Glucose Transporter Type 1 Deficiency Syndrome (GLUT1)

GLUT1 deficiency syndrome (Glut1DS), De Vivo disease and the SLC2A1-related allelic spectrum, including exercise-induced paroxysmal forms.

Literature and regulatory information updated: 6 October 2026.

0. Initial Reference Table

1. Introduction

Glucose transporter type 1 deficiency syndrome is a disorder of cerebral energy metabolism. Glucose is the brain's essential fuel and enters the parenchyma through GLUT1, expressed in endothelial cells forming the blood–brain barrier and in astrocyte membranes; GLUT3 predominates in neurons. When GLUT1 function is insufficient, glucose crossing the barrier cannot meet the brain's requirements, which rise rapidly after birth, peak in early childhood and remain high until approximately 10 years of age, subsequently declining throughout adolescence. The result is a chronic cerebral “energy crisis”, with intermittent worsening during fasting, exertion and metabolic stress [1, 7].

The foundational article was published in September 1991 in the New England Journal of Medicine by Darryl De Vivo and colleagues from Columbia University. They described two children with seizures beginning in the third month of life, refractory to antiseizure medications, developmental delay and, in one case, deceleration of head growth. Repeated lumbar punctures consistently showed low CSF glucose, with CSF/blood glucose ratios between 0.19 and 0.35 (normal 0.65) and CSF lactate at or below the lower limit of normal; usual causes of hypoglycorrhachia, such as bacterial meningitis, subarachnoid haemorrhage and hypoglycaemia, were absent. Erythrocyte studies demonstrated reduced cytochalasin B binding, consistent with lower transporter density, and, in the first patient, 3-O-methyl-D-glucose uptake of approximately 60% of that observed in the parents and a control. The authors postulated impaired glucose transport across the blood–brain barrier and treated the children with a ketogenic diet: seizures ceased within four and seven days, respectively [14]. Nyhan's Atlas dates the first description to 1951, a typographical error: the reference cited by the Atlas entry itself is the 1991 article [14, 15].

The molecular basis was established in the late 1990s: a heterozygous SLC2A1 deletion was found in one of the original patients, and in 2000, 15 heterozygous patients were described, establishing autosomal dominant disease caused by haploinsufficiency. In 2010, the first family with autosomal recessive inheritance was reported [1, 15]. In 2008, two European groups demonstrated that paroxysmal exercise-induced dyskinesia, alone or associated with epilepsy, results from SLC2A1 variants, and in 2011, paroxysmal choreoathetosis with spasticity (DYT9) was shown to be allelic to DYT18 [16–18]. Early-onset absence epilepsy, epilepsy with myoclonic–atonic seizures and stomatin-deficient cryohydrocytosis followed, so Glut1DS now denotes a phenotypic continuum rather than a single presentation [1, 19].

The clinical relevance of this condition derives from three facts. First, there is a mechanism-directed intervention, ketogenic dietary therapy, which provides ketone bodies as alternative fuel, carried by the monocarboxylate transporter MCT1 independently of GLUT1. Second, timing matters: patients treated earlier have better outcomes, and deceleration of head growth in early childhood is considered a sign of irreversible injury. Third, diagnosis remains delayed, particularly in mild forms and adults [1, 7, 20]. In 2020, a group of 23 experts from ten countries published the first disease-specific international consensus on diagnosis and treatment, which serves as this article's reference, supplemented by general recommendations on ketogenic therapies in children and adults [7, 10, 11].

2. Molecular Genetics and Pathophysiology

2.1. Gene, protein and transport reserve

SLC2A1 lies at 1p34.2 and encodes GLUT1, a member of the SLC2 family of facilitative hexose transporters, which transports glucose down its concentration gradient independently of insulin. The protein contains 492 amino acids; the foundational article already noted its abundance in erythrocytes, cerebral microvessels and placenta and the organisation of the gene into 10 exons [3, 14]. In erythrocytes, GLUT1 accounts for approximately 5% of membrane protein, allowing transporter function to be studied in a blood sample [21]. The MANE Select reference transcript is NM_006516.4 (protein NP_006507.2); the literature and GeneReviews still use NM_006516.2, with no difference in numbering of the variants discussed here, as verified using VariantValidator [4].

Under normal conditions, glucose transport across the barrier is not the rate-limiting step in cerebral metabolism: approximately half the transported glucose returns to the blood, while intracellular phosphorylation by hexokinase limits glycolysis. A deficient transporter narrows this reserve, and a greater proportion of entering glucose is consumed, explaining the simultaneous reduction in CSF glucose and lactate [1, 14]. The cerebral metabolic rate of glucose is low during fetal life, rises linearly after birth, peaks around three years, remains high throughout the first decade and falls in the second. This explains the usually asymptomatic fetal and neonatal periods, onset in the first months and the tendency for seizures to diminish in adolescence [1]. Saudubray adds that immature endothelial tight junctions during fetal and neonatal life may allow some paracellular glucose transport [21].

The consequences extend beyond inadequate fuel. In murine models, GLUT1 haploinsufficiency disrupts cerebral angiogenesis during development, and the reduced capillary network worsens neuroglycopenia; GLUT1 deficiency also triggers neuroinflammation. These effects can be prevented by early, but not late, transporter replacement, supporting the concept of a therapeutic window [1, 7]. Saudubray also describes cerebral and retinal microvascular defects and reduced brain growth as consequences of the deficiency [21].

2.2. Residual function, variant type and severity

Clinical severity correlates with residual transporter function. Mild and intermittent phenotypes (epilepsy, dyskinesias and ataxia) are associated with reductions of 25–35% in GLUT1 function, while severe infantile-onset forms are associated with greater reductions, possibly 40–75%. The erythrocyte uptake assay, a surrogate measure of residual function, shows a mean uptake of approximately 50% of control values in the classic form [1]. In a study of 109 suspected cases with compatible clinical features and hypoglycorrhachia, mean erythrocyte uptake correlated with clinical severity (R² = 0.94) [22].

The international consensus and GeneReviews summarise correlations by variant class: missense variants are predominantly associated with mild or moderate phenotypes; splicing, nonsense, insertion, deletion and exon deletion variants with moderate or severe phenotypes; and whole-gene deletions with severe phenotypes [1, 7]. In a series of 57 patients, mutation type was associated with intellectual disability severity and complex movement disorders, while the CSF/blood glucose ratio was associated with mutation type and phenotype [23]. A systematic review published in July 2026, comprising 96 patients with movement disorders, found that non-missense variants were associated with earlier onset and more ataxia, while missense variants were independently associated with paroxysmal dyskinesias (odds ratio 3.02; 95% confidence interval [CI] 1.09–8.35) [24]. A series of 16 patients in 2005 found no obvious correlations between phenotype, genotype and biochemical measures, illustrating that these associations are group trends rather than individual predictors [25]. Five of the 21 pathogenic variants known at the time clustered in a vulnerable region of the fourth transmembrane domain, encoded by exon 4 [1].

2.3. Variant spectrum and mechanisms

Most variants are de novo and of various types (missense, nonsense, splicing, small insertions and deletions, exon deletions and microdeletions encompassing the entire gene), distributed throughout the gene [1, 21]. Non-coding variants detectable by genome but not exome sequencing have been described [1]. In 2026, a functional study demonstrated that the synonymous variant NM_006516.4:c.972G>A (p.Ser324=), adjacent to the intron 7 splice donor site, activates two cryptic sites and produces approximately 40% abnormal transcripts and 60% normal transcripts; this change had not been detected by whole-blood RNA analysis and segregated with epilepsy and paroxysmal dyskinesia phenotypes in three families [26]. Synonymous variants should therefore not be automatically filtered out when the phenotype is suggestive.

The predominant mechanism is loss of function with haploinsufficiency [1]. Some variants add an effect on erythrocytes: the deletion of four conserved amino acids in the pore region, NM_006516.4:c.844_855del p.(Gln282_Ser285del), described in a family with paroxysmal exercise-induced dyskinesia over three generations, reduces glucose transport and creates a cation leak that alters intracellular sodium, potassium and calcium concentrations, with haemolytic anaemia and echinocytosis; other variants from the same series (A275T and G314S, historical numbering, = NM_006516.4:c.823G>A p.(Ala275Thr) and c.940G>A p.(Gly314Ser)) reduced transport without altering cation permeability [17]. In stomatin-deficient cryohydrocytosis, SLC2A1 variants cause both loss of glucose transport and a cold-induced cation leak [19]. The consensus also mentions hypotheses of destabilisation of the native interactions of GLUT1, which functions as a tetramer,, protein aggregation and defects of transcription, processing and trafficking, as well as an influence of non-coding RNA genes [7].

2.4. Inheritance, mosaicism and recessive forms

Glut1DS is predominantly autosomal dominant; approximately 90% of patients have a de novo variant and approximately 10% inherited it from a parent. The transmitting parent may be mildly affected or unaffected, potentially explained by parental somatic mosaicism; both somatic and gonadal mosaicism have been described. In dominant inheritance, penetrance is described as complete, and mosaicism attenuates clinical severity; mosaicism should be investigated in an asymptomatic or minimally symptomatic parent carrying the child's variant [1]. GeneReviews also mentions possible reduced penetrance in a heterozygous parent among explanations for sibling risk when parents have not been tested [1].

Autosomal recessive inheritance has been documented in two families: one consanguineous and another in which a severely affected child was compound heterozygous for NM_006516.4:c.377G>T, p.(Arg126Leu), and c.766_767delinsGT, p.(Lys256Val). The asymptomatic mother carried p.(Lys256Val) and had 87% residual erythrocyte uptake (values above 74% correlate with a normal clinical state); the child had 37%. In Xenopus oocytes, p.(Arg126Leu) was more deleterious than p.(Lys256Val). The authors interpreted this as synergy between two missense variants with partial effects [1]. Their coordinates were confirmed using VariantValidator (GRCh38: NC_000001.11:g.42930765C>A and g.42929693_42929694delinsAC) [4]. The following table brings together variants cited in this article with normalised nomenclature and current ClinVar classifications.

HGVS normalisation and genomic coordinates for the first five variants were obtained using VariantValidator; the synonymous variant is described using the nomenclature of the original article, which agrees with its ClinVar record [4, 26, 28]. The discrepancy for p.(Lys256Val) illustrates an interpretative principle: a hypomorphic allele that causes disease only in combination with another allele may remain classified as a variant of uncertain significance in databases, and that classification should not be changed in a report solely on the basis of the severity of the compound heterozygous case.

2.5. From Transporter Dysfunction to Symptoms

The clinical presentation reflects the interaction between residual function and energy requirements. Manifestations are influenced by meal patterns, physical activity and intercurrent illness; seizures typically occur in the morning before the first meal and resolve after food intake, while movement disorders worsen with fasting, exertion, infection and emotional stress [1]. Exercise-induced dyskinesia has been interpreted as a consequence of an episodic energy deficit in the basal ganglia during exertion [17]. In forms involving cation leakage, the energy deficit may be accompanied by haemolysis [21]. Ketone bodies cross the barrier through MCT1, independently of GLUT1, and can partially replace glucose as a substrate for cerebral energy metabolism; this underlies ketogenic therapy, discussed in Section 7 [1, 29].

3. Clinical Spectrum and Manifestations by Organ System

3.1. An Age-Dependent Continuum

The cardinal features of the classic form are severe epilepsy, a complex movement disorder and developmental delay following an uncomplicated pregnancy and neonatal period [1, 21]. The pattern evolves with age: paroxysmal eye–head movements, drug-resistant seizures, deceleration of head growth and developmental delay predominate in infancy; persistent and paroxysmal movement disorders, dysarthria and intellectual disability predominate in older children, adolescents and adults, while seizures diminish or resolve [1, 7]. In a series of 57 patients, the classic phenotype (84%) comprised early onset before two years of age (65%) and late onset (18%); 15% had a non-classic phenotype with intellectual disability and a movement disorder without epilepsy, and one adult had minimal symptoms [23]. The Atlas describes a classic phenotype in 81% of 16 patients (infantile seizures, acquired microcephaly and spasticity), as well as seizure-free cases with dystonia, dysarthria and ataxia [15]. Saudubray recognises variants with only one or two cardinal features, such as isolated early-onset absence epilepsy or an isolated movement disorder, and adults with minimal symptoms [21].

3.2. Central Nervous System: Epilepsy

Seizures usually begin before six months of age, are refractory to antiseizure medications and may be preceded by episodes of apnoea, cyanosis or abnormal eye movements [1, 21]. All seizure types are possible (generalised tonic–clonic, focal, myoclonic, atonic, typical and atypical absence seizures), with generalised seizures more frequent than focal seizures [1, 7]. Frequency ranges from several seizures daily to seizures separated by weeks or months [1]. The international consensus notes that catastrophic epilepsy syndromes, such as Lennox–Gastaut syndrome, have not been described [7]. In the Scottish population-based cohort, six of the seven patients with SLC2A1 variants presented with generalised seizures (three tonic–clonic, two myoclonic and one absence), with a median age at onset of 12 months; Glut1DS was suspected before the genetic result in only two, and six had developed subtle gait ataxia by reassessment [8].

Two epilepsy syndromes warrant particular attention. Early-onset absence epilepsy, before four years of age, is characteristic: typical absences with generalised 3 Hz spike-and-wave discharges [1, 7]. Saudubray attributes up to 10% of early-onset absence epilepsy cases to Glut1DS [21]; a Danish and European study identified variants in 5 of 50 patients (10%) with absence epilepsy [30]. The second is epilepsy with myoclonic–atonic seizures (Doose syndrome): Saudubray and Oxford attribute approximately 5% of these cases to Glut1DS and recommend excluding it [21, 31], but the same Danish study found no SLC2A1 variants in 120 patients with myoclonic–astatic epilepsy without an associated movement disorder [30]. This discrepancy probably reflects cohort selection; in practice, the association of myoclonic–atonic seizures with a movement disorder, worsening during fasting or a dietary response strengthens suspicion. The Orphanet entry for epilepsy with myoclonic absences, a condition distinct from Doose syndrome, regards SLC2A1 variants as one of its most frequent monogenic causes [6].

The spectrum also includes susceptibility to idiopathic generalised epilepsy, including childhood absence epilepsy (OMIM 614847) [3], and families with dominantly inherited epilepsy and marked intrafamilial variability, from asymptomatic carriers to diverse epilepsy presentations [1, 15].

3.3. Central Nervous System: Eye Movements and Movement Disorders

Paroxysmal eye–head movements are the second most frequent early sign after seizures [7]. In a Columbia University series of 101 patients, they were documented in 32 (32%); among 18 with detailed descriptions, they began before six months in 15 (83%) and preceded seizures in 10 of 16 patients with both types of episode. Episodes were brief (usually less than 5 minutes), with preserved awareness; video analysis showed rapid, multidirectional saccades separated by fixation intervals of 200 to 800 ms and accompanied by head movement in the same direction approximately 30–60 ms later. They disappeared between 3 and 6 years of age in seven of eight patients with known follow-up. The authors proposed the term paroxysmal eye–head saccades and distinguished them from opsoclonus, which lacks fixation intervals and head movement in the same direction [32]. Saudubray regards these movements as pathognomonic and transient features of the first months of life [33].

Movement disorders are the predominant manifestation in adolescence and adulthood. In a study of videos and records from 57 patients, 89% had gait impairment (predominantly spastic–ataxic or ataxic), 86% limb action dystonia, 75% mild chorea, 70% cerebellar action tremor, 28% non-epileptic paroxysmal episodes (ataxia, weakness, parkinsonism and non-kinesigenic dyskinesias), 21% dyspraxia and 16% myoclonus [1, 34]. Ataxia is predominantly axial rather than appendicular and becomes apparent when the child starts walking; chorea mainly affects the face and distal upper limbs; dyspraxia, including oculomotor and orobuccal dyspraxia, is under-recognised [7]. Paroxysmal movement disorders affect approximately 75% of patients, and triggers include fasting, exercise, emotional stress, fever, fatigue, inadequate ketosis, sleep deprivation, temperature changes and medications [7]. In the 2026 systematic review, movement disorders were the initial manifestation in 34.4% of 96 cases, with paroxysmal dyskinesias (63.5%) and ataxia (47.9%) the most frequent phenotypes; patients presenting with motor manifestations had later onset and less cognitive impairment [24].

Dystonic tremor of the limbs and voice may be the only manifestation, as in a mother and daughter described in 2011; electrophysiological recordings showed a 6–8.5 Hz tremor with bursts lasting 80–170 ms [1, 15]. Saudubray includes head tremor among the typical signs, alongside vitamin E deficiency [33].

3.4. Allelic Paroxysmal Disorders

Paroxysmal exercise-induced dyskinesia (DYT18, now PxMD-SLC2A1 in Movement Disorder Society nomenclature) is characterised by onset beyond early childhood, normal head circumference, a normal neurological examination between episodes and a less pronounced reduction in CSF glucose than in the classic form [1, 21]. The Orphanet entry describes episodes triggered by 15 to 60 minutes of exercise, lasting 5 minutes to 2 hours, usually confined to the exercised limbs, bilateral, exacerbated by cold, stress, fatigue and sleep deprivation, and occurring from once daily to once monthly; onset may occur between 1 and 30 years of age [5]. In a five-generation family and three nuclear families, dyskinesia (choreoathetosis, dystonia or both, predominantly involving the lower limbs) coexisted with generalised epilepsy, the median CSF/blood glucose ratio was 0.52, and three patients responded to the ketogenic diet [16]. The Atlas describes involuntary leg movements after walking or arm movements after writing, transmission over three and five generations, and a normal EEG in up to 43% of patients [15].

Paroxysmal choreoathetosis with spasticity (DYT9) combines predominantly exercise-induced dyskinesia, beginning between 1 and 15 years of age, with progressive spastic paraparesis beginning in young adulthood, mild ataxia, mild-to-moderate cognitive impairment and epileptic seizures; no causal variants were found in a sample of 139 patients with hereditary spastic paraparesis without paroxysmal dyskinesia [1, 18].

Stomatin-deficient cryohydrocytosis combines haemolytic anaemia with cold-induced cation leakage, hepatosplenomegaly, cataracts, seizures, intellectual disability and a movement disorder [19]. Pseudohyperkalaemia is among the rare manifestations recorded in the international consensus [7]. Haemolytic anaemia with reticulocytosis and echinocytosis may also accompany exercise-induced dyskinesia [15, 17].

3.5. Development, Language and Behaviour

Intellectual disability ranges from learning difficulties to severe disability, and minimally affected patients have IQs at the lower end of the normal range [1]. Language is affected in all patients, with expression more impaired than comprehension; dysarthria is frequent and accompanied by dysfluency [1]. Performance abilities are more affected than verbal abilities, with visuospatial and visuomotor deficits, while adaptive social behaviour is a relative strength [7]. Features of attention-deficit/hyperactivity disorder (ADHD) are present in most patients and sometimes require medication; autism spectrum disorders appear under-represented, and mood disorders are uncommon [1]. Microcephaly was present in 32 of 58 patients: mild (more than 1 standard deviation below the mean) in 14, moderate (more than 2) in 10 and severe (more than 3) in 8 [1].

3.6. Other Neurological and Systemic Manifestations

Non-motor paroxysmal episodes include migraine, behavioural disturbances, cyclical vomiting and sleep episodes [1, 7]. Writer's cramp, intermittent ataxia, complete paralysis, parkinsonism, painful nocturnal muscle cramps, alternating hemiplegia, hemiplegic migraine and stroke-like episodes with hemiparesis, dysarthria or aphasia have been described [7, 33]. Hepatosplenomegaly, periventricular calcifications, cerebral atrophy, pseudohyperkalaemia, cataracts and retinal dysfunction are rare [7]. Saudubray reports a thin retina with reduced vascularisation and a normal electroretinogram, and includes SLC2A1 deficiency among the causes of hypertrophic osteoarthropathy (pachydermia) [33].

3.7. Adults

Adult data are more recent. In a series of 32 adults from two Milan centres (median age 32 years; 68.8% women), the median age at onset was 2 years and the diagnostic delay was 18 years; 62% were diagnosed in adulthood and 31% only after their child's diagnosis. With two exceptions, all had a history of typical symptoms that had gone unrecognised or were too mild to prompt consultation, and psychosocial difficulties were identified in 42% [20]. This observation shifts the focus in adults from an “atypical” presentation to lack of clinical recognition.

The French protocol groups adult presentations into four pictures: a mild chronic encephalopathy, with rare seizures, mild intellectual disability and variable degrees of spasticity and ataxia, often recognised only after a child has been diagnosed; a residual severe epileptic encephalopathy, beginning in early childhood and tending to improve after puberty; an epilepsy resembling the idiopathic generalised epilepsies (absences or isolated generalised tonic–clonic seizures), in which coexisting paroxysmal exercise-induced dyskinesia points to the diagnosis; and paroxysmal exercise-induced dyskinesia, isolated or accompanied by fixed signs (dysarthria, ataxia, spasticity, dystonia) [12]. In this last picture, what suggests Glut1DS is the demonstration that episodes are triggered by prolonged exertion and worsened by fasting, which separates them from the paroxysmal kinesigenic and non-kinesigenic dyskinesias of lesional or other genetic causes. In most patients, symptoms stabilise after puberty [12].

4. Clinical Recognition and Warning Signs

Recognition depends on integrating signs that appear non-specific in isolation. The international consensus lists the following as suggestive: any unexplained movement disorder with spasticity, dystonia and ataxia; complete seizure control with ketogenic therapy in a child with drug-resistant epilepsy; unexplained paroxysmal episodes at any age; absence epilepsy before four years of age; and myoclonic–atonic epilepsy [7]. In the survey of 13 centres, paroxysmal eye–head movements in infants were considered specific even in isolation by 12 (92%), and absence epilepsy before four years by 10 (77%); conversely, unexplained developmental delay was never considered sufficient in isolation, and alternating hemiplegia and stroke-like episodes warranted investigation only when combined with other signs [7].

Source: Table 1 of the international consensus; percentages refer to 13 centres, and not all centres answered every question [7].

The relationship with meals is a warning sign across presentations. Seizures occurring in the morning before breakfast and resolving after a meal, an abnormal preprandial EEG that improves after eating, gait ataxia or dyspraxia that worsens before meals, and a movement disorder exacerbated by fasting should prompt measurement of CSF glucose [1, 7, 33]. Saudubray includes Glut1DS among causes of recurrent ataxic episodes, in which ataxia may be the only sign and is usually worse before meals, and dystonic episodes that may resemble cramps [33]. Exercise-triggered paroxysmal dystonia is considered highly suggestive of Glut1DS in adults, although it can occur in pyruvate dehydrogenase deficiency [35]. Other warning signs include deceleration of head growth, clinical deterioration with phenobarbital, failure to respond to several antiseizure medications, haemolytic anaemia associated with exertion or cold, and a family history of epilepsy, paroxysmal dyskinesia or migraine across several generations [1, 15, 17, 36].

The French PNDS organises suspicion in infants and young children around three types of initial manifestation. Paroxysmal eye-head movements, highly suggestive but not pathognomonic, occur mainly before 6 months of age and may precede seizures. Epileptic seizures are the first sign in many patients and are usually drug-resistant, a definition the protocol takes from the ILAE (International League Against Epilepsy): persistence of seizures despite two appropriate antiseizure medications at correct doses, a situation that applies to 20–30% of epilepsies of all causes. Particularly suggestive are early-onset drug-resistant epilepsy with progressive microcephaly, pyramidal signs or abnormal movements; typical absences beginning, in most cases, before 2 years of age (sometimes up to 4); epilepsies with myoclonus; and, more rarely, myoclonic-atonic epilepsy. Finally, any epilepsy associated with a paroxysmal (dyskinesia, dystonia) or fixed (ataxia, dyspraxia, pyramidal signs) motor disorder should raise the diagnosis [12]. Later, paroxysmal motor disturbances (involuntary movements, myoclonus, ataxia, dyskinesia, stroke-like weakness or paralysis) and non-motor disturbances (hemiplegic migraine or migraine with aura, behavioural disturbances, drowsiness) appear, together with persistent signs. Any neurological symptom that follows a pattern related to fasting, meals or exercise should prompt investigation; the protocol adds emotional stress, fever, fatigue, insufficient ketosis, sleep deprivation and temperature changes as triggers [12].

In adults, the main barrier is lack of recognition: in the Milan series, almost all patients had typical symptoms that had not been recognised as significant, and one third were diagnosed after their child [20]. A child's diagnosis should therefore prompt targeted assessment of the parents, even when the family history appears negative [1].

5. Diagnosis: Criteria, Tests and Algorithms

5.1. International Consensus Criteria

Diagnosis rests on three pillars: a characteristic clinical presentation, hypoglycorrhachia documented by lumbar puncture, and a pathogenic (or likely pathogenic) SLC2A1 variant, usually heterozygous [1, 7]. All consensus centres agreed that a definitive diagnosis requires all three criteria; when one is negative or uncertain, most considered two of three sufficient in the presence of characteristic clinical features. In an asymptomatic individual, the combination of hypoglycorrhachia and a pathogenic variant justifies the diagnosis only if the individual is a presymptomatic member of an affected family [7]. The following table reproduces the logic of consensus Table 2, read from the published image.

Source: Table 2 (figure) of the international consensus [7].

The French PNDS reproduces this three-criterion grid without changes, including the indications for starting the diet, and recommends that, when the picture is suggestive, the patient be referred to a specialist department for lumbar puncture and sequencing; confirmation is the responsibility of the paediatric neurologist, the specialist paediatrician, the neurologist or the epileptologist [12]. The same text accepts the diagnosis without an SLC2A1 variant on the basis of hypoglycorrhachia and a characteristic clinical picture, particularly when there is a favourable response to the ketogenic diet [12].

5.2. Lumbar Puncture: Preanalytical Conditions and Interpretation

Hypoglycorrhachia with normoglycaemia is the single most important laboratory finding [1]. Lumbar puncture should be performed in the postabsorptive state: GeneReviews specifies a four-hour fast, the international consensus four to six hours, and Saudubray four to six hours, in a stable metabolic state and outside a seizure episode. Blood glucose should be sampled immediately before lumbar puncture to prevent stress hyperglycaemia from increasing the denominator [1, 7, 21]. Reference values are independent of sex but age-dependent [7]. Other causes of hypoglycorrhachia must be excluded: hypoglycaemia, meningitis, subarachnoid haemorrhage and ventriculoperitoneal shunts [7]. In Glut1DS, CSF lactate is always low-normal or low, distinguishing this condition from other disorders of cerebral energy metabolism, particularly mitochondrial disorders; cell counts and protein are typically normal [7, 21].

Sources: [1, 7, 12, 16, 21, 23, 25].

The ratio threshold adopted by the French protocol (0.45) lies between that of GeneReviews (0.4) and that of Saudubray (0.5), which shows that no single cut-off value is universal. The PNDS also insists on repeating the lumbar puncture when the result is equivocal, particularly if the preceding fast was too short or if the ratio was borderline [12].

The Atlas notes that some reports, particularly of exercise-induced dyskinesia, describe CSF glucose of 2.6–2.8 mmol/L with ratios of 0.5–0.59, and that hypoglycorrhachia may not be present in every patient at every assessment [15]. Differences between sources reflect cohorts, disease stages and methods: a borderline value must be interpreted alongside simultaneous blood glucose, fasting status and clinical features, and cannot be taken to exclude the disorder when the phenotype is suggestive [1, 7].

5.3. Molecular Tests and Their Limitations

The approaches recommended by GeneReviews are exome sequencing (the most widely used) or genome sequencing, or a multigene panel including SLC2A1 and genes relevant to the differential diagnosis; single-gene analysis followed by deletion/duplication testing is rarely useful and is not recommended, and a panel may be too restrictive given the phenotypic spectrum [1]. Sequencing identifies approximately 90% of pathogenic variants and targeted deletion/duplication analysis (quantitative PCR, MLPA [multiplex ligation-dependent probe amplification] or targeted microarray) approximately 10%; exome and genome sequencing can detect deletions/duplications by read depth, but with lower sensitivity [1]. The consensus reports 81–89% of patients confirmed by sequencing and 11–14% by deletion/duplication analysis [7]. These proportions describe the distribution of detected variants and not the clinical sensitivity of the test: in 10–15% of patients with a Glut1DS-like presentation no SLC2A1 variant is found, even with MLPA, and the Orphanet entry estimates that 10% of patients with typical clinical and CSF findings have no molecular diagnosis [2, 21]. Some of these cases may result from non-coding alterations, synonymous variants that alter splicing or defects of transcriptional regulation, such as those suggested for the PURA gene [1, 21, 26]. The French protocol places the proportion of patients without an SLC2A1 abnormality at 5–10% and lists, among possible mechanisms, intronic sequences and defects of GLUT1 transcription, translation, activation or trafficking [12].

The technical annexes of the PNDS specify the limits of each method. Sequencing (NGS panel or Sanger) covers the 10 exons and the exon–intron junctions; Sanger sequencing does not detect large heterozygous deletions, which account for approximately 10% of variants, and an NGS panel detects them only if the laboratory performs copy number variation (CNV) analysis, which not all laboratories do; large deletions are confirmed by MLPA, FISH or quantitative PCR. In France, the reference laboratory is the genetics laboratory of the Bichat-Claude Bernard Hospital (Paris), which has studied the gene since 2005, and there is a targeted 11-gene panel for Glut1DS and its differential diagnosis; the gene is also included in panels for intellectual disability, epilepsy, epileptic encephalopathy, dystonia, ataxia, microcephaly and paroxysmal channelopathies [12]. For a variant of uncertain significance, the protocol calls for segregation analysis and a METAglut1 test to support causality; GeneReviews merely states that a heterozygous variant of uncertain significance neither establishes nor excludes the diagnosis [1, 12].

The report should specify the gene, transcript and version (NM_006516.4), HGVS c. and p. nomenclature, zygosity, classification according to ACMG/AMP criteria (American College of Medical Genetics and Genomics / Association for Molecular Pathology), the method and its limitations (deletion detection, non-coding region coverage and mosaicism). A heterozygous variant of uncertain significance neither establishes nor excludes the diagnosis [1]. A dietary response or partial segregation within a family does not, by itself, reclassify a variant of uncertain significance; functional studies and complete segregation analysis are investigational tools and should be described as such [1, 7].

5.4. Blood-Based Functional Tests

Uptake of 3-O-methyl-D-glucose by isolated erythrocytes measures transporter function. Patient values range from 35% to 74% of control values, with a mean of approximately 50%; the result should be abnormal in all confirmed cases, with one documented exception, and a pathogenic variant is detected molecularly in more than 95% of patients with low uptake [1]. In the series of 109 suspected cases, increasing the abnormality threshold from 60% to 74% yielded 99% sensitivity and 100% specificity [22]. Approximately 3% of patients have normal uptake in the assay at 4 °C, associated with p.(Thr295Met), which retains normal function at low temperature [1]. The assay is available only for research (Colleen Giblin Laboratory, Columbia University Irving Medical Center), and neither Western blot nor erythrocyte uptake testing is commercially available [1, 21]. In the absence of a pathogenic variant, the consensus considers uptake between 35% and 74% of control values diagnostic [7]. Measurement of GLUT1 on the erythrocyte surface by flow cytometry (METAglut1 test) has been described as diagnostically valuable, but the consensus did not consider it established as a standard tool [7, 21]. A registered study of 636 participants evaluated this test for early diagnosis, with no results posted in the registry [37].

The French protocol, which postdates the consensus, assigns this test a screening role. In France, METAglut1 is performed by the Cerba laboratory and by the biochemistry department of the Bichat Hospital. In an initial study of 30 patients, sensitivity was 78%; the PNDS describes the multicentre validation as a cohort of 549 patients with suspected disease and 78 patients with an established diagnosis, with sensitivity of approximately 80%, specificity above 99%, positive predictive value above 80% and negative predictive value above 97% [12]. The published study (33 French centres) analysed 428 patients in the prospective cohort, 15 of whom had newly diagnosed Glut1DS, and 67 patients in the retrospective cohort with a previous diagnosis; it confirmed a sensitivity of 80% and specificity above 99%, substantial agreement with CSF glucose and detection of patients with SLC2A1 mosaicism and with variants of uncertain significance [38]. The PNDS denominators do not correspond to the patients analysed in the article. The PNDS concludes that the test cannot yet serve as definitive proof but that, because sampling is easy, it is useful for screening; a negative result does not obviate lumbar puncture when the phenotype is suggestive, and its value in patients without an SLC2A1 abnormality remains undetermined [12]. For adults, the protocol proposes the following sequence: positive test, confirmatory genetic testing; negative test with strong suspicion, lumbar puncture, sequencing and glucose uptake study; negative test with weak suspicion, look for another cause [12]. With a sensitivity of approximately 80%, one in five patients would have a false-negative result, which justifies this hierarchy.

5.5. EEG and Imaging

The interictal EEG is frequently normal. Focal slowing and epileptiform discharges predominate in infants, while generalised 2.5–4 Hz spike-and-wave activity appears from two years of age; improvement in the preprandial EEG after a meal has diagnostic value when present [7, 21]. GeneReviews recommends pre- and postprandial EEG to characterise seizures and to distinguish seizures from non-epileptic events [1]; the French PNDS considers it an optional test, useful mainly when symptoms follow the rhythm of meals, and stresses that its diagnostic value has not been evaluated [12]. Magnetic resonance imaging (MRI) serves mainly to exclude other encephalopathies: approximately one in four patients has non-specific changes (hyperintensity of subcortical U-fibres, prominent perivascular spaces and delayed myelination) [7]. Fluorodeoxyglucose positron emission tomography (FDG-PET) shows diffuse cortical, cerebellar and thalamic hypometabolism, with relative preservation or apparent increase in the striatum, especially the caudate, and a more marked deficit in the mesial temporal lobe; this pattern appears early and persists into adulthood, irrespective of severity and diet, but its sensitivity and specificity have not been established, and the consensus does not regard it as a standard test [1, 7]. MR spectroscopy has been used in isolated cases [7].

