of which may lead to coma and death, sometimes preceded by unexplained vomiting; symptomatic hypoglycaemia in some; and pulmonary hypertension in chronic lactic acidosis, which may be a cause of death [78]. In a series of 22 patients with OXPHOS deficiency, 9 had a severe neonatal presentation with liver failure, usually rapidly fatal, with hypotonia, psychomotor delay, seizures and hypoglycaemia; onset between 2 and 18 months was also frequently fatal [78].

The neonatal cohorts describe the natural history without targeted treatment. In Japan, between 2004 and 2020, 294 of 1023 patients with mitochondrial disease (28.7%) had neonatal onset; of the 281 analysed, 69.0% had multisystem disease, 13.5% cardiomyopathy, 9.3% Leigh syndrome and 8.2% liver disease [19]. The mean birth weight was 2346 g, with 35.5% preterm and 35.3% small for gestational age; 73.5% of the initial symptoms appeared in the first two days, mainly respiratory distress (25.9%), neonatal asphyxia (14.3%) and poor feeding or feed refusal (10.6%); hyperlactataemia above 2.1 mM was present in 86.2% [19]. In the Czech Republic and Austria, 129 of 461 patients (28%) had neonatal onset, with prematurity, IUGR and hypotonia requiring ventilation in one third, cardiomyopathy in 40%, neonatal seizures in 16%, Leigh syndrome in 15%, raised lactate in 87% and hyperammonaemia in 22 of 52 newborns tested; complex V deficiency (31) and combined deficiencies (53) were the most frequent [20].

Infancy and early childhood. The Atlas describes, from later infancy, episodic or chronic ataxia triggered by stress or infection, often with episodic neurological degeneration, and Leigh syndrome [78]. Leigh syndrome is the most frequent presentation of mitochondrial disease in childhood, with more than 100 genes in the two genomes [1]. Early-onset leucoencephalopathies are associated with mitochondrial aminoacyl-tRNA synthetases (DARS2, EARS2, AARS2, MARS2, KARS1), ISCA2 and other proteins of lipoate metabolism, complex IV assembly factors (COA7, COA8) and complex I subunits and assembly factors [1]. Isolated organ involvement is possible: epileptic encephalopathy, HCM or acute liver failure [1].

Primary CoQ10 deficiency (treatable)

Primary CoQ10 deficiency falls within the differential. The traditional classification into five phenotypes (encephalomyopathy, cerebellar ataxia, severe infantile multisystem disease, SRNS and isolated myopathy) is considered outdated: the spectrum is broader and overlapping, and there are no patients with molecularly confirmed primary deficiency with isolated myopathy [14]. Age at onset ranges from birth to the seventh decade in COQ2 [14]. In infants, Saudubray describes multisystem disease with rapidly progressive nephropathy, often to end-stage kidney disease (ESKD), with deafness, optic atrophy, ataxia, dystonia, weakness and stroke-like episodes; some forms have prenatal onset and a poor response to treatment [1].

SRNS is the hallmark of primary CoQ10 deficiency, unusual in mitochondrial diseases; without high-dose CoQ10 it usually progresses to ESKD; it has been described with PDSS1, PDSS2, COQ2, COQ6 and COQ8B, and not with COQ4, COQ8A or COQ9 [14]. It usually manifests as proteinuria in infancy, often followed by encephalomyopathy with seizures and stroke-like episodes, severe neurological deficit and death; some patients have only SRNS in the first or second decade, with slow progression [14]. HCM occurs in neonatal-onset COQ2, in COQ4 (sometimes prenatal) and in COQ9 with neonatal lactic acidosis [14]. Sensorineural deafness is common in COQ6, retinopathy occurs with PDSS1, COQ2, COQ4 and COQ5, and optic atrophy with PDSS1, PDSS2 and COQ2 [14].

Source: GeneReviews, Primary Coenzyme Q10 Deficiency Overview, Table 1 [14].

Early-onset forms are concentrated in COQ4, COQ9, PDSS2 and neonatal COQ2. In COQ4, the original series of five patients included four with prenatal or neonatal onset and death within hours, and a fifth with slowly progressive neurodegeneration from 10 months, all with marked lactic acidaemia [78]; the largest cohort, of 11 patients from southern China, had five with classic neonatal encephalocardiomyopathy, and prenatal HCM due to COQ4 worsened despite CoQ10 [30]. PDSS2 has fewer than ten patients described, with SRNS ranging from infantile proteinuria to ESKD, collapsing-sclerosing renal lesions, severe infantile Leigh syndrome and a lethal neonatal form with SRNS, HCM, encephalomyopathy and lactic acidosis [31]. COQ8A presents between 18 months and 3 years with weakness and exercise intolerance, followed by cerebellar ataxia with severe cerebellar atrophy (SCAR9) [14]. COQ8B causes SRNS in the first and second decades, frequently progressing to ESKD [14]. In OMIM, COQ8B corresponds to nephrotic syndrome type 9 (NPHS9): in the initial series of 15 patients from 8 families, renal biopsy showed focal segmental glomerulosclerosis or its collapsing variant [29]. PDSS2 was identified in an infant with fatal encephalomyopathy and nephrotic syndrome (López 2006), and COQ4 has a founder variant in Ashkenazi Jews and a recurrent variant in southern China (section 2) [26, 28].

ACAD9 deficiency (treatable)

ACAD9 deficiency falls within the differential. The largest cohort, of 70 patients from 50 families, describes cardiomyopathy in 85%, usually hypertrophic, muscle weakness in 75% and exercise intolerance in 72%, usually with lactic acidosis [23, 33]. Age at onset divides the disease: most present in the first year, and in that subgroup 50% do not survive to 2 years; with later onset, 10-year survival exceeds 90% [23]. The brain is usually spared: severe intellectual disability in 1 of 70 and severe delay in 4, with more than 70% of patients performing activities of daily living like their peers [33]. The spectrum includes Leigh syndrome, macrocephaly, liver disease, Reye-like episodes with encephalopathy and seizures [33]. Saudubray summarises the presentation as HCM, myopathy and lactic acidaemia in infancy or childhood, with neurodevelopmental problems in some [75].

ATP synthase deficiency due to TMEM70

In the series of 48 patients from Europe, Turkey and Israel, onset was neonatal in 41, in infancy in 6 and at 2 years in 1 [36, 86]. At presentation, poor feeding, hypotonia, lethargy, respiratory and heart failure predominated, with the biochemical triad of lactic acidosis, 3-MGA and hyperammonaemia [36]. During follow-up, developmental delay was recorded in 98%, hypotonia in 95%, faltering growth in 94%, short stature in 89%, non-progressive cardiomyopathy in 89%, microcephaly in 71%, facial dysmorphism in 66%, hypospadias in 50% of boys, persistent pulmonary hypertension of the newborn in 22% and Wolff-Parkinson-White syndrome in 13%. There were acute metabolic crises in 24 survivors, often followed by regression [36]. Saudubray adds pulmonary hypertension and describes hyperammonaemia with high citrulline [90, 91].

GRACILE and other BCS1L phenotypes

GRACILE is a lethal AR disorder of the Finnish disease heritage: fetal growth restriction, tubular aminoaciduria, cholestasis, hepatic iron overload and severe neonatal lactic acidosis, with death of most in the neonatal period or early infancy [40]. Complex III deficiency nuclear type 1 manifests as encephalopathy, regression, ataxia, severe faltering growth, hepatic dysfunction, tubulopathy, weakness and exercise intolerance, and a fraction falls within the Leigh spectrum (less than 1% of nuclear Leigh syndrome), with deafness, proximal tubulopathy, liver involvement and pili torti [16, 40]. Björnstad syndrome, which is allelic, combines congenital sensorineural deafness and pili torti, without multiorgan failure [40]. Other early-onset complex III deficiencies include TTC19 (ataxia, severe olivopontocerebellar atrophy, survival to 20 or 30 years) and UQCRQ [16]. According to OMIM, about half of infants with GRACILE die in the first days of life despite intensive care, and the rest by 4 months; three British infants with BCS1L variants had a similar picture, with complex III deficiency and additional neurological symptoms [38]. In complex III deficiency nuclear type 1, onset is neonatal, with lactic acidosis, hypotonia, hypoglycaemia, encephalopathy, hepatopathy and tubulopathy, and BCS1L variants were found in one third of the Turkish patients with complex III deficiency in the de Lonlay series (2001) [7]. Intravenous apotransferrin followed by exchange transfusion, tested in two infants with GRACILE (Fellman 2000), transiently normalised transferrin saturation, but both died at 10 and 8 weeks; it is not an established treatment [38].

Isolated complex I deficiencies

Isolated complex I deficiency has an unknown prevalence (ORPHA:2609) [50]. Subunit and assembly genes mainly cause early Leigh syndrome, with features that guide the diagnosis: NDUFS4, early onset and HCM; NDUFS1 and NDUFV1, cystic leucoencephalopathy; NDUFS2, optic atrophy and HCM; NDUFS8, leucodystrophy and HCM; FOXRED1, seizures, myoclonus and a slowly progressive course; NDUFAF5, fatal infantile lactic acidosis or later onset with survival to young adulthood; NDUFAF6, dystonia [16, 49]. There is no targeted therapy for any complex I subunit gene [49]. NUBPL causes complex leucoencephalopathy with progressive cerebellar atrophy [1]. ACAD9, an assembly factor, is the treatable exception [23]. Isolated complex I deficiency is the most frequent enzymatic defect of OXPHOS [6]. In a series of 157 patients with a respiratory chain defect (von Kleist-Retzow 1998), the deficiency was of complex I in 33%, IV in 28%, I+IV in 28%, II in 4% and III in 7%; deficiencies of I and of I+IV were more often associated with cardiomyopathy (p<0.01) and liver failure (p<0.05), and parental consanguinity was 20% in IV deficiency, 28% in I+IV deficiency and 76% in North African families [8]. In a cohort of 109 children with isolated complex I deficiency from 101 families (Swalwell 2011), 29% of probands had a pathogenic mtDNA variant (21 in MT-ND genes and 8 in tRNA genes) and 38% an inferred nuclear defect; the most frequent presentation was Leigh or Leigh-like syndrome, and the median age at onset was 12 months in mtDNA disease and 3 months in nuclear disease, with overlap [6].

Isolated complex IV deficiencies: SCO2, SURF1 and others

The classic presentation of SCO2 deficiency is fatal infantile hypertrophic cardioencephalomyopathy: HCM in the neonatal period or first months, severe hypotonia and weakness, feeding difficulties, regression and lactic acidosis, with death of most in the first year from cardiorespiratory failure [43]. Anterior horn involvement may mimic spinal muscular atrophy, and there is a febrile syndrome without infection [16, 43]. In a national series, one compound heterozygote had neonatal cardiomyopathy with encephalopathy and five E140K homozygotes had progressive encephalopathy between 2 and 6 months, all with a poor prognosis [43]. At the mild end, compound heterozygotes for E140K or D135G with another missense variant near the CxxxC motif have early-onset axonal motor neuropathy without cardiomyopathy, which justifies cardiac assessment in SCO2 neuropathy [43]. OMIM also describes nystagmus, fasciculations, dystonia, electroencephalogram abnormalities and imaging consistent with Leigh syndrome, with neurogenic muscle atrophy more frequent than myopathy; three infants homozygous for E140K (Jaksch 2001) presented between 3 and 6 months and only developed HCM at 8, 11 and 18 months, with two of them dying at 8 and 13 months [42].

SURF1 deficiency is the most frequent isolated nuclear cause of Leigh syndrome (about 50% of Leigh syndrome with COX deficiency and about 30% of nuclear Leigh syndrome, according to the 2025 GeneReviews); the Oxford Desk Reference gives 75% of Leigh syndrome with COX deficiency, which corresponds to 10–20% of all Leigh syndrome [11, 16]. In the multicentre study of 44 patients, presentation was in infancy in 73% (median 9.5 months), with faltering growth (95%), hypotonia (93%), regression (71%), central respiratory failure (78%, median 31 months), hypertrichosis (41%) and a median survival of 5.4 years [46, 92].

Other early-onset complex IV genes: COX10 (deafness, HCM, anaemia), COX15 (seizures, retinopathy, HCM, hepatic steatosis), PET100 (prominent myoclonus and seizures, spastic tetraparesis, 50% survival to 20 years) and LRPPRC, responsible for the French-Canadian variant of Leigh syndrome, with crises of acidosis and ketosis, stroke-like episodes and a median age at death of 1.6 years [16, 48]. Fatal infantile COX deficiency (ORPHA:1561) has 16 recorded cases or families and a prevalence below 1/1,000,000 [44]. In SURF1 deficiency (MC4DN1), OMIM describes rapidly progressive neurodegeneration with loss of motor and cognitive skills between about 5 and 18 months, after normal initial development, with oculomotor abnormalities, deafness, apnoeas and death in childhood, often from central respiratory failure [8]. In LRPPRC deficiency (MC4DN5), OMIM adds episodic hypoglycaemia and mild facial dysmorphism; outside Quebec, 10 patients from 7 families have been described, four of them of Indian or Pakistani origin with the same homozygous variant, with almost complete loss of complex IV subunits (Oláhová 2015) [47].

Combined deficiencies due to translation defects (COXPD)

GFM1 (COXPD1). OMIM describes onset at or shortly after birth, with growth retardation, microcephaly, hypertonia with axial hypotonia, encephalopathy, cardiomyopathy and hepatic dysfunction, and death in the first weeks or years [5]. The classic picture is a rapidly progressive neonatal or early infantile hepatoencephalopathy, with lactic acidosis and liver failure, often fatal in the first months; there are also Leigh forms with axial hypotonia, spasticity and refractory seizures, dystonia, polymicrogyria and severe intellectual disability [51]. Seizures are more prominent than in TRMU deficiency [15].

TSFM (COXPD3). Severe infantile multisystem disease with metabolic acidosis, as encephalomyopathy or HCM; late forms progress from infantile cardiomyopathy to juvenile Leigh syndrome with neuropathy and optic atrophy [54].

TUFM (COXPD4). Hypotonia, respiratory failure, regression and lactic acidosis in infancy, with an initial presentation similar to that of TRMU deficiency [15]. Four infants from two families were identified by exome sequencing of 142 Japanese patients with childhood-onset respiratory chain disease, and complementation with normal TUFM restored the assembly of complexes I and IV in fibroblasts (Kohda 2016) [55].

MTO1 (COXPD10). In an international review of 35 cases, at presentation there was lactic acidosis in 62% and HCM in 44%; over the course of the disease, lactic acidosis in 100%, HCM in 79%, global delay or intellectual disability in 97%, seizures in 34% and optic atrophy in 52%. Two sisters without cardiomyopathy show that a normal echocardiogram does not exclude the diagnosis [59]. In 5 patients treated with a ketogenic diet, there was a favourable effect on seizures in 2 (O'Byrne 2018), a case series finding without a comparison group [58].

AARS2 (COXPD8). HCM with perinatal or infantile onset, hypotonia, myopathy, often pulmonary hypoplasia, sometimes non-immune hydrops, with the liver spared; mean age at death of 2.1 months and the oldest survivor aged 11 months [57].

ELAC2 (COXPD17). Severe infantile HCM in the first year with hypotonia, faltering growth and lactic acidosis; in a series of five patients from three families, the cardiomyopathy was mild to moderate in four and the neurological picture prominent [61].

Others. GTPBP3 is associated with early onset with HCM, MTFMT with cystic leucoencephalopathy with a milder course, MRPS34 with microcephaly, NARS2 with deafness and EARS2 with leucoencephalopathy with thalamus and brainstem involvement and high lactate (LTBL), in which magnetic resonance imaging (MRI) may improve [1, 16]. TSFM, MRPS22, MRPL44, MRPS14 and TMEM70 are among the causes of neonatal HCM [91]. Mitochondrial aminoacyl-tRNA synthetases usually cause severe combined OXPHOS disease, with relatively consistent phenotypes per gene: pontocerebellar hypoplasia type 6 (RARS2), leucoencephalopathy with brainstem and spinal cord involvement and lactate elevation (DARS2), Alpers syndrome (CARS2, FARS2, NARS2, PARS2), HUPRA (hyperuricaemia, pulmonary hypertension, renal failure and alkalosis; SARS2) and Perrault syndrome [1, 89]. The OMIM series also includes neonatal COXPD (COXPD2, 21 and 30), a prenatal form (COXPD39), hypotonia with lactic acidaemia and hyperammonaemia (COXPD5) and severe X-linked encephalomyopathy (COXPD6), all with an Orphanet prevalence below 1/1,000,000 [4].

Reversible infantile respiratory chain deficiency

Reversible infantile myopathy due to MT-TE. At a few weeks of life, a rapidly progressive myopathy appears with hypotonia, profound weakness and severe lactic acidosis; feeding by nasogastric tube or gastrostomy is usually necessary and some patients need ventilation until 12–18 months; gradual recovery begins at around 6 months [1]. Biopsy shows increased lipid and glycogen, ragged-red fibres and COX-negative fibres; the acidosis and histology normalise, with mild residual myopathy in adulthood; two siblings had basal ganglia lesions [66]. OMIM places spontaneous recovery after the first year. In one patient (DiMauro 1983), muscle COX activity rose from about 8% of control at 1 month to 47% at 7 months and to normal values at 36 months, and COX-positive fibres from less than 5% to all fibres; lactate normalised at 14 months, walking began at 16 months and at 33 months mild proximal weakness persisted with normal cognition [65]. Horvath and colleagues (2009) found homoplasmic m.14674T>C in all 17 patients from 12 families, with unaffected carriers, and muscle glutamate tRNA levels of 16–30% of control between 1 and 3 months, which then improved; m.14674T>G was described in two Japanese patients [65]. NAMDC describes spontaneous improvement up to 3 years and proposes the term "reversible infantile myopathy with respiratory chain deficiency" [13].

Transient infantile liver failure due to TRMU. Without treatment, symptoms begin between 2 and 4 months with transient acute hepatic dysfunction (raised transaminases and gamma-glutamyltransferase, synthetic failure, hepatomegaly, cholestatic jaundice), severe and persistent lactic acidosis, hypoglycaemia, hyperammonaemia and faltering growth [15]. Among 60 untreated symptomatic children, there was lactic acidosis in 44, acute hepatic dysfunction in 43, jaundice in 34, faltering growth in 33, vomiting or diarrhoea in 28, neurodevelopmental delay in 24, hypotonia in 15 to 20, hepatomegaly in 13, cardiomyopathy in 5 and seizures in 4; hypoglycaemia, present in 11 of 28, persisted despite dextrose and continuous feeding [15]. With support, liver function improves or normalises, lactate usually improves without normalising, and neurological dysfunction, cardiomyopathy and respiratory failure may persist or appear; 23% have recurrent episodes of liver failure in the first year [15]. MRI may show lesions of the thalami, putamen, basal ganglia and brainstem, with a lactate peak on spectroscopy in the acute phase, and may normalise after recovery; Leigh syndrome has been described in two patients and is probably underestimated [15]. Hyperalaninaemia with lactic aciduria distinguishes chronic lactic acidosis from that secondary to acute liver failure [15]. In the Zeharia series (2009), of seven Yemenite Jewish families, all infants were born at term and were admitted between 2 and 4 months; improvement began after 2–3 weeks of support, liver function normalised in 3–4 months and liver volume only between 3 months and 3 years; the authors proposed a window of risk between 1 and 4 months, linked to the limited availability of cysteine in that period [64].

Multiple mitochondrial dysfunctions syndromes (MMDS) and lipoate defects

BOLA3 (about a dozen patients) causes a neonatal form with hypotonia, respiratory failure, HCM and leucodystrophy, and an infantile form with motor regression, HCM, seizures, spasticity and optic atrophy, with death of most in the first 2 years [74]. NFU1 (20 patients) causes a neonatal form with bradycardia, apnoea and pulmonary arterial hypertension, with death by 4 months, and an infantile form usually fatal before 2 years, with often cavitating leucodystrophy [74]. ISCA2 (21 patients) usually begins between 3 and 7 months, after normal initial development, with almost constant optic atrophy, periventricular leucodystrophy sparing the U-fibres and a vegetative state within 1 to 2 years [74]. IBA57 (28 patients) has a neonatal form with polymicrogyria and arthrogryposis, an infantile form and a late form of spastic paraplegia with optic atrophy [74]. GLRX5 is the only iron-sulphur defect with sideroblastic anaemia [74]. Among the lipoate defects, LIAS causes a neonatal picture resembling severe non-ketotic hyperglycinaemia, LIPT2 encephalopathy with raised glycine and lactate, and LIPT1 raised lactate without raised glycine, with a fatal neonatal form or Leigh syndrome [74]. The term "atypical non-ketotic hyperglycinaemia" is historical and should be abandoned [74]. OMIM numbers these syndromes from MMDS1 to MMDS9B (table in section 1): NFU1 is MMDS1, with a founder variant in 9 Spanish patients with fatal disease; BOLA3 is MMDS2, with raised serum glycine and glycine in the cerebrospinal fluid (CSF) without the marked elevation of classic non-ketotic hyperglycinaemia; ISCA2 is MMDS4, with raised CSF glycine, complex II deficiency and a founder variant in Arab families; and ISCA1 is MMDS5, with early seizures, severe leucodystrophy and death in early childhood [9, 71–73].

Flavin deficiencies (treatable)

FAD synthase deficiency (FLAD1; 12 lineages) causes generalised weakness from the neonatal period to adulthood, dysarthria, dysphagia, scoliosis or respiratory failure, with cardiomyopathy and arrhythmias described, an acylcarnitine and organic acid profile of multiple acyl-CoA dehydrogenase deficiency and complex II (and I) deficiency [75]. Mitochondrial FAD transporter deficiency (SLC25A32) causes muscle weakness in childhood [75]. Riboflavin improved clinically and biochemically the two patients with SLC25A32 and some with FLAD1; others did not respond and several died in infancy [75].

Sengers, MEGD(H)EL and other 3-methylglutaconic acidurias

Sengers syndrome (AGK) has four forms: late, with HCM, congenital cataract, myopathy, exercise intolerance and normal mental development, with survival to the fourth or fifth decade; fatal neonatal encephalomyopathy; fatal neonatal hepatic dysfunction; and isolated cataract [67]. Of the 29 patients reviewed by Haghighi in 2014, 17 died (59%), 15 between 2 days and 11 months; tachydyspnoea and raised lactate were associated with a high risk of death [67]. Urinary 3-methylglutaconic acid is raised in 70% and complex I deficiency is constant in the myocardium [67]. In MEGD(H)EL (SERAC1), an acronym for 3-methylglutaconic aciduria, deafness, hepatopathy, encephalopathy and Leigh-like lesions, severe neonatal hepatic dysfunction is accompanied by hypoglycaemia (48%), hypotonia (91%), progressive spasticity and dystonia (82%), moderate to severe intellectual disability (88%), bilateral basal ganglia involvement on MRI (98%) and 3-MGA in 98% [67]. CLPB causes intellectual disability, congenital neutropenia, progressive cerebral atrophy, movement disorder and cataracts, and duplications of the ATAD3 locus cause neonatal multivisceral distress, fatal cardiomyopathy, corneal clouding and hyperlactataemia with 3-MGA [84, 91].

mtDNA depletion and succinyl-CoA ligase deficiency

mtDNA depletion syndromes are AR, begin in infancy or early childhood and, in most cases, progress rapidly to death in childhood, with hepatocerebral (DGUOK, POLG, TWNK, MPV17, SUCLG1), myopathic (TK2) and encephalomyopathic (SUCLA2, SUCLG1, RRM2B, ABAT, MGME1) forms [1]. SUCLA2 deficiency has a homogeneous picture of early-onset encephalomyopathy, dystonia, deafness and Leigh-like MRI, with mild MMA; SUCLG1 deficiency has a severe neonatal form with multiorgan failure, and HCM and liver involvement are exclusive to SUCLG1 [70]. OMIM adds, in MTDPS5 (SUCLA2), external ophthalmoplegia and variable renal tubular dysfunction, and in MTDPS9 (SUCLG1) survival to 12 years of a compound heterozygous patient with residual enzyme activity [68, 69].

Infantile mitochondrial cardiomyopathy

Respiratory chain disorders are a frequent cause of neonatal HCM, with heart failure due to hypertrophic-dilated cardiomyopathy, hypotonia, weakness, faltering growth and severe lactic acidosis [91]. HCM is the most common form, followed by DCM and left ventricular non-compaction; it is typically concentric and non-obstructive, with systolic dysfunction more frequent than in sarcomeric HCM [18]. DCM accounts for about 30% of mitochondrial cardiomyopathies, often secondary to previous HCM [18]. Specific early-onset causes include ACAD9, TMEM70, SCO2, AARS2, MTO1, ELAC2, TSFM, GTPBP3, BOLA3, AGK, COQ2, COQ4, COQ9, MRPL44 and MIPEP [14, 18, 33, 36, 43, 54, 57, 59, 61, 63, 67, 74]. According to the 2017 MMS standards of care, cardiac involvement was present in 30% of more than 200 patients with genetically confirmed PMD, and conduction defects in more than 10% [93]. Barth syndrome (TAFAZZIN) falls within the differential. In COXPD31 (MIPEP), OMIM describes left ventricular non-compaction as the main feature, with global delay and severe hypotonia, and more variably seizures and cataract (Eldomery 2016, four children); in COXPD8 (AARS2), the liver is spared and the deficiencies of I, III and IV are in the heart, muscle and brain [56, 62].

Mitochondrial pyrophosphatase deficiency (PPA2)

Mild PPA2 deficiency causes sudden death from heart failure after small amounts of alcohol; siblings with the same genotype without severe symptoms maintained strict avoidance of alcohol [1].

Treated and untreated disease

The natural history described in the cohorts corresponds, for most types, to disease without targeted therapy. The exceptions for which there are data on treated disease are ACAD9 deficiency (deaths in 7 of 22 treated versus 16 of 17 untreated patients with onset in the first year) [33], TRMU deficiency (survival beyond acute decompensation of 84% with L-cysteine and/or N-acetylcysteine versus 61% without) [15] and primary CoQ10 deficiency (response dependent on gene and severity) [14]; these data are in section 7.

4. Clinical Recognition and Red Flags

When to suspect

Suspicion rests on involvement of two or more apparently unrelated organs, on deterioration or disproportionate symptoms during periods of increased energy demand (infections, vaccination, exercise), on maternal inheritance and on consanguinity [18]. Any clinical clue from the central nervous system, muscle, eyes, ears, heart, pancreas, kidney or bone marrow combined with lactic acidosis is highly suggestive, and certain combinations (diabetes and deafness) are very suggestive [11]. Regression associated with infections is typical [1]. GeneReviews recommends considering mitochondrial dysfunction in the differential diagnosis of any progressive multisystem disorder and justifies a full assessment in a complex neurological picture or in a single neurological manifestation with involvement of another system [2]. Many children with childhood-onset encephalomyopathy are isolated cases, without a family history [2].

Red flags by age and context

In the newborn, congenital lactic acidosis presents with tachypnoea and a sepsis-like picture [1]; in the neonatal cohorts, respiratory distress, neonatal asphyxia, faltering growth, hypotonia, apnoea, cardiomyopathy, heart failure and hepatopathy predominate, with onset in the first two days in 73.5% [19], and neonatal hyperammonaemia and cardiomyopathy stand out in the Czech-Austrian cohort [20]. IUGR, prematurity and congenital anomalies are frequent [1, 19]. In infants, unexplained vomiting may herald episodes of acidosis with hyperventilation [78], and stepwise regression after infection or metabolic stress, with dysphagia, hypotonia, dystonia, ataxia, ophthalmoparesis, nystagmus and optic atrophy, suggests Leigh syndrome [1]. Acute liver failure in the first months, especially if recurrent or accompanied by lactic acidosis, points to TRMU, mtDNA depletion, GFM1 or SERAC1 [15, 91]. Neonatal HCM with severe lactic acidosis points to respiratory chain disease [91].

Urgent signs

Urgent signs are severe metabolic acidosis with an increased anion gap and dehydration [90], severe hyperammonaemia with lactic acidosis (TMEM70 deficiency, the only congenital lactic acidosis in which extracorporeal clearance should be considered) [90], heart failure, persistent hypoglycaemia, respiratory failure (hypotonia, central apnoea), pulmonary hypertension and any acute change in neurological status, which requires repeat neuroimaging [15, 21, 78]. Suspicion of mitochondrial disease should not delay treatment of treatable causes of lactic acidosis, such as acquired thiamine deficiency, in which an intravenous infusion of 100–200 mg of thiamine may be life-saving [90].

Clues to the treatable forms

Newborn screening

The sources consulted do not describe newborn screening programmes targeting early-onset mitochondrial encephalomyopathies. Biotinidase deficiency, which falls within the differential of lactic acidosis, is screened for in the neonatal period [16]. The Czech-Austrian cohort proposed a flowchart for critically ill newborns that allows direct molecular analysis, without biopsy, when the picture points to Alpers, Barth, Leigh syndrome due to MT-ATP6, Pearson, SCO1, SCO2, TMEM70, ATP5E, SUCLG1 or pyruvate dehydrogenase deficiency [20].

5. Diagnosis: Criteria, Tests and Algorithms

Formal diagnostic criteria

The 1996 adult criteria (Walker) were modified for children by Bernier and colleagues in 2002: applied retrospectively to 146 consecutive children referred for suspected respiratory chain disease, with "definite", "probable" and "possible" categories, they included paediatric clinical and histological features and a more sensitive coding of enzymes and functional studies, and agreed with the previously assigned diagnostic certainty in 99% of patients [13, 95]. In the same year, Wolf and Smeitink proposed the Mitochondrial Disease Criteria (MDC) for infants and children, which score clinical, metabolic, imaging and histopathological features and, separately, muscle biochemistry; in 30 children, the modified adult criteria and the MDC gave similar results, and no patient reached the "definite" category with the original adult criteria [96]. Morava and colleagues validated in 2006 the MDC score ("Nijmegen") in 61 children with multisystem disease and muscle biopsy: all 44 with genetically confirmed OXPHOS disease had a score above 6 ("probable") and 73% above 8 ("definite"), whereas none of the 17 with non-mitochondrial disease reached "definite"; the score is applicable before biopsy [97]. The modified Nijmegen score remains in use to classify paediatric cohorts [98].

The 2022 NAMDC consensus points out that a definitive diagnosis usually requires five types of data: biochemistry (with a marker such as citrate synthase), histology, metabolic markers, neuroimaging and molecular genetics [13]. In the genomic era, the diagnosis is mainly molecular, and the clinical-biochemical criteria serve to select patients, interpret variants of uncertain significance (VUS) and classify cases without a molecular diagnosis. For Leigh syndrome, GeneReviews adopts the 2023 McCormick criteria (typical imaging, a neurological feature and a biochemical or mitochondrial abnormality) [16].

NAMDC defines minimum criteria for each syndrome, built on a list of manifestations by system; encephalopathy requires at least one central neurological manifestation (cerebellar or sensory ataxia, chorea, dementia, parkinsonism, seizures, or stroke or stroke-like lesions) [13]. The table reproduces the criteria for the syndromes relevant to early onset.

Source: Emmanuele and colleagues, Tables 1 and 3 [13].

Initial biochemical investigation

The 2015 MMS consensus recommends, at the initial blood assessment, full blood count, CK, transaminases, albumin, lactate and pyruvate, amino acids and acylcarnitines, and urine organic acids; postprandial lactate is more sensitive than fasting lactate, the L/P ratio is only of value when lactate is raised, and 3-methylglutaconic acid should be quantified in plasma and urine; if cerebrospinal fluid (CSF) is collected, lactate, pyruvate, amino acids and 5-methyltetrahydrofolate should be measured [21]. The consensus used a Delphi method with 19 specialists, on literature of Oxford Centre for Evidence-Based Medicine grade 3 or lower, and the recommendations are consensus-based, without an individual grade of evidence [21].

Lactate may be appropriately raised in a crying child or after a seizure, and the most common cause of raised lactate is sampling technique; free-flowing blood is collected without a tourniquet, ideally through a venous catheter for multiple samples [1, 78]. Secondary causes should be excluded: hypoxia, hypoventilation, shock, hypoperfusion, sepsis, cardiac, pulmonary or liver disease, severe anaemia, anaerobic exercise and seizures [78]. In neonatal intensive care, glucose infusion above the capacity for utilisation produces artefactual hyperlactataemia, which corrects when the glucose concentration is reduced to 5% or even 2.5% [78, 90].

Specific metabolites point to groups of genes, as summarised in the table.

Source: Saudubray 2022, Table 10.5 [1].

Low citrulline and arginine occur in maternally inherited Leigh syndrome, neuropathy, ataxia and retinitis pigmentosa (NARP) syndrome, MELAS and Pearson syndrome; some patients with MT-ATP6 have raised C3 or C5-OH, mimicking multiple carboxylase deficiency [1]. CK is raised mainly in cytochrome b disorders and in TK2 deficiency [21]. Plasma CoQ10 is useless, because it reflects diet, and leucocyte CoQ10 is inadequate for diagnosing primary deficiency, which requires muscle CoQ10 [14, 21].

Multisystem assessment

Screening for organ involvement includes echocardiogram and electrocardiogram (ECG), renal tubular and endocrine function (pituitary, thyroid, parathyroid, pancreas, adrenal), ophthalmological and audiological assessment, repeated over time [1]. In cardiology, 12-lead ECG and transthoracic echocardiogram; the ECG is abnormal in 68% of adults with PMD (Limongelli 2010, according to Clinical Cardiogenetics), with T-wave inversion as the most common abnormality; HCM is defined by unexplained left ventricular thickening greater than 15 mm or +2 standard deviations (Z); cardiac MRI is the reference standard for function and fibrosis, and Holter monitoring studies conduction and arrhythmias [18].

Neuroimaging

The MMS recommends brain MRI when there is central nervous system involvement, with spectroscopy (MRS) for parenchymal lactate; stroke-like lesions outside a vascular territory, diffuse white matter disease and bilateral involvement of the deep grey nuclei are classic, but neither sensitive nor specific, and neuroimaging is never an absolute criterion [21]. Some images point to specific genes.

Source: Saudubray 2022, Table 10.3 [1].

Genetic tests and order of use

Saudubray recommends considering exome or genome sequencing early in the work-up, depending on availability, and leaving functional studies on biopsy until later, or performing them in parallel in acutely ill patients [1]. The 2015 MMS prefers whole mitochondrial genome sequencing to point mutation panels; recommends testing another tissue when suspicion of mtDNA disease is strong and blood is negative; considers urine more informative than blood for some heteroplasmic variants; recommends NGS with complete coverage of the known nuclear genes, advises against single-gene testing and recommends considering the exome if the panel is negative [21]. Trio sequencing is recommended for interpretation [18]. GeneReviews describes two options: a multigene panel guided by inheritance and distinguishing features, or comprehensive genomic testing (exome, genome, mitochondrial sequencing) that analyses nDNA and mtDNA simultaneously [2]. Targeted variant or single-gene tests are justified only in very characteristic pictures with few variants [1].

The yield varies with cohort selection. In an Australian paediatric cohort of 40 children, trio genome sequencing from blood gave a definitive diagnosis in 55% and a probable or definitive diagnosis in 67%, with causal mtDNA variants in 4 [98]. In the 100,000 Genomes Project, in 345 patients with suspected mitochondrial disease, a definitive or probable diagnosis was obtained in 31% of families, and 63% of the diagnoses were non-mitochondrial disorders (developmental disorders, epileptic encephalopathies, other metabolic disorders, cardiomyopathies, leucodystrophies) that would have escaped a targeted approach [100]. Reanalysis of undiagnosed exomes from patients with multiple respiratory chain deficiencies gave a possible or probable diagnosis in 8 of 14 (57%) [101]. In the Japanese neonatal cohort, based mainly on biochemical diagnosis, a molecular diagnosis was obtained in 29.9%, 81% of the variants were nuclear and the yield was highest in combined deficiencies (50.7%) [19].

Paediatric "genetics-first" algorithm (Wortmann 2017)

In the classic approach, called "biopsy first" or function-to-gene, clinical characterisation and OXPHOS enzymology in tissue guided Sanger sequencing of candidate genes, one by one. Muscle complex I deficiency, the most frequent biochemical signature of mitochondrial diseases, shows the limits of that pathway: no clinical sign is pathognomonic of CI deficiency or of a gene; Leigh syndrome had more than 75 monogenic causes in 2016 (Lake 2016, cited by the guideline), a number now above 100 (section 3); siblings with the same variant and the same environment may have different courses, as in m.3243A>G disease; metabolites (lactate in blood or CSF, urinary Krebs cycle metabolites and 3-methylglutaconic acid) reflect non-specific mitochondrial dysfunction; and neuroimaging suggests the gene only in a subgroup, such as the rapidly progressive leucoencephalopathy of NDUFS1 deficiency. CI deficiency in muscle may be isolated, which suggests a subunit defect, or combined with deficiency of complexes III, IV and V, which points to more general defects (supercomplex formation, translation), broadening the list of candidates. Therefore, despite repeated investigations in blood, urine and CSF, MRI and MRS and exercise testing, most children studied by this route remained without a genetic diagnosis [22].

The guideline of the Salzburg, Munich and Bern group replaces that pathway with the "genetics-first" approach, from which reverse phenotyping (from gene to function) follows. It is an article from a single group, funded by the E-Rare GENOMIT project, without a formal consensus method or grading of recommendations [22]. Before ordering the test, the physician should ask whether the mitochondrial dysfunction is inherited rather than acquired, with a detailed family history, complete physical and neurological examination and targeted tests (metabolic work-up, MRI, audiometry, echocardiogram, exercise testing). Suspicion rests on three types of data: suggestive personal or family history, suggestive multisystem signs and symptoms, and suggestive metabolites (serum lactate, plasma alanine, urinary lactate, Krebs cycle metabolites, 3-methylglutaconic acid). The patient is considered stable when there is no suspicion of a disorder requiring its own management, such as the ketogenic diet in PDHc deficiency, or of a disorder with tissue specificity, such as isolated myopathy. In the stable child, blood from the proband, a skin biopsy for fibroblast culture from the proband and blood from the parents are collected, and exome sequencing is performed on leucocyte DNA (a few millilitres of blood in EDTA), of the proband or as a trio, in all patients with suspected disease, whatever their age, sex or disease stage; genome sequencing from a dried blood spot has also been used successfully [22].

The trio is the preferred option for two reasons: it immediately proves the segregation of the candidate variants and identifies disorders that mimic mitochondrial disease, often de novo, such as copy number variants (microdeletion syndromes), multiple malformation syndromes or hereditary intellectual disability. Alternatively, targeted genetics of up to 3 genes is acceptable when the clinical and metabolic combination is very suggestive, such as MELAS (MT-TL1), Alpers-Huttenlocher syndrome (POLG) or Barth syndrome (TAZ). For practical and cost-effectiveness reasons, the exome should use a capture kit with coverage above 97% and good ability to detect copy number variants. The diagnostic yield of large exome studies is about 40%, reaching 60% in selected subgroups, such as CI deficiency; the exome, including interpretation, takes from several weeks to 3 months [22].

Interpretation requires clinical, genetic and biochemical knowledge of mitochondrial diseases. Each person, affected or not, has about 21,000 coding variants in their roughly 20,000 genes, of which about 100 are private, mostly normal variants. In-house exome databases (Munich's had more than 10,000 in 2017) and population databases (ExAC, now gnomAD, then with 120,000 exomes and 15,000 genomes) are essential: a frequent variant, above a maximum allele frequency of 1%, or one observed in homozygosity in those databases is unlikely to be causal, with the caveat that it may be a low-penetrance variant [22].

The timing of muscle biopsy is decided according to the following table.

Source: Wortmann 2017, Table 1 and Figure 2 [22].

In patients with a rapidly progressive or potentially fatal course, the guideline allows empirical cofactor therapy to be considered while the work-up proceeds (regimen and dose comparison in section 7) [22]. The exome result falls into one of four scenarios, which define the next step.

Source: Wortmann 2017, text and Figure 2 [22].

With functional confirmation, the diagnosis of mitochondrial disease is established. Without it, the guideline proposes a case conference with three questions (is mitochondrial disease still likely? is an inherited disease still likely? has the affected tissue been studied?) and, after it, the search for other diagnoses or a restart of the pathway with new tests. In scenario 4, RNA sequencing of fibroblasts to detect hidden splicing variants gave a molecular diagnosis in about 10% of cases in the work cited by the guideline (Kremer, 2016 preprint), and even synonymous variants may alter splicing, as in a case of infantile Barth syndrome due to a synonymous substitution in TAZ [22].

The guideline diverges from the other sources of this article on three points. It starts with the exome and does not mention mtDNA analysis, which the 2015 MMS recommends by whole mitochondrial genome sequencing, with testing of another tissue if suspicion is strong and blood is negative [21]; it accepts targeted genetics of up to 3 genes, whereas the MMS advises against single-gene testing [21]; and it proposes systematic skin biopsy to store fibroblasts, whereas the MMS reserves biopsy for cases without DNA confirmation and points out that fibroblasts may give false negatives [21], a limitation the guideline itself acknowledges by preferring muscle in scenario 4 [22]. The yield of about 40% is consistent with that of later cohorts, of 31–67% depending on selection, and the high proportion of non-mitochondrial diagnoses in the 100,000 Genomes Project confirms the advantage of the trio in detecting disorders that mimic mitochondrial disease [98, 100].

Limitations of the tests

A negative result does not exclude the disease. Low heteroplasmy in blood is detectable down to 5–10%, or 1–2% with recent methods; heteroplasmic variants may be undetectable in blood and require muscle or urinary epithelium, and some large-scale deletions are detectable only in muscle [2, 21]. The heteroplasmy of many mtDNA variants tends to decrease in leucocytes over a lifetime and may increase in post-mitotic tissues, such as muscle, and in pure myopathy due to mtDNA mosaicism the variant may be detectable only in muscle DNA [22]. mtDNA depletion is not assessed by mtDNA NGS and requires quantification by real-time polymerase chain reaction (PCR) in tissue; rearrangements require long-range, real-time or droplet digital PCR, preferably in muscle [1, 21]. Short-read exome and genome sequencing may miss deletions and duplications of more than 8–10 bp, expansions and epigenetic changes, with false negatives varying by region [2]. Tissue specificity (different thresholds, isoforms, X inactivation, heteroplasmy, somatic mosaicism) makes it necessary to reassess the affected tissue when genetics in blood is negative; in POLG, for example, muscle may be normal and liver abnormal [1]. Skin biopsy is not useful in TRMU deficiency, because the protein is not required for mitochondrial function in skin [15], and in CoQ10 deficiency there are genetically proven patients with normal CoQ10 in muscle and fibroblasts [14].

Biopsy, histochemistry and enzymology

The MMS recommends muscle (and/or liver) biopsy when the diagnosis is not confirmed by DNA, preferably open, from the vastus lateralis, with histochemistry for COX, SDH, NADH-tetrazolium reductase (NADH-TR), combined SDH/COX and electron microscopy, the latter strongly recommended in paediatrics, because the histological findings are often limited; extra tissue should be frozen [21]. Tissue biochemistry does not always distinguish primary disease from secondary dysfunction: respiratory chain enzymology (spectrophotometry of complexes I to IV) is performed on frozen tissue or fresh mitochondria from the affected tissue, normalised to citrate synthase and/or complex II; activities above 20% of the control mean are interpreted with caution, PMD is not diagnosed on tissue biochemistry alone and normal enzymology does not exclude it [21]. ATP synthesis by ATP synthase requires intact mitochondria; the combined I+III and II+III activities detect CoQ10 deficiency; native gel electrophoresis (BN-PAGE) studies assembly and supercomplexes [1]. Samples have their own requirements: fresh tissue on wet ice for respiration and cell culture, tissue frozen in liquid nitrogen for enzymology, histochemistry and BN-PAGE, fixed tissue for electron microscopy, EDTA blood for genetics, and urine, buccal swab or hair roots for mtDNA [1]. Fibroblasts may give false negatives, and the buccal smear should not be first line [21]. The 2017 paediatric guideline adds skin biopsy for fibroblast culture and storage in all patients with suspected disease, at the time of the initial genetic work-up, and defines when muscle biopsy is performed in parallel with genetics or after it (subsection on the paediatric algorithm); the practical details of muscle biopsy are in an online supplement to that article [22]. COX deficiency in SURF1 deficiency is more marked in fibroblasts than in muscle [16], and in TRMU deficiency hepatic complex IV is always the lowest [15].

Algorithm for congenital lactic acidaemia

The Atlas algorithm for congenital lactic acidaemia follows these steps: document raised lactate, pyruvate or alanine in blood, CSF or urine; exclude organic acidaemias and β-oxidation defects by organic acids and acylcarnitines; study mtDNA in blood; without an mtDNA abnormality, distinguish a gluconeogenesis defect from an oxidative defect by a prolonged fast with glucagon (0.5 mg intramuscularly at 6 hours of fasting, blood glucose at 15, 30, 45, 60 and 90 minutes, fasting up to 18–24 hours if euglycaemic and a second glucagon dose at the end); if the fast fails, study biotinidase, carboxylases and loading tests; if the fast is passed, study the pyruvate dehydrogenase complex in fibroblasts and, with raised L/P and 3-hydroxybutyrate/acetoacetate ratios, perform a muscle biopsy for the respiratory chain complexes and mtDNA [78]. This algorithm predates the widespread use of exome and genome sequencing, which now come in early [1, 18]. A practical integration of the sources is as follows: (1) stabilisation and initial biochemistry during a crisis, with reserve samples; (2) exclusion of secondary causes and of organic acidaemias, β-oxidation defects, biotinidase deficiency and pyruvate carboxylase deficiency; (3) whole mtDNA sequencing and trio exome or genome sequencing, urgently in the critically ill patient; (4) biopsy of affected tissue with enzymology, CoQ10 and mtDNA quantification if genetics is inconclusive or to validate a VUS; (5) immediate start of the low-risk empirical therapies that cover treatable forms (section 7) [1, 18, 21]. The 2017 paediatric guideline moves skin biopsy to step 3, to store fibroblasts, and brings muscle biopsy forward to that same step in the five situations of the paediatric algorithm table [22].

Content of the genetic report and interpretation

The report should state the gene, the reference transcript with version, the variant in HGVS nomenclature, the zygosity or, for mtDNA, the heteroplasmy and the tissue analysed, and the class according to the American College of Medical Genetics and Genomics and the Association for Molecular Pathology (ACMG/AMP). Segregation helps to classify a VUS: a de novo variant in an AD gene with healthy parents is often likely pathogenic, and a VUS in trans with a pathogenic variant in an AR gene gains weight; with no variant identified but high suspicion, first-degree relatives may be assessed clinically [18]. Tissue biochemistry is now used mainly for cases without a molecular diagnosis and to confirm VUS [14]. In the clinical case in section 10, the missense variant was classified as a VUS (PM2, PP3) and the deletion as likely pathogenic (PVS1, PM2), and the reduced mtDNA copy number supported the interpretation [63].

Prenatal and preimplantation diagnosis

In nuclear disease, prenatal diagnosis and preimplantation genetic testing (PGT) are straightforward when the variants are known, as in any Mendelian disorder; biochemical prenatal diagnosis in cultured amniocytes is possible when the biochemical abnormality is known, but biochemical methods have often failed [1, 2, 78]. In mtDNA, the load in chorionic villi may not reflect other fetal tissues and may change with development; m.8993T>G and m.8993T>C have a more uniform and stable distribution, which has allowed prenatal diagnosis by amniocentesis (about 15–18 weeks) or chorionic villus sampling (about 10–12 weeks) [2]. Detailed counselling is in section 8.

6. Differential Diagnosis

The differential has two levels: distinguishing primary mitochondrial disease from secondary causes of lactic acidosis and from other metabolic and non-metabolic disorders with the same phenotype, and distinguishing the types of the class from one another, especially the treatable ones. The table brings together the main differentials.

Secondary and iatrogenic lactic acidosis

The first step is to confirm that the rise in lactate is real and primary: the most common cause is sampling technique, and secondary causes (hypoxia, hypoventilation, shock, hypoperfusion, severe anaemia, seizures) must be excluded [78]. Two errors deserve attention in the newborn and the infant in intensive care: hyperlactataemia sustained by a high glucose infusion, which corrects when the concentration is reduced to 5% or 2.5%, and acquired thiamine deficiency, which mimics congenital lactic acidosis with heart failure and responds to intravenous thiamine [90]. D-lactic aciduria, due to intestinal bacteria in malabsorption or short bowel syndromes, may cause systemic acidosis and coma and requires a specific D-lactate assay [78]. In the approach to metabolic acidosis in inborn errors of metabolism, the criteria are: bicarbonate below 18 mmol/L with pH below 7.38, most with an increased anion gap and ketonuria as the main key; without ketosis, consider pyruvate dehydrogenase deficiency and β-oxidation and gluconeogenesis defects; with ketosis, organic acidaemias and congenital lactic acidoses such as pyruvate carboxylase deficiency and multiple carboxylase deficiency [91].

Other metabolic disorders

Organic acidaemias (propionic, methylmalonic, isovaleric), HMG-CoA lyase deficiency and pyroglutamic aciduria cause secondary lactic acidaemia and are excluded by organic acids; multiple carboxylase deficiency acts directly on pyruvate carboxylase; β-oxidation defects are excluded by acylcarnitines [78]. Patients with respiratory chain defects may have secondary changes in fatty acid oxidation that mimic short-chain (SCAD; C4) or medium-chain (MCAD; C6, C8, C10) acyl-CoA dehydrogenase deficiency in fibroblasts [78], and the reverse also occurs: multiple acyl-CoA dehydrogenase deficiency causes secondary CoQ10 deficiency [14]. The pattern of lactate after a meal helps: it rises more than twofold in pyruvate dehydrogenase deficiency and in glycogen storage diseases 0, III and VI/IX, falls in glycogen storage disease I and in gluconeogenesis defects, and in primary respiratory chain defects the redox state may worsen, with paradoxical ketonaemia [78].

Cardiomyopathy in infancy

In the presence of HCM in an infant, the differential includes sarcomeric genes, inborn errors of metabolism, neuromuscular diseases and syndromes [18]. The combination of heart failure with hypertrophic-dilated cardiomyopathy, hypotonia, weakness and faltering growth suggests β-oxidation defects (with hypoglycaemia), respiratory chain disorders (with severe lactic acidosis), Pompe disease (vacuolated lymphocytes) or fatal congenital cardiac glycogenosis due to PRKAG2; 3-MGA suggests Barth syndrome and 2-oxoglutaric acid oxoglutarate dehydrogenase deficiency [91]. Hyperammonaemia with raised lactate and cardiomyopathy, with high citrulline, points to TMEM70 [91].

Liver failure in infancy

The GeneReviews on TRMU deficiency gives a targeted differential: DGUOK (similar onset, metabolic profile and progression), glycogen storage disease Ia (some patients with TRMU were initially thought to have glycogen storage disease), GFM1 (more prominent seizures), methylmalonic and propionic acidaemia (first weeks), MPV17 (later onset, acral ulcers, more neurological signs), MTO1 (HCM), MT-TE (reversible disease, mainly myopathic), POLG (the most common mitochondrial cause of liver failure in infancy), SERAC1 (similar reversible liver failure and metabolic profile), TUFM and Leigh syndrome [15]. Neonatal liver failure also has non-mitochondrial causes, such as galactosaemia, hereditary fructose intolerance, tyrosinaemia type I, neonatal haemochromatosis, transaldolase deficiency, alpha-1 antitrypsin deficiency and LCHAD deficiency, and recurrent acute liver failure triggered by fever suggests NBAS or LARS1; GRACILE belongs to this group [1, 91]. Recurrent and neonatal acute liver failure of monogenic cause falls within this differential.

Differential within the class

The distinction between types rests on combinations of signs: SRNS (CoQ10), HCM with preserved cognition (ACAD9), hyperammonaemia with 3-MGA and hypospadias (TMEM70), cholestasis with iron overload and aminoaciduria in a newborn with IUGR (GRACILE), HCM with a spinal muscular atrophy-like picture and fever without infection (SCO2), liver failure at 2–4 months (TRMU), profound myopathy with a COX-negative biopsy (MT-TE), congenital cataract with HCM (Sengers), neonatal hepatopathy with a putaminal "eye" (MEGDEL), raised glycine with cavitating leucoencephalopathy and pulmonary hypertension (MMDS), mild MMA with dystonia and deafness (SUCLA2) and hypertrichosis with central respiratory failure (SURF1) [14, 15, 33, 36, 40, 43, 46, 66, 67, 70, 74]. The overlap is large and the definitive diagnosis is genetic. Combined oxidative phosphorylation deficiencies due to defects of mitochondrial translation (COXPD) belong to this group.

7. Treatments and Risk Management Strategies

Principles and quality of the evidence

There is no curative treatment for the great majority of mitochondrial diseases, and GeneReviews states that no treatments are known to alter the course of mitochondrial disease in general [1, 2]. The exceptions with targeted therapy are rare and concentrated in defects of cofactors and of accessible metabolic pathways: primary CoQ10 deficiency, ACAD9 deficiency, flavin deficiencies, TRMU deficiency, PPA2 deficiency, defects of thiamine transport and activation, and TK2 deficiency [1, 16, 24]. Supportive care is the mainstay of treatment: screening for and proactive treatment of complications by a multidisciplinary team, with follow-up at a tertiary centre [1, 18, 93].

The evidence is mostly consensus-based. The 2015 MMS diagnostic and treatment consensus and the 2017 standards of care used Delphi methods on low-grade literature [21, 93]; the 2022 Australian adaptation was produced without a new Delphi process [99]; the 2020 drug safety consensus corresponds to level 4 evidence [103]; the supplement doses come from a 2009 table of opinion from MMS members, because the 2015 consensus does not give doses (these would be in an online supplement that was not accessed) [21, 104]; and the doses for cofactor defects come from the 176th European Neuromuscular Centre (ENMC) workshop, observational cohorts and reviews [15, 23, 105]. The randomised trials available in the general paediatric population with PMD were negative (table of trials). Almost all supplements and cofactors are used off-label and have no Food and Drug Administration (FDA), European Medicines Agency (EMA) or INFARMED label for the mitochondrial indication [31, 104].

Acute treatment: metabolic crisis and intercurrent illness

The written emergency plan, updated at each visit, should summarise the diagnosis, symptoms, baseline cardiorespiratory function, nutritional needs, medication, allergies, contraindicated therapies and the anticipatory plan; patients should be triaged and transferred early to a centre able to manage multisystem dysfunction [93, 99]. The precipitants are catabolism, infection, fasting, anaesthesia and surgery [93].

Most newborns with primary hyperlactataemia present with acute metabolic acidosis with an increased anion gap and dehydration, and support is similar to that for branched-chain organic acidurias; it is usually enough to lower lactate to a level without severe acidosis, and no defect requires toxin removal, except TMEM70 deficiency [90]. The acidosis may require large amounts of sodium bicarbonate, and sodium citrate may be ineffective in oxidative defects [78].

Acute assessment includes routine biochemistry, glucose, transaminases and lactate, with attention to cardiac and neurological decompensation; electrolytes (sodium, potassium, calcium, magnesium, phosphate), with tubular and adrenal studies if the imbalances recur; echocardiogram if there is haemodynamic instability; and repeat neuroimaging in the event of any neurological change [21, 93]. Potentially toxic drugs are stopped and metabolic abnormalities corrected [21]. In bone marrow failure, transfusions are used, and granulocyte colony-stimulating factor (G-CSF) in severe neutropenia [93].

Hyperammonaemic crises in TMEM70 deficiency. TMEM70 deficiency is the only congenital lactic acidosis in which extracorporeal clearance should be considered [90]. In the series of 48 patients, crises responded to glucose infusion and lipid emulsion, ammonia scavengers and haemodiafiltration, and the outcome depended mainly on adequate treatment of hyperammonaemic crises in the neonatal period and early childhood [36].

Severe thiamine deficiency. In Shoshin beriberi, in preterm infants on parenteral nutrition without B vitamins and in leukaemia, an intravenous infusion of 100–200 mg of thiamine may be life-saving [90].

Anaesthesia and the perioperative period

Almost all general anaesthetics reduce mitochondrial function, especially volatile agents and propofol [21]. The recommendations converge on preventing catabolism: minimise preoperative fasting, place the patient first on the list, add glucose to intravenous fluids (unless on a ketogenic diet or with an adverse reaction to high glucose), avoid lactate-containing solutions and avoid hypothermia, hypoglycaemia, acidosis and postoperative vomiting [16, 21, 99, 103]. Anaesthesia may worsen respiratory function, so it is monitored before, during and after; neuromuscular blocking agents require caution and monitoring in the myopathic phenotype; local anaesthetics are well tolerated; and there is no clear link with malignant hyperthermia, although similar reactions have been described [16, 21, 99]. Some patients, especially those with reduced complex I capacity, are more sensitive to volatile agents, and sevoflurane appears to be better tolerated than isoflurane or halothane (class IV) [104]. The preoperative respiratory assessment includes oximetry and blood gases if saturation is below 95%, preoperative non-invasive ventilation if the cough is weak, there are recurrent pneumonias or the maximal expiratory pressure is low, and extubation to non-invasive ventilation [93].

The societies diverge on propofol and volatile agents. The 2015 MMS asks for caution with volatile agents and advises avoiding propofol or limiting it to short procedures [21]; the 2022 Australian adaptation advises avoiding volatile agents and propofol [99]; the 2020 Delphi consensus considers volatile agents and propofol safe for induction (the latter with non-strong consensus, mean 3.81) and propofol infusion up to 48 hours in refractory status epilepticus [103]. The GeneReviews on the Leigh spectrum points out that prolonged propofol infusion for maintenance may increase the risk of lactic acidosis [16], and the GeneReviews on TRMU deficiency includes it among the mitochondrial inhibitors to avoid [15].

Drugs to avoid or to use with caution

Sources: MMS 2015 [21]; 2020 international Delphi consensus, with 16 experts and two representatives of International Mito Patients (IMP), on 46 drugs or classes [103]. The 2017 MMS precautions table also includes vigabatrin, because it inhibits the mitochondrial nucleoside salvage pathway, and stresses that, except for valproate in POLG-related disorders, none of these drugs is contraindicated and all may be used with caution [93].

The 2020 consensus concluded that the 46 drugs or classes assessed are generally safe, with restrictions for specific molecular defects and situations, and pointed out that only for 18 of 44 drugs was the preclinical toxic concentration comparable to therapeutic levels and that infection is a much greater risk than antibiotics [103]. It also recommends adjusting the dose in renal failure and using acidifying drugs with caution [103]. The practical recommendation is to prescribe on an individual basis, according to the 2020 consensus [1, 16].

Chronic supportive treatment by organ

Epilepsy. Seizures occur in 35–60% of infants, children and adolescents with confirmed disease; treatment is similar to that in the general population, with a low threshold for electroencephalography (EEG), and epilepsia partialis continua should raise suspicion of a metabolic stroke [93]. Levetiracetam and clobazam appear to be the most effective antiepileptic drugs in mitochondrial epilepsy [1].

Movement disorders. Dystonia is treated with trihexyphenidyl, baclofen, tetrabenazine and gabapentin, alone or in combination, starting at a low dose; botulinum toxin is used with caution, because it may worsen weakness, and never in the head or neck, because of the risk of dysphagia [46]. Deep brain stimulation may be considered [93].

Cardiology. Treatment of cardiomyopathy follows the European Society of Cardiology (ESC) guidelines for HCM and heart failure (diuretics, angiotensin-converting enzyme inhibitors, ACE inhibitors, and a beta-blocker), with atenolol as the preferred beta-blocker because it does not inhibit mitochondrial function in vitro [18]. Ablation is recommended in supraventricular tachycardia and Wolff-Parkinson-White syndrome, a low threshold for pacemaker implantation, an implantable cardioverter-defibrillator if there is a risk of sudden death, left ventricular thickness greater than 30 mm or ventricular tachycardia, and heart transplantation in end-stage failure, weighing the comorbidities and the natural history of the specific disease [18, 93]. In ACAD9 deficiency, of the four transplanted patients, the two with onset in the first year died despite transplantation and the two with onset after the first year developed normally [33]. In TMEM70 deficiency, ECG surveillance looks for pre-excitation [36].

Respiratory and nutrition. Children with more severe disease have earlier and more severe respiratory complications; baseline assessment, oximetry at every visit and polysomnography to be considered at diagnosis are recommended, with physiotherapy and cough assistance and limitation of opioids [93]. Gastrostomy is performed with a low threshold, surveillance for malnutrition at least every 6 months, and acute pseudo-obstruction is treated aggressively [46, 93].

Kidney, liver, endocrine, blood and senses. Tubulopathy requires electrolyte replacement, sometimes in large volumes, with early gastrostomy if necessary; dialysis is considered palliative and ideally accompanied by assessment for renal transplantation [93]. Saudubray also lists ptosis surgery, hearing aids or cochlear implants, hormone replacement (growth hormone, thyroxine, insulin, hydrocortisone), transfusions and, occasionally, organ transplantation for isolated end-stage involvement in an otherwise healthy child [1].

Diet. In oxidative defects, often glucose-sensitive, lactate falls with a high-fat diet (at least 50% of calories from fat, without having to be ketogenic) [78]. Clinical Cardiogenetics mentions the ketogenic diet in disorders due to complex I defects [18], and the Australian adaptation considers a high-fat diet in refractory epilepsy, in complex I deficiency or in an acute setting, after β-oxidation defects have been excluded [99]. On the other hand, the 2015 MMS found no evidence for adjusting the diet on the basis of respiratory chain results [21], and the GeneReviews on the Leigh spectrum found no evidence of efficacy of the ketogenic diet outside pyruvate dehydrogenase deficiency and drug-resistant epilepsy [46]. The ketogenic diet is standard treatment in pyruvate dehydrogenase deficiency and is contraindicated in β-oxidation defects and in pyruvate carboxylase deficiency [104].

Exercise and vaccination. Supervised, progressive aerobic training started at low intensity, after cardiac screening, increases mitochondrial content and strength without a negative heteroplasmic shift [21]; the Australian adaptation proposes moderate intensity at 70% of maximum heart rate, except with arrhythmia or cardiopulmonary involvement [99]. The MMS recommends vaccinating all patients, because the risk of severe infection outweighs the theoretical risk of decompensation [93, 104].

Targeted therapies for the treatable types

Primary CoQ10 deficiency: ubidecarenone (CoQ10)

CoQ10 (ubidecarenone, ubiquinone) is marketed as a food supplement and is not a medicine approved by the FDA or the EMA; no regulatory label defines dose, indication or age, and it crosses the blood-brain barrier poorly [31]. There are no published clinical practice guidelines for primary CoQ10 deficiency [14]. Treatment should start as early as possible, because it may limit progression and reverse some manifestations, whereas established severe neurological or renal damage does not reverse; at-risk siblings should be assessed early so that CoQ10 can be started [14]. With SRNS, angiotensin-converting enzyme inhibitors may be combined with CoQ10, and renal transplantation is an option in ESKD [14].

The sources diverge on the dose. The reference is that of the ENMC 176 workshop consensus: 10–30 mg/kg/day in children and 1200–3000 mg/day in adults, orally, which appear effective in patients with COQ2 variants, especially for neurological and renal manifestations; less than 5% of oral CoQ10 reaches the plasma [105]. By comparison, the 2023 GeneReviews gives 5–50 mg/kg/day, with soluble formulations apparently more bioavailable [14], as does the Brazilian book on genetic syndromes [106]; Saudubray, citing the same workshop, gives at least 30 mg/kg/day in childhood, up to 3 g/day in adults [1, 107]; and the 2009 MMS opinion table, for mitochondrial disease in general, gives ubiquinone 10–30 mg/kg/day in two doses in children and 300–2400 mg/day in two to three doses in adults, or ubiquinol 2–8 mg/kg/day in two doses in children and 50–600 mg/day in adults, with better absorption with meals [104]. The 2009 MMS describes no increase in plasma levels with 3000 mg/day compared with 2400 mg/day in adults (class IV) and the absence of an intravenous formulation [104]. Ubiquinol, considered by the 2015 MMS to be the most bioavailable form, with dose adjustment [21], has no dose or efficacy data in primary deficiency according to GeneReviews [14]. The adverse effects described are wakefulness and sleep disturbance, and CoQ10 may reduce warfarin concentrations [104]. CoQ10 levels in plasma and leucocytes are used to monitor absorption and adherence [21].

The response is highly variable. In COQ2, it is poor in the severe forms of the first months and good, neurological and renal, in those with late onset; in COQ4, patients with a null variant did not respond and those with two missense variants had a partial response; COQ6 p.Gly255Arg or p.Ala353Asp homozygotes responded well, with a better renal response in COQ6; a truncating COQ8B homozygote had resolution of oedema and marked improvement in proteinuria; in PDSS2 and COQ9 the neonatal forms respond poorly; in COQ8A most respond poorly, with three favourable responses of ataxia, dystonia and myoclonus [14, 105]. Prenatal HCM due to COQ4 and multisystem disease due to COQ9 worsened despite CoQ10 [14, 30]. A 2022 systematic review identified 303 patients, 89 with treatment data, and improvement after CoQ10 in 24 (27.0%), always partial (ataxia in 12, proteinuria in 5), with deterioration after withdrawal in 5, concluding that there is "very weak" evidence of efficacy [108]. In COQ8A, the Traschütz series (2020) brought together 30 patients treated with CoQ10 at a mean dose of 11 mg/kg/day, lower than the ENMC 176 reference dose: a clinical response was reported in 13, no response in 15 and adverse effects in 2 (anorexia, diarrhoea); in the 11 patients with longitudinal assessment by the SARA scale (Scale for the Assessment and Rating of Ataxia), the mean annual change was −0.88 points, with no association between response and age at onset, disease duration, ataxia severity, dose or variant type [27]. The OMIM summary for COQ10D4, based on the initial descriptions (Mollet 2008; Lagier-Tourenne 2008), states that there was no significant neurological improvement, which partly diverges from these observational data [27]. In COQ4, a 12-month-old infant treated with CoQ10 and thiamine had resolution of respiratory distress, crying episodes and seizures (Lu 2019, isolated report) [28]. Idebenone and other short-chain quinone analogues have been associated with clinical deterioration in CoQ10 deficiency and should not replace CoQ10, contrary to what the Oxford Desk Reference suggests [11, 14].

ACAD9 deficiency: riboflavin

Riboflavin increases mitochondrial FAD and improves the folding and stability of ACAD9 and complex I assembly [33]. In the cohort of 70 patients, there were data on clinical effect in 31: benefit in 20 (65%), no effect in 11 (35%), with no deterioration or adverse effects; complex I improved in 9 of 15 fibroblast lines (14–109%); among patients with onset in the first year, 7 of 22 treated and 16 of 17 untreated died (p = 5.34×10⁻⁵) [23, 33]. The evidence is retrospective and observational, collected by a survey of physicians, with a risk of survivorship selection bias and dose and duration data in a minority; there is no randomised trial [23, 33]. The 2017 MMS recommends considering riboflavin in myopathy due to ACAD9 deficiency, without giving a dose [93]. In the first report of a response (Haack 2010), a boy with hypotonia, cardiac hypertrophy, lactic acidosis and complex I deficiency from birth received riboflavin and had normal psychomotor development at 5 years [32].

The reference dose is that proposed by the cohort of Repp and colleagues: a trial of riboflavin at 20 mg/kg/day, maximum 200 mg/day, in all patients with this diagnosis, and starting it in a phenotypically compatible patient while genetics is under way [23]. By comparison, Saudubray attributes to the same study a dose of 100–400 mg/day orally, which matches its drug table (100–400 mg/day in two to three doses for the riboflavin indications, including ACAD9) [1, 107], and the 2009 MMS opinion table gives 50–400 mg/day for mitochondrial disease in general [104]. At high doses, riboflavin may cause anorexia and nausea, alters the colour and smell of urine and is given at bedtime [104]. Riboflavin has no FDA or EMA label for this indication. CoQ10, biotin and L-carnitine had only anecdotal positive effects in ACAD9 deficiency, and bezafibrate increased complex I in fibroblasts without confirmation in humans [33]. Pregnancy proceeded normally in a woman on riboflavin since the age of 4 years (section 8) [33].

Flavin deficiencies (FLAD1, SLC25A32)

Riboflavin improved clinically and biochemically the two patients with SLC25A32 and some with FLAD1 [75]. Saudubray's drug table includes the FAD transporter defect and FAD synthase deficiency among the indications for riboflavin, at 100–400 mg/day in two to three doses, with doses up to 80 mg/kg/day used in deficiencies of the riboflavin transporters RFVT2 and RFVT3 [107]. Riboflavin, possibly with CoQ10, is the treatment of multiple acyl-CoA dehydrogenase deficiency (ETFDH) [93, 105].

TRMU deficiency: L-cysteine and N-acetylcysteine

L-cysteine, with or without N-acetylcysteine (NAC), should be started as soon as the condition is suspected; NAC is recommended with cysteine, because it improves the redox potential, especially in the first year [15]. The doses used are L-cysteine 85–300 mg/kg/day (target of 300 mg/kg/day in total) and NAC 70–150 mg/kg/day (Murali 2021, according to GeneReviews); L-cysteine-enriched formulas may help, and supplementation is continued at least throughout the first year [15]. This is off-label use, with no maximum dose defined in the source beyond the upper limits of these ranges [15]. Survival beyond acute decompensation was 84% with L-cysteine and/or NAC and 61% without (Vogel 2023), but supplementation does not influence native liver survival and there is insufficient evidence to state that it fully prevents hepatic dysfunction [15]. Two siblings treated before symptoms had a milder course: one, with maternal L-cysteine 500 mg twice daily in the third trimester and, after birth, NAC 150 mg/kg/day with L-cysteine 140 mg/kg/day, then NAC 110 mg/kg/day and L-cysteine 85 mg/kg/day, had lactic acidosis, hypoglycaemia and hyperammonaemia from 2 months, without acute liver failure, and was asymptomatic at 3 years, with persistently raised lactate; the other, on cysteine from birth, had micronodular cirrhosis at 26 months [15]. Taurine is theoretically beneficial, with no published data [15].

Orthotopic liver transplantation was performed in 11 children (median 4 months), with no influence on overall survival and with two post-transplant deaths; it is decided case by case and may be considered before targeted therapy in p.(Met1?) patients, who are less responsive [15]. Support includes standard treatment of hypoglycaemia, lactic acidosis, coagulopathy and hyperammonaemia, gastrostomy if necessary, support of cardiomyopathy until recovery and respiratory support; drugs that increase metabolic demand (corticosteroids), mitochondrial inhibitors (valproate, prolonged propofol infusion), fasting and high-concentration dextrose should be avoided [15].

Reversible myopathy due to MT-TE

Treatment is supportive; the molecular diagnosis justifies maintaining ventilatory and nutritional support during the acute phase rather than withdrawing it, because recovery is expected and the prognosis excellent [66]. Some patients need prolonged ventilation, up to 18 months [1]. Aminoglycosides, linezolid, tobacco, alcohol, valproate and dichloroacetate should be avoided, and metformin used with caution [66].

PPA2 deficiency

Strict avoidance of alcohol is the only measure described and was associated with the absence of severe symptoms in siblings with the same genotype [1].

Defects of thiamine transport and activation

In biotin-thiamine-responsive basal ganglia disease (SLC19A3), the manifestations typically resolve within days with biotin and thiamine; in SLC25A19 deficiency, thiamine is lifelong and increased by 25% during fever, surgery or acute decompensation; in TPK1 deficiency, thiamine produces a response in about 50%, more with early diagnosis [16]. Doses diverge between sources. The reference is the BTBGD GeneReviews, updated in 2025: biotin 5–10 mg/kg/day and thiamine up to 40 mg/kg/day (maximum 1500 mg/day), orally, from diagnosis and lifelong, with thiamine at double dose and intravenously in acute decompensations; adults are usually maintained on biotin 600 mg/day and thiamine 1500 mg/day; with an insufficient response, thiamine may be increased up to 75 mg/kg/day in the acute phase, with subsequent reduction to about 30 mg/kg/day (Aldosari 2025, according to GeneReviews); treatment in the first month after onset significantly improves the outcome, and thiamine, alone or with biotin, gives better results than biotin alone [109]. By comparison, a Polish series citing the literature gives biotin 5–10 mg/kg/day and thiamine 10–40 mg/kg/day (300–900 mg/day) [94], and the Saudubray table thiamine 100–400 mg/day and biotin 5–20 mg/day [107]. In the Polish series, a patient initially labelled as Leigh syndrome improved within days on a "mitochondrial cocktail" that included thiamine 300 mg/day and biotin 20 mg/day, and another relapsed with dystonia after the parents stopped thiamine during a respiratory infection [94]. This experience underpins an early empirical trial of thiamine and biotin in infantile encephalopathy with basal ganglia lesions. Clinical Cardiogenetics, citing Distelmaier and colleagues (2017), states that supplementation with thiamine, riboflavin, biotin and CoQ10 may be used empirically in critically ill patients with PMD while awaiting the molecular diagnosis [18].

TK2 deficiency: doxecitine and doxribtimine

In 2025, the phase 2b trial of sonlicromanol still stated that there was no approved treatment for mitochondrial disease, except LHON [110]. Kygevvi® (doxecitine and doxribtimine) was approved after that and is indicated, in the USA, for thymidine kinase 2 deficiency (TK2d) in adults and children with symptom onset up to 12 years of age [24] and, in the European Union, in paediatric and adult patients with genetically confirmed TK2d and symptom onset up to 12 years of age [111, 112]. Doses and evidence are summarised in the regulatory and trials tables; treatment of TK2 deficiency follows the recommendations of the ENMC 232 workshop for mtDNA maintenance disorders [113].

Supplements and cofactors in mitochondrial disease in general ("mitochondrial cocktail")

Vitamin and cofactor "cocktails" are advocated by some centres, especially in the USA; some have a logical rationale (riboflavin in complex I deficiency, thiamine in pyruvate dehydrogenase deficiency) and many are non-specific antioxidants; a Cochrane review found little evidence to support or refute any vitamin or cofactor [1, 21]. The societies diverge. The 2015 MMS recommends offering CoQ10 to most patients, and not only in primary deficiency, offering α-lipoic acid and riboflavin, considering folinic acid with central nervous system manifestations (and giving it routinely with documented CSF deficiency), using L-carnitine only with documented deficiency, starting supplements one at a time and correcting deficiencies [21]. The 2022 Australian adaptation considers the evidence insufficient outside specific conditions (thiamine with the ketogenic diet in pyruvate dehydrogenase deficiency and CoQ10 in primary CoQ10 deficiency) [99]. Saudubray does not list the components owing to lack of evidence [1]. A 2025 opinion article questions whether the "cocktail" should be a default option [114]. The table reproduces the doses from the 2009 opinion table of MMS members, the only source with systematic paediatric doses, and adds doses from other sources when they diverge; the level of evidence is expert opinion (class IV) unless stated otherwise.

Source of the MMS 2009 columns: Parikh and colleagues, Table 2, opinion article by members of the Mitochondrial Medicine Society; all indications are off-label use, except L-carnitine, which the table states is FDA-approved for metabolic disorders [104]. CoQ10 and a B vitamin are the components most used in an initial "cocktail" [104].

Empirical therapy in the critically ill child under investigation. The 2017 paediatric diagnostic guideline allows, in patients with a rapidly progressive or potentially fatal course and while the genetic work-up proceeds, consideration of an empirical regimen of thiamine 20 mg/kg/day, biotin 5 mg/kg/day, riboflavin 20 mg/kg/day and CoQ10 15 mg/kg/day, supported by a 2017 review of treatable mitochondrial diseases (Distelmaier and colleagues); the recommendation is conditional and does not state the route, number of doses, maximum dose or duration [22]. It is the reference for this scenario, as it is the only source in this article with a weight-based empirical paediatric regimen. By comparison: riboflavin matches the dose of the therapeutic trial in ACAD9 deficiency, which adds a maximum of 200 mg/day [23]; CoQ10 lies within the range of the ENMC 176 consensus (10–30 mg/kg/day) and of the 2009 MMS table for ubiquinone [104, 105]; biotin matches the lower limit of the GeneReviews on biotin-thiamine-responsive basal ganglia disease (5–10 mg/kg/day), and thiamine corresponds to half the maximum of that GeneReviews (up to 40 mg/kg/day; up to 75 mg/kg/day in the acute phase) [109]; the 2022 Australian adaptation does not recommend supplements outside specific conditions [99]. All of these uses are off-label. Once the diagnosis is confirmed, only what the condition justifies is continued, at the doses in the relevant subsections.

Dichloroacetate

Dichloroacetate activates the pyruvate dehydrogenase complex by inhibiting its kinase and lowers lactate, pyruvate and alanine; it improves lactate in congenital lactic acidosis, but neurological improvement was elusive in most, it is not recommended in gluconeogenesis defects (it may cause hypoglycaemia) and peripheral neuropathy may appear or worsen [78]. The first randomised, double-blind trial in congenital lactic acidosis (43 patients aged 0.9 to 19 years) used 12.5 mg/kg every 12 hours (25 mg/kg/day) for 6 months and showed no difference in global assessment of efficacy, growth or intercurrent illnesses; it attenuated the postprandial rise in lactate and was well tolerated, with reduced clearance on chronic use [115]. The reference dose is that of the trial, 25 mg/kg/day in two doses, with no maximum dose defined in the sources, which matches that of Saudubray's drug table and the 2014 review [107, 115, 116]. By comparison, the emergency treatment chapter of the same Saudubray gives 50 mg/kg/day in one or two doses, with little clinical effect beyond reduction of tachypnoea and with reversible sensorimotor peripheral neuropathy limiting chronic use [90]. A three-year randomised crossover trial in MELAS was stopped because of toxic peripheral neuropathy, and its use declined [104]. Dichloroacetate has orphan drug status in the USA, with no marketing authorisation identified in the sources [104], and should be avoided in the MT-TE defect [66].

Regulatory status

In Portugal, the centralised European Commission authorisation of Kygevvi® applies directly [111]. In Brazil, the ANVISA open medicines data consulted on 27-10-2026 contain no registration of doxecitine and doxribtimine, elamipretide, idebenone or dichloroacetate; ubidecarenone (CoQ10) and levocarnitine have only inactive registrations; acetylcysteine (for example, Fluimucil®), thiamine, biotin, thioctic acid (α-lipoic acid; Thioctacid®) and calcium folinate have active registrations for other indications, so their use in the disorders of this article is off-label [119].

Clinical trials and studies with a comparison group

These studies do not allow indirect comparisons between them: they differ in population, design and endpoint. The randomised trials in heterogeneous paediatric populations (dichloroacetate, CoQ10, vatiquinone) were negative, and genetic heterogeneity, the unpredictable course and the widespread empirical use of CoQ10 hampered recruitment and the choice of endpoints [1, 121]. The data on treated disease in the treatable forms come from retrospective cohorts, with a risk of survivorship selection bias (ACAD9) [23], and from studies with external controls (TK2); for Kygevvi, the 86% relative reduction in the risk of death corresponds to an absolute difference of 3.8% versus 35.9% deaths in the matched pairs [24]. In the European SmPC, among patients with onset up to 12 years of age treated for more than 5 years, loss of at least one motor milestone went from 82.1% before treatment to 26.3% after, and 83.9% of those who had lost a milestone regained it after starting treatment, a before-after comparison without a control group [112]. The paediatric outcome consensus statements propose natural history studies with assessments every 6 months, and every 3 months below 4 years of age or in rapidly progressive disease [125], and the ENMC 280 workshop defined outcome measures for primary mitochondrial myopathies in adults and children [25].

Developments in the last 6–12 months and ongoing programmes

Experimental approaches include antioxidants (CoQ10 analogues, N-acetylcysteine), stimulation of mitochondrial biogenesis (bezafibrate, vitamin B3 analogues), membrane lipid stabilisers, rapamycin and analogues, nucleosides in TK2 deficiency and gene therapy with adeno-associated viruses or lentiviruses for nuclear genes [1]. Development is slow because of the difficulty of running trials in heterogeneous disorders with an unpredictable course [1].

Divergences between guidelines and consensus statements

The main divergences, already presented in the tables, are four. First, supplements: the 2015 MMS recommends offering CoQ10, α-lipoic acid and riboflavin to most patients, and the 2022 Australian adaptation considers the evidence insufficient outside primary CoQ10 deficiency and pyruvate dehydrogenase deficiency [21, 99]. Second, anaesthesia: caution with volatile agents and propofol limited to short procedures (MMS 2015), avoidance of volatile agents and propofol (Australia 2022) or both safe for induction (Delphi 2020) [21, 99, 103]. Third, drugs: a broad list of drugs to avoid in the 2015 MMS, most considered safe in the 2020 consensus, and a specific list of antiepileptic drugs to avoid in infantile hepatocerebral disease in the Australian adaptation [21, 99, 103]. Fourth, fluids: 5–10% dextrose, escalating up to 20% with insulin (MMS 2009), versus a 5% limit (Australia 2022) [99, 104]. The doses of CoQ10, riboflavin in ACAD9, dichloroacetate, L-carnitine and biotin diverge between sources, as described; the reference adopted is the dose of the consensus or study cited, and the others are presented as comparison.

Pregnancy and lactation

Pregnancy in women with PMD has a variable clinical course, with an increased risk of cardiac deterioration, preterm birth, pre-eclampsia and magnesium sulphate toxicity [18]. The 2017 MMS recommends high-risk obstetric surveillance, early and late oral glucose tolerance testing, medication review and fetal surveillance when there is a risk of transmission [93]. A woman with ACAD9 deficiency on riboflavin since the age of 4 years had a normal pregnancy and a healthy newborn at 26 years [33]. In TRMU deficiency, the mother of an affected fetus received L-cysteine 500 mg twice daily in the third trimester, and no offspring of affected women have yet been described [15]. The Kygevvi label has no human data in pregnancy and describes malformations in rabbits at exposures much higher than the maximum human dose [24]. The sources consulted do not describe the safety of supplements during lactation.

Transition to adulthood

Treated forms and those with later onset survive into adulthood: in ACAD9 deficiency with onset after the first year, 10-year survival exceeds 90% [23], the late form of Sengers syndrome survives to the fourth or fifth decade [67], and in TRMU deficiency one transplanted patient was followed up to 13 years and another retained the native liver at 17 [15]. The standards of care provide for care coordination, social and school support and routine psychiatric assessment (screening for depression and anxiety) [93]. The sources consulted do not describe structured transition programmes for these disorders.

8. Reproductive Options and Family Counselling

Principle: identify the genome and the gene first

Counselling depends on the molecular diagnosis. Almost all childhood-onset nuclear mitochondrial disease is AR, with a few X-linked exceptions, and counselling, prenatal diagnosis and PGT are straightforward when the variants are known [1]. In mtDNA disease, the risk depends on the type of variant, the load and the genetic bottleneck, and is difficult to estimate precisely [1]. Without a molecular diagnosis, the risks are empirical: in a large study of Leigh syndrome with a respiratory chain defect of unknown molecular basis, the recurrence risk lay between 1/8 and 1/4, much higher if the mother has an unrecognised mtDNA mutation [11]. DNA banking is indicated in patients without a molecular diagnosis [14].

Recurrence risks by inheritance

Recurrence risks for siblings in SURF1 Leigh syndrome are 25%, and the old empirical risk of 33% for Leigh syndrome in general (van Erven 1987) is probably overestimated [11]. COXPD1 due to GFM1 is recessive and should not be confused with maternal inheritance: the recurrence risk is 25% [51].

Cascade testing and at-risk relatives

In mtDNA disease, cascade screening follows only the maternal line, and the children of male carriers do not need testing; in AR disease, siblings are tested [18]. In primary CoQ10 deficiency, the proband's siblings should be assessed as early as possible so that CoQ10 can be started early: by molecular testing if the variants are known or, if the diagnosis is only biochemical, by CoQ10 in fibroblasts [14]. In TRMU deficiency, a newborn sibling without prenatal diagnosis should receive L-cysteine until the result of family testing is known [15]. In ACAD9 deficiency, intrafamilial observations (first child dying before 2 years without riboflavin; siblings treated from diagnosis surviving) reinforce the value of early diagnosis of siblings [33]. In the absence of an identified variant with high suspicion, first-degree relatives may be assessed clinically (lactate, amino acids, blood glucose, organic acids, periodic cardiological assessment if the proband has cardiomyopathy) [18]. Predictive testing of maternal relatives in mtDNA disease carries high uncertainty [11].

Prenatal diagnosis and PGT

In nuclear disease, once the variants have been identified, prenatal diagnosis by chorionic villus sampling or amniocentesis and PGT are offered [2, 18]. In TRMU deficiency, prenatal diagnosis also serves to guide maternal L-cysteine supplementation and to start cysteine at birth [15]. Knowledge of the variant allows accurate prenatal diagnosis where biochemical methods have often failed [78].

In mtDNA disease, interpretation is complex: the load in amniocytes or villi may not correspond to that of fetal tissues and may change in utero or after birth through mitotic segregation [18]. Data from fetal tissues suggest little tissue variation, and the load in villi or amniotic fluid appears to predict the load in most tissues at birth; the difficulty lies in predicting the phenotype [11]. For de novo point mutations or carriers with a low load, the recurrence risk is low and prenatal diagnosis may be offered [18]. In PGT, embryos with a load below a mutation-specific threshold or below a general expression threshold of 18% (Smeets 2015) are transferred [18]. In the integrated Newcastle programme (United Kingdom), 39 heteroplasmic women, predominantly of white European ancestry, underwent PGT after intracytoplasmic sperm injection (ICSI): on day 3 a single blastomere was removed from embryos with 6 or more cells (262 embryos, 84.8% of zygotes, from 37 patients), heteroplasmy was quantified by pyrosequencing (detection limit of 3%) and embryos with heteroplasmy below 30% were preferentially used, with decisions adjusted to the variant and the family history [132]. This operational threshold of 30% is higher than the general threshold of 18% of Smeets 2015; the authors associate heteroplasmy above 60% with an increased risk of symptoms, on the basis of the systematic review by Hellebrekers 2012 [18, 132]. Of the blastocysts of "optimal" or "good" quality, 42.4% had heteroplasmy above 30% and were excluded, so the best-quality embryo could not be used in 25% of fresh transfers; 80% of patients had embryos for transfer, and the lack of low-load embryos was the main reason for no transfer (5 patients) [132]. There was a clinical pregnancy (ultrasound at 7 weeks) in 16 of 39 patients (41%) and in 39.2% of transfers; 3 of the 20 clinical pregnancies ended in miscarriage (15%) and 18 children were born (9 female and 9 male), from 42 of 51 single-embryo transfers [132]. The clinical pregnancies after PGT were distributed among MT-ATP6 variants (4), MT-ND variants not associated with LHON (4), m.3243A>G (6) and m.8344A>G (2) [132]. Neonatal blood heteroplasmy was measured in only 10 of the 18 children, because not all families accepted the measurement: it was undetectable in the 8 born from single-embryo transfer whose blastomere had no detectable variant (variants m.10158T>C in MT-ND3, m.13094T>C in MT-ND5, m.3243A>G in MT-TL1, m.8993T>G and m.8993T>C in MT-ATP6), undetectable in a child born after transfer of two embryos with 9% and 24% in the blastomere (m.3243A>G) and 7% in a child born after transfer of two embryos with undetectable and 5% blastomeres (m.3688G>A in MT-ND1) [81, 132]. Trophectoderm biopsy at the blastocyst stage, instead of the blastomere on day 3, could improve results, but there are reports of clinically relevant discrepancy between the load in the trophectoderm and that of the child [132]. Oocyte donation avoids transmission [1, 18].

Mitochondrial donation

The United Kingdom legislated in 2015 for two mitochondrial donation techniques (pronuclear transfer and spindle transfer), and the Human Fertilisation and Embryology Authority licensed the first clinic for pronuclear transfer in March 2017 [1]. In Newcastle, pronuclear transfer was offered to women with homoplasmic pathogenic variants or with high heteroplasmy, and PGT to heteroplasmic women: there was a clinical pregnancy in 8 of 22 (36%) with pronuclear transfer and in 16 of 39 (41%) with PGT, with 8 live births and one ongoing pregnancy after pronuclear transfer and 18 live births after PGT [132]. Blood heteroplasmy in the 8 babies from pronuclear transfer ranged from undetectable to 16%, with a level of the maternal variant 95–100% lower in 6 and 76.5–88% lower in 2, relative to the enucleated zygote or maternal urine; the abstract of the article gives 77.6–100%, but the values in Table 1 (68% → 16% and 99.9% → 12%) give 76.5% and 88% [132]. All were healthy at birth; one child had hyperlipidaemia and cardiac arrhythmia and another infantile myoclonic epilepsy with spontaneous remission, all with normal development to date [133]. No congenital anomalies were reported at birth, but a cardiac anomaly was detected in one child born after pronuclear transfer (m.4300A>G in MT-TI) and in another born after PGT (m.3243A>G), both with maternal variant heteroplasmy of 5% or less; gestational age and birth weight did not differ between the two groups [81, 132]. Eligibility for pronuclear transfer rested on the probability of obtaining embryos with heteroplasmy below 30%: it was offered to carriers of homoplasmic variants (12) and to some with high heteroplasmy (13; m.3243A>G in 5 and m.8344A>G in 4, including 3 women with previous PGT cycles in the study), whereas PGT was offered in most cases of heteroplasmy (39); each treatment required case-by-case HFEA approval, granted to 32 women, of whom 25 underwent oocyte retrieval [132]. In the 60 women who started treatment, the variants were in protein-coding genes (27: MT-ND in 17, of which 11 primary LHON variants and 6 not associated with LHON, and MT-ATP6 in 9) or in mitochondrial transfer RNA genes, mt-tRNA (33; m.3243A>G in 21 women, 15 of whom underwent PGT), homoplasmic in 12 and heteroplasmic in 48; one woman had a large mtDNA deletion, NC_012920.1:m.7529_14025del; LHON carriers formed the largest pronuclear transfer group (44%) [132]. In the 10 heteroplasmic women (67–98%) with enucleated zygotes analysed (67 zygotes), 70% had no zygote below 30%; overall, 4.5% of zygotes were below 30% and 67.2% above 60%, which indicates that PGT would hardly have benefited them [132]. In the procedure, the patients' oocytes were vitrified until fresh donor oocytes were available (survival after warming of 99.4%, 325 of 327), fertilised by ICSI, and the pronuclei were transferred 8–13 hours later to the enucleated donor zygote; normal fertilisation was 48.5% in the oocytes of the pronuclear transfer patients, versus 63.8% in those of the donors and 66.5% in those of the PGT patients, and pronuclear replacement succeeded in 127 of 160 attempts (79.4%), with 122 intact embryos the following day [132]. Blastocyst formation was similar to that of PGT (45.1% and 46.3% on day 5), but the clinical pregnancy rate per transfer was lower (20% versus 39.2%), owing to more losses after a positive pregnancy test; all 40 transfers were of a single embryo and none of the 8 clinical pregnancies ended in miscarriage [132]. The 8 clinical pregnancies occurred in 5 carriers of primary LHON variants and in 3 of rare mt-tRNA variants; none of the carriers of m.3243A>G (3) or of m.8344A>G (4) who underwent pronuclear transfer had a clinical pregnancy [132]. Six of the 21 women with a live birth (by either technique) had previously had an affected child [132].

The children born after pronuclear transfer in Newcastle had the following individual results, with mtDNA positions referred to the revised Cambridge reference sequence (NC_012920.1) and pyrosequencing with a detection limit of 3% [81, 132].

In the 62 pronuclear transfer embryos unsuitable for transfer, high-throughput sequencing showed low heteroplasmy (undetectable to 3.9%; mean 1.18%), regardless of whether the patient's zygote was homoplasmic or heteroplasmic; the cytoplasmic carryover score correlated with the heteroplasmy of arrested embryos, but the two newborns with 12% and 16% had low scores (1 and 2), comparable to those of procedures that resulted in less than 1%, and the proposed causes are a replicative advantage of maternal mtDNA or an uneven distribution of heteroplasmy between embryonic lineages [132]. United Kingdom regulations do not allow heteroplasmy to be measured in embryos obtained by pronuclear transfer, and none of the patients opted for prenatal diagnosis [132]. The authors point out that residual maternal mtDNA became homoplasmic in about 20% of embryonic stem cell lines derived from mitochondrial donation embryos and that 1 of 6 babies born after spindle transfer for infertility had maternal heteroplasmy of 40–60%; mitochondrial donation is therefore presented as a risk reduction strategy, not a risk elimination strategy, with mandatory follow-up of heteroplasmy stability over time and across tissues [132]. An 18-month follow-up study is recruiting, with completion expected in 2027 [134]. Before that, an apparently healthy baby was born in Mexico, by spindle transfer, to a mother with m.8993T>G, with no long-term follow-up published [1]. In Australia there was a public deliberation on public funding of mitochondrial donation [135]. Maeve's Law authorises maternal spindle transfer and pronuclear transfer only to reduce the risk of mitochondrial disease, and not to treat infertility, in the mitoHOPE pilot programme (Monash University); each technique requires approval by an ethics committee and a licence from the Embryo Research Licensing Committee in three phases (preclinical research and training, research and training for the clinical trial, and pilot trial); the programme received its first preclinical licence, began spindle transfer in the laboratory with donated oocytes and, in the update of 18-08-2026, estimated that the first participants would enter the pilot trial in the second half of 2027, depending on the licences [136]. These techniques are intended for mtDNA disease and not for the nuclear forms that dominate infantile-onset disease.

Counselling of minors

Testing of minor siblings is justified when there is a treatment that benefits from early initiation (CoQ10, TRMU, ACAD9) [14, 15, 33]. For carrier status of an AR disorder, which has no consequence for the child's health, the sources consulted do not describe specific recommendations for these disorders.

9. Special Considerations: Population Diversity and Epidemiology

Prevalence and incidence

There are no precise data on the frequency of each OXPHOS defect, because registries are incomplete and many patients lack a genetic diagnosis [1]. The available studies measure different quantities, which prevents them from being added together.

Leigh syndrome, of all causes, has a birth prevalence of 2.8/100,000 in Europe by expert estimate and an annual incidence of 1.3/100,000 in Australia, according to Orphanet [46]. The specific forms are rare or ultra-rare: GRACILE has a validated birth prevalence of 2.0/100,000 in Finland (about 1 in 50,000 births) and a point prevalence below 1/1,000,000 in Europe [40, 41], whereas OMIM cites the estimate of Fellman (1998), based on 14 cases diagnosed between 1985 and 1997, of 1 in 56,000 Finnish newborns, with an allele frequency of 0.004 and a carrier frequency of 1 in 120, probably underestimated owing to underdiagnosis [38]; ACAD9 deficiency has 23 validated cases in Orphanet and a published cohort of 70 patients [33, 34]; fatal infantile COX deficiency has 16 cases or families [44]; TRMU deficiency, 62 individuals from 56 families [15]; and almost all COXPD have an Orphanet prevalence below 1/1,000,000 [4]. Isolated complex I deficiencies and TMEM70 deficiency have an unknown prevalence [37, 50].

mtDNA mutations explain only 15–30% of OXPHOS defects in childhood [1], and in the Japanese neonatal cohort 81% of the variants identified were nuclear [19]. In the London and Salzburg experience, POLG and SURF1 are the most prevalent nuclear genes [1]. In the European population, m.3243A>G is found at low load in 0.2% of people, as is homoplasmic m.1555A>G [1].

Founder populations

In founder populations, the local frequency may far exceed the global one. In Saguenay-Lac-Saint-Jean, the birth prevalence of French-Canadian COX deficiency was estimated at 1 in 2473, with 1 carrier in 28 (De Braekeleer 1991), and the incidence at 1 in 2063 live births between 1979 and 1990, with 1 carrier in 23 (Morin 1993); genealogical reconstruction of 54 obligate carriers pointed to 17th-century European settlers [47]. In the Faroe Islands, the incidence of the encephalomyopathic form due to SUCLA2 was estimated at 1 in 1700 (Ostergaard 2007) or 1 in 2500, with 2% carriers (Carrozzo 2007) [68]. Parental consanguinity was 76% in the North African families of a French series of 157 patients with a respiratory chain defect [8].

Japan and Central Europe: neonatal cohorts

The population-based Japanese cohort (2004–2020) and the Czech-Austrian cohort are the largest series of neonatal-onset mitochondrial disease, with 28.7% and 28% neonatal onset, respectively [19, 20]. The Japanese cohort, based mainly on biochemical diagnosis, identified nuclear variants in 36 genes, including ACAD9, BOLA3, COQ4, GFM1, LRPPRC, TMEM70 and TUFM [19]; the Czech-Austrian cohort highlighted the frequency of neonatal cardiomyopathy and hyperammonaemia and complex V deficiency as the most frequent isolated deficiency [20].

Portugal and Brazil

The sources consulted provide no epidemiological data from Portugal or Brazil for these disorders. The Brazilian book on genetic syndromes reports, for CoQ10 deficiency, scant epidemiological data and an incidence below 1:100,000, without a population-based study, and classifies it in ICD-10 under G71.3 (mitochondrial myopathy) [106].

Special populations

Consanguinity increases the frequency of AR forms, as shown by the prevalence in Lebanese Australian families [17]. Preterm and small-for-gestational-age newborns are over-represented in the neonatal forms [19]. Pregnant women with PMD and carriers of mtDNA variants require high-risk obstetric surveillance (section 7) [93].

Reference networks and patient organisations

In the European Union, mitochondrial diseases are covered by MetabERN (European reference network for hereditary metabolic disorders; https://metab.ern-net.eu/) and ERN EURO-NMD (neuromuscular diseases; https://ern-euro-nmd.eu/), whose mitochondrial disease working group organised the ENMC 280 workshop on diagnostic criteria and outcome measures [25]. The patient organisations listed by GeneReviews and the sources include International Mito Patients (IMP; https://www.mitopatients.org/), the United Mitochondrial Disease Foundation (UMDF; https://www.umdf.org/), The Lily Foundation (United Kingdom), the Charlie Gard Foundation, Mitocon (Italy), the Champ Foundation (single deletion and Pearson syndrome) and the patient contact registry of the North American Mitochondrial Disease Consortium [2]. IMP took part in the 2020 drug safety consensus [103]. The sources consulted do not identify specific Portuguese or Brazilian organisations.

10. Illustrative Clinical Case

Main case: COXPD31 due to MIPEP variants (Wang and colleagues, 2022)

An 8-month-old boy, the only child of healthy, non-consanguineous Chinese parents, born at term after a normal pregnancy and delivery, was admitted with mild coma and poor response to stimuli [63]. He had a temperature of 37 °C, heart rate of 180/min, respiratory rate of 45/min and blood pressure of 75/45 mmHg, with respiratory distress and severe lactic acidaemia (lactate 17.7 mmol/L; pH 7.138) that required mechanical ventilation [63]. Psychomotor development was delayed and he could not sit at 8 months owing to global hypotonia; he had a short penis and cryptorchidism [63].

The echocardiogram showed HCM, with a left ventricular posterior wall of 6 mm and an interventricular septum of 13.2 mm (Z-scores of 3.29 and 21.83), an ejection fraction of 50% and a minimal pericardial effusion (Figure 1 of the article: https://pmc.ncbi.nlm.nih.gov/articles/PMC9884671/figure/F1/) [63]. Brain MRI showed abnormal signal in the thalami and dorsal brainstem; B-type natriuretic peptide was 2613 pg/mL (upper limit 100) and transaminases were mildly raised (alanine aminotransferase 121 U/L, aspartate aminotransferase 105 U/L); metabolic screening showed elevation of several acylcarnitines and of 3-hydroxybutyrate [63].

On suspicion of mitochondrial disease, the child received high-dose CoQ10, L-carnitine, B complex and other supplements; the pH normalised, but lactate remained at 8–9 mmol/L, and the child died 2 months after admission [63]. Trio exome sequencing with copy number variation analysis identified compound heterozygosity in MIPEP: the paternal missense variant NM_005932.4:c.1081T>A p.(Tyr361Asn), classified as a VUS (PM2, PP3), and a maternal 1.12 Mb deletion (chr13:23777833-24895906) encompassing the entire MIPEP gene, classified as likely pathogenic (PVS1, PM2) and confirmed by quantitative PCR of exons 1, 10 and 19; the mtDNA copy number in blood was reduced by 71.5% relative to three controls (Figure 2 of the article, with pedigree, sequencing, mtDNA copy number, structural model, orthologue alignment and quantitative PCR: https://pmc.ncbi.nlm.nih.gov/articles/PMC9884671/figure/F2/) [63, 87]. The diagnosis was combined OXPHOS deficiency type 31 (COXPD31; OMIM 617228), AR [63].

In the authors' review, there were 7 MIPEP variants in 5 published patients; 4 had cardiac involvement (left ventricular non-compaction, HCM or non-compaction with DCM), 3 of these died before 2 years from progressive heart failure or sudden death, and one patient without heart disease survived more than 20 years [63].

Clinical interpretation. The case brings together the features of the class: onset in the first year, encephalopathy with thalamic and brainstem lesions, hypotonia, HCM and severe lactic acidosis, with deficiency of a mitochondrial quality control gene (Table 10.4 in section 2) [1]. The supplement "cocktail" used in the report does not correspond to a uniform recommendation: the 2015 MMS recommends offering CoQ10 to most patients, the Australian adaptation considers the evidence insufficient outside specific conditions, and there is no evidence of benefit in combined deficiencies due to quality control defects [21, 99]. Interpretation of the VUS rested on phase (in trans with a deletion) and on a functional marker (reduced mtDNA copy number), as recommended by the approach in section 5 [18].

Complementary case: fatal HCM due to MRPL44

A 15-month-old girl arrived at the emergency department in shock and did not survive resuscitation; autopsy showed mild facial dysmorphism, hepatomegaly due to steatosis and a greatly enlarged heart with severe left and mild right ventricular hypertrophy, with swollen mitochondria on electron microscopy of the myocardium (Figure 11.2 of Clinical Cardiogenetics) [18]. In the previous week she had been unwell, and on the day before she drank less and had fewer wet nappies; she had mild feeding difficulties, normal motor development and mild language delay [18]. Trio exome sequencing identified the homozygous missense variant MRPL44 c.467T>G p.(Leu156Arg), previously described in two Finnish sisters with HCM and hepatic steatosis, one who died at 6 months with a respiratory infection and the other asymptomatic in adolescence [18]. The case illustrates decompensation during an intercurrent illness and the variable expressivity of the same genotype.

11. Prognosis and Long-Term Follow-up

Prognosis

Many early-onset mitochondrial diseases lead to death in infancy and early childhood, but the natural history is highly variable, even with the same genetic defect (Keshavan and Rahman 2018, according to Saudubray) [1]. Age at first symptoms is a major prognostic factor: onset before 6 months is associated with a 10-fold higher risk of death (Debray 2007, 73 paediatric patients, according to Clinical Cardiogenetics) [18]. Cardiomyopathy worsens the prognosis: mortality of 71% with cardiomyopathy versus 26% without (Holmgren 2003) and survival to 16 years of 16% versus 95% (Scaglia 2004), according to the same book, with mortality of up to 84% in mitochondrial cardiomyopathy despite support [18].

In the Japanese neonatal cohort, overall mortality was 44.8%, neonatal mortality 23%, median survival 1.86 years (95% CI 0.90–2.95) and 1-year survival 51.8%; mortality was 60.5% in the cardiomyopathic form and 23.1% in Leigh syndrome, which had better survival (HR 0.15; 95% CI 0.04–0.63) [19]. Molecular diagnosis was associated with an HR of 1.87 for death according to the authors' discussion, but the corresponding figure legend states the opposite, an internal inconsistency of the article [19].

Surveillance

The table brings together the intervals from the 2017 MMS standards of care, Clinical Cardiogenetics and the GeneReviews on the treatable types. The cardiac intervals are the same as in section 7. The 2017 MMS screening table groups the tests into three columns (at diagnosis, at 1–2-year intervals and according to symptoms or disease type), whereas the text of the same document recommends annual testing for endocrine biochemistry, transaminases, full blood count, CK in myopathy, ophthalmology and orthopaedics; the table follows the text and flags the tests that the MMS reserves for clinical indication [93].

In natural history studies, the outcome consensus statements propose assessments every 6 months, and every 3 months below 4 years of age or in rapidly progressive disease [125].

12. Summary and Final Considerations

Early-onset mitochondrial encephalomyopathies are a class of OXPHOS disorders caused by variants in mtDNA or, in most children, in nuclear genes with autosomal recessive inheritance. They present with congenital lactic acidosis, encephalopathy, hypotonia, cardiomyopathy, hepatopathy or combinations of these, often in the first days of life, and have a severe prognosis, especially when they begin before 6 months or are accompanied by cardiomyopathy [1, 18, 19].

Several classifications coexist: biochemical by complex and enzymatic pattern, genetic by genome and mechanism, nosological (COXPD, MMDS and other OMIM and Orphanet series), metabolic (ICIMD) and syndromic, harmonised by NAMDC [1, 12, 13, 83]. None suffices on its own, and the definitive diagnosis is now mainly genetic, with whole mtDNA sequencing and trio exome or genome sequencing early in the work-up, leaving biopsy and enzymology for cases without a diagnosis or with a VUS [1, 21].

The practical value of early diagnosis lies in the treatable forms: CoQ10 in primary CoQ10 deficiency, riboflavin in ACAD9 deficiency and in flavin deficiencies, L-cysteine with N-acetylcysteine in TRMU deficiency, alcohol avoidance in PPA2 deficiency, biotin and thiamine in thiamine transport defects and nucleosides in TK2 deficiency, the latter authorised by the FDA and the European Union [1, 14–16, 23, 24, 111]. The evidence for these therapies comes from consensus statements, retrospective cohorts and studies with external controls, and the doses diverge between sources; the reference dose is that of the consensus or the specific study. The molecular diagnosis also prevents wrong decisions, such as withdrawing support in reversible myopathy due to MT-TE [66].

In the remaining types, treatment is multisystem supportive care, with prevention of catabolism, a written emergency plan, perioperative care and prescribing guided by the drug safety consensus, which restricts mainly valproate in POLG and liver disease and aminoglycosides in predisposing mtDNA variants [93, 103]. Randomised trials in heterogeneous paediatric populations were negative, and recent developments are concentrated on therapies targeting specific mechanisms and on mitochondrial donation for mtDNA disease [115, 122, 132]. Leigh syndrome, MELAS, MERRF, mtDNA deletions, LHON and POLG and depletion disorders fall within the differential.

13. References

1. 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 (Tables 10.1–10.5). doi:10.1007/978-3-662-63123-2_10
2. Chinnery PF. Primary mitochondrial disorders overview. 2000 Jun 8 [updated 2021 Jul 29]. In: GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK1224/
3. Orphanet. Mitochondrial disease (ORPHA:68380). https://www.orpha.net/en/disease/detail/68380
4. GeneGuidance. Conditions linked by term (COXPD, MMDS, MTDPS) with OMIM/ORPHA. Version of 2026-09-26.
5. OMIM® Online Mendelian Inheritance in Man. #609060 COMBINED OXIDATIVE PHOSPHORYLATION DEFICIENCY 1; COXPD1. Johns Hopkins University; last edited 9 May 2025. https://omim.org/entry/609060. Accessed 27 October 2026.
6. OMIM® Online Mendelian Inheritance in Man. #252010 MITOCHONDRIAL COMPLEX I DEFICIENCY, NUCLEAR TYPE 1; MC1DN1. Johns Hopkins University; last edited 18 December 2023. https://omim.org/entry/252010. Accessed 27 October 2026.
7. OMIM® Online Mendelian Inheritance in Man. #124000 MITOCHONDRIAL COMPLEX III DEFICIENCY, NUCLEAR TYPE 1; MC3DN1. Johns Hopkins University; last edited 17 February 2020. https://omim.org/entry/124000. Accessed 27 October 2026.
8. OMIM® Online Mendelian Inheritance in Man. #220110 MITOCHONDRIAL COMPLEX IV DEFICIENCY, NUCLEAR TYPE 1; MC4DN1. Johns Hopkins University; last edited 2 February 2026. https://omim.org/entry/220110. Accessed 27 October 2026.
9. OMIM® Online Mendelian Inheritance in Man. #605711 MULTIPLE MITOCHONDRIAL DYSFUNCTIONS SYNDROME 1; MMDS1. Johns Hopkins University; last edited 12 September 2024. https://omim.org/entry/605711. Accessed 27 October 2026.
10. OMIM® Online Mendelian Inheritance in Man. #607426 COENZYME Q10 DEFICIENCY, PRIMARY, 1; COQ10D1. Johns Hopkins University; last edited 13 March 2025. https://omim.org/entry/607426. Accessed 27 October 2026.
11. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. p. 490–495 (Mitochondrial DNA diseases).
12. Frazier AE, Thorburn DR, Compton AG. Mitochondrial energy generation disorders: genes, mechanisms, and clues to pathology. J Biol Chem. 2019;294(14):5386-5395. doi:10.1074/jbc.R117.809194. PMID: 29233888; PMC6462508.
13. Emmanuele V, Ganesh J, Vladutiu G, Haas R, Kerr D, Saneto RP, et al. Time to harmonize mitochondrial syndrome nomenclature and classification: A consensus from the North American Mitochondrial Disease Consortium (NAMDC). Mol Genet Metab. 2022;136(2):125-131. doi:10.1016/j.ymgme.2022.05.001. PMID: 35606253; PMC9341219.
14. Salviati L, Trevisson E, Agosto C, Doimo M, Navas P. Primary coenzyme Q10 deficiency overview. 2017 Jan 26 [updated 2023 Jun 8]. In: GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK410087/
15. Reinhart M, Muraresku C, Ganetzky R. TRMU deficiency. 2023 May 11. In: GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK591557/
16. Rahman S, Thorburn DR, Ball M. Nuclear gene-encoded Leigh syndrome spectrum overview. 2015 Oct 1 [updated 2025 May 1]. In: GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. Available from: https://www.ncbi.nlm.nih.gov/books/NBK320989/
17. Skladal D, Halliday J, Thorburn DR. Minimum birth prevalence of mitochondrial respiratory chain disorders in children. Brain. 2003;126(Pt 8):1905-12. doi:10.1093/brain/awg170. PMID: 12805096.
18. Blank AC, Breur JMPJ, Fuchs SA, Koop K, Baas AF. Mitochondrial cardiomyopathies. In: Baars HF, van der Smagt JJ, Doevendans PAFM, van Tintelen JP, editors. Clinical Cardiogenetics. 3rd ed. Cham: Springer; 2020. p. 179–190. doi:10.1007/978-3-030-45457-9_11
19. Ebihara T, Nagatomo T, Sugiyama Y, Tsuruoka T, Osone Y, Shimura M, et al. Neonatal-onset mitochondrial disease: clinical features, molecular diagnosis and prognosis. Arch Dis Child Fetal Neonatal Ed. 2022;107(3):329-334. doi:10.1136/archdischild-2021-321633. PMID: 34625524; PMC9046829.
20. Honzik T, Tesarova M, Magner M, Mayr J, Jesina P, Vesela K, et al. Neonatal onset of mitochondrial disorders in 129 patients: clinical and laboratory characteristics and a new approach to diagnosis. J Inherit Metab Dis. 2012;35(5):749-59. doi:10.1007/s10545-011-9440-3. PMID: 22231385.
21. Parikh S, Goldstein A, Koenig MK, Scaglia F, Enns GM, Saneto R, 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. PMID: 25503498; PMC5000852.
22. Wortmann SB, Mayr JA, Nuoffer JM, Prokisch H, Sperl W. A Guideline for the Diagnosis of Pediatric Mitochondrial Disease: The Value of Muscle and Skin Biopsies in the Genetics Era. Neuropediatrics. 2017;48(4):309-314. doi:10.1055/s-0037-1603776. PMID: 28599323.
23. Repp BM, Mastantuono E, Alston CL, Schiff M, Haack TB, Rötig A, et al. Clinical, biochemical and genetic spectrum of 70 patients with ACAD9 deficiency: is riboflavin supplementation effective? Orphanet J Rare Dis. 2018;13(1):120. doi:10.1186/s13023-018-0784-8. PMID: 30025539; PMC6053715.
24. UCB Inc. KYGEVVI (doxecitine and doxribtimine) powder, for oral solution. Prescribing information. NDA 219792, initial US approval 2025-11-03. DailyMed setid 7d5ce670-9314-4fd7-8b0a-1447910dcf9e (version 4, published 2026-10-19). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=7d5ce670-9314-4fd7-8b0a-1447910dcf9e
25. Mancuso M, Lopriore P, Semmler L, Kornblum C; 280th ENMC workshop study group. 280th ENMC International Workshop: The ERN EURO-NMD mitochondrial diseases working group; diagnostic criteria and outcome measures in primary mitochondrial myopathies. Hoofddorp, the Netherlands, 22-24 November 2024. Neuromuscul Disord. 2025;50:105340. doi:10.1016/j.nmd.2025.105340. PMID: 40273815.
26. OMIM® Online Mendelian Inheritance in Man. #614652 COENZYME Q10 DEFICIENCY, PRIMARY, 3; COQ10D3. Johns Hopkins University; last edited 27 March 2015. https://omim.org/entry/614652. Accessed 27 October 2026.
27. OMIM® Online Mendelian Inheritance in Man. #612016 COENZYME Q10 DEFICIENCY, PRIMARY, 4; COQ10D4. Johns Hopkins University; last edited 25 September 2020. https://omim.org/entry/612016. Accessed 27 October 2026.
28. OMIM® Online Mendelian Inheritance in Man. #616276 COENZYME Q10 DEFICIENCY, PRIMARY, 7; COQ10D7. Johns Hopkins University; last edited 23 January 2024. https://omim.org/entry/616276. Accessed 27 October 2026.
29. OMIM® Online Mendelian Inheritance in Man. #615573 NEPHROTIC SYNDROME, TYPE 9; NPHS9. Johns Hopkins University; last edited 7 November 2016. https://omim.org/entry/615573. Accessed 27 October 2026.
30. GeneGuidance. COQ4 — clinical curation (primary CoQ10 deficiency type 7). Version of 2026-09-26.
31. GeneGuidance. PDSS2 — clinical curation (primary CoQ10 deficiency type 3). Version of 2026-09-26.
32. OMIM® Online Mendelian Inheritance in Man. #611126 MITOCHONDRIAL COMPLEX I DEFICIENCY, NUCLEAR TYPE 20; MC1DN20. Johns Hopkins University; last edited 23 June 2022. https://omim.org/entry/611126. Accessed 27 October 2026.
33. GeneGuidance. ACAD9 — clinical curation of gene and condition (complex I deficiency due to ACAD9). Version of 2026-09-26.
34. Orphanet. Acyl-CoA dehydrogenase 9 deficiency (ORPHA:99901). https://www.orpha.net/en/disease/detail/99901
35. OMIM® Online Mendelian Inheritance in Man. #614052 MITOCHONDRIAL COMPLEX V (ATP SYNTHASE) DEFICIENCY, NUCLEAR TYPE 2; MC5DN2. Johns Hopkins University; last edited 4 May 2023. https://omim.org/entry/614052. Accessed 27 October 2026.
36. GeneGuidance. TMEM70 — clinical curation (complex V deficiency nuclear type 2). Version of 2026-09-26.
37. Orphanet. TMEM70-related mitochondrial encephalo-cardio-myopathy (ORPHA:1194). https://www.orpha.net/en/disease/detail/1194
38. OMIM® Online Mendelian Inheritance in Man. #603358 GRACILE SYNDROME. Johns Hopkins University; last edited 18 September 2023. https://omim.org/entry/603358. Accessed 27 October 2026.
39. OMIM® Online Mendelian Inheritance in Man. #262000 BJORNSTAD SYNDROME; BJS. Johns Hopkins University; last edited 11 January 2022. https://omim.org/entry/262000. Accessed 27 October 2026.
40. GeneGuidance. BCS1L — clinical curation (GRACILE, complex III deficiency, Björnstad). Version of 2026-09-26.
41. Orphanet. GRACILE syndrome (ORPHA:53693). https://www.orpha.net/en/disease/detail/53693
42. OMIM® Online Mendelian Inheritance in Man. #604377 MITOCHONDRIAL COMPLEX IV DEFICIENCY, NUCLEAR TYPE 2; MC4DN2. Johns Hopkins University; last edited 2 November 2020. https://omim.org/entry/604377. Accessed 27 October 2026.
43. GeneGuidance. SCO2 — clinical curation (fatal infantile cardioencephalomyopathy due to COX deficiency). Version of 2026-09-26.
44. Orphanet. Isolated cytochrome C oxidase deficiency (ORPHA:254905); Fatal infantile cytochrome C oxidase deficiency (ORPHA:1561). https://www.orpha.net/en/disease/detail/254905
45. OMIM® Online Mendelian Inheritance in Man. #256000 LEIGH SYNDROME, NUCLEAR; NULS. Johns Hopkins University; last edited 14 May 2024. https://omim.org/entry/256000. Accessed 27 October 2026.
46. GeneGuidance. SURF1 — clinical curation (Leigh syndrome due to COX deficiency; CMT4K). Version of 2026-09-26.
47. OMIM® Online Mendelian Inheritance in Man. #220111 MITOCHONDRIAL COMPLEX IV DEFICIENCY, NUCLEAR TYPE 5; MC4DN5. Johns Hopkins University; last edited 2 November 2020. https://omim.org/entry/220111. Accessed 27 October 2026.
48. GeneGuidance. LRPPRC — clinical curation (Leigh franco-canadiano). Version of 2026-09-26.
49. GeneGuidance. NDUFS4 — clinical curation (complex I deficiency/Leigh syndrome). Version of 2026-09-26.
50. Orphanet. Isolated complex I deficiency (ORPHA:2609). https://www.orpha.net/en/disease/detail/2609
51. GeneGuidance. GFM1 — clinical curation (COXPD1). Version of 2026-09-26.
52. Orphanet. Mitochondrial myopathy with reversible cytochrome C oxidase deficiency (ORPHA:254864); Hepatoencephalopathy due to combined oxidative phosphorylation defect type 1 (ORPHA:137681). https://www.orpha.net/en/disease/detail/254864
53. OMIM® Online Mendelian Inheritance in Man. #610505 COMBINED OXIDATIVE PHOSPHORYLATION DEFICIENCY 3; COXPD3. Johns Hopkins University; last edited 22 December 2023. https://omim.org/entry/610505. Accessed 27 October 2026.
54. GeneGuidance. TSFM — clinical curation (COXPD3). Version of 2026-09-26.
55. OMIM® Online Mendelian Inheritance in Man. #610678 COMBINED OXIDATIVE PHOSPHORYLATION DEFICIENCY 4; COXPD4. Johns Hopkins University; last edited 2 December 2016. https://omim.org/entry/610678. Accessed 27 October 2026.
56. OMIM® Online Mendelian Inheritance in Man. #614096 COMBINED OXIDATIVE PHOSPHORYLATION DEFICIENCY 8; COXPD8. Johns Hopkins University; last edited 7 December 2023. https://omim.org/entry/614096. Accessed 27 October 2026.
57. GeneGuidance. AARS2 — clinical curation (COXPD8; leucoencephalopathy with ovarian failure). Version of 2026-09-26.
58. OMIM® Online Mendelian Inheritance in Man. #614702 COMBINED OXIDATIVE PHOSPHORYLATION DEFICIENCY 10; COXPD10. Johns Hopkins University; last edited 22 December 2023. https://omim.org/entry/614702. Accessed 27 October 2026.
59. GeneGuidance. MTO1 — clinical curation (COXPD10). Version of 2026-09-26.
60. OMIM® Online Mendelian Inheritance in Man. #615440 COMBINED OXIDATIVE PHOSPHORYLATION DEFICIENCY 17; COXPD17. Johns Hopkins University; last edited 22 December 2023. https://omim.org/entry/615440. Accessed 27 October 2026.
61. GeneGuidance. ELAC2 — clinical curation (COXPD17). Version of 2026-09-26.
62. OMIM® Online Mendelian Inheritance in Man. #617228 COMBINED OXIDATIVE PHOSPHORYLATION DEFICIENCY 31; COXPD31. Johns Hopkins University; last edited 22 December 2021. https://omim.org/entry/617228. Accessed 27 October 2026.
63. Wang L, Lu P, Yin J, Xu K, Xiang D, Zhang Z, et al. Case report: Rare novel MIPEP compound heterozygous variants presenting with hypertrophic cardiomyopathy, severe lactic acidosis and hypotonia in a Chinese infant. Front Cardiovasc Med. 2022;9:1095882. doi:10.3389/fcvm.2022.1095882. PMID: 36727025; PMC9884671.
64. OMIM® Online Mendelian Inheritance in Man. #613070 LIVER FAILURE, INFANTILE, TRANSIENT; LFIT. Johns Hopkins University; last edited 14 November 2022. https://omim.org/entry/613070. Accessed 27 October 2026.
65. OMIM® Online Mendelian Inheritance in Man. #500009 MITOCHONDRIAL MYOPATHY, INFANTILE, TRANSIENT; MMIT. Johns Hopkins University; last edited 27 October 2011. https://omim.org/entry/500009. Accessed 27 October 2026.
66. GeneGuidance. MT-TE — clinical curation (reversible infantile respiratory chain deficiency). Version of 2026-09-26.
67. Lamari F, Rossignol F, Mitchell GA. Disorders of intracellular triglyceride and phospholipid metabolism. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 35, sections 35.3.6–35.3.7. doi:10.1007/978-3-662-63123-2_35
68. OMIM® Online Mendelian Inheritance in Man. #612073 MITOCHONDRIAL DNA DEPLETION SYNDROME 5 (ENCEPHALOMYOPATHIC WITH OR WITHOUT METHYLMALONIC ACIDURIA); MTDPS5. Johns Hopkins University; last edited 13 July 2017. https://omim.org/entry/612073. Accessed 27 October 2026.
69. OMIM® Online Mendelian Inheritance in Man. #245400 MITOCHONDRIAL DNA DEPLETION SYNDROME 9 (ENCEPHALOMYOPATHIC TYPE WITH METHYLMALONIC ACIDURIA); MTDPS9. Johns Hopkins University; last edited 12 August 2024. https://omim.org/entry/245400. Accessed 27 October 2026.
70. Brivet M, Gaignard P, Schiff M. Disorders of pyruvate metabolism and the tricarboxylic acid cycle. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 11, section 11.6. doi:10.1007/978-3-662-63123-2_11
71. OMIM® Online Mendelian Inheritance in Man. #614299 MULTIPLE MITOCHONDRIAL DYSFUNCTIONS SYNDROME 2 WITH HYPERGLYCINEMIA; MMDS2. Johns Hopkins University; last edited 8 July 2025. https://omim.org/entry/614299. Accessed 27 October 2026.
72. OMIM® Online Mendelian Inheritance in Man. #616370 MULTIPLE MITOCHONDRIAL DYSFUNCTIONS SYNDROME 4; MMDS4. Johns Hopkins University; last edited 18 June 2018. https://omim.org/entry/616370. Accessed 27 October 2026.
73. OMIM® Online Mendelian Inheritance in Man. #617613 MULTIPLE MITOCHONDRIAL DYSFUNCTIONS SYNDROME 5; MMDS5. Johns Hopkins University; last edited 9 July 2020. https://omim.org/entry/617613. Accessed 27 October 2026.
74. Van Hove JLK, Van Coster R. Nonketotic hyperglycinaemia and lipoate deficiency disorders. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 23. doi:10.1007/978-3-662-63123-2_23
75. Morris AAM, Spiekerkoetter U. Disorders of mitochondrial fatty acid oxidation & 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, sections 12.1 (ACAD9) and 12.2 (FLAD1, SLC25A32). doi:10.1007/978-3-662-63123-2_12
76. OMIM® Online Mendelian Inheritance in Man. #616277 MITOCHONDRIAL SHORT-CHAIN ENOYL-CoA HYDRATASE 1 DEFICIENCY; ECHS1D. Johns Hopkins University; last edited 5 September 2025. https://omim.org/entry/616277. Accessed 27 October 2026.
77. Luft R, Ikkos D, Palmieri G, Ernster L, Afzelius B. A case of severe hypermetabolism of nonthyroid origin with a defect in the maintenance of mitochondrial respiratory control: a correlated clinical, biochemical, and morphological study. J Clin Invest. 1962;41(9):1776-804. doi:10.1172/JCI104637. PMID: 14467237; PMC291101.
78. Nyhan WL, Hoffmann GF, Al-Aqeel AI, Barshop BA. Introduction to lactic acidemias. In: Atlas of Inherited Metabolic Diseases. 4th ed. Boca Raton: CRC Press; 2020. p. 353–363.
79. Holt IJ, Harding AE, Morgan-Hughes JA. Deletions of muscle mitochondrial DNA in patients with mitochondrial myopathies. Nature. 1988;331(6158):717-9. doi:10.1038/331717a0. PMID: 2830540.
80. Wallace DC, Singh G, Lott MT, Hodge JA, Schurr TG, Lezza AM, et al. Mitochondrial DNA mutation associated with Leber's hereditary optic neuropathy. Science. 1988;242(4884):1427-30. doi:10.1126/science.3201231. PMID: 3201231.
81. Landrum MJ, Chitipiralla S, Brown GR, Chen C, Gu B, Hart J, et al. ClinVar: improvements to accessing data. Nucleic Acids Res. 2020;48(D1):D835-D844. doi:10.1093/nar/gkz972. PMID: 31777943; PMC6943040. VCV records accessed on 28 October 2026: https://www.ncbi.nlm.nih.gov/clinvar/
82. Bourgeron T, Rustin P, Chretien D, Birch-Machin M, Bourgeois M, Viegas-Péquignot E, et al. Mutation of a nuclear succinate dehydrogenase gene results in mitochondrial respiratory chain deficiency. Nat Genet. 1995;11(2):144-9. doi:10.1038/ng1095-144. PMID: 7550341.
83. Ferreira CR, Rahman S, Keller M, Zschocke J; ICIMD Advisory Group. An international classification of inherited metabolic disorders (ICIMD). J Inherit Metab Dis. 2021;44(1):164-177. doi:10.1002/jimd.12348. PMID: 33340416; PMC9021760.
84. Schiff M, Brassier A, Dionisi-Vici C. Branched-chain organic acidurias/acidaemias. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 18, section 18.3 (Table 18.1). doi:10.1007/978-3-662-63123-2_18
85. Sandmann C, Gudapati S, Mayr JA, Abou Jamra R, Kovacevic A, Syrbe S. Cardiomyopathy and mitochondrial encephalomyopathy in a female child associated with a heterozygous X-linked AIFM1 variant. Mol Cell Pediatr. 2026;13(1):34. doi:10.1186/s40348-026-00246-z
86. Magner M, Dvorakova V, Tesarova M, Mazurova S, Hansikova H, Zahorec M, et al. TMEM70 deficiency: long-term outcome of 48 patients. J Inherit Metab Dis. 2015;38(3):417-26. doi:10.1007/s10545-014-9774-8. PMID: 25326274.
87. Freeman PJ, Hart RK, Gretton LJ, Brookes AJ, Dalgleish R. VariantValidator: Accurate validation, mapping, and formatting of sequence variation descriptions. Hum Mutat. 2018;39(1):61-68. doi:10.1002/humu.23348. PMID: 28967166; PMC5765404. Available from: https://variantvalidator.org/
88. Banerjee R, Purhonen J, Sultana N, Ros O, Nieminen AI, Kietz C, et al. Hepatic gene replacement improves energy metabolism and survival in a mouse model of neonatal mitochondrial disease GRACILE syndrome. Mol Ther. 2026;34(8):4518-4530. doi:10.1016/j.ymthe.2026.04.044. PMID: 42035262; PMC13464136.
89. Ferreira CR, Darling A, Vockley J. Disorders of nucleic acid metabolism, tRNA metabolism and ribosomal biogenesis. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 39, section 39.2.3. doi:10.1007/978-3-662-63123-2_39
90. Schiff M, Mochel F, Dionisi-Vici C. Emergency treatments. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 4, section 4.2.5. doi:10.1007/978-3-662-63123-2_4
91. Saudubray JM, García-Cazorla Á. Clinical approach to inborn errors of metabolism in paediatrics. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 1, sections 1.1.3.2, 1.3.5.2, 1.3.7.1, 1.4.12–1.4.13. doi:10.1007/978-3-662-63123-2_1
92. Wedatilake Y, Brown RM, McFarland R, Yaplito-Lee J, Morris AA, Champion M, et al. SURF1 deficiency: a multi-centre natural history study. Orphanet J Rare Dis. 2013;8:96. doi:10.1186/1750-1172-8-96. PMID: 23829769.
93. Parikh S, Goldstein A, Karaa A, Koenig MK, Anselm I, Brunel-Guitton C, et al. Patient care standards for primary mitochondrial disease: a consensus statement from the Mitochondrial Medicine Society. Genet Med. 2017;19(12):1380-1397. doi:10.1038/gim.2017.107. PMID: 28749475; PMC7804217.
94. Wesół-Kucharska D, Greczan M, Kaczor M, Pajdowska M, Piekutowska-Abramczuk D, Ciara E, et al. Early treatment of biotin-thiamine-responsive basal ganglia disease improves the prognosis. Mol Genet Metab Rep. 2021;29:100801. doi:10.1016/j.ymgmr.2021.100801. PMID: 34631424; PMC8488057.
95. Bernier FP, Boneh A, Dennett X, Chow CW, Cleary MA, Thorburn DR. Diagnostic criteria for respiratory chain disorders in adults and children. Neurology. 2002;59(9):1406-11. doi:10.1212/01.wnl.0000033795.17156.00. PMID: 12427892.
96. Wolf NI, Smeitink JA. Mitochondrial disorders: a proposal for consensus diagnostic criteria in infants and children. Neurology. 2002;59(9):1402-5. doi:10.1212/01.wnl.0000031795.91814.d8. PMID: 12427891.
97. Morava E, van den Heuvel L, Hol F, de Vries MC, Hogeveen M, Rodenburg RJ, et al. Mitochondrial disease criteria: diagnostic applications in children. Neurology. 2006;67(10):1823-6. doi:10.1212/01.wnl.0000244435.27645.54. PMID: 17130416.
98. Riley LG, Cowley MJ, Gayevskiy V, Minoche AE, Puttick C, Thorburn DR, et al. The diagnostic utility of genome sequencing in a pediatric cohort with suspected mitochondrial disease. Genet Med. 2020;22(7):1254-1261. doi:10.1038/s41436-020-0793-6. PMID: 32313153.
99. Sue CM, Balasubramaniam S, Bratkovic D, Bonifant C, Christodoulou J, Coman D, et al. Patient care standards for primary mitochondrial disease in Australia: an Australian adaptation of the Mitochondrial Medicine Society recommendations. Intern Med J. 2022;52(1):110-120. doi:10.1111/imj.15505. PMID: 34505344; PMC9299181.
100. Schon KR, Horvath R, Wei W, Calabrese C, Tucci A, Ibañez K, et al. Use of whole genome sequencing to determine genetic basis of suspected mitochondrial disorders: cohort study. BMJ. 2021;375:e066288. doi:10.1136/bmj-2021-066288. PMID: 34732400; PMC8565085.
101. Olimpio C, Paramonov I, Matalonga L, Laurie S, Schon K, Polavarapu K, et al. Increased Diagnostic Yield by Reanalysis of Whole Exome Sequencing Data in Mitochondrial Disease. J Neuromuscul Dis. 2024;11(4):767-775. doi:10.3233/JND-240020. PMID: 38759022; PMC11307028.
102. Nyhan WL, Hoffmann GF, Al-Aqeel AI, Barshop BA. Fructose-1,6-diphosphatase deficiency. In: Atlas of Inherited Metabolic Diseases. 4th ed. Boca Raton: CRC Press; 2020. p. 371–375.
103. De Vries MC, Brown DA, Allen ME, Bindoff L, Gorman GS, Karaa A, 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):800-818. doi:10.1002/jimd.12196. PMID: 32030781; PMC7383489.
104. Parikh S, Saneto R, Falk MJ, Anselm I, Cohen BH, Haas R, et al. A modern approach to the treatment of mitochondrial disease. Curr Treat Options Neurol. 2009;11(6):414-30. doi:10.1007/s11940-009-0046-0. PMID: 19891905; PMC3561461.
105. Rahman S, Clarke CF, Hirano M. 176th ENMC International Workshop: diagnosis and treatment of coenzyme Q₁₀ deficiency. Neuromuscul Disord. 2012;22(1):76-86. doi:10.1016/j.nmd.2011.05.001. PMID: 21723727; PMC3222743.
106. Maia Filho JM, Medeiros LO, Ribeiro EM. Deficiência de coenzima Q10 (CoQ10). In: Ribeiro EM, coord. Síndromes Genéticas: Reconhecer para Tratar. Fortaleza: EdUnichristus; 2021. p. 127–129.
107. Morris AAM, Jones S. Medications used in the treatment of inborn errors of metabolism. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 45, Table 45.1. doi:10.1007/978-3-662-63123-2_45
108. Wang Y, Hekimi S. The efficacy of coenzyme Q(10) treatment in alleviating the symptoms of primary coenzyme Q(10) deficiency: A systematic review. J Cell Mol Med. 2022;26(17):4635-4644. doi:10.1111/jcmm.17488. PMID: 35985679; PMC9443948.
109. Tabarki B, Al-Hashem A, Ortigoza-Escobar JD, Alsharhan H, Alfadhel M. Biotin-thiamine-responsive basal ganglia disease. 2013 Nov 21 [updated 2025 Jan 9]. In: Adam MP, Bick S, Mirzaa GM, et al., editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. PMID: 24260777. Available from: https://www.ncbi.nlm.nih.gov/books/NBK169615/
110. Smeitink J, van Es J, Bosman B, Janssen MCH, Klopstock T, Gorman G, et al. Phase 2b program with sonlicromanol in patients with mitochondrial disease due to m.3243A>G mutation. Brain. 2025;148(3):896-907. doi:10.1093/brain/awae277. PMID: 39501914; PMC11884763.
111. European Medicines Agency. Kygevvi (doxecitine/doxribtimine): EPAR. EMEA/H/C/005119. CHMP opinion 2026-01-29; authorisation 2026-03-26. Available from: https://www.ema.europa.eu/en/medicines/human/EPAR/kygevvi
112. European Medicines Agency. Kygevvi (doxecitine/doxribtimine) 2 g/2 g powder for oral solution: Summary of Product Characteristics (Annex I). UCB Pharma; first authorisation 26 March 2026. Available from: https://www.ema.europa.eu/en/documents/product-information/kygevvi-epar-product-information_en.pdf
113. López-Gómez C, Cámara Y, Hirano M, Martí R; 232nd ENMC Workshop Participants. 232nd ENMC international workshop: Recommendations for treatment of mitochondrial DNA maintenance disorders. 16 - 18 June 2017, Heemskerk, The Netherlands. Neuromuscul Disord. 2022;32(7):609-620. doi:10.1016/j.nmd.2022.05.008. PMID: 35641351.
114. Stacpoole PW, Cederbaum SD. Should the "mitochondrial cocktail" be a default option? An opinion. Mol Genet Metab. 2025;146(4):109264. doi:10.1016/j.ymgme.2025.109264. PMID: 41161075.
115. Stacpoole PW, Kerr DS, Barnes C, Bunch ST, Carney PR, Fennell EM, et al. Controlled clinical trial of dichloroacetate for treatment of congenital lactic acidosis in children. Pediatrics. 2006;117(5):1519-31. doi:10.1542/peds.2005-1226. PMID: 16651305.
116. Avula S, Parikh S, Demarest S, Kurz J, Gropman A. Treatment of mitochondrial disorders. Curr Treat Options Neurol. 2014;16(6):292. doi:10.1007/s11940-014-0292-7. PMID: 24700433; PMC4067597.
117. Stealth BioTherapeutics Inc. FORZINITY (elamipretide) injection, for subcutaneous use. Prescribing information. NDA 215244, US accelerated approval 2025-09-19. openFDA label set_id 146bf34c-76f2-48db-ac07-fb29cce2cd75 (effective 2025-12-10). Available from: https://dailymed.nlm.nih.gov/dailymed/drugInfo.cfm?setid=146bf34c-76f2-48db-ac07-fb29cce2cd75
118. European Medicines Agency. Medicines output — report of authorised medicines. Available from: https://www.ema.europa.eu/en/documents/report/medicines-output-medicines_json-report_en.json
119. Agência Nacional de Vigilância Sanitária (ANVISA). Dados abertos: medicamentos registados no Brasil (DADOS_ABERTOS_MEDICAMENTOS.csv) [accessed 2026 Oct 27]. Available from: https://dados.anvisa.gov.br/dados/DADOS_ABERTOS_MEDICAMENTOS.csv
120. ClinicalTrials.gov. NCT00432744: Phase 3 Trial of Coenzyme Q10 in Mitochondrial Disease. Available from: https://clinicaltrials.gov/study/NCT00432744
121. Stacpoole PW, deGrauw TJ, Feigenbaum AS, Hoppel C, Kerr DS, McCandless SE, et al. Design and implementation of the first randomized controlled trial of coenzyme CoQ₁₀ in children with primary mitochondrial diseases. Mitochondrion. 2012;12(6):623-9. doi:10.1016/j.mito.2012.09.005. PMID: 23022402; PMC3975832.
122. ClinicalTrials.gov. NCT04378075: Efficacy and Safety Study of Vatiquinone for the Treatment of Mitochondrial Disease Subjects With Refractory Epilepsy (MIT-E). Available from: https://clinicaltrials.gov/study/NCT04378075
123. ClinicalTrials.gov. NCT05162768: Elamipretide in Subjects With Primary Mitochondrial Disease Resulting From Pathogenic Nuclear DNA Mutations (NuPower). Available from: https://clinicaltrials.gov/study/NCT05162768
124. ClinicalTrials.gov. NCT03845712: A Phase 2 Open-Label Study of Continuation Treatment With Combination Pyrimidine Nucleosides in Patients With Thymidine Kinase 2 Deficiency (TK2). Available from: https://clinicaltrials.gov/study/NCT03845712
125. Koene S, van Bon L, Bertini E, Jimenez-Moreno C, van der Giessen L, de Groot I, et al. Outcome measures for children with mitochondrial disease: consensus recommendations for future studies from a Delphi-based international workshop. J Inherit Metab Dis. 2018;41(6):1267-1273. doi:10.1007/s10545-018-0229-5. PMID: 30027425; PMC6326961.
126. Akinci G, Sardarzada J, Topaloglu H. Transforming the natural course of infantile onset thymidine kinase 2 deficiency through early nucleoside replacement therapy. J Neuromuscul Dis. 2026 Apr 7:22143602261433223 (online ahead of print). doi:10.1177/22143602261433223. PMID: 41944166; PMC13438941.
127. ClinicalTrials.gov. NCT06451757: KHENERFIN — Sonlicromanol in Adult Subjects With m.3243A>G. Available from: https://clinicaltrials.gov/study/NCT06451757
128. Smeitink J, van Maanen R, de Boer L, Ruiterkamp G, Renkema H. A randomised placebo-controlled, double-blind phase II study to explore the safety, efficacy, and pharmacokinetics of sonlicromanol in children with genetically confirmed mitochondrial disease and motor symptoms ("KHENERGYC"). BMC Neurol. 2022;22(1):158. doi:10.1186/s12883-022-02685-3. PMID: 35477351; PMC9044835.
129. ClinicalTrials.gov. NCT05650229 (FALCON) and NCT07514338: KL1333 in Patients With Primary Mitochondrial Disease. Available from: https://clinicaltrials.gov/study/NCT05650229
130. ClinicalTrials.gov. API v2 search (sonlicromanol, vatiquinone, elamipretide, EPI-743, doxecitine, deoxythymidine, mitochondrial donation, pronuclear transfer), 27 October 2026. Available from: https://clinicaltrials.gov/
131. ClinicalTrials.gov. API v2 search (mitochondrial conditions × AAV/gene therapy/gene transfer), 27 October 2026. Available from: https://clinicaltrials.gov/
132. Hyslop LA, Blakely EL, Aushev M, Marley J, Takeda Y, Pyle A, et al. Mitochondrial Donation and Preimplantation Genetic Testing for mtDNA Disease. N Engl J Med. 2025;393(5):438-449. doi:10.1056/NEJMoa2415539. PMID: 40673696; PMC7617940.
133. McFarland R, Hyslop LA, Feeney C, Pillai RN, Blakely EL, Moody E, et al. Mitochondrial Donation in a Reproductive Care Pathway for mtDNA Disease. N Engl J Med. 2025;393(5):461-468. doi:10.1056/NEJMoa2503658. PMID: 40689593; PMC7617939.
134. ClinicalTrials.gov. NCT04113447: Mitochondrial Donation: An 18 Month Outcome Study. Available from: https://clinicaltrials.gov/study/NCT04113447
135. Newson AJ, Williams J, Fuscaldo G, Hill A, Kneebone E, Ludlow K, et al. Public funding for mitochondrial donation: An Australian public deliberation. BMC Med Ethics. 2025;26(1):131. doi:10.1186/s12910-025-01284-4. PMID: 41063113; PMC12509408.
136. Monash University. The mitoHOPE Program: progress [Internet]. Melbourne: Monash University; updated 2026 Aug 18 [accessed 2026 Oct 27]. Available from: https://www.monash.edu/medicine/mitohope/progress
137. Gorman GS, Schaefer AM, Ng Y, Gomez N, Blakely EL, Alston CL, et al. Prevalence of nuclear and mitochondrial DNA mutations related to adult mitochondrial disease. Ann Neurol. 2015;77(5):753-9. doi:10.1002/ana.24362. PMID: 25652200; PMC4737121.