Migraine, Brain Glucose Metabolism and the “Neuroenergetic” Hypothesis: A Scoping Review

Hypoglycemia was considered a precipitating factor in migraine headaches as far back as 1935. In the light of new supportive evidence, particularly studies investigating glucose homeostasis in migraine patients, recent research has brought the hypoglycemic hypothesis back into play. The evidence suggesting that the cerebral energy deficit may characterize people who are particularly susceptible to postprandial (or reactive) hypoglycemia, because of altered insulin sensitivity, will be detailed in the following paragraphs.

In 1998, Bonora et al. estimated that the prevalence of insulin resistance was of 65.9% for subjects with impaired glucose tolerance, 83.9% for non-insulin-dependent diabetes mellitus sufferers, 53.5% for people with hypercholesterolemia, 84.2% for those with hypertriglyceridemia, 88.1% in subjects with low HDL cholesterol, 62.8% in hyperuricemia and 58.0% in hypertension. These percentages are impressive and show how prevalent insulin resistance is in the general population. Indeed, about one third of American adults have impaired glycemic homeostasis^, and young people are not spared either. This can be seen in a 2006 population-based study on non-diabetic US adolescents (12-19 years old), which reported that about 13% of them were insulin resistant. Postprandial hypoglycemia in insulin resistant subjects is associated with impaired first phase glucose-stimulated insulin response and a compensatory increased late insulin response.^,, Half a century ago, it was hypothesized that an increase in insulin sensitivity might also underlie postprandial hypoglycaemia, and this was later demonstrated with the hyperinsulinemic normoglycemic glucose clamp technique. Nowadays, it is scientific knowledge that subjects with high insulin sensitivity have an increased glucose absorption which leads to postprandial hypoglycaemia.^, Moreover, this altered insulin sensitivity leads to postprandial hypoglycemia, which, in turn, causes increased food intake and weight^,, – in particular if high insulin sensitivity is combined with a high acute insulin response to glucose (high AIRg)^,, – and may represent an event preceding insulin resistance.^, Postprandial hypoglycemia is not uncommon, both in the general population and in those with diseases that alter the glucose metabolism. Table 1 reports some significant studies on the incidence of postprandial hypoglycemia and shows that a 2-h OGTT does not suffice to fully understand the real prevalence of those suffering from postprandial hypoglycemia, but that a 4- or 5-hour (OGTT) may be required.^,,

As described in Table 1, three studies carried out OGTTs and observed that subjects with diseases characterized by insulin resistance, ie, polycystic ovary syndrome and obesity, had high incidence of reactive hypoglycemia: 32%, 50% and 64%. Noteworthy, was the high incidence of postprandial hypoglycemia reported also in young, normal weight (on average) subjects, without a diagnosis of abnormal glucose metabolism: 24.4%, 23.8%, 10% and 18.8%. Other studies reported in Table 1^,, carried out OGTTs to identify reactive hypoglycemia in thousands of diabetes-free subjects or subjects with normal glucose tolerance.

No association was observed between reactive hypoglycemia at the 2-h OGTT and insulin resistance. Conversely, a glycemia of < 3.3 mmol/l detected at the 2-h OGTT, was associated with a younger age, higher insulin sensitivity and a lower body mass index. This is further evidence supporting that both high insulin sensitivity and insulin resistance may lead to the development of postprandial hypoglycemia. Moreover, high insulin sensitivity appears to be the most frequent cause of postprandial hypoglycemia, being probably implicated in 50 to 70% of all cases.^,