5.6. Practical Algorithm

1. Clinical suspicion based on the signs in Section 4, with particular attention to their relationship with meals and exertion [1, 7].
2. Lumbar puncture after a 4–6-hour fast, with blood glucose sampled immediately beforehand; measure CSF glucose, lactate, cells and protein, and exclude other causes of hypoglycorrhachia [1, 7].
3. In parallel, molecular testing with SLC2A1 sequencing and deletion/duplication detection, ideally by exome or genome sequencing, interpreted according to ACMG/AMP criteria [1, 3].
4. Classify according to the three consensus criteria (confirmed, probable or possible) and initiate ketogenic therapy in confirmed and probable cases, considering it in possible cases [7].
5. If molecular testing is negative despite typical clinical and CSF findings, consider non-coding region analysis, RNA studies, a research erythrocyte functional assay and differential diagnosis genes (PURA, HK1) [1, 21, 26].
6. After diagnosis, assess disease extent (neurological assessment with MRI, movement assessment, pre- and postprandial EEG, rehabilitation, language, development, psychology, neuropsychiatry from 12 months, genetic counselling and family support) [1].
7. The French PNDS sets as the systematic initial assessment a general and neurological examination with head circumference, blood glucose followed by lumbar puncture (the fastest route, with a same-day result), genetic testing, a genetics clinic appointment after the result, an age-appropriate neuropsychological assessment (language, motor skills, visuomotor integration, attention, memory, executive functions, socio-emotional behaviour) and a pregnancy test in women of childbearing age before the diet or antiseizure medications, in order to choose compatible contraception. Depending on the context, it adds MRI and fasting EEG, blood and CSF lactate (normal or low in Glut1DS, sometimes raised in mitochondrial disorders), amino acids and organic acids to screen for disorders of intermediary metabolism, and assessment of renal and hepatic function [12].

Oxford recommends simultaneous CSF and blood glucose measurement when investigating seizures with developmental delay [31]. GeneGuidance summarises the practical rule: neither a low ratio should end the investigation nor a negative genetic test exclude the diagnosis [3]. Newborn screening is not established; the consensus and GeneReviews regard it as desirable given the effects of early intervention in animal models [1, 7].

6. Differential Diagnosis

The differential diagnosis depends on the presentation. The table integrates GeneReviews, Saudubray's clinical approach chapters and sources on paroxysmal forms.

Three practical points deserve note. First, CSF lactate separates Glut1DS from mitochondrial disorders, including pyruvate dehydrogenase deficiency, another first-line indication for the ketogenic diet [7, 10]. Second, the ketogenic diet is contraindicated in defects of fatty acid oxidation, ketogenesis and ketolysis, in pyruvate carboxylase deficiency and in porphyria, so the metabolic differential diagnosis should be reviewed before starting the diet in epilepsy without a confirmed aetiology [10, 29]. Third, GeneReviews records patients with a severe phenotype, typical CSF findings and normal erythrocyte uptake without an SLC2A1 variant, which suggests another, as yet unidentified, gene [1].

The French PNDS adds two practical recommendations. When abnormal movements may have another cause, it proposes CSF neurotransmitter measurement and a gene panel; and the French panel targeted at Glut1DS and its differential diagnosis (epilepsy, episodic ataxia and paroxysmal dyskinesias) includes, in addition to SLC2A1, the genes SCN1A, ADCY5, CACNA1A, ATP1A3, KCNA1, PRRT2, SCN2A, KCNQ2, SLC45A1 and PURA, which summarises the entities that the clinical picture may mimic [12]. The same protocol includes in the differential diagnosis episodes of paroxysmal neurological dysfunction that respond to or are prevented by carbohydrate intake (alternating hemiparesis, ataxia, cognitive dysfunction, seizures) and the other causes of myoclonic epilepsy in infancy [12].

7. Treatments and Risk Management Strategies

7.1. Principles

There is no cure for Glut1DS and no medicine specifically approved for this disease [1, 9]. The mechanism-directed intervention is ketogenic dietary therapy (KDT), introduced in 1991, which replaces much of the carbohydrate and protein with fat and creates chronic ketosis: ketone bodies produced in the liver cross the blood–brain barrier via MCT1 and provide a supplementary fuel for cerebral energy metabolism [1, 21, 29]. The international consensus defines KDT as first-line treatment for epilepsy, movement disorders and developmental impairment in Glut1DS, to be started at diagnosis, at any age and as early as possible, with the highest tolerated degree of ketosis, and to be maintained into adolescence and adulthood [7]. The Orphanet entry, reviewed with participation of the EpiCARE network, describes KDT as the gold standard and adds that symptoms may return when the diet is stopped [2]. Patients develop a mild compensated metabolic acidosis when in ketosis [1]. The French PNDS follows the same approach: classic ketogenic diet or modified Atkins diet as first line at any age, started as early as possible, prepared and followed by a multidisciplinary team with dual medical and dietetic supervision, and never started outside that setting; the stated goals are to control seizures and paroxysmal episodes, to stop or at least reduce antiseizure medications, to limit permanent motor and cognitive signs and to reduce the impact on school, social and family life [12].

The international consensus clearly opposed discontinuing KDT to enrol patients in trials of new therapies (12 of 13 centres), and 96% of International Ketogenic Diet Study Group members considered that KDT should be preferred to triheptanoin as first-line treatment, even in a patient without seizures and with a movement disorder alone [7, 10].

7.2. Ketogenic Therapy Regimens

The diet may be initiated as an outpatient or inpatient; an initial fast has become optional because it does not improve ketosis or seizure control at three months and is associated with greater weight loss, hypoglycaemia and acidosis [10]. Before initiation, a consultation is recommended to identify metabolic contraindications (primary carnitine deficiency, carnitine palmitoyltransferase I or II deficiency, translocase deficiency, beta-oxidation defects, pyruvate carboxylase deficiency and porphyria), swallowing difficulties, kidney stones, hypercholesterolaemia, reflux, constipation and cardiomyopathy, and obtain baseline investigations (blood count, liver and renal function, fasting lipid profile, acylcarnitine profile and vitamin D, among others) [10]. In adults, an acylcarnitine profile, urinary organic acids and serum amino acids are requested only when there is clinical suspicion of a metabolic disorder, and pregnancy testing should be considered in women of childbearing potential [7, 11].

The French protocol departs from the general recommendations on the setting for initiation: it recommends admission preferably at any age, and systematically before 1 year of age, with a team trained in the diet, also because admission facilitates family education; the International Ketogenic Diet Study Group recommendations accept initiation as an outpatient or inpatient [10, 12]. The energy distribution presented by the PNDS allows the modalities to be compared with a usual diet:

Source: Figure 1 and Section 4.3.1 of the PNDS, which refers to the 2018 recommendations [10, 12].

The percentages refer to energy. The 20–27% of energy from carbohydrate that the PNDS assigns to the modified Atkins diet exceeds the initiation limit of 10–15 g/day of the general recommendations (previous table), which is expressed in grams and describes the initial phase; the PNDS itself provides for a progressive increase in carbohydrate over time. For the initial prescription, the reference is the gram limit of the 2018 recommendations, and the PNDS composition should be read as an average distribution and not as an initiation regimen [10, 12]. The PNDS maintains, as does the consensus, the rejection of the low glycaemic index treatment, whose ketogenic capacity it considers insufficient to provide an alternative fuel [7, 12].

The protocol brings together in a table the contraindications to the diet in children, which it attributes to the 2018 recommendations. The table does not match the original: the 2018 recommendations regard as absolute only primary carnitine deficiency, CPT I or II and carnitine translocase deficiencies, beta-oxidation defects, pyruvate carboxylase deficiency and porphyria, and as relative the inability to maintain adequate nutrition, an identified surgical focus, non-adherence and concomitant use of propofol, because of the risk of propofol infusion syndrome; glycogen storage diseases, heart disease, organ failure, hyperinsulinism, growth retardation, severe reflux and familial hypercholesterolaemia are additions by the PNDS, and propofol does not appear in its table [10, 12]. It notes that these disorders rarely coexist with Glut1DS, but that a clinical picture suggesting them warrants the opinion of an inherited metabolic disease centre before starting the diet [10, 12].

Source: Table 2 of the PNDS [12]; add concomitant propofol as a relative contraindication [10].

7.3. Ketosis Monitoring and Duration

All patients with Glut1DS should monitor ketosis by beta-hydroxybutyrate (BHB) in capillary or venous blood; urinary ketones measure only acetoacetate, qualitatively, and strongly positive ketonuria may coexist with low blood ketone levels [1, 7]. Targets differ slightly between sources: the international consensus recommends blood BHB of 2–5 mmol/L, GeneReviews 3–5 mmol/L, measured daily, weekly or as needed [1, 7]. The recommendation for adults with epilepsy in general, which prioritises symptom control rather than ketosis itself, should not be transposed to Glut1DS without careful consideration, because in this disease ketosis is the therapeutic substrate itself [7, 11]. Unlike drug-resistant epilepsy in general, in which the diet is often discontinued after two years, in Glut1DS the diet has no end date: the International Ketogenic Diet Study Group recommends maintaining it at least until puberty, possibly with benefit into adulthood, and the disease-specific consensus recommends continuing it for as long as it is tolerated [7, 10]. Saudubray indicates maintenance throughout childhood and adolescence and states that in adults the indication is less clear, but that MAD is increasingly recommended [21]. The French PNDS adopts the consensus target of 2–5 mmol/L and states that, in theory, the diet should never be stopped and should be continued, if possible, into adulthood; it acknowledges that, in practice, this duration is not always achieved, because of the difficulty of maintaining the diet for years, and allows, after childhood and with a specialist team, a switch to the modified Atkins diet [12]. It adds that the type and frequency of ketosis measurements vary between teams [12].

7.4. Observed Efficacy

There is no randomised placebo-controlled trial of KDT in Glut1DS; evidence comes from series, surveys and consensus statements [3]. The following figures describe different populations and cannot be compared directly.

The diet effectively controls seizures and improves gait, but seizures may recur despite good adherence, and cognitive and neurobehavioural manifestations may persist despite seizure control [1, 7]. Its effects on development and movement are less pronounced than those on epilepsy, and cognitive dysfunction persists throughout life without evidence of progressive deterioration [7]. The Atlas records that the diet did not improve cognitive function in a 2005 series, and that all four patients in another series became seizure-free, with kidney stones in one [15]. In a study of 57 patients, the 40 receiving dietary treatment had less gait impairment but more dystonia, chorea, tremor, myoclonus, dyspraxia and paroxysmal episodes than the 17 on a conventional diet, and poor adherence and low ketonuria appeared to trigger episodes; this was a cross-sectional comparison subject to confounding by indication [34]. Saudubray notes that seizures diminish with age and dysarthria and movement disorders, particularly paroxysmal dystonia, become the main problem, often poorly controlled by the diet [21].

In 2025, an expert group proposed an operational definition of ketogenic diet resistance: confirmed therapeutic ketosis (blood BHB ≥2.0–2.5 mmol/L on repeated measurements), verified adherence, sufficient duration (at least 3 months, longer for cognitive or motor goals), and no improvement on standardised measures; EEG may be an adjunctive marker [43]. This proposal derives from a narrative review and expert opinion and has not yet been validated.

7.5. Supplements and Carnitine

All ketogenic diets require multivitamin and calcium supplementation [21]. The International Ketogenic Diet Study Group recommends a multivitamin with minerals (including trace elements, particularly selenium), calcium and vitamin D, preferably in carbohydrate-free preparations; oral citrate has class III evidence for kidney stone prevention, with divided opinions on empirical use [10].

Carnitine is the main area of disagreement. GeneReviews recommends dietary supplementation with 50 mg/kg/day of L-carnitine in all patients on KDT because the diet is low in this cofactor [1]. The international consensus considered routine supplementation controversial: most centres measure carnitine regularly, while others supplement routinely, and monitoring is recommended because patients may develop secondary deficiency [7]. In general paediatric recommendations, 84% of centres supplement only when levels are low or symptoms occur [10]. The US Carnitor® (levocarnitine) label covers primary carnitine deficiency and secondary deficiency due to an inborn error of metabolism; the recommended oral dose for infants and children is 50 to 100 mg/kg/day in divided doses, starting at 50 mg/kg/day, with a maximum of 3 g/day; reported adverse effects are predominantly gastrointestinal (nausea, vomiting and abdominal cramps) [44]. Empirical supplementation in a child without documented deficiency therefore represents an expert recommendation rather than a labelled indication. In the published case of a newborn girl with Glut1DS, carnitine was started at 25 mg/kg because of a borderline level and known maternal deficiency [27].

7.6. Adverse Effects of Ketogenic Therapy and Their Management

The French PNDS separates the effects to be monitored by phase. At initiation or in the short term: acidosis, increased seizure frequency, constipation, excessive ketosis, worsening of gastro-oesophageal reflux, fatigue, hypoglycaemia, food refusal and vomiting. In the long term: renal stones, vitamin D deficiency, secondary carnitine deficiency, reduced bone mineral density, dyslipidaemia, fractures, weight loss or insufficient weight gain and growth disturbances [12]. In adults, the protocol considers data on bone loss and fractures to be scarce, especially in patients taking enzyme-inducing antiseizure medications, and the cardiovascular risk to be low: most studies show normalisation of LDL cholesterol in the long term, no increase in carotid intima–media thickness and no worsening of cardiovascular risk biomarkers [12].

The disease-specific consensus considers that prolonged use in Glut1DS may make adverse effects more apparent than short-term use in drug-resistant epilepsy, and recommends regular monitoring for kidney stones, growth impairment and cardiovascular risk (blood pressure, fasting lipid profile and transcranial Doppler from 10 years of age) [7]. Long-term adverse effects and the long-term prognosis of KDT in Glut1DS remain largely unknown [7].

7.7. Antiseizure Medications and Substances to Avoid

Antiseizure medications are generally ineffective or provide limited benefit without KDT and do not treat the metabolic defect; the consensus recommends no particular medication and expressed concern about possible harm when they are used alone [1, 7]. Nevertheless, in patients with an incomplete dietary response, an adjunctive antiseizure medication was effective in 10 of 17 non-responders in a 2026 series, with no specific medication superior to the others [40]. Consensus centres used levetiracetam (9/13) and valproate (9/13, excluding girls after menarche) as adjuncts, followed by lamotrigine (5/13); some considered ethosuximide, carbamazepine, oxcarbazepine and zonisamide ineffective [7]. Saudubray suggests considering ethosuximide for early-onset absence epilepsy [21].

Phenytoin and carbamazepine did not inhibit transport in vitro, and prolonged phenytoin exposure increased transport by 30–60% in vitro; the Atlas regards these medications as preferable, but the consensus emphasises that the clinical relevance of these findings is uncertain [7, 15].

The French PNDS adds a position that diverges from the avoidance lists. It lists barbiturates, diazepam, sodium valproate, chloral hydrate and ethanol as non-specific GLUT1 inhibitors in vitro, but states that these experimental data have no demonstrated clinical application and do not constitute a contraindication to their use, whereas GeneReviews, the Orphanet entry and Saudubray recommend avoiding at least barbiturates [1, 2, 12, 21]. On carbonic anhydrase inhibitors, the protocol agrees with the other sources: acetazolamide, topiramate and sultiame may worsen the acidosis of a strict diet, and topiramate, acetazolamide and zonisamide increase the risk of urinary stones, particularly in combination with the diet [10, 12]. It does not recommend any particular antiseizure medication and maintains them only while the diet does not fully control seizures [12]. In practice, the more restrictive position (avoiding barbiturates) protects against a plausible risk without depriving the patient of alternatives, but the divergence shows that avoidance rests on in vitro data and parental reports, not on trials.

For paroxysmal exertion-induced dystonias and dyskinesias, the PNDS recommends first adjusting the diet until a balance that prevents episodes is found, including a meal before physical activity; beyond the diet, management is symptomatic, follows the usual practice for movement disorders, takes into account the child's plasticity, growth and educational plan, and no particular medicine is recommended; medicines are maintained only if the diet does not fully control the manifestations [12].

7.8. Alternative, Adjunctive and Investigational Therapies

Triheptanoin. A triglyceride of seven-carbon fatty acids that provides four- and five-carbon ketone bodies and anaplerotic substrates [1, 9]. Following favourable open-label findings, controlled trials were negative (table below) [9, 45].

The limitations of these trials require careful interpretation. Both excluded patients on a ketogenic diet and therefore neither test triheptanoin as an adjunct to KDT nor compare it with KDT. The absence subgroup finding is exploratory, with post hoc EEG reclassification and only one patient with isolated absences in the placebo group. Safflower oil is a calorie- and fat-containing placebo, not metabolically inert [9, 45]. The 2020 consensus recorded that 8 of 13 centres had used triheptanoin for refractory paroxysmal episodes on an intention-to-treat basis, and recent reviews describe its efficacy as inconsistent [7, 46].

Ketone salts and esters. These provide fuel without dietary restriction and achieve plasma concentrations similar to those produced by the diet, but they are expensive, unpalatable and, in the case of salts, cause sodium overload; daily amounts of up to 400 g have been reported [7, 21]. Saudubray's ketogenesis chapter noted in 2022 that there were no reports of ketone esters in Glut1DS, and the consensus made no recommendation on oral ketones or ketoesters [7, 29].

Acetazolamide, levodopa and alpha-lipoic acid. Refractory paroxysmal episodes were treated with acetazolamide by 12 of 13 centres and alpha-lipoic acid by 2, and isolated reports describe benefit from acetazolamide and levodopa; the consensus could not make recommendations on treating paroxysmal dyskinesia [1, 7]. A 2026 systematic review in children (22 studies, 290 patients) found the most consistently favourable results with MAD variants and acetazolamide, based on low-quality evidence [46]. Acetazolamide is also a carbonic anhydrase inhibitor, requiring monitoring for acidosis and kidney stones when combined with the diet [1, 10]. The Atlas cites increased GLUT1 transcription induced by thioctic (alpha-lipoic) acid in patient fibroblasts, a laboratory rather than a clinical finding [15].

Diazoxide. It inhibits insulin secretion and has been proposed to induce mild hyperglycaemia and increase the glucose gradient towards the brain. Experience is anecdotal: in a 14-year-old girl with dietary resistance, titration guided by continuous glucose monitoring (doses of 7.3 to 9.0 mg/kg/day over eight months) was associated with disappearance of electroclinical seizures but complicated by fluid retention treated with hydrochlorothiazide and a 13.7 kg weight gain; earlier attempts in five patients had caused persistent hyperglycaemia of 300–500 mg/dL [1, 47]. This is off-label use without a controlled trial; these doses describe a case and are not a recommendation.

Other strategies. Intravenous sodium lactate has been tested in two patients (NCT04112862), and red cell exchange transfusion, intended to replace GLUT1-deficient erythrocytes with donor erythrocytes, is registered as a suspended mechanistic study (NCT04137692) [37]. Intermittent glucose administration and lactate have been evaluated only in small studies at high risk of bias [46]. Increasing carbohydrate intake in the usual diet has been described as an option if KDT is intolerable [7]. The French PNDS groups among exploratory alternatives, which it does not recommend, decanoic acid (a ten-carbon fatty acid, C10), D,L-3-hydroxybutyrate (an exogenous ketone body) and triheptanoin; all aim to provide fuel to the brain without major dietary restriction, and none is validated for first-line use [12].

Gene-directed therapies. In mice, transfer of murine Glut1 in an adeno-associated virus serotype 9 (AAV9) vector increased RNA and protein, restored CSF glucose when given early, normalised brain size, reduced seizures and corrected motor deficits, with benefit lasting more than 12 months; administration in adult mice did not improve symptoms. In a porcine model, injection into the cisterna magna produced high expression in the cerebral endothelium, but the tropism of AAV9 for human endothelium remains to be determined, and there is no clinical trial of gene therapy [1, 7]. In 2025, transgenic expression of the human genomic GLUT1 locus in mice increased brain GLUT1, CSF glucose and motor performance and reduced seizures, partly through stimulation of murine Slc2a1 by a long non-coding RNA included in the transgene [48]. Saudubray and the consensus also mention small-molecule strategies to increase expression of the normal allele [7, 21]. All these data are preclinical. The French protocol of 2024 records the same situation: benefits of AAV9 gene replacement in mice, without clinical application [12].

7.9. Regulatory Status

Levocarnitine has a US label for primary and secondary carnitine deficiency (Section 7.5), but its empirical use during KDT in Glut1DS is an expert recommendation rather than a labelled indication [1, 44]. No authorisation application, Committee for Medicinal Products for Human Use (CHMP) opinion or PDUFA date for a Glut1DS-specific therapy was identified in the sources consulted. This statement is limited to the sources searched and does not establish the absence of ongoing programmes.

7.10. Developments in the Last 12 Months (October 2025 to October 2026)

The L-fucose trial administers 500 mg/kg up to a maximum of 10 g, three times daily, according to the registered protocol; this regimen is experimental and is not treatment guidance [37]. A prospective natural history registry of epilepsy–dyskinesia syndromes including SLC2A1 is also recruiting (NCT06967727) [37].

7.11. Supportive Care, Rehabilitation and Intercurrent Illness

GeneReviews recommends multidisciplinary care involving neurology familiar with KDT, physical and rehabilitation medicine, physiotherapy, occupational therapy, speech and language therapy, clinical genetics and genetic counselling [1]. Spasticity is managed with physiotherapy and stretching to prevent contractures and falls, together with positioning and mobility devices; hypertonia or dystonia may require specialist input for baclofen, tizanidine, botulinum toxin, antiparkinsonian agents or orthopaedic surgery. Dysarthria warrants speech and language therapy and, when expression is severely limited, augmentative and alternative communication, which does not impede speech development. Oromotor dysfunction should be assessed at every visit, with a swallowing study for choking, poor weight gain or respiratory infections, and nasogastric tube feeding or gastrostomy in severe cases. ADHD features may warrant medication, and interventions used for autism spectrum disorder may help some children [1]. The early intervention and educational plans described in GeneReviews reflect the US system and should be adapted to each country's system [1].

During intercurrent illness, hospital admission and surgery, the main risk is inadvertent loss of ketosis through glucose-containing fluids or carbohydrate-containing medicines, which should be avoided; the family should help monitor these exposures [1]. Saudubray includes Glut1DS among causes of intractable seizures with a specific treatment, the ketogenic diet [51].

The French PNDS distributes roles between the multidisciplinary hospital team, which sets the follow-up schedule, and the general practitioner. The latter is responsible for referring the patient to paediatric or adult neurology to confirm the diagnosis, ensuring that follow-up is provided by a team familiar with the disease and the diet, coordinating medical and allied health care, participating in monitoring for complications and adverse effects in liaison with the reference team, providing the usual care of any child or adult and treating intercurrent illnesses. Among the professionals involved in treatment, the protocol includes the emergency physician for paroxysmal manifestations, a dietitian experienced in the ketogenic diet, nursing staff and social services [12].

The diagnosis is disclosed by the physician who established it, in the presence of the patient and, if a minor, of both parents whenever possible. The PNDS asks that the cause, the mechanisms and the symptoms that may appear with age be explained, that the goals be defined (maximal control of seizures and paroxysmal manifestations with few or no adverse effects and improvement of motor and cognitive disturbances), and that the family be warned that the diet does not always control all manifestations. Social support and questions concerning the immediate circle are addressed at the same appointment. Apart from the demands of the diet, the disease should not impose restrictions on daily life: the child's educational rules remain unchanged, and the advised adaptations are those for any epilepsy [12].

The diet should be explained to the child, the parents and those who care for the child in various settings (family, friends, school, sport), within a full therapeutic education programme, with one or more sessions for parents on the disease, its manifestations and the diet (adverse effects and adherence). For the school or a medico-social institution, a document is drawn up describing the diet, its practical application and what to do in the event of an unexpected deviation; in France, this takes the form of an individualised reception plan (PAI, projet d'accueil individualisé). The family should have one written document for health professionals and another for themselves, and should inform any physician of the ongoing diet, so that treatments containing carbohydrate are avoided. The protocol summarises the management of the most frequent intercurrent situations [12]:

Source: Section 4.4 of the PNDS [12].

For neurodevelopmental and learning disorders, the PNDS regards the diet as the baseline treatment, although its effect on cognitive abilities is difficult to measure, and follows the usual approach to these disorders: a neuropsychological assessment identifies the domains affected and guides support, and in France an application to the MDPH (maison départementale des personnes handicapées) is generally recommended [12].

8. Reproductive Options and Family Counselling

8.1. Risks to Relatives

In the autosomal dominant form, approximately 90% of probands have a de novo variant and approximately 10% inherited it from a parent, who may have mild or no manifestations or somatic mosaicism [1]. The family history may appear negative because affected relatives have not been recognised; a variant can therefore be described as de novo only after testing the parents [1]. If the proband has a molecular diagnosis, the parents should be tested for the variant; otherwise, GeneReviews proposes comparing parental erythrocyte glucose uptake with control values [1].

Discussing de novo epileptic encephalopathies in general, Oxford notes that identifying parental mosaicism requires adjustment of the recurrence risk, and that phenotypic variability complicates prediction when a variant is inherited from a parent with few signs [31]. In Glut1DS, mosaicism attenuates severity, explaining parents with a minimal phenotype and children with the classic form [1].

8.2. Assessment of At-Risk Relatives

Assessment of newborns, infants and other at-risk relatives aims to identify those who would benefit from KDT as early as possible, because early initiation, ideally during the first year, improves seizure control and neurological prognosis [1]. If the familial variant is known, molecular testing is used; otherwise, lumbar puncture may be performed in relatives with suggestive clinical features [1]. An asymptomatic relative with a pathogenic variant and hypoglycorrhachia may be considered presymptomatically affected in the familial context; treatment of asymptomatic carriers and individuals with few symptoms was an area of consensus disagreement, and decisions should be individualised [7].

8.3. Reproductive Options

Once the variant (or variants) has been identified in an affected relative, prenatal diagnosis and preimplantation genetic testing are possible [1]. The Orphanet entry on exercise-induced dyskinesia mentions amniocentesis (usually at 15–18 weeks) and chorionic villus sampling (at 10–12 weeks) [5]. GeneReviews recommends discussing genetic risk and options before pregnancy and offering counselling to affected or at-risk young adults, acknowledging that perspectives on prenatal and preimplantation diagnosis vary among professionals and families [1]. Intrafamilial variability and the effect of early treatment should form part of the discussion, because the same variant may be associated with very different phenotypes [15, 27]. The French PNDS, which estimates that approximately 90% of cases are sporadic, accepts discussion of prenatal diagnosis in familial forms (50% risk) within the framework of the multidisciplinary prenatal diagnosis centres (CPDPN), which should be consulted; testing is carried out in the reference laboratory or in the laboratory that identified the index case, generally by Sanger sequencing or, for large deletions, by quantitative PCR. Like GeneReviews, it considers the preconception period the most appropriate time to estimate risk and discuss prenatal and preimplantation testing, and recommends counselling for affected or at-risk young adults [12].

8.4. Pregnancy in a Woman with Glut1DS

Experience with KDT in pregnancy is limited [1]. The adult recommendations did not reach consensus on avoiding the diet in pregnancy and list it as a relative contraindication; they describe two cases of intrauterine exposure in epilepsy, one with a normal infant at 12 months and the other with bilateral malformations of the auricle of unknown significance [11]. The French PNDS, which cites the same two cases (both outside Glut1DS), specifies that, in the second, the mother was taking lamotrigine with a modified Atkins diet and that the cause of the auricular anomaly (medicine, diet or advanced maternal age) could not be determined; the infant had normal development up to 8 months. The protocol adds that animal studies of a high-fat diet during gestation are not favourable, recommends considering a pregnancy test before starting the diet (included in the systematic initial assessment of women of childbearing age, also to choose contraception compatible with antiseizure medications) and asks that women be informed of the scarcity of data and followed closely in a specialist setting if they become pregnant [12]. The disease-specific consensus identified the use of KDT in pregnant women with Glut1DS as an area of disagreement and recommends considering a pregnancy test before starting the diet in women of childbearing age, given the unknown teratogenic risk [7]. The case described in Section 10.2 shows that maintaining the modified diet during pregnancy was possible, with obstetric and carnitine monitoring [27].

9. Special Considerations: Population Diversity and Epidemiology

9.1. Frequency and What Each Estimate Measures

If the Scottish estimate is correct, GeneReviews projects 150–200 new diagnoses annually in the United States and 5,000–5,500 worldwide [1]. The triheptanoin trial article estimated a point prevalence of approximately 8,500–12,000 patients in the United States and Europe, extrapolated from birth incidence [9]. These projections are calculations based on a single cohort of young children rather than measurements of prevalence. Mild, motor-predominant or late-onset forms are systematically under-represented in studies selecting for early epilepsy, so the true frequency is probably higher [3, 8]. In the Scottish cohort, only one of seven patients underwent diagnostic lumbar puncture, and Glut1DS was suspected before the genetic result in only two, illustrating the current role of molecular diagnosis through epilepsy panels [8].

9.2. Regional Data and Diagnostic Delay

The main series come from the United States (Columbia, New York), Europe (the Netherlands, Germany, Italy, Denmark and the United Kingdom) and Australia [1, 23, 36]. The sources consulted do not describe a founder effect, consistent with the predominance of de novo variants distributed throughout the gene [1, 21]. Diagnostic delay is an indicator of access: a mean of 6.6 years in the 2010 European series (range 1 month to 16 years) [23]; a median of 49 months in the 2026 Argentine series, whose authors attribute underdiagnosis in Latin America to limited awareness and restricted access to genetic testing [41]; and 18 years in Italian adults [20]. In the 2026 Chinese series, median age at diagnosis was 32.5 months, with 98.2% de novo variants [42]. No data from Portugal or Brazil were found in the sources consulted; the Orphanet entry has a Portuguese summary, and one physician in the 2010 European series worked in Portugal, but those sources contain no published national data [2, 23]. This absence does not establish a lack of cases or national studies.

9.3. Special Populations

Young children are the group in whom treatment has the greatest impact, and ClinGen assigned Glut1DS a strong actionability assertion only in the paediatric context (document AC1036), according to GeneGuidance [3]. Adults experience late diagnosis, psychosocial needs and lower tolerance of the classic diet [10, 20]. Considerations for women of childbearing potential and pregnant women are described in Section 8.4. Patients with haemolytic forms should be assessed for anaemia and cold exposure [17, 19].

9.4. European Networks and Patient Organisations

The relevant European Reference Network is ERN EpiCARE, dedicated to rare and complex epilepsies, with 61 teams in 24 European Union countries and Norway; EpiCARE experts reviewed the Glut1DS Orphanet entry [2, 13]. Organisations cited in the sources include the Glut1 Deficiency Foundation (United States), the European Glut1DS Foundation and Matthew's Friends (United Kingdom), the latter two acknowledged by the international consensus for their participation in its review [1, 7]. The G1D patient registry (UT Southwestern Medical Center, Dallas) and epilepsy organisations such as the Epilepsy Foundation and American Epilepsy Society appear among GeneReviews resources [1]. The Orphanet entry lists patient organisations by country [2].

In France, patients are followed within the network of reference centres for rare epilepsies (CRéER), with a coordinating centre at AP-HM (Marseille) and 12 constituent centres (Robert-Debré, sponsor of the PNDS, Necker and Pitié-Salpêtrière, in Paris; Lille, Amiens, Angers, Bordeaux, Nancy, Rennes, Strasbourg, Toulouse and Lyon), and in 17 competence centres, two of them in Réunion. The health filières involved are DéfiScience (rare diseases of brain development and intellectual disability) and G2M (inherited metabolic diseases). The disease falls under ALD no. 17 (inherited metabolic diseases requiring prolonged specialised treatment), which affects reimbursement of some ketogenic products: the application for exemption from the ticket modérateur (patient co-payment) should be systematic and itemise the products used, and the costs of non-reimbursed products should be included in the MDPH file. The French family association is the Association sur le syndrome du déficit en GLUT1, which took part in the review of the protocol; the PNDS also lists Matthew's Friends and GLUT1 UK, the International Neurological Ketogenic Society, the associations Epilepsie France and EFAPPE, and the ketogenic recipes of CRéER Robert-Debré [12]. The reference laboratory for SLC2A1 and the METAglut1 testing laboratories are described in Section 5.

10. Illustrative Clinical Case

10.1. Paroxysmal Eye–Head Movements as the Presenting Sign

This case was published by Pearson and colleagues in Neurology in 2017 within a Columbia University series of 101 patients with confirmed Glut1DS [32]. A girl had her first episode of unusual eye movements at 14 weeks of age: over three minutes, her eyes converged, followed by repetitive eye movements accompanied by head turning. By 15 months, she had experienced another 14 episodes lasting 1 to 10 minutes, with preserved alertness. Her mother described them as follows: “It is as if she is watching something moving quickly past her; she does not turn her head repetitively or in a pattern; she seems to turn her head to try to focus her eyes on something.” Episodes appeared to improve with feeding. Because of these movements, she was investigated for opsoclonus–myoclonus syndrome at 27 months; lumbar puncture revealed low CSF glucose, and Glut1DS was confirmed by erythrocyte glucose uptake assay and SLC2A1 analysis [32].

Analysis of the series' videos showed that, with every change in gaze, the eye movement was followed by head movement in the same direction, with fixation between movements (Figure 1 of the original article, eye–head saccades in frames at 150 ms intervals); frame-by-frame timing in four patients showed irregular intervals between saccades, predominantly 200 to 800 ms, peaking at 400–500 ms (Figure 2 of the original article) [32].

The case illustrates three points. The earliest sign may not be a seizure and is informative only when described precisely or recorded on video; in the series, parents and physicians used imprecise terms (“eye rolling”), and head movement was documented in only half the records despite being present in almost all videos. Confusion with opsoclonus delayed diagnosis by more than two years. The response to feeding provided a pathophysiological clue, as with preprandial ataxia [1, 32]. The study was retrospective, and the 32% frequency may be an underestimate because patients were not systematically asked about these episodes [32].

10.2. A Mother and Daughter: From Pregnancy to Preschool Age

In 2021, Kramer and Smith described a woman diagnosed at 18 years after a long history of non-classic symptoms: migraine and movement abnormalities from 2 years, developmental milestones at the late end of the normal range, a previous diagnosis of developmental coordination disorder, ataxia and dystonia, and no seizures on several EEGs. CSF glucose was 38 mg/dL with blood glucose of 101 mg/dL, and genetic testing identified SLC2A1 c.377G>A, p.(Arg126His). Full-scale IQ was 82. With a modified Atkins diet, episodes of ataxia, dystonia and migraine resolved and school performance improved [27].

She became pregnant at 19, with prenatal care beginning between 25 and 28 weeks. Before pregnancy was confirmed, she already had increased dystonic movements, headaches, anaemia and low ketonuria. She received high-risk obstetric care with fortnightly growth ultrasound scans and cardiotocography, normal apart from two transient pseudosinusoidal traces. During pregnancy, frequent abnormal movement episodes persisted, particularly on days of anxiety and when ketosis was difficult to maintain (ketonuria +1 and BHB 0.12 mmol/L); carnitine was low and normalised only after delivery despite supplementation. On the advice of a consulted expert, she maintained the diet throughout pregnancy, and breast milk or donor milk was recommended for the neonatal period [27].

The girl was born at term, with Apgar scores of 7 and 9, a weight of 2.56 kg and a normal head circumference of 33 cm; initial blood glucose of 37 mg/dL stabilised with donor milk, and neurological examination was normal. At 12 days, lumbar puncture showed CSF glucose of 28 mg/dL with blood glucose of 85 mg/dL (ratio 0.32). She was admitted to neonatal intensive care for diet initiation over a week, reaching a ratio of 2:1 and BHB of 1.31 mmol/L within 48 hours; a low ratio was chosen because the EEG was normal and the maternal phenotype mild. Carnitine was started at 25 mg/kg for a borderline level. Genetic testing confirmed the same maternal variant. Until 6 months, feeding combined infant formula and 4:1 ketogenic formula (overall ratio 2–2.5:1); BHB, below 2 mmol/L until 6 months, subsequently increased to 3–4 mmol/L without a dietary change. Transition to solids was difficult, requiring feeding therapy between 18 months and 4 years, and at 18 months she was referred to nephrology for persistently low bicarbonate and elevated uric acid, treated with sodium citrate. At 5 years, while maintaining an estimated overall ratio of 2:1, development was entirely normal, with no seizures or abnormal movements and appropriate growth [27]. GeneReviews adds that, at 10 years, the child continues to develop normally, with no reported seizures or movement disorder [1].

This report contains no figures. The decisions made (low ratio, carnitine dose and donor milk use) reflect an individual case and are not general recommendations. Its significance lies in demonstrating adult diagnosis through mild symptoms, the possibility of presymptomatic diagnosis and treatment when the familial variant is known, and practical difficulties in maintaining ketosis during pregnancy and infancy [1, 27].

11. Prognosis and Long-Term Follow-Up

11.1. Natural History and Treated Disease

Glut1DS is a chronic condition dominated by epilepsy in childhood and movement disorders thereafter. During a mean follow-up of 14.2 years (8.9 to 23.6) in 13 patients, longitudinal measures remained stable, epilepsy improved throughout childhood, dystonia appeared in childhood or adolescence, and earlier diet initiation correlated with better outcomes on some measures [52]. Seizures become uncommon after adolescence, while adult intellectual disability, speech impairment and neurobehavioural manifestations reflect previous structural injury to the immature brain [1]. Eye–head movements usually disappear by 6 years [32]. Saudubray describes a favourable prognosis when the diet is started early, with continued progress in speech and mobility and stabilisation after puberty; however, dysarthria and movement disorders, particularly paroxysmal dystonia, become the main problem with age, and seizures and varying degrees of impairment may persist, sometimes requiring a supported environment [21]. Paroxysmal episodes usually diminish in frequency and severity in adulthood, and there is no evidence of progressive cognitive deterioration [7]. Deceleration of head growth in early childhood is considered a sign of irreversible cerebral dysfunction [7].

These data predominantly describe treated patients with variable treatment initiation; comparable untreated natural history data are lacking, and experience with the earliest diagnosed patients who have reached adulthood remains limited [7, 21]. The absence of cognitive deterioration in small series does not imply zero risk of decline, and the long-term effects of decades of dietary treatment remain to be characterised [7].

11.2. Surveillance Plan

The French schedule departs from that of the 2018 recommendations on two points. After the second year, it proposes annual visits, laboratory tests and renal ultrasound, whereas the International Ketogenic Diet Study Group provides for six-monthly visits and laboratory tests every 6–12 months and does not consider renal ultrasound routine in all centres; and it places bone assessment from 12 months of diet onwards, on an optional basis, whereas the paediatric recommendations propose it after 2 years and the adult recommendations every 5 years [10–12]. The PNDS also measures response to treatment by the impact of symptoms and not only by their frequency: one monthly seizure that takes the patient to the emergency department weighs more than daily seizures lasting seconds without risk of accident. At each visit, all paroxysmal symptoms (seizures, paroxysmal movement disorders, fatigability, migraine with aura) and permanent symptoms (motor, cognitive and learning) are reviewed, and the family is informed that new symptoms may appear with age [12].

11.3. Transition to Adulthood

Transition from paediatric to adult care, including the diet, is one of the challenges identified by the consensus, which recommends a formal plan [7]. GeneReviews refers to the American Epilepsy Society tools and the Child Neurology Foundation principles [1]. The Milan adult series proposes a multidisciplinary model with paediatric and adult neurologists, rehabilitation, clinical psychology, nutrition and dietetics, to support motor, cognitive and emotional function and adherence to the diet [20]. The French PNDS likewise considers essential a transition plan covering the diet and medical care during the move from the paediatrician to the adult specialist, and points out that monitoring must keep pace with the changing manifestations with age, from seizures in early childhood to dystonia and paroxysmal exercise-induced dyskinesia in adolescence and adulthood, with cognitive dysfunction that may persist throughout life and late effects of the diet (growth retardation, nephrolithiasis, cardiovascular risk) to be monitored [12].

12. Summary and Final Considerations

GLUT1 deficiency syndrome is an encephalopathy caused by inadequate fuel availability, predominantly due to SLC2A1 haploinsufficiency, which reduces glucose transport across the blood–brain barrier and into astrocytes [1, 7]. Since De Vivo's description in 1991, the spectrum has expanded from infantile epileptic encephalopathy with acquired microcephaly to exercise-induced paroxysmal forms, early-onset absence epilepsy, choreoathetosis with spasticity and forms with haemolysis, without any of these presentations being exclusive [1, 14]. Their common feature is worsening symptoms when glucose delivery to the brain fails to meet demand: fasting, exertion, illness and stress [1].

Diagnosis combines clinical assessment, lumbar puncture under controlled conditions and molecular testing, each with its own limitations: CSF glucose thresholds vary between sources, approximately 10% of patients have no identifiable variant, and non-coding or synonymous variants may escape routine methods [1, 7, 26]. Ketogenic therapy is the targeted treatment, started early and maintained throughout life while tolerated, controlling seizures in most patients and providing smaller, more variable benefits for movement and cognition; approximately four in ten patients in a recent series did not become seizure-free with the diet alone [7, 40]. Triheptanoin failed in two controlled trials, diazoxide and acetazolamide are supported by anecdotal experience, and gene-directed therapies remain preclinical [9, 45, 47, 48].

Sources diverge on practical points that clinicians should know: routine carnitine supplementation, use of valproate during the diet, the blood ketone target and the duration of fasting before lumbar puncture [1, 7, 10]. The French protocol of 2024 adds its own divergences: it prefers initiation of the diet as an inpatient, considers that in vitro GLUT1 inhibitors (barbiturates, valproate, diazepam) are not contraindicated and proposes annual monitoring after the second year [12]. Current priorities are early diagnosis, including assessment of relatives and, in the future, newborn screening, management of the transition to adulthood and research into therapies that restore transporter function [7, 20].

13. References

1. Wang D, Sands T, Tang M, Monani U, De Vivo D. Glucose Transporter Type 1 Deficiency Syndrome. In: Adam MP, Bick S, Mirzaa GM, Wallace SE, Amemiya A, editors. GeneReviews® [Internet]. Seattle (WA): University of Washington, Seattle; 1993–2026. First published 30 July 2002; comprehensive update 6 March 2025. NBK1430.
2. Orphanet. Classic glucose transporter type 1 deficiency syndrome. ORPHA:71277 [Internet]. Last updated September 2023. Disease entry. Accessed 6 October 2026.
3. GeneGuidance. SLC2A1 gene: identifiers, clinical curation, frequency, ClinGen actionability and management library [internal database]. Extracted 26 September 2026.
4. VariantValidator. Normalisation of SLC2A1 variants on transcripts NM_006516.2 and NM_006516.4 (MANE Select), GRCh38 VariantValidator. Accessed 6 October 2026.
5. Orphanet. Paroxysmal exertion-induced dyskinesia. ORPHA:98811 [Internet]. Last updated November 2013. Disease entry. Accessed 6 October 2026.
6. Orphanet. Epilepsy with myoclonic absences. ORPHA:86911 [Internet]. Last updated December 2021. Disease entry. Accessed 6 October 2026.
7. Klepper J, Akman C, Armeno M, Auvin S, Cervenka M, Cross HJ, et al. Glut1 Deficiency Syndrome (Glut1DS): state of the art in 2020 and recommendations of the international Glut1DS study group. Epilepsia Open. 2020;5(3):354–65. doi:10.1002/epi4.12414. PMID:32913944.
8. Symonds JD, Zuberi SM, Stewart K, McLellan A, O'Regan M, MacLeod S, et al. Incidence and phenotypes of childhood-onset genetic epilepsies: a prospective population-based national cohort. Brain. 2019;142(8):2303–18. doi:10.1093/brain/awz195. PMID:31302675.
9. Striano P, Auvin S, Collins A, Horvath R, Scheffer IE, Tzadok M, et al. A randomized, double-blind trial of triheptanoin for drug-resistant epilepsy in glucose transporter 1 deficiency syndrome. Epilepsia. 2022;63(7):1748–60. doi:10.1111/epi.17263. PMID:35441706.
10. Kossoff EH, Zupec-Kania BA, Auvin S, Ballaban-Gil KR, Christina Bergqvist AG, Blackford R, et al. Optimal clinical management of children receiving dietary therapies for epilepsy: updated recommendations of the International Ketogenic Diet Study Group. Epilepsia Open. 2018;3(2):175–92. doi:10.1002/epi4.12225. PMID:29881797.
11. Cervenka MC, Wood S, Bagary M, Balabanov A, Bercovici E, Brown MG, et al. International recommendations for the management of adults treated with ketogenic diet therapies. Neurol Clin Pract. 2021;11(5):385–97. doi:10.1212/CPJ.0000000000001007. PMID:34840865.
12. Haute Autorité de Santé. Protocole National de Diagnostic et de Soins (PNDS): Déficit en transporteur de glucose GLUT1. Centre de référence des épilepsies rares de l'hôpital Robert-Debré; filières DéfiScience et G2M. Saint-Denis La Plaine: HAS; 2024 Nov 7 [cited 2026 Nov 6]. Available from: https://www.has-sante.fr/jcms/p_3555044
13. EpiCARE — European Reference Network for rare and complex epilepsies [Internet]. Official website. Accessed 6 October 2026.
14. De Vivo DC, Trifiletti RR, Jacobson RI, Ronen GM, Behmand RA, Harik SI. Defective glucose transport across the blood-brain barrier as a cause of persistent hypoglycorrhachia, seizures, and developmental delay. N Engl J Med. 1991;325(10):703–9. doi:10.1056/NEJM199109053251006. PMID:1714544.
15. Nyhan WL, Hoffmann GF, Al-Aqeel AI, Barshop BA. Atlas of Inherited Metabolic Diseases. 4th ed. Boca Raton: CRC Press; 2020. Chapter 104, GLUT1 deficiency, p. 793–795. Local chapter.
16. Suls A, Dedeken P, Goffin K, Van Esch H, Dupont P, Cassiman D, et al. Paroxysmal exercise-induced dyskinesia and epilepsy is due to mutations in SLC2A1, encoding the glucose transporter GLUT1. Brain. 2008;131(Pt 7):1831–44. doi:10.1093/brain/awn113. PMID:18577546.
17. Weber YG, Storch A, Wuttke TV, Brockmann K, Kempfle J, Maljevic S, et al. GLUT1 mutations are a cause of paroxysmal exertion-induced dyskinesias and induce hemolytic anemia by a cation leak. J Clin Invest. 2008;118(6):2157–68. doi:10.1172/JCI34438. PMID:18451999.
18. Weber YG, Kamm C, Suls A, Kempfle J, Kotschet K, Schüle R, et al. Paroxysmal choreoathetosis/spasticity (DYT9) is caused by a GLUT1 defect. Neurology. 2011;77(10):959–64. doi:10.1212/WNL.0b013e31822e0479. PMID:21832227.
19. Flatt JF, Guizouarn H, Burton NM, Borgese F, Tomlinson RJ, Forsyth RJ, et al. Stomatin-deficient cryohydrocytosis results from mutations in SLC2A1: a novel form of GLUT1 deficiency syndrome. Blood. 2011;118(19):5267–77. doi:10.1182/blood-2010-12-326645. PMID:21791420.
20. Previtali R, Adami L, Benvenuto C, Gianola L, Segarizzi C, Benzoni C, et al. GLUT1 deficiency syndrome in adulthood: lost in diagnosis. BMJ Neurol Open. 2026;8(1):e001337. doi:10.1136/bmjno-2025-001337. PMID:41561681.
21. Santer R, Klepper J. Disorders of glucose and monocarboxylate transporters. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 8, Sections 8.1 (introduction), 8.3.1–8.3.5, 8.4.2 and 8.6. doi:10.1007/978-3-662-63123-2_8.
22. Yang H, Wang D, Engelstad K, Bagay L, Wei Y, Rotstein M, et al. Glut1 deficiency syndrome and erythrocyte glucose uptake assay. Ann Neurol. 2011;70(6):996–1005. doi:10.1002/ana.22640. PMID:22190371.
23. Leen WG, Klepper J, Verbeek MM, Leferink M, Hofste T, van Engelen BG, et al. Glucose transporter-1 deficiency syndrome: the expanding clinical and genetic spectrum of a treatable disorder. Brain. 2010;133(Pt 3):655–70. doi:10.1093/brain/awp336. PMID:20129935.
24. Bercini Z, Gessani A, Rossi J, Bassi MC, Fioravanti V, Di Rauso G, et al. Movement disorders in GLUT1 deficiency syndrome: a systematic review of the literature. J Neurol. 2026;273(8):446. doi:10.1007/s00415-026-13976-x. PMID:42406126.
25. Wang D, Pascual JM, Yang H, Engelstad K, Jhung S, Sun RP, et al. Glut-1 deficiency syndrome: clinical, genetic, and therapeutic aspects. Ann Neurol. 2005;57(1):111–8. doi:10.1002/ana.20331. PMID:15622525.
26. Higgins AT, Wang F, Pickrell WO, Johnston AJ, Natarajan J, Hamandi K, et al. A synonymous SLC2A1 variant causes familial epilepsy and paroxysmal exercise-induced dyskinesia by creating aberrant mosaic splicing patterns. Neurobiol Dis. 2026;225:107433. doi:10.1016/j.nbd.2026.107433. PMID:42086118.
27. Kramer J, Smith L. Ketogenic diet in Glut 1 deficiency through the life cycle: pregnancy to neonate to preschooler. Child Neurol Open. 2021;8:2329048X211034655. doi:10.1177/2329048X211034655. PMID:34541019.
28. National Center for Biotechnology Information. ClinVar: SLC2A1 variants on transcript NM_006516.4 — VCV000016111 (c.377G>A), VCV000016109 (c.377G>T), VCV000016108 (c.766_767delinsGT), VCV000207229 (c.884C>T), VCV000198842 (c.997C>T) and VCV000207202 (c.972G>A). Accessed 6 October 2026.
29. Morris AAM. Disorders of ketogenesis and ketolysis. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 13, Sections 13.5 and 13.6. doi:10.1007/978-3-662-63123-2_13.
30. Larsen J, Johannesen KM, Ek J, Tang S, Marini C, Blichfeldt S, et al. The role of SLC2A1 mutations in myoclonic astatic epilepsy and absence epilepsy, and the estimated frequency of GLUT1 deficiency syndrome. Epilepsia. 2015;56(12):e203–8. doi:10.1111/epi.13222. PMID:26537434.
31. Firth HV, Hurst JA. Oxford Desk Reference: Clinical Genetics and Genomics. 2nd ed. Oxford: Oxford University Press; 2017. Dystonia (Table 2.9), p. 140; Seizures with developmental delay, p. 308; Epilepsy in infants and children, p. 412–414.
32. Pearson TS, Pons R, Engelstad K, Kane SA, Goldberg ME, De Vivo DC. Paroxysmal eye-head movements in Glut1 deficiency syndrome. Neurology. 2017;88(17):1666–73. doi:10.1212/WNL.0000000000003867. PMID:28341645. PMCID:PMC5405761.
33. 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 1.3.2.1, 1.3.4, 1.4.2, 1.4.6.4, 1.5 (Tables 1.17, 1.19, 1.29), 1.5.1.1, 1.5.1.2, 1.5.2.2, 1.5.2.3, 1.5.2.6, 1.6.2 and 1.6.5. doi:10.1007/978-3-662-63123-2_1.
34. Pons R, Collins A, Rotstein M, Engelstad K, De Vivo DC. The spectrum of movement disorders in Glut-1 deficiency. Mov Disord. 2010;25(3):275–81. doi:10.1002/mds.22808. PMID:20063428.
35. Mochel F, Sedel F. Inborn errors of metabolism in adults: a diagnostic approach to neurological and psychiatric presentations. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 2, Sections 2.1, 2.2, 2.2.5, 2.3.3, 2.3.6 and 2.3.9. doi:10.1007/978-3-662-63123-2_2.
36. Coman DJ, Sinclair KG, Burke CJ, Appleton DB, Pelekanos JT, O'Neil CM, et al. Seizures, ataxia, developmental delay and the general paediatrician: glucose transporter 1 deficiency syndrome. J Paediatr Child Health. 2006;42(5):263–7. doi:10.1111/j.1440-1754.2006.00852.x. PMID:16712556.
37. ClinicalTrials.gov. Records NCT01993186, NCT02960217, NCT03181399, NCT03722212, NCT04112862, NCT04137692, NCT06967727 and NCT07432490 [Internet]. Bethesda (MD): National Library of Medicine. Search for “GLUT1 deficiency”. Accessed 6 October 2026.
38. Mochel F, Gras D, Luton MP, Nizou M, Giovannini D, Delattre C, et al. Prospective multicenter validation of a simple blood test for the diagnosis of Glut1 deficiency syndrome. Neurology. 2023;100(23):e2360-e2373. doi:10.1212/WNL.0000000000207296. PMID:37076312; PMCID:PMC10256121.
39. García-Cazorla Á, Artuch R, Pearl PL. Disorders of neurotransmission. In: Saudubray JM, Baumgartner MR, García-Cazorla Á, Walter J, editors. Inborn Metabolic Diseases: Diagnosis and Treatment. 7th ed. Berlin: Springer; 2022. Chapter 30, Section 30.1.5. doi:10.1007/978-3-662-63123-2_30.
40. Klepper J, Runkel E, Kiesel L. "Not always the magic bullet" — insufficient seizure control by ketogenic dietary therapies in Glut1 Deficiency Syndrome. Epilepsia Open. 2026;11(3):847–58. doi:10.1002/epi4.70243. PMID:41830506.
41. Armeno ML, Massaro M, Boccoli J, Caballero E, Chacon S, Diz M, et al. Glucose transporter type 1 deficiency syndrome: phenotypes, molecular findings, and ketogenic therapy implementation in Argentina. Epileptic Disord. 2026;28(3):734–43. doi:10.1002/epd2.70187. PMID:41626780.
42. Wang M, Liu L, Shang L, Huang D, Li Q, Guo H, et al. Efficacy and safety evaluation of ketogenic diet therapy in patients with glucose transporter type 1 deficiency syndrome. Front Nutr. 2026;13:1920521. doi:10.3389/fnut.2026.1920521. PMID:42824025.
43. Falsaperla R, Sortino V, Kluger GJ, Herberhold T, Rüegger A, Striano P, et al. Toward a working definition of ketogenic diet resistance in GLUT1 deficiency syndrome. Epileptic Disord. 2025;27(6):1247–54. doi:10.1002/epd2.70107. PMID:41026139.
44. U.S. Food and Drug Administration. CARNITOR (levocarnitine) tablets and oral solution — prescribing information (NDA 018948 and NDA 019257). openFDA, label dated 30 October 2024. Accessed 6 October 2026.
45. De Giorgis V, Bhatia KP, Boespflug-Tanguy O, Gras D, Marina AD, Desurkar A, et al. Triheptanoin did not show benefit versus placebo for the treatment of paroxysmal movement disorders in Glut1 deficiency syndrome: results of a randomized phase 3 study. Mov Disord. 2024;39(8):1386–96. doi:10.1002/mds.29822. PMID:38725190.
46. Falsaperla R, Ben Amer S, Sortino V. Beyond the classic ketogenic diet: a systematic review of alternative and adjunctive therapies for GLUT1 deficiency syndrome. Epilepsy Res. 2026;228:107926. doi:10.1016/j.eplepsyres.2026.107926. PMID:42800429.
47. Logel SN, Connor EL, Hsu DA, Fenske RJ, Paloian NJ, De Vivo DC. Exploring diazoxide and continuous glucose monitoring as treatment for Glut1 deficiency syndrome. Ann Clin Transl Neurol. 2021;8(11):2205–9. doi:10.1002/acn3.51462. PMID:34612610.
48. Tang M, Teng S, Kim AY, Peng Y, Monani UR. Transgenic expression of the human Glucose Transporter1 (GLUT1) gene locus reduces disease burden in Glut1 deficiency syndrome model mice. Neurobiol Dis. 2025;214:107047. doi:10.1016/j.nbd.2025.107047. PMID:40754252.
49. U.S. Food and Drug Administration. DOJOLVI (triheptanoin) oral liquid — prescribing information (NDA 213687), version dated 10 June 2025. openFDA. Accessed 6 October 2026.
50. European Medicines Agency. Medicines — list of human medicines (JSON report) [Internet]; and European Commission. Union Register of active orphan designations [Internet]. EMA list; Community register. Accessed 6 October 2026.
51. 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 4.2.6. doi:10.1007/978-3-662-63123-2_4.
52. Alter AS, Engelstad K, Hinton VJ, Montes J, Pearson TS, Akman CI, et al. Long-term clinical course of Glut1 deficiency syndrome. J Child Neurol. 2015;30(2):160–9. doi:10.1177/0883073814531822. PMID:24789115.