Glucose transporter type 1 deficiency syndrome (GLUT1-DS) is caused by impaired glucose transport through the blood–brain barrier and inherited in an autosomal dominant trait, which results from mutations of SLC2A1 gene encoding GLUT1 (Brockmann et al., 2001; Klepper et al., 2001). Autosomal recessive transmission has also been described in rare cases, whereas homozygosity showed embryonic lethality in a mouse model (Wang et al., 2006; Klepper et al., 2009; Rotstein et al., 2010). The classic phenotype is characterized by refractory infantile epilepsy, developmental delay, acquired microcephaly, and complex movement disorders, such as ataxia, dystonia, and spasticity, which was first described by De Vivo et al. in 1991 (De Vivo et al., 1991). Over the past few decades, its clinical spectrum has broadened to include non-classic phenotypes, such as paroxysmal exercise-induced dyskinesia and epilepsy, paroxysmal choreoathetosis with spasticity, atypical childhood absence epilepsy, and myoclonic astatic epilepsy (Wang et al., 2018). Additionally, paroxysmal non-epileptic events, including intermittent ataxia, paroxysmal eyeball movements, dysarthria, migraine, alternating hemiplegia, spastic paraparesis, and periodic confusion (De Giorgis and Veggiotti, 2013; Pearson et al., 2013; Nicita et al., 2019), and some extraneurologic features, like hemolytic anemia and cataracts, have also been described (Weber et al., 2008; Flatt et al., 2011; Bawazir et al., 2012). Despite its complex and variable clinical manifestations, GLUT1-DS can be diagnosed based on the presence of hypoglycorrhachia and pathogenic variants of SLC2A1 gene. Ketogenic diet (KD) is recognized as the most effective treatment, which can provide ketone bodies to serve as an alternative fuel for the brain, and the earlier it is used, the better the prognosis will be (Kass et al., 2016).

In 2008, Bao et al. diagnosed a Chinese GLUT1-DS patient for the first time in mainland China and then reported three Chinese GLUT1-DS ones in 2012 (Liu et al., 2012). However, most (90%) detected SCL2A1 gene mutations are sporadic and de novo, whereas familial cases are rare (Pearson et al., 2013). Here we reported a Chinese family with three members affected with genetically confirmed GLUT1-DS and evaluated the efficacy of KD. Meanwhile, we analyzed and summarized the clinical and genetic characteristics of GLUT1-DS in the newly identified and previously reported Chinese GLUT1-DS cases.

Among the 30 patients, there were 18 males and 12 females. The median age of onset was 8.5 months (range, 33 days to 10 years). Epileptic seizures were reported in 25 patients (83.3%), presenting as the first symptom in 23 patients (76.7%) during infancy. The seizures were of mixed types, including 11 cases of generalized tonic–clonic (GTC) (36.7%), nine cases of focal (30%), four cases of tonic (13.3%), three cases of absence (10%), two cases of atonic (6.7%), one case of spasm (3.3%), one case of myoclonic–atonic (3.3%), and one case of febrile seizures (3.3%). Most seizures were resistant to antiseizure medications (ASMs), such as phenobarbital, valproic acid, lamotrigine, levetiracetam, and oxcarbazepine. Movement disorders were reported in 20 patients (66.7%), including ataxia in 15 cases (50.0%), dystonia in 14 cases (46.7%), gait disturbance in eight cases (26.7%), dysarthria in seven cases (23.3%), paroxysmal hypokinesia in three cases (10.0%), tremor in two cases (6.7%), paroxysmal limb paralysis in two cases (6.7%), and paroxysmal eye movement disorder in two cases (6.7%). Developmental delay was observed in 25 patients (83.3%), and microcephaly was found only in six patients (20%). The clinical characteristics of 30 Chinese patients with GLUT1-DS are summarized in Table 1.

A fasting lumber puncture was performed in 24 patients, but only 17 patients reported definite values of CSF analysis, with low CSF glucose at 1.1–2.6 mmol/L (median, 1.63 mmol/L) and low CSF/blood glucose ratio of 0.215–0.484 (median, 0.340).

EEG was performed in 28 of 30 patients, showing generalized or focal epileptiform discharges or background slowing in 19 cases. No abnormalities were found in the remaining nine cases. Brain MRI was performed in 25 patients. Only nine cases exhibited abnormalities, such as mild brain atrophy, bilateral ventricle enlargement, hypomyelination, abnormal signals in the bilateral frontal cortex, widening of the bilateral frontotemporal sulci, and cerebral dysplasia.

Genetic testing was performed in 28 patients. Except for one patient who was negative for SLC2A1 gene mutation, 27 patients carried pathogenic mutations in the SLC2A1 gene, including 10 with missense, nine with frameshift, three with nonsense, three with large fragment deletion, and two with splice-site mutations (Table 2).

Twenty-three patients were treated with KD, 21 of whom became free from seizures and movement disorders. Two cases gave up KD treatment for intolerance, and their symptoms were not improved. Besides this, four patients were given a regular diet with increased frequency or advised to eat candy, three of whom had a significant improvement in their outcomes. The other three patients refused any treatment for their mild symptoms or were still counseled for KD treatment.

Glucose, the most essential energy source for brain metabolism, enters into the brain facilitated by GLUT1. The cerebral metabolic rate for glucose was low at birth and then increased rapidly to reach adult level by 2 years old. At about 3 to 4 years old, it reaches a peak that exceeds adult level by over two folds, keeps up until 9 years of age, and then gradually decreases to adult rate again by the age of 20 years (Chugani et al., 1987; Chugani, 1998). Consequently, the seizures generally occur at 1–6 months after birth as the most common initial symptom of GLUT1-DS and gradually reduce in adolescence, whereas these rarely appear during the neonatal period, but cyanotic or apnea episodes and paroxysmal eyeball movements—named aberrant gaze saccades—may occur before seizures in infancy (Pearson et al., 2017), which are easily ignored or misdiagnosed as focal seizures. Among the 30 patients in our study, the earliest onset age was 33 days, and the median age was 8.5 months. About 76.7% of the patients presented epileptic seizures as the first symptom.

Approximately 90% of GLUT1-DS patients have seizures of various types, including GTC, absence, focal, myoclonic, atonic, tonic, and spasm seizures, in which GTC and absence seizures are the most common (Pong et al., 2012). Although seizures are often resistant to ASMs, some of which even inhibit the GLUT1 function in vitro, such as phenobarbital, valproate, chloral hydrate, and diazepam (Klepper et al., 1999; Klepper et al., 2003; Wong et al., 2005), there are still about 8% of patients who respond positively to them (Pong et al., 2012). In this study, 83.3% of the 30 patients had epileptic seizures, and the seizure types in some patients evolved with age, with GTC (36.7%) and focal (30%) seizures mostly observed. Some patients achieved partial relief of seizures and motor disorders by taking ASMs, such as oxcarbazepine and levetiracetam. However, the epileptic seizures in such GLUT1-DS patients are not closely related to the epileptiform discharges on EEG. Our study showed that one patient without seizures was found to be with epileptiform discharges, while seven cases with seizures had normal EEG findings.

Compared with epileptic seizures, movement disorders are more common in older children and tend to replace seizures over the years (Leen et al., 2014). In this study, 66.7% of the 30 patients had movement disorders, with ataxia (50.0%), dystonia (46.7%), gait disturbance (26.7%), and dysarthria (23.3%) being more commonly observed.

Since infancy, the developing brain of GLUT1-DS patients is unable to obtain enough energy for its growth, leading to cerebral dysfunction and limited head growth. The majority of patients with GLUT1-DS have experienced some degree of cognitive impairment, ranging from mild learning disability to severe intellectual impairment. In this study, developmental delay was observed in 83.3% of the 30 patients, while microcephaly was not a key sign, only observed in 20% of patients. Furthermore, one patient with microcephaly had no developmental delay, indicating that microcephaly was not always accompanied by developmental delay.

The laboratory hallmark for GLUT1-DS is low glucose of the CSF (hypoglycorrhachia) when normoglycemic. Generally, fasting CSF glucose values are less than 3.3 mmol/L (60 mg/dl). In the majority (>90%) of cases, the values are less than 2.2 mmol/L (40 mg/dl). Another less reliable biomarker is the glucose ratio of CSF to blood below 0.45 (De Vivo and Wang, 2008; Yang et al., 2011; Klepper, 2012; Leen et al., 2013), but normal CSF glucose has also been reported in some patients with milder phenotypes. In addition, the CSF lactate level from normal to low (<1.4 mmol/L) can help differentiate GLUT1-DS from other causes of hypoglycorrhachia, such as meningitis, encephalitis, subarachnoidal haemorrhage, and some metabolic disorders, in which the CSF lactate level is elevated (>2.0 mmol/L) (Chow, 2005). Leen et al. (2013) suggested that it was unlikely to diagnose GLUT1-DS if the CSF lactate was increased. The CSF glucose and CSF/blood glucose ratio were low in this study, ranging 1.1–2.6 mmol/L (median 1.63 mmol/L) and 0.215–0.484 (median 0.340), respectively, which was consistent with that described above.

To date, SLC2A1 is recognized as the only gene associated with GLUT1-DS mapped to the short arm of chromosome 1 (1p34.2), with a length of about 35 kb and containing 10 exons (Shows et al., 1987; Fukumoto et al., 1988). Its coded protein GLUT1 is a transmembrane glycoprotein and composed of 492 amino acids. It has 12 transmembrane domains that span the plasma membrane in an α-helical structure, forming a pore for the transport of glucose and other substrates, and is mainly distributed on brain endothelial cells, astrocytes, and erythrocytes (Mueckler et al., 1985). Since the first elucidation of its genetic–pathogenic roles (Seidner et al., 1998), more than 200 SLC2A1 gene mutations have been described, including missense, nonsense, frameshift, splice-site, and large fragment deletion mutations. Several mutation hotspots within the SLC2A1 gene have been identified, such as Asn34, Gly91, Ser113, Arg126, Arg153, Arg264, Thr295, and Arg333 (Leen et al., 2010). However, about 5–10% of GLUT1-DS cases were reported with absence of mutations in the SLC2A1 gene (Yang et al., 2011; Klepper, 2013). In our study, there was one case with a negative SLC2A1 gene mutation as well. Therefore, some studies have suggested that GLUT1-DS may have undetectable variants in the non-coding regions or downstream of the DNA that could affect posttranscriptional modifications, such as mRNA splicing, protein assembly, folding, transportation, intracellular storage, and activation (Klepper and Leiendecker, 2007).

Furthermore, the phenotypes and symptom severity of GLUT1 patients with the same mutations were heterogeneous even among affected family members sharing the same mutation, indicating that secondary genes (such as PURA gene) and other modifying proteins might be involved in glucose transport or that DNA variant regulatory elements of the wild-type allele may modulate the expression level of GLUT1 (Leen et al., 2010; Mayorga et al., 2018). Mayorga et al. (2018) reported a patient with a PURA gene mutation that had marked hypoglycorrhachia, overlapping the clinical findings with GLUT1-DS, though no variants were found on the SLC2A1 gene. By western blot assays, they confirmed a significantly reduced expression of GLUT1 in the peripheral blood cells of the patient compared to controls. Based on the known functions of PURA as a transcriptional and translational regulator, they proposed GLUT1 as a new PURA target, and mutations in PURA can decrease GLUT1 expression.

Though familial cases were usually correlated with milder manifestations compared to sporadic de novo cases (De Giorgis et al., 2015), attention should be paid to prenatal genetic counseling in familial cases because of phenotypic heterogeneity—even if a patient has very few and mild symptoms, and it is possible that offspring with the same mutation will be severely affected. Herein the familial cases in this study, who carried a nonsense mutation c.73C > T (p.Gln25*) and inherited in an autosomal dominant manner, presented mild to moderate clinical phenotypes with overlapping but variable symptoms. The clinical manifestations of the father of the proband were mild, with only episodic hypokinesia but no intelligence or language disorder, which might be related to the late onset of the disease, while the phenotypes of the proband and her brother were moderate, including three main symptoms, and even evolved with age.

KD is the gold standard treatment for GLUT1-DS, which should be started as early as possible and maintained at least until adolescence to meet the increased energy demand of the developing brain (Klepper, 2012; Kass et al., 2016). In our study, most patients were free of seizures and movement disorders after initiation of KD, and intelligence development and the head circumference of some patients recovered to normal levels. The other three patients treated with a regular diet of increased frequency also showed an improvement that might be classified as “carbohydrate-responsive” phenotype.

In conclusion, the diagnosis of GLUT1-DS is challenging due to its clinical heterogeneity. When children presented with unexplained epileptic seizures, developmental delay, and complicated movement disorder, GLUT1-DS should be suspected and a fasting lumbar puncture or genetic testing is recommended. Once the diagnosis of GLUT1-DS is confirmed, KD should be administered earlier and timely in order to gain a better prognosis.