Clinical evidence suggests that migraine can, to a large extent, be generated by postprandial hypoglycemia. Indeed, data from clinical practice has taught us that the most frequent triggering factor reported by migraineurs is fasting and that migraine is more likely to occur in susceptible persons when there is insulin resistance.^, Hockaday et al. carried out a study where 50g of glucose was given to 10 migraineurs whose attacks were associated with fasting, after a 10-hour fast. A total of 6 of 10 of them had a migraine attack within 8 hours of the glucose test, as the hyperglycemic effect of cortisol requires a few hours to appear. Luyckx et al. observed that 30 of 47 patients who had reactive hypoglycemia had reported signs of a so-called “neuroglycopenia” occurring from 2 to 4 hours after a meal in their everyday life. They had signs and symptoms which included weakness, faintness, headache, irritability, anxiety, nervousness, palpitations, inward trembling, vertigo, hunger, and syncope. In another study, 74 migraineurs who reported that fasting had triggered their attacks, had a 5-h OGTT with 100g of glucose. A curve, consistent with reactive hypoglycemia values, was observed in 56 of 74 (76%) of them. Wilkinson reported that 11 of 13 (85%) subjects, who seemed to have ‘headaches of a migrainoid nature’ had induced headaches during a 5-hour OGTT. The headache began 3-4 h into the test when the glucose level dropped to its lowest (ie, < 3.3 mmol/L). A serum glucose level below 3.3 mmol/L within a few hours of glucose ingestion is considered to be a sign of reactive hypoglycemia. A review of international literature indicates that, more often than not, there are two main specific dietary factors, ie, fasting and a relatively mild reactive hypoglycemia, which may follow large (ie, 100 g) carbohydrate ingestions, which induce migraine in sufferers and more generalized headaches in the general population.

The classical signs of hypoglycemia include: blurred vision, headache, confusion, depression, tremors, anxiety, hunger, palpitations, sweating, nausea, dizziness and weakness.^,, Similar symptoms have been observed in subjects with insulinoma,^,,,, a rare condition that can lead to postprandial hypoglycemia. The patients enrolled into these studies complained of typical hypoglycemia symptoms, such as dizziness, sweating, confusion, irritability and blurred vision, 2 hours after a meal. All the symptoms reported by people with insulinoma may be attributed to an insufficient supply of glucose to the brain. Their comorbidities were epilepsy, anxiety, depression and, interestingly, migraine. The hypoglycemia symptoms match most of the non headache symptoms of migraine, including tiredness and/or weariness, difficulty in concentration, blurred vision, light sensitivity, intolerance and/or irritability, hunger and/or food craving and dizziness. Moreover, estro-progestinic drugs induce hyperinsulinism and hypoglycemia, which might explain the frequent worsening of migraine in patients on these drugs. In line with this, Granella et al. reported a more severe migraine in 25% of patients without aura and in 56% of those with aura on estro-progestinic drugs.

Intriguingly, an experimental study demonstrated that administration of insulin as well as glucagon, a peptide hormone produced by alpha cells of the pancreas that counteracts insulin actions, significantly modulate the neuronal firing in the trigeminocervical-complex, a key structure in the pathogenesis of the migraine attack. This suggests that there is a potential neurobiological link between migraine and altered glucose homeostasis.

Biochemical studies highlight similarities in the metabolism observed during a migraine attack and a hypoglycemic state. Indeed, it was observed that the levels of free fatty acid, ketone bodies, glycerol and cortisol, were increased in the venous blood samples of migraineurs during an attack. The same metabolic pattern was observed during fasting or glucoprivation, in the general population. Several studies reported a higher frequency of altered insulin sensitivity in both episodic migraine (EM) and chronic migraine (CM). Bhoi et al. observed that insulin resistance correlated with the duration of migraine attacks. A case-control study identified a significant insulin resistance prevalence in CM with a three-fold higher probability of having insulin resistance than the EM group, where an insulin resistance prevalence similar to that of the control group was observed. This association remained constant also after adjustment for the confounding variables commonly associated with a higher insulin resistance status.

There is an increasing amount of data on insulin and brain insulin resistance, which evidence important features of migraine, dementia and other neurodegenerative disorders.^, Under physiological conditions, the regulatory mechanisms in the blood-brain barrier, astrocytes and neurons provide an efficient supply of energy during neuronal activation.

Current literature reports that the human brain is an insulin sensitive organ and, as such, may become insulin resistant.^, Indeed, some researchers suggests that peripheral insulin resistance can extend to the brain, