Brain Glucose Supply and the Syndrome of Infantile Neuroglycopenia

ORIGINAL CONTRIBUTION Brain Glucose Supply and the Syndrome of Infantile Neuroglycopenia

Juan M. Pascual, MD, PhD; Dong Wang, MD; Veronica Hinton, PhD; Kristin Engelstad, BS; Chitra M. Saxena, MPH; Ronald L. Van Heertum, MD; Darryl C. De Vivo, MD

Objective: To describe neuroglycopenia as a specific syn- abnormalities and reduced thalamocortical glucose up- drome caused by insufficient glucose availability during take despite subsequent supply of energetic substrate. brain development. Conclusions: When neuroglycopenia—the lack of ad- Design: Neurologic examinations, neuropsychologic equate glucose supply to the nervous system—occurs in tests, biochemical methods, and functional imaging. the developing brain, thalamic and cortical metabolism mature aberrantly, causing epilepsy associated with other Participants: Patients afflicted by genetic mutation of characteristic neurologic and behavioral disturbances, a the cerebral glucose transporter type 1 and a patient af- pattern also reflected in functional images, as if there were flicted by persistent infantile hypoglycemia (hyperinsu- a temporal window during which glucose were crucial linism) matched to her healthy twin. for brain development. When maturation is complete, glucose merely serves as a fuel, and then, when deficient, it Results: The hallmark of the phenotype is the combina- only causes unrelated disturbances. tion of infantile epilepsy and cerebellar and pyramidal tract dysfunction, together with permanent neuropsychologic Arch Neurol. 2007;64:507-513

IVING ORGANISMS HARNESS fects of both substrate-mediated modula- and exchange energy via the tion and of mutation-induced intrage- formation and breakdown of nomic changes on the development of the bonds found in certain com- nervous system are unknown in most neu- pounds. Not all substances rometabolic diseases, partly because they withL the potential to restore high-energy need not act exclusively and partly be- bonds are efficient fuels, because only a cause of oversimplified gene/protein func- few are recognized by cells. Thus, energy tion paradigms that do not describe metabolism accurately. CME course available at Brain energy serves 2 purposes: devel- www.archneurol.com opment and activity maintenance. After birth, brain metabolism relies predomi- metabolism is inextricably dependent on nantly on glycolysis, which is stimulated Author Affiliations: Colleen at an accelerated rate. During childhood, Giblin Research Laboratories, molecular recognition. Most energetic Neurological Institute of New compounds function as molecular sig- the cerebral uptake of glucose increases, exceeding that of the newborn by 3-fold York (Drs Pascual, Wang, and nals in addition to fuels, because they regu- and that of the adult by 2-fold.2,3 In the De Vivo, and Ms Engelstad), late their own metabolism, acting di- adult, the cerebral metabolic rate for glu- Departments of Neurology rectly on enzymes and genes. Therefore, (Drs Pascual, Wang, and cose stabilizes and, at this time, can be as- De Vivo, and Ms Engelstad) and disorders of energy metabolism cause sec- sumed to mainly reflect maintenance con- Pediatrics (Drs Pascual and ondary genetic dysregulation, similar to the sumption. The large difference between the De Vivo), Department of way deficiencies of transcription or tro- highest (child) and lowest (neonate) glu- Radiology, Kreitchman PET phic factors alter gene expression. An ad- cose consumption includes expenditures Center (Ms Saxena and ditional mechanism, of not only meta- in development. Dr Van Heertum), and bolic disorders but of all genetic diseases, Department of Neurology, Numerous energy-consuming pro- Sergievsky Center (Dr Hinton), involves widespread effects exerted by a cesses are limited or preferentially circum- College of Physicians and mutant gene on the rest of the genome at scribed to infancy and childhood, but Surgeons, Columbia University, the pretranslational level, regardless of the among them the formation and consoli- New York, New York. function of the gene product.1 The ef- dation of neural circuits—which include

(REPRINTED) ARCH NEUROL / VOL 64, APR 2007 WWW.ARCHNEUROL.COM 507

©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 the division and movement of neurons and precursors, leptic encephalopathy that was refractory to anticonvulsants. and the generation and pruning of synapses—are asso- From age 3 to 15 months, head circumference growth had ciated with a high energetic demand.4 This energy must decelerated from the 50th percentile to 2 cm below the third ultimately be derived from glucose. There are two in- percentile, where she remained all of her life. Studies before stances where the cerebral access of glucose is compro- the participant aged 15 months included a computed tomography scan of the head; electroencephalography; and examina- mised for a sufficiently prolonged period of time to im- tion of visual- and auditory-evoked potentials; urine oligosac- pact development: one is mutation of the glucose charides and amino acids; blood gases, pH, ammonia, lactate, 5 transporter type 1 (GLUT1) of the blood-brain barrier, pyruvate, amino acids, organic acids, lysosomal hydrolases; and the other is chronic congenital (or early infantile) and cerebrospinal fluid lactate, pyruvate, and amino acids. All hypoglycemia. Neither resembles acute hypoglycemia, study results were normal. At age 15 months, a routine blood which causes a different type of encephalopathy domi- chemistry survey obtained while fasting during a hospital nated by cellular injury. admission for further evaluation revealed a blood glucose of 15 The classic phenotype of GLUT1 deficiency is mg/dL (0.83 mmol/L). Additional blood glucose determina- known6-10; most patients who are GLUT1 deficient mani- tions confirmed hypoglycemia. During a brief (6-hour) fast, fest encephalopathy dominated by hyperexcitability (epi- blood glucose concentration fell from 92 to 32 mg/dL (5.1 to 1.8 mmol/L). At the end of the fast, a glucagon stimulation test lepsy) and abnormalities of cognition and motor con- resulted in a marked glycemia response peaking at 134 mg/dL trol that respond to a ketogenic diet to a varying degree. (7.4 mmol/L), suggesting hyperinsulinism. Diazoxide and In particular, seizures are typically controllable with this ephedrine failed to control glycemia and the patient ultimately diet, whereas other patients experience a transient im- necessitated a nasogastric glucose drip to prevent seizures provement in neurologic performance after a carbohy- associated with hypoglycemia. Several additional maneuvers drate load. Independent of these interventions, most pa- also suggested hyperinsulinism: a ketogenic diet failed to gen- tients exhibit a persistent, residual encephalopathy erate ketone bodies despite decreasing glucose levels, and a characterized by the constellation of motor and lan- leucine load failed to alter glycemia. Insulin levels were more guage dysfunction and mental retardation together with often normal than elevated. A subtotal pancreatectomy (85%) thalamocortical hypometabolism and accentuated basal revealed a pancreatic tail adenoma and islet cell material throughout the body of the organ, consistent with nesidioblas- ganglia metabolism as detected by 18F-2-deoxyglucose 11 tosis. On discharge several weeks later, the patient’s neurologic positron emission tomography (PET). This peculiar performance was vastly improved with residual psychomotor imaging pattern appears to be imprinted on the brain in abnormalities; language had not yet developed. Muscle tone infancy and remains essentially immutable, indicating that and reflexes were increased, and plantar responses were flexor. deficiency of GLUT1 causes an abnormal neural matu- Prominent ataxia limited standing to brief periods of time. Gly- rational and functional pattern during the earlier devel- cemic levels were normal or high normal (Ͻ130 mg/dL [Ͻ7.2 opmental interval. mmol/L]). Seizures all but ceased, and rare residual convul- We set out to elucidate whether this aberrant pattern sions subsided during the next few years. Electroencephalogra- was caused by the genetic deficit of GLUT1 (ie, by vir- phy results were normal under normoglycemic conditions. At tue of the mutation of the glucose carrier irrespective of age 26 years, the patient’s brain magnetic resonance imaging results were normal. its function as a transporter) or whether it was simply Participant 2, a 16-year-old adolescent boy, had GLUT1 de- the consequence of diminished brain glucose availabil- ficiency. His early development was characterized by frequent ity during a specific developmental interval despite a nor- generalized seizures that were refractory to anticonvulsants since mal GLUT1 gene. The clinical, neuropsychologic, and PET age 18 months. Additional features were ataxia, dysarthria, and study of twins with normal GLUT1 expression, one of difficulties with limb action due to pyramidal tract dysfunc- whom was afflicted by chronic congenital hypoglyce- tion. Sitting was accomplished at 1 year of age and ambulation mia, supported the latter mechanism, indicating that glu- was achieved at 2 years. Head circumference growth deceler- cose, rather than GLUT1, is required for the develop- ated and remained at the third percentile. Additional seizures ment of normal excitability in certain cerebral regions consisting of loss of postural tone became prominent after 3 during a critical developmental interval. years of age and occurred routinely each morning within 20 minutes of waking. Brain magnetic resonance imaging and computed tomography results were normal at 15 years of age, as METHODS had been a variety of serial analytic investigations performed in blood and urine at different ages. He carried the mutation R126H in 1 GLUT1 allele, and his cerebrospinal fluid glucose PARTICIPANTS concentration was 38 mg/dL (2.1 mmol/L).12 Participant 3, the 23-year-old twin sister of participant 1, Informed consent conforming to the guidelines of the institu- was healthy and served as a control. She was born just before tional review board of Columbia University and the Columbia her sister. Participant 4, aged 20 years, was an unrelated healthy Kreitchman PET Center was obtained from all research par- man who also served as a control. ticipants. Patients who are GLUT1 deficient (one of whom is subsequently described further), a congenital hypoglycemia patient, and the healthy twin sister of the latter were assessed for PET SCANNING signs and symptoms of encephalopathy, erythrocyte glucose uptake (mediated by GLUT1), and sequence analysis of the GLUT1 Participants fasted for at least 8 hours prior to injection of the gene if erythrocyte glucose uptake was abnormal. radiopharmaceutical. Fasting glucose level was 79 to 100 Participant 1, a 23-year-old woman, experienced unrecog- mg/dL (4.4-5.6 mmol/L) in all cases. Intravenous access was nized hypoglycemia earlier in her life. Seizures were present obtained at least 15 minutes prior to the radiopharmaceutical since 3 months of age and were followed by a progressive epi- administration. The participants were then injected with 0.14

(REPRINTED) ARCH NEUROL / VOL 64, APR 2007 WWW.ARCHNEUROL.COM 508

Figure. Positron emission tomography in congenital hypoglycemia (participant 1, aged 23 years), GLUT1 deficiency (participant 2, aged 16 years), and healthy controls (participants 3 and 4, aged 23 and 20 years, respectively). Participant 3 was the twin sister of participant 1. Participants 1 and 2 exhibit diminished cortical and thalamic glucose uptake, with a relatively enhanced basal ganglia signal. Participants 3 and 4 illustrate normal distribution of the tracer. Axial-oblique scans were tilted to include the maximum cross-sectional area of both basal ganglia and thalamus, demarcated by red and blue trapezoids. The images displayed represent 1-cm-thick slices (twice the acquisition thickness), which was chosen to facilitate the visualization of a larger brain volume for illustration purposes.

mCi of 18F-2-deoxyglucose per kilogram of body weight and ERYTHROCYTE UPTAKE ASSAY scanned after a period of 30 minutes following injection. Stud- ies were acquired on a Siemens RS ECAT EXACT HRϩ Function of GLUT1 was assessed in erythrocytes. Blood samples (Siemens Medical Solutions, Malvern, Pa) with full-width half- were collected in sodium-heparin or citrate-phosphate- maximum=4 mm. Each study was acquired using a multiframe dextrose solution, and uptake of 0.5 mmol of 14C-labeled 3-O- technique and autoattenuation correction in a dynamic scan methyl-D-glucose (stock activity 1 mCi/mL) per liter of blood mode (4 frames at 480 seconds per frame) with filter backpro- into erythrocytes was measured at 4°C and a pH of 7.4. 3-O- jection. Reconstruction with autoattenuation correction was methyl-D-glucose influx was terminated at 5-second inter- achieved using a Hann filter (cutoff, 0.40 cycles per pixel). vals; washed cells were lysed, and uptake was quantitated by Postreconstruction transverse, oblique transverse, and coronal liquid scintillation counting. Data were expressed as the natu- and sagittal plane images with a slice thickness of 0.50 cm were ral logarithm of the ratio of intracellular radioactivity at vari- then produced and displayed using both an inverted gray-scale ous times and at equilibrium vs time; 3-O-methyl-D-glucose map and a rainbow (16-step) map. Images were interpreted vi- uptake was expressed as the slope of the resulting curve.13 sually on transverse brain image file (slice thickness, 2.54 mm). They were compared with a healthy control by a nuclear neuroradiologist, who identified the main cerebral structures and NEUROPSYCHOLOGIC STUDY evaluated uptake qualitatively. The images in the Figure were resliced at 1-cm thickness for illustration purposes. Participants 1, 2, and 3 were given a battery of neuropsychologic tests, including measures of intelligence (the Wechsler ADDITIONAL CLINICAL STUDIES Adult Intelligence Scale, Third Edition), vocabulary (the Pea- body Picture Vocabulary Test, Third Edition), and visual construction (the Developmental Test of Visual-Motor Integra- Selected participants received standard electroencephalo- tion). The object assembly subtest of the Developmental Test grams. Magnetic resonance imaging scans were performed using of Visual-Motor Integration was not administered. The moth- 1.5-T scanners. The following imaging sequences were ob- ers of participants 1, 2, and 3 completed a measure of adaptive tained in the axial, coronal, and sagittal planes: T1-weighted function (the Vineland Adaptive Behavior Scales). axial and sagittal proton-density images; and T2-weighted axial, fluid-attenuated inversion recovery axial, and diffusion- weighted axial images. RESULTS

GENETIC ANALYSIS NEUROLOGIC FUNCTION

DNA was extracted from participants’ blood after consent for Participant 1 was afflicted by a residual encephalopathy genetic testing had been obtained according to standard meth- at 23 years of age when her PET scan was performed. She ods described elsewhere.5,8 Genomic DNA was purified and was receiving special education. Her neurologic exami- quantified, and samples were subjected to gel electrophoresis nation results were abnormal, showing hypertonicity, prior to polymerase chain reaction. Appropriate polymerase ataxia, dysarthria, hyperreflexia with ankle clonus, and chain reaction primers were designed to yield DNA fragments Babinski signs. Participant 2 had residual epileptic en- spanning the entire GLUT1 coding region and intron-exon boundaries in chromosome 1. DNA was automatically se- cephalopathy at age 16 years when he underwent PET quenced and mutations were confirmed by sequencing both scanning. He was receiving special education and was em- strands. All the patients with GLUT1 deficiency described herein, ployed part-time. His neurologic examination was domi- including participant 2 and our prior series of patients studied nated by spasticity, ataxia, dysarthria, hyperreflexia with by PET,11 carried 1 mutation in 1 allele. ankle clonus, and Babinski signs. Later in his life, inter-

(REPRINTED) ARCH NEUROL / VOL 64, APR 2007 WWW.ARCHNEUROL.COM 509

Syndrome Cause Course Manifestation Key Analyte Neuroglycopenia GLUT1 deficiency syndrome Static Epilepsy CSF glucose (Ͻ40 mg/dL Spasticity [2.2 mmol/L]) Dysarthria Chronic congenital Microcephaly Blood glucose hypoglycemia Movement disorders Learning disabilities Acute hypoglycemia Various Reversible Behavioral abnormalities Blood glucose Seizures Hemiparesis Coma

Abbreviations: CSF, cerebrospinal fluid; GLUT1, glucose transporter type 1.

PET SCANS Table 2. Neuropsychologic Assessment of Neuroglycopenia The functional images of participants 1 and 2 were al- Test Participant 1 Participant 2 Participant 3 most identical (Figure). They were qualitatively ana- Wechsler scale WAIS-III WISC-III WAIS-III lyzed by a nuclear neuroradiologist unaware of geno- Standard Standard Standard type or clinical manifestations. Mild to moderate Score* Score* Score* Verbal IQ 62 60 97 hypometabolic activity was noted in the parietal regions Performance IQ 73 46 113 bilaterally. More marked hypometabolism was identi- Full-scale IQ 64 49 104 fied in the temporal regions bilaterally and especially me- Wechsler subtest WAIS-III WISC-III WAIS-III sially. Hypometabolic activity was identified in the thala- Scaled Scaled Scaled mus. This overall hypometabolic pattern contrasted with Score† Score† Score† a relative enhancement of uptake displayed by the basal Picture completion 8 1 12 Vocabulary 3 1 9 ganglia. These results are typical of GLUT1 defi- 11 Digit/symbol coding 5 1 15 ciency. Participants 3 and 4 had normal PET-scan re- Similarities 5 3 10 sults and were used as similar age controls. Block design 7 1 11 Arithmetic 3 6 11 NEUROPSYCHOLOGIC PERFORMANCE Matrix reasoning 4 NA 14 Digit span 4 7 8 Information 6 4 11 Participants 1 and 3 were simultaneously assessed for Picture arrangement 4 1 8 intelligence (the Wechsler Adult Intelligence Scale), Comprehension 3 1 9 vocabulary (the Peabody Picture Vocabulary Test), and Letter/number 3NA12 visual construction (the Developmental Test of Visual- sequencing Motor Integration). Their mother completed the Vine- Other tests Standard Standard Standard land Adaptive Behavior Scales as a means of determin- Score* Score* Score* PPVT-III 66 47 105 ing each participant’s adaptive capabilities. Both VMI 61 55 65 individuals applied themselves to the testing, differing VABS composite 83 49 119 in their demeanor. Participant 3 was serious, answered questions thoroughly, spoke articulately, and worked Abbreviations: NA, not applicable; PPVT-III, Peabody Picture Vocabulary Test, quickly through the tests. Participant 1 was friendly and Third Edition; VABS, Vineland Adaptive Behavior Scales; VMI, Developmental Test of Visual-Motor Integration; WAIS-III, Wechsler Adult Intelligence Scale, laughed easily. Her speech was indistinct because of poor Third Edition; WISC-III, Wechsler Intelligence Scales for Children, Third Edition. articulation (substitutions of w sounds for r sounds) and *Standard scores have a population mean ± SD of 100 ± 15. prosody, with each word being uniformly stressed with- †Scaled scores have a population mean ± SD of 10 ± 3. out inflection. Additionally, she displayed semantic para- phrasia; for example, she described the “bodies of the face” instead of the “parts of the face,” and the “bristle of the knife” instead of the “point of the knife.” Participant 3 ventions that aimed to increase glycemia (up to 160 mg/dL had a full-scale IQ score of 104 (61st percentile, in the [8.9 mmol/L]), such as diazoxide combined with an oral average range) while the full-scale IQ score for partici- glucose load (75 g), resulted in improved alertness and pant 1 was 64 (first percentile, in the very low range) psychomotor coordination. Cornstarch supplementa- (Table 2). Participant 3 performed within normal lim- tion together with diazoxide limited seizures to occa- its while participant 1 demonstrated an impaired perfor- sions of physical and emotional stress but caused an un- mance across all subtest and composite scores on the desired weight gain. The abnormalities exhibited by Wechsler Adult Intelligence Scale. Both twins showed rela- participants 1 and 2 are included in Table 1. Partici- tive strength on the nonverbal items; participant 3 scored pant 3 was examined simultaneously with her twin sis- in the high average range, while her sister scored within ter (participant 1) and found to be neurologically healthy. the borderline range (participant 3, performance IQ=113

(REPRINTED) ARCH NEUROL / VOL 64, APR 2007 WWW.ARCHNEUROL.COM 510

©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 [81st percentile]; participant 1, performance IQ=73 sufficient, with adaptive skills within the normal [fourth percentile]). This difference was also noted on range. Both individuals exhibited a friendly demeanor, the vocabulary measure, where participant 3 had an av- and each attempted the test items in a playful manner, erage performance and participant 1 had an impaired per- a feature characteristic of GLUT1 deficiency. formance. However, the twins’ performances on the drawing test did not differ significantly. Both twins performed poorly on the Developmental Test of Visual-Motor In- COMMENT tegration, though the reasons behind this are likely different; the performance of participant 3 was quick and The combination of infantile epilepsy, mental retarda- careless, and she made errors on easier items yet cor- tion, and abnormal movement coordination and tone, to- rectly copied more complex ones, while participant 1 was gether with a pattern of cerebral–gray matter abnormali- more careful on the easy items but was unable to repro- ties detectable by PET scanning—all due to a persistent duce the more complex ones. According to their mother, decrease in glucose access to the developing brain— both women had good adaptive behavior skills, with par- constitutes the syndrome of neuroglycopenia. This disor- ticipant 3 scoring in the high average range and partici- der can be caused by 2 mechanisms: persistent hypogly- pant 1 scoring in the low average range (Vineland Adap- cemia and molecular deficits in the GLUT1 gene. Both are tive Behavior Scales composite scores=119 [90th eminently recognizable and treatable diseases. In in- percentile] and 83 [13th percentile], respectively). fancy, the fundamental manifestation of neuroglycope- Participant 2 was tested when he was aged 16 years. nia is epilepsy that is refractory to medications and decel- He was tested with the Wechsler Intelligence Scales for eration of head growth. Abnormalities of ocular movement Children, Third Edition, as well as the Peabody Picture resembling opsoclonus (which can be particularly promi- Vocabulary Test, and the Developmental Test of Visual- nent in GLUT1 deficiency), alterations of muscle tone, ar- Motor Integration; his mother completed the Vineland ticulatory language dysfunction, and movement disor- Adaptive Behavior Scales. His manner during the evalu- ders (including ataxia and dystonia) are commonly ation was cheerful and friendly. He was attentive and per- associated.7 After childhood, the residual clinical pattern severed even when the material became difficult for him. is dominated by spasticity, ataxia, and language difficul- His speech was notably difficult to understand; he spoke ties, with or without epilepsy. In GLUT1 deficiency, these haltingly, pausing between syllables, and made numer- clinical features are often associated with generalized 2.5- ous articulation errors. For example, he pronounced r to 3.5-Hz spike-wave electroencephalographic dis- sounds as w sounds (eg, saying “twee” for “tree”) and charges14 and with normal brain structure as assessed by frequently simplified syllabic complexity at the end of magnetic resonance imaging. Measures of neuropsycho- words (eg, “tootbwu” for “toothbrush”). His perfor- logic performance are invariably abnormal. mance across all the measures was significantly im- A subset of patients with GLUT1 deficiency are able paired (Table 2). His full-scale IQ and his performance to overcome a variety of milder developmental and learn- IQ scores were both 49 (0.1 percentile, in the extremely ing disabilities, and these patients give rise to the famil- low range). His verbal IQ score was also impaired (ver- ial transmission of the disease in an autosomal- bal IQ=60; 0.4 percentile, in the extremely low range). dominant fashion.15,16 These patients with a milder On both the drawing test and the test of receptive vo- phenotype generally suffer from isolated dyslexia, men- cabulary, the participant had marked difficulty and also tal retardation, ataxia, or dystonia and constitute an in- scored in the extremely low range. According to his moth- creasingly recognized group, as the practice of perform- er’s responses on the Vineland Adaptive Behavior Scales, ing a lumbar puncture (documenting hypoglycorrhachia) his adaptive behavior skills were also impaired (Vine- for the study of these neurologic abnormalities is receiv- land Adaptive Behavior Scales composite score=49; 0.03 ing increased consideration. Thus, our observation of epi- percentile, in the extremely low range). lepsy in all cases of GLUT1 deficiency as originally re- In summary, participants 1 and 2 had similarly, sig- ported constituted an ascertainment bias arising from the nificantly impaired cognitive skills across all the test initial recognition of the most severely affected patients. measures, especially when compared with participant Because glucose transport appears to be rate limiting for 3. Both had full-scale IQ scores in the extremely low brain metabolism,17 chronic congenital hypoglycemia is range. Participant 1 scored somewhat better than par- more likely to produce a severe neuroglycopenia pheno- ticipant 2 on visuospatial items (such as those consti- type, as reflected by the study of participant 1. Despite in- tuting the performance IQ score), yet their scores on terprandial normoglycemia, a subsequent (and probably verbal tests were comparable. Both had extremely lim- consistent) preprandial and postprandial decrease in glu- ited receptive vocabularies and verbal skills. Both cose concentration caused encephalopathy as severe as that spoke in complete sentences, yet the construction of experienced by some hemizygous patients with GLUT1 de- the sentences was very simple, and when asked to ficiency or as that caused by mutations that effectively abol- answer complex questions, both had difficulty ish the function of 1 GLUT1 allele. As an autosomal- responding. Both also had unusual speech characteris- dominant condition, GLUT1 deficiency is associated with tics, including poor articulation and restricted a normal residual GLUT1 allele, supplying at least 50% of prosody. In general, participant 2 was more impaired, transport capacity in the context of normoglycemia.5 and his speech difficulties were considerably more Acute hypoglycemia, a related entity, is character- pronounced. Further, participant 2 had deficient daily ized by divergent pathologic and clinical features. The living skills, whereas participant 1 was more self- fundamental difference from neuroglycopenia is a tem-

(REPRINTED) ARCH NEUROL / VOL 64, APR 2007 WWW.ARCHNEUROL.COM 511

©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 poral one in terms of both its duration and its special re- manifestations of neuroglycopenia may not appear evident lation to cerebral maturation; neuroglycopenia is a per- until the corticothalamic unit is functionally active and ca- sistent phenomenon that occurs during development. pable of influencing behavior. More studies of an animal Acute hypoglycemia is a reversible process, unless it is model of GLUT1 deficiency will be necessary to address this prolonged and profound enough to cause excitotoxic- question. ity. When severe, hypoglycemia causes neuronal death in experimental animals after glucose levels have fallen GLUCOSE AS A DEVELOPMENTAL SIGNAL below 18 mg/dL (0.1 mmol/L) and the electroencepha- logram remains isoelectric for about 30 minutes.18 At that time, glutamate and aspartate are indiscriminately re- The syndrome of neuroglycopenia is characterized by a par- leased into the extracellular space, occupying available tial and persistent deprivation of substrate while the brain excitatory amino acid receptors.19 Necrosis of the den- develops. Our view, in light of the current results, consid- tate gyrus of the hippocampus can occur, and a predi- ers two possible mechanisms by which decreased brain glu- lection for the superficial layers of the cortex is some- cose can cause neurologic disturbance: one is decreased fuel times seen.20 In milder cases, the clinical manifestations (energy), and the other is abnormal thalamocortical matu- of hypoglycemia are again distinguishable from neuro- ration (development). Both are expanded upon herein, rec- glycopenia, as they include altered behavior and con- ognizing that these are simplified hypotheses and that ad- sciousness, or coma, seizures, or hemiparesis that sub- ditional alternatives may be feasible. side by restoring blood glucose. Recovery is almost Several considerations underlie our current view of invariably complete without detectable structural or func- glucose as an energetic substrate. Of all cellular and ex- tional sequelae detectable by neurologic examination21 tracellular compartments, the interstitial fluid is the least or PET.22-24 Transient neonatal hypoglycemia is not likely to contain glucose and carbohydrates in GLUT1 thought to be associated with abnormalities of cerebral deficiency. Most likely, there is reduced availability of glucose metabolism when resolved.25 interstitial medium glucose to both astrocytes and neurons, the former of which rely on GLUT1 for their glu- A PERSISTENT PATTERN OF ABNORMALITIES cose uptake, while glucose transporter type 3, the high- affinity neuronal transporter, may remain fully active, Both the clinical and PET imaging features of neuroglyco- effectively capturing all available glucose into the neu- penia leave an indelible mark on the brain; neurobehavioral ron. If the calculations of Barros et al17 and the experi- difficulties persist and thalamocortical glucose uptake re- mental observations of Magistretti and colleagues29 are mains decreased. In this abnormal background, the basal correct, the astrocyte, which probably is not a signifi- ganglia signal appears enhanced. It is not known whether cant barrier to plasma-interstitium glucose flux, would the basal ganglia metabolic rate is absolutely increased or be deprived of glucose, which in turn could result in a whether it is also decreased but appears higher than the con- decrease in the amount of lactate produced for subse- trasting surrounding structures. Nevertheless, these PET quent delivery to the neuron. From this perspective, both findings are stereotyped regardless of disease severity, or GLUT1 deficiency and hypoglycemia would primarily im- type and duration of therapy, and persist into adulthood.11 pair astrocytic lactate production, resulting in down- For example, a minimally symptomatic patient with GLUT1 stream neuronal dysfunction manifested (for reasons that deficiency, the 39-year-old father of a 7-year-old patient with are still unclear) as seizures, the immediately observ- genetically transmitted severe epileptic encephalopathy, able phenomena hallmark to these conditions. was only afflicted by dyslexia. His PET scan was indistin- From a developmental perspective, additional obser- guishable from all other patients with GLUT1 deficiency, vations may be relevant. In carnivores, developing thala- despite his advancing age and mild disease severity (J.M.P., mocortical axons first approach their appropriate corti- unpublished data, 2003). In contrast, the time of appear- cal regions and then wait in the subplate region before ance of the imaging abnormalities is not known. Neurogly- invading the cortex.30 They then progress tangentially to- copenia presents clinically only in infancy or later, when ward the cingulate, changing direction. The waiting pe- the phenotype is one of mild developmental and learning riod that these axons experience as well as the change of disabilities. Two explanations are possible. The prenatal and direction that they undergo suggest that the subplate con- neonatalblood-brainbarrierisimmatureandallowsthepen- tains important signals for thalamocortical circuit devel- etrationofsubstances(eitherdirectlyorbyspecializedmecha- opment,31 including glycosaminoglycans, which can serve nisms) that are later excluded. According to this view, mol- as axonal guidance cues.32,33 We hypothesize that accu- ecules such as glucose might circumvent the blood-brain mulation of cerebral glycosaminoglycans may be im- barrier during the first weeks of postnatal life as other small paired in GLUT1 deficiency. During maturation, collat- molecules do until the blood-brain barrier matures. This eral projections of thalamocortical and corticothalamic interpretation is unlikely because the cerebral metabolic rate fibers enter the reticular nucleus of the thalamus, which of glucose is low perinatally, which is in good correlation occupies a central place in the synchronization of net- with both the low rate of glucose transport across the blood- works of neurons.34 The activity of the reticular nucleus brain barrier after birth and the diminished GLUT1 den- on thalamic relay cells results in mode switching from sity typical of that developmental period.26 Only later, when tonic to repetitive burst firing that is propagated to and GLUT1 expression increases, the metabolic rate for glucose from the cortex, inducing resonance.35 This form of syn- isstimulatedandthebrainbecomesapredominantlyglucose- chronization may constitute the basis of approximately consuming organ.27,28 Alternatively (and more likely), the 3-Hz spike-wave epilepsy in neuroglycopenia.

(REPRINTED) ARCH NEUROL / VOL 64, APR 2007 WWW.ARCHNEUROL.COM 512

©2007 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/28/2021 haploinsufficiency of the blood-brain barrier hexose carrier. Nat Genet. 1998; CONCLUSIONS 18:188-191. 6. De Vivo DC, Trifiletti RR, Jacobson RI, Ronen GM, Behmand RA, Harik SI. De- Neuroglycopenia, a syndrome associated with selective fective glucose transport across the blood-brain barrier as a cause of persistent neural deficits, can result from GLUT1 deficiency or early hypoglycorrhachia, seizures, and developmental delay. N Engl J Med. 1991; 325:703-709. hypoglycemia, which phenocopy one another. In in- 7. Wang D, Pascual JM, Yang H, et al. Glut-1 deficiency syndrome: clinical, ge- fancy, the state of neuroglycopenia predominantly causes netic, and therapeutic aspects. Ann Neurol. 2005;57:111-118. hyperexcitability and is accompanied and followed by re- 8. Wang D, Kranz-Eble P, De Vivo DC. Mutational analysis of GLUT1 (SLC2A1) in sidual encephalopathy with marked pyramidal and cer- Glut-1 deficiency syndrome. Hum Mutat. 2000;16:224-231. ebellar dysfunction. While the mechanisms of neurogly- 9. Pascual JM, Wang D, Lecumberri B, et al. GLUT1 deficiency and other glucose transporter diseases. Eur J Endocrinol. 2004;150:627-633. copenic brain development and function remain to be 10. De Vivo DC, Wang D, Pascual JM, Ho YY. Glucose transporter protein syndromes. elucidated, glucose may act in a dual capacity, both as Int Rev Neurobiol. 2002;51:259-288. fuel and as a signaling molecule, causing a more selec- 11. Pascual JM, Van Heertum RL, Wang D, Engelstad K, De Vivo DC. Imaging the meta- tive spectrum of abnormalities than those due to global bolic footprint of Glut1 deficiency on the brain. Ann Neurol. 2002;52:458-464. 12. von Moers A, Brockmann K, Wang D, et al. EEG features of glut-1 deficiency unavailability of energetic substrate. syndrome. Epilepsia. 2002;43:941-945. 13. Klepper J, Garcia-Alvarez M, O’Driscoll KR, et al. Erythrocyte 3-O-methyl-D- Accepted for Publication: October 19, 2006. glucose uptake assay for diagnosis of glucose-transporter-protein syndrome. Published Online: February 12, 2007 (doi:10.1001/ J Clin Lab Anal. 1999;13:116-121. archneur.64.4.noc60165). 14. Leary LD, Wang D, Nordli DR Jr, Engelstad K, De Vivo DC. Seizure characteriza- tion and electroencephalographic features in Glut-1 deficiency syndrome. Epilepsia. Correspondence: Juan M. Pascual, MD, PhD, Depart- 2003;44:701-707. ments of Neurology, Physiology, and Pediatrics, Univer- 15. Brockmann K, Korenke G, Moers A, et al. Epilepsy with seizures after fasting and sity of Texas Southwestern Medical Center, 5323 Harry retardation: the first familial cases of glucose transporter protein (glut1) deficiency. Hines Blvd, Mail Code 8813, Dallas, TX 75390-8813 Eur J Paediatr Neurol. 1999;3:A90-A91. 16. Brockmann K, Wang D, Korenke CG, et al. Autosomal dominant glut-1 defi- ([email protected]). ciency syndrome and familial epilepsy. Ann Neurol. 2001;50:476-485. Author Contributions: Study concept and design: Pascual, 17. Barros LF, Porras OH, Bittner CX. Why glucose transport in the brain matters for Wang, Hinton, Van Heertum, and De Vivo. Acquisition PET. Trends Neurosci. 2005;28:117-119. of data: Pascual, Hinton, Engelstad, Saxena, and 18. Auer RN. Hypoglycemic brain damage. Metab Brain Dis. 2004;19:169-175. Van Heertum. Analysis and interpretation of data: Pascual, 19. Ludolph AC, Riepe M, Ullrich K. Excitotoxicity, energy metabolism and neurodegeneration. J Inherit Metab Dis. 1993;16:716-723. Wang, Hinton, and De Vivo. Drafting of the manuscript: 20. Auer RN, Wieloch T, Olsson Y, Siesjo BK. The distribution of hypoglycemic brain Pascual. Critical revision of the manuscript for important damage. Acta Neuropathol (Berl). 1984;64:177-191. intellectual content: Pascual, Wang, Hinton, Van Heertum, 21. Malouf R, Brust JC. Hypoglycemia: causes, neurological manifestations, and and De Vivo. Statistical analysis: Pascual, Wang, and outcome. Ann Neurol. 1985;17:421-430. 22. Segel SA, Fanelli CG, Dence CS, et al. Blood-to-brain glucose transport, cerebral Hinton. Obtained funding: Pascual, Hinton, Van Heertum, glucose metabolism, and cerebral blood flow are not increased after hypoglycemia. and De Vivo. Administrative, technical, and material sup- Diabetes. 2001;50:1911-1917. port: Pascual, Wang, Hinton, Engelstad, and Saxena. Study 23. Fanelli CG, Dence CS, Markham J, et al. Blood-to-brain glucose transport and supervision: Pascual, Van Heertum, and De Vivo. cerebral glucose metabolism are not reduced in poorly controlled type 1 diabetes. Financial Disclosure: None reported. Diabetes. 1998;47:1444-1450. 24. Chabriat H, Sachon C, Levasseur M, et al. Brain metabolism after recurrent in- Funding/Support: This work was supported by grants sulin induced hypoglycaemic episodes: a PET study. J Neurol Neurosurg Psychiatry. NS037949 (Pascual, Wang, and De Vivo) and NS001698 1994;57:1360-1365. (Pascual and Wang) from the National Institute of Neu- 25. Kinnala A, Nuutila P, Ruotsalainen U, et al. Cerebral metabolic rate for glucose rological Disorders and Stroke. after neonatal hypoglycaemia. Early Hum Dev. 1997;49:63-72. 26. Vannucci RC, Vannucci SJ. Glucose metabolism in the developing brain. Semin Acknowledgment: We appreciate the assistance of the Perinatol. 2000;24:107-115. Giblin and Will Foundations. We also thank Hong Yang, 27. Cornford EM, Hyman S, Pardridge WM. An electron microscopic immunogold MD, Dalina Stiner, MS, and Carolina Regus, BS, at the analysis of developmental up-regulation of the blood-brain barrier GLUT1 glu- Colleen Giblin Laboratories at Columbia University and cose transporter. J Cereb Blood Flow Metab. 1993;13:841-854. John Kim, BS, at the Columbia Kreitchman PET Center. 28. Cornford EM, Cornford ME. Nutrient transport and the blood-brain barrier in developing animals. Fed Proc. 1986;45:2065-2072. We also thank Felipe Barros, MD, PhD, for his insight- 29. Pellerin L, Magistretti PJ. Neuroenergetics: calling upon astrocytes to satisfy hun- ful discussions. gry neurons. Neuroscientist. 2004;10:53-62. 30. Auladell C, Perez-Sust P, Super H, Soriano E. The early development of thalamocortical and corticothalamic projections in the mouse. Anat Embryol (Berl). REFERENCES 2000;201:169-179. 31. Richards LJ, Koester SE, Tuttle R, O’Leary DD. Directed growth of early cortical 1. Carlson CS, Eberle MA, Kruglyak L, Nickerson DA. Mapping complex disease loci axons is influenced by a chemoattractant released from an intermediate target. in whole-genome association studies. Nature. 2004;429:446-452. J Neurosci. 1997;17:2445-2458. 2. Chugani HT, Phelps ME. Maturational changes in cerebral function in infants de- 32. Kennedy TE. Cellular mechanisms of netrin function: long-range and short- termined by 18FDG positron emission tomography. Science. 1986;231: range actions. Biochem Cell Biol. 2000;78:569-575. 840-843. 33. Holt CE, Dickson BJ. Sugar codes for axons? Neuron. 2005;46:169-172. 3. Chugani HT, Phelps ME, Mazziotta JC. Positron emission tomography study of 34. Rhodes PA, Llinas R. A model of thalamocortical relay cells. J Physiol. 2005;565: human brain functional development. Ann Neurol. 1987;22:487-497. 765-781. 4. Yamada H, Sadato N, Konishi Y, et al. A milestone for normal development of 35. Llinas RR, Leznik E, Urbano FJ. Temporal binding via cortical coincidence de- the infantile brain detected by functional MRI. Neurology. 2000;55: tection of specific and nonspecific thalamocortical inputs: a voltage-dependent 218-223. dye-imaging study in mouse brain slices. Proc Natl Acad Sci U S A. 2002;99: 5. Seidner G, Alvarez MG, Yeh JI, et al. GLUT-1 deficiency syndrome caused by 449-454.

(REPRINTED) ARCH NEUROL / VOL 64, APR 2007 WWW.ARCHNEUROL.COM 513