The Nervous System · Part One · How It Is Built and Fueled
Lesson 14 / 61
Glucose and the Brain: Carriers, Sensing Neurons and the Defended Range
One sugar, no reserve, and a carrier that decides who gets fed.
Glucose is the sugar the brain runs on, and it reaches neurons only through carrier proteins in the vessel wall. The brain stores almost none of it. Safety comes instead from transport capacity that runs about twice the rate of use, and from carriers whose isoform decides which cell is fed. The Unified Model of Tone reads brain fuel as one state held by carrier capacity and alarm placement together.
Carriers
GLUT1 at the barrier and glia, GLUT3 on neurons
Transport half-saturation
About 4.8 mM plasma glucose, human brain
Brain glucose pool
About 1.0 micromole per gram at euglycemia
Glycogen store
Minimal, held mainly in astrocytes
Cerebral glucose metabolism
The set of reactions that consume glucose inside the brain. Glycolysis splits it to pyruvate, the citric acid cycle strips the carbon, and the respiratory chain converts the yield to ATP. A branch at the first step diverts glucose-6-phosphate into the pentose phosphate pathway, which makes ribose-5-phosphate for nucleotides and NADPH for antioxidant defense. Transmitter carbon leaves the cycle as glutamate, GABA, glycine and D-serine.
Glucose and tone
The Unified Model of Tone reads a glucose value as one sample of a curve, and reads the curve against where the alarms are set. Two numbers carry the information. The excursion is how far the value travels after a meal or a fast and how long it takes to come home. The placement is the glucose at which counterregulation commits, measured against the glucose at which performance drops. Transport capacity sets the first. Recent exposure sets the second.
01Glucose at the barrier
Glucose reaches the brain only through carrier proteins, and the isoform decides which cell it feeds
Glucose crosses no brain membrane on its own. Facilitative carriers move it, and the isoform map is fixed. GLUT1 sits at high concentration in the blood-brain barrier as a heavily glycosylated 55 kilodalton form. A less glycosylated 45 kilodalton form sits in the parenchyma, predominantly glia. GLUT3 is the neuronal carrier and GLUT5 sits on microglia Vannucci 1997.
Human transport has been measured directly. Proton magnetic resonance in four healthy subjects aged 18 to 22 put the plasma glucose concentration for half maximal transport at 4.8 plus or minus 2.4 mM Gruetter 1996. Maximal transport ran 0.80 plus or minus 0.45 micromol per gram per minute. Cerebral glucose consumption in the same fits came out at 0.32 plus or minus 0.16 micromol per gram per minute. Brain glucose at euglycemia was taken as 1.0 micromole per gram from carbon-13 measurement.
A half-saturation constant near 4.8 mM has a consequence for daily life. Ordinary blood sugar sits close to the concentration at which the carrier is working at half its maximum, so the barrier operates on the steep part of its own curve. Small falls in plasma glucose translate into real falls in delivery.
Transport capacity runs about twice the rate of use
The margin the brain keeps is in the carrier rather than the pantry. Five healthy volunteers held plasma glucose near 17 mmol/l for about two hours. Fitting transport and utilization together put maximum capacity across the blood-brain barrier at nearly twice maximum cerebral glucose utilization Shestov 2011. That surplus is the entire reserve.
Storage adds almost nothing to it. Brain glycogen is minimal and lives mainly in astrocytes, and the size of that tank is counted on Neuroglia and the Neurovascular Unit. The brain committed to constant resupply over deep reserves, so its whole safety margin is a transport number.
Carrier density can be the limiting step outright. In developing rat brain, the maturational rise in cerebral glucose utilization tracked the expression of non-vascular GLUT1 and GLUT3 more closely than the barrier rate constants. The authors concluded that transporter expression is rate limiting for glucose use in early postnatal life Vannucci 1994. Circuitry can be finished while the supply line is not.
02Findings
What the research shows
What the measurements show about brain glucose: how much carrier stands between blood and neuron, where the alarms sit, and what a shortfall takes first.
03Fuel and building blocks
Glucose pays for the pumps and supplies the carbon that becomes transmitters
Glucose does two jobs in the neuron and the second is easy to miss. Glycolysis and the citric acid cycle turn it into ATP, which rebuilds the sodium and potassium gradients that every impulse spends. The same carbon skeletons become glutamate, GABA, glycine and D-serine.
Carbon-13 spectroscopy tracks the split. Human measurements put citric acid cycle flux at 0.74 micromol per minute per gram Morris 2003. That fits positron emission tomography rates of cerebral glucose consumption of 0.3 to 0.4 micromol per minute per gram, since each glucose molecule yields two pyruvate. Glial recycling of glutamate accounted for roughly 50 percent of that cycle flux.
Half the cycle is therefore transmitter traffic rather than housekeeping. A fall in glucose supply withdraws the raw material of communication at the same moment it withdraws the energy to send it. Brain activation raised cycle flux 60 percent in the same body of work, close to the rise in cerebral glucose consumption seen on imaging. Where the pyruvate finally goes is counted on Mitochondria and ATP.
The pentose phosphate branch buys defense rather than ATP
Some glucose leaves the pathway at the first step and never becomes ATP at all. Glucose-6-phosphate dehydrogenase diverts it into the pentose phosphate pathway, which yields nucleotide sugars and the NADPH that keeps glutathione reduced. In rat hippocampal CA1, severe hypobaric hypoxia lowered glucose-6-phosphate dehydrogenase activity and cut NADPH levels, promoting oxidative stress and neuronal loss Vetrovoy 2020. Topotecan at 5 mg/kg normalized the enzyme and restored total glutathione.
That branch explains a clinical pattern. Disruption of normal glucose metabolism forms the pathophysiological basis of many brain disorders Mergenthaler 2013. Part of the reason is arithmetic. A glucose shortfall withdraws the cell's chemical defense in the same stroke that it withdraws the cell's power, so the tissue meets oxidative damage with less to answer it.
04Glucose use during activity
Active cortex takes up far more glucose than it burns, so glucose flow reports signaling rather than oxidation
Activation breaks the ratio between glucose and oxygen. At rest the human brain consumes the two in a molar ratio of 4.1 to 1. During focal physiological neural activity glucose uptake rose 51 percent and blood flow rose 50 percent while oxygen consumption rose 5 percent Fox 1988. The molar ratio for those increments came out at 0.4 to 1.
The extra glucose is not being oxidized. Whatever the active tissue buys with it, the purchase is not immediate ATP from the respiratory chain, and the local blood flow answering that demand is being regulated for purposes other than oxidative metabolism.
The resting stoichiometry is uniform across the cortex. In normal human brain at rest the oxygen to glucose index came out near 5.3 and the oxygen extraction fraction near 0.4 Hyder 2016. Maximum ATP turnover ran near 9.4 micromol per gram per minute, and no resting network stood apart from the gray matter mean. Resting cortex oxidizes nearly everything it takes, which is what makes the activation surplus stand out.
Glutamate uptake is the signal that routes the glucose
The address on the delivery is written by the transmitter. Glutamate stimulates glucose utilization and lactate production in astrocytes through its sodium-dependent uptake system and not through its receptors, and the coupling step is the sodium-potassium ATPase Pellerin 1994. Sodium carried in with glutamate raises the pump's workload, and the pump's workload calls for glycolysis.
Glucose consumption therefore marks the synapse where transmitter was just cleared. An image built on labeled glucose is reading synaptic traffic and its cleanup, which is why lactate rising in an activated cortex signals handling rather than starvation. Glutamate and its clearance are set out on GABA, Glutamate and the Balance.
Local glucose handling reports a local state
The reading is sensitive enough to find an epileptic focus between seizures. In 24 patients with temporal lobe epilepsy studied inter-ictally, the oxygen-glucose index ran higher on the side of the focus than on the other side Zhang 2022. The 18 healthy controls stayed bilaterally symmetrical. Where labeled glucose alone disagreed with invasive electrophysiology the index agreed with it, and the reverse also held.
A patch of cortex holding an abnormal state shows that state in how it spends sugar, with nothing happening at the time. Glucose handling is a local readout of local organization, not a whole-organ average. The local setting it reports is the central integrative state, set out on The Neuron and the Central Integrative State.
05Glucose sensing neurons
Dedicated neurons in the hypothalamus read glucose itself and change their firing rate
Some neurons treat glucose as a message rather than a fuel. They sit in the hypothalamus and the brainstem, and they govern the nutrient axis running between the body and the brain. In the ventromedial hypothalamic nucleus, glucose-excited neurons raise their action potential frequency and glucose-inhibited neurons lower theirs as glucose rises from 0.1 to 2.5 mmol/l Song 2005. Glucose-excited cells sense it through the ATP-sensitive potassium channel. Glucose opens a chloride channel in the glucose-inhibited cells.
The population is small and mixed. Ventromedial neurons were recorded in slices from 14 to 21 day old male Sprague-Dawley rats Song 2001. Of those cells, 14 percent were directly inhibited by a fall in glucose and 3 percent were directly excited by it. Another 14 percent were excited presynaptically by a fall. Of the rest, 11 percent were presynaptically inhibited when glucose rose above 2.5 mmol/l and 8 percent were presynaptically excited.
Two thirds of the glucose sensing in that nucleus is therefore indirect, carried on synapses rather than on the cell that reports. Fuel status reaches the output as a network state and not as a single meter reading. In a rodent model of diet-induced obesity and type 2 diabetes, every subtype was fewer in number and answered glucose abnormally.
Lactate reaches the same cells and does not act on them uniformly. Lactate reversed the inhibitory effect of decreased glucose on glucose-excited neurons by closing the same potassium channel Song 2005. Adding 0.5 mmol/l lactate raised their firing rate even where a rise from 2.5 to 5 mmol/l glucose did not. It also raised the firing rate of glucose-inhibited neurons in both low and 2.5 mmol/l glucose. Non-glucosensing neurons were unaffected. A substitute fuel is read as a different message by the two halves of the sensor.
The counterregulatory answer leaves through the splanchnic nerve
Sensing buys nothing without an output, and this output is a spinal one. Insulin-induced hypoglycemia raised adrenal tyrosine hydroxylase mRNA three to five fold at 5 hours in rats, and no such rise occurred after the splanchnic nerve was cut Vietor 1996. Severing one nerve removed the adrenal gland's ability to rebuild the epinephrine it was spending.
The line from a low blood sugar to a surge of epinephrine runs through preganglionic fibers leaving the thoracic cord. A metabolic emergency and a segmental sympathetic outflow are the same circuit at this point, which is why one cut abolishes the whole response. The wider map is set out on The Autonomic Nervous System, and the hormonal arm on The HPA Axis and Cortisol.
06Thresholds for hypoglycemia
The body defends blood glucose in a staged order, and every warning fires before thinking fails
Hypoglycemia triggers a ranked sequence rather than one alarm. Stepped clamps in 10 normal volunteers put the start of glucagon and epinephrine secretion at 68 plus or minus 1 mg/dl Mitrakou 1991. Norepinephrine began at 65 plus or minus 1 and growth hormone at 67 plus or minus 2. Autonomic symptoms began lower, at 58 plus or minus 2 mg/dl.
The last two steps are the ones the person notices. Neuroglycopenic symptoms began at 51 plus or minus 3 mg/dl and cognitive function began to deteriorate at 49 plus or minus 2. Two reviews put the same order in millimolar terms, and they agree to within a tenth. Falling insulin secretion appears at 4.5 to 4.6 mmol/l and counterregulatory hormones at 3.6 to 3.8 Cryer 1997 Cryer 1993. Symptoms arrive near 3.0 and cognitive dysfunction at 2.6 to 2.7.
Roughly 19 mg/dl separates the first hormone from the first mistake. The brain is defended on a forecast, and that gap between the alarm and the failure is the margin the whole arrangement exists to hold open.
The thresholds move with recent experience
These glycemic thresholds are dynamic rather than static, and they vary with recent antecedent glycemia Cryer 1997. Among 112 patients on long-term intensive insulin therapy, the group holding HbA1c at or below 5.5 percent showed high thresholds and impaired adrenaline responses Pampanelli 1996. That group was also unaware of hypoglycemia, with cognitive dysfunction delayed to the lowest plateau tested at 2.3 mmol/l.
Exposure explains the shift. Mild symptomatic episodes ran 54.5 per patient-year in that tightest group, against 33.7 and 20.4 per patient-year in the two groups with higher HbA1c, and episode frequency correlated with HbA1c at r = -0.82. Repeated lows teach the system to stop calling them low.
The warning gate and the failure gate can slide apart
The two gates are placed separately. A clamp took 8 patients with well controlled diabetes, 9 with poor control and 10 healthy people from 5.0 to 2.2 mmol/L. Thresholds for adrenergic symptoms and for the hormones epinephrine, norepinephrine, cortisol and growth hormone all sat lower in the well controlled group Widom 1990. Median thresholds for visual-spatial, visual-motor and global cognitive dysfunction did not differ.
The Unified Model of Tone reads that result as a staged defense whose outer gate has moved. The warning carries the adjustable setting, and repeated exposure slides it down toward the level at which cognition already fails. Nothing about the failure point moved. The danger sits entirely in the narrowing gap between the two.
That reading commits the model to something testable. If the warning and the failure were independent systems rather than two placements on one defended range, then restoring the first would leave the second untouched. An input that moved one would leave the other where it was. The evidence runs the other way. Awareness of hypoglycemia is restored by scrupulous avoidance of hypoglycemia, and that avoidance also at least partially restores the reduced epinephrine response Cryer 1997.
Complete avoidance makes the point cleanly. After pancreatic islet transplantation removes hypoglycemia altogether, counterregulatory responses and autonomic symptom generation come back together Rickels 2019. The model predicts that an input which lowers exposure moves the hormone threshold and the symptom threshold in the same direction, in patients whose thresholds start too low and in patients whose thresholds start too high, though compensation can hold one threshold while the other travels.
07Transport failure and dysglycemia
A neuron meets whatever crossed the carrier, which is why brain fuel fails at a normal blood sugar
One gene at the barrier can starve a brain surrounded by sugar. Patients with GLUT1 deficiency syndrome show infantile seizures, delayed development and acquired microcephaly alongside normal circulating blood glucose, low-to-normal cerebrospinal fluid lactate and persistently low cerebrospinal fluid glucose Seidner 1998. Hemizygosity of GLUT1 and nonsense mutations truncating the protein both produce it.
Clinical neurochemistry already had a name for that situation. It separates true fasting hypoglycemia from reactive postprandial hypoglycemia, and it calls the third case hypoglucocytosis, where blood readings look normal while glucose is not reaching the cell. GLUT1 deficiency is the genetic proof that the category is real.
The diagnostic numbers are specific. A cerebrospinal fluid glucose below 50 mg/dl, or a fluid to blood ratio below 0.60, points to the disorder and warrants sequencing SLC2A1 De Giorgis 2013. Published cutoffs sit a little apart, and a ratio below 0.5 is also reported as the diagnostic line Ramm-Pettersen 2014. The 0.60 figure is the trigger to sequence rather than the diagnosis itself. Two patients diagnosed in late childhood at ages 10 and 15 both carried ratios of 0.41 against a reference range of 0.65 plus or minus 0.1 Gramer 2012.
Carrier function is reduced rather than absent. Measured as labeled 3-O-methylglucose influx into red blood cells, mean uptake in 11 patients ran 53 percent of that in 30 controls Klepper 2003. Half a carrier is enough to produce epilepsy, a complex movement disorder and intellectual impairment while the venous glucose reads normal.
The same carrier can be blocked by things already circulating. Ethanol, diazepam and chloral hydrate each inhibited GLUT1-mediated transport in that work, while carbamazepine and phenytoin did not. What reaches the neuron depends on what else is competing for the door.
Chronic high glucose lowers the carrier's capacity
Persistent hyperglycemia turns the carrier down. In nonobese diabetic mice hyperglycemic for 4 to 6 days, maximal transport velocity across the blood-brain barrier measured 989 plus or minus 214 nmol per minute per gram Cornford 1995. The permeability surface area product fell against normoglycemic mice. The half-saturation constant held at 5.80 plus or minus 1.38 mmol/l.
The barrier lowered its throughput without changing what it recognizes. A high number in the blood therefore fails to guarantee a high number at the neuron. The interface rescales to its recent history, which is the mechanism behind a brain that is simultaneously bathed in glucose and short of it.
Dysglycemia shows up as brain hypometabolism
The rescaling is visible on imaging in people. Among 749 nondemented participants with a median age of 79.0 years, 20.6 percent had diabetes. Labeled glucose hypometabolism in the Alzheimer signature region appeared in 48.1 percent of them, against 28.9 percent of the rest Roberts 2014. The odds ratio was 2.28, with a 95 percent confidence interval of 1.56 to 3.33.
Amyloid did not follow. The retention ratio for labeled amyloid gave an odds ratio of 1.03 with a confidence interval spanning 0.71 to 1.51, so the fuel channel moved while the protein deposit stayed put. Dysglycemia is read here as a delivery problem with neurological consequences. The metabolic syndrome cluster of abdominal obesity, high triglycerides, low HDL and high fasting glucose names a pattern that bears on how well the brain is fed.
08Fuel switching and the range
The brain can run on ketones, and both ends of the glucose range injure it
Glucose is not the only substrate the brain accepts. Catheterization of cerebral vessels in three obese patients through 5 to 6 weeks of starvation showed beta-hydroxybutyrate and acetoacetate replacing glucose as the predominant fuel for brain metabolism Owen 1967. Fuel choice is a setting the body can change, given days.
That switch is why early diagnosis of GLUT1 deficiency matters. It allows prompt initiation of a ketogenic diet, which supplies a substrate that does not need the faulty carrier De Giorgis 2013. Compliance in this disorder runs better than in the other conditions treated the same way. Routing around a broken door is a different act from forcing more through it.
Both directions of the glucose range injure a developing brain. Operational definitions place neonatal hypoglycemia at 47 mg/dL or below and neonatal hyperglycemia above 150 to 180 mg/dL, and the relationship between dysglycemia and adverse neurological outcome is u-shaped Lagacé 2024. Glycemic lability has itself been recognized as a key factor in adverse neurodevelopmental outcomes.
Swinging across the range costs something on its own, which is the strongest argument against reading a single value. A fasting number reports one point on a curve. What the curve reports is the excursion after a load, the time taken to return, and whether the alarms fired at the level they were built for.
Fuel and delivery rank first among the neurochemical priorities because every other function is charged against them, and that ranking is set out on The Four Neurochemical Priorities. Transmitters, plasticity and tone all assume a reliable line behind them. When the line is steady the neuron is free to do its work, and when it wavers the whole hierarchy feels it from the bottom up.
A neuron never meets the number on the lab slip. It meets whatever crossed its carrier into its own cytoplasm.
09Tone
How this system expresses tone
Glucose keeps almost no reserve, so its whole margin sits in two places: how much carrier stands at the barrier, and how far below normal the alarm has been set.
Load
Holding resting potentials is charged against a brain pool of about 1.0 micromole per gram, a standing bill paid by delivery rather than by storage.
Prediction
Insulin secretion falls first, near 4.5 mmol/l, while glucose is still inside the physiological range and nothing has yet gone wrong.
Constraint
One faulty copy of the GLUT1 gene caps a whole brain, holding cerebrospinal fluid glucose near 0.41 of the blood value.
Coupling: active cortex raised glucose uptake 51 percent while oxygen consumption rose 5 percent, so glucose flow marks the synapse that just cleared transmitter. Time course: adrenal tyrosine hydroxylase mRNA rose three to five fold five hours after insulin-induced hypoglycemia in rats, which is how long resupply of the hormone takes to begin. Gain: four to six days of hyperglycemia in the nonobese diabetic mouse cut maximal transport to 989 nmol per minute per gram, leaving the half-saturation constant unchanged. Oscillation: glucose-excited and glucose-inhibited neurons move their firing rates in opposite directions across one rise from 0.1 to 2.5 mmol/l in rat slices. Input quality: ethanol, diazepam and chloral hydrate each inhibit GLUT1, so what reaches a neuron depends on what else is crossing the same carrier. Set point: counterregulation commits at 68 mg/dl in healthy volunteers while cognition holds to 49, which places the defended level about 19 mg/dl above the failure point.
10Across the library
How this page relates to the rest of the library
The neighboring pages that pick up this supply line at either end.
The other half of the delivery contract. Substrate with no oxygen buys almost nothing, and the brain holds no reserve of either against a shortfall.
Where the pyruvate goes. The citric acid cycle and the proton gradient turn the carbon delivered here into the ATP the membrane pumps spend.
The astrocyte side of the supply. Brain glycogen is stored there, and the lactate handoff that keeps an axon conducting without glucose is counted there.
Labeled glucose and blood-flow imaging treated as an instrument. That page carries the reference values, the cohorts scanned, and what the scans predicted about outcome.
Where fuel and delivery sit in the search order, and why a failure this far upstream produces many symptoms at once.
What happens when supply stops outright. Pumps lose their ATP, the membrane depolarizes, and calcium finishes the cell. That page also carries the isoelectric EEG evidence that damage tracks minutes of silence rather than the blood sugar.
11Frequently asked
Questions about this topic
Why does the brain need so much glucose?
The brain runs almost entirely on glucose and keeps essentially no store of it. Every impulse spends sodium and potassium gradients that ATP has to rebuild, and the same glucose carbon becomes glutamate, GABA, glycine and D-serine. Carbon-13 measurements put human citric acid cycle flux at 0.74 micromol per minute per gram, and glial recycling of glutamate accounts for about half of it. Fuel and message draw on one supply, so a shortfall costs signaling and synthesis at once, and the loss shows in speed before it shows in structure.
What are GLUT1 and GLUT3?
GLUT1 and GLUT3 are facilitative glucose carriers, and each serves a different cell. GLUT1 appears at high concentration in the blood-brain barrier as a heavily glycosylated 55 kilodalton form, and as a 45 kilodalton form in the brain parenchyma, predominantly glia. GLUT3 is the neuronal carrier and GLUT5 sits on microglia. Glucose crosses no brain membrane without one of them, so the density and kinetics of these proteins set how much sugar any given cell can actually receive. A carrier fault starves a brain at a normal blood sugar.
Can blood sugar be normal while the brain is short of glucose?
Yes, and GLUT1 deficiency syndrome is the published proof. Patients have infantile seizures, developmental delay and acquired microcephaly with normal circulating blood glucose and persistently low cerebrospinal fluid glucose. A cerebrospinal fluid glucose below 50 mg/dl, or a fluid to blood ratio below 0.60, points to the disorder. Two patients diagnosed at ages 10 and 15 carried ratios of 0.41 against a reference of 0.65. A neuron experiences whatever crossed its carrier, and a venous sample cannot report that. Transport is the fact the cell lives on.
What is hypoglycemia unawareness?
Hypoglycemia unawareness is the loss of warning symptoms before cognition fails, and it comes from repeated exposure to low glucose. Glycemic thresholds are dynamic rather than static and shift with recent antecedent glycemia. Among 112 patients on long-term intensive insulin therapy, those holding HbA1c at or below 5.5 percent showed high thresholds, blunted adrenaline responses and 54.5 mild episodes per patient-year. Avoiding hypoglycemia restores the warning, and awareness returns along with the epinephrine response. The alarm moved, and exposure is what moves it back.
How does the brain sense a falling blood sugar?
Dedicated glucose-sensing neurons in the ventromedial hypothalamus change their firing rate with glucose itself. Glucose-excited cells raise their action potential frequency as glucose rises from 0.1 to 2.5 mmol/l through the ATP-sensitive potassium channel, and glucose-inhibited cells lower theirs through a chloride channel. In rat slices only 14 percent were directly excited by glucose and 3 percent directly inhibited, while another third responded presynaptically. Most of the sensing is a network property rather than one meter. The answer then leaves through the splanchnic nerve to the adrenal medulla.
Does diabetes affect the brain's glucose use?
Yes, and the change is measurable on imaging. Among 749 nondemented adults with a median age of 79, glucose hypometabolism in the Alzheimer signature region appeared in 48.1 percent of diabetic participants against 28.9 percent of the rest, an odds ratio of 2.28. Amyloid retention did not differ, at an odds ratio of 1.03. In hyperglycemic mice the barrier itself lowered its maximal transport velocity while its half-saturation constant held. A high number in the blood therefore fails to guarantee a high number at the neuron.
Can the brain run on anything other than glucose?
It can run on ketone bodies. Catheterization of cerebral vessels in three obese patients through five to six weeks of starvation showed beta-hydroxybutyrate and acetoacetate replacing glucose as the predominant brain fuel. That switch is why a ketogenic diet is started promptly in GLUT1 deficiency syndrome, since the substrate crosses without the faulty carrier. Ketones supplement the supply rather than replace the machinery, because the sugar still supplies most transmitter carbon. Fasting shifts the fuel, and the shift takes days to arrive.
Does the brain use more glucose when you concentrate?
Yes, and it takes up more than it burns. During focal neural activity in humans glucose uptake rose 51 percent and blood flow rose 50 percent while oxygen consumption rose only 5 percent. At rest the brain consumes the two in a molar ratio near 4.1 to 1, so the extra sugar during activity is not being oxidized. Glutamate clearance by astrocytes is what routes it, working through the sodium-potassium pump. A glucose image therefore reports signaling traffic and its cleanup rather than oxidative demand.
12The sources
References
Sources: primary literature, linked inline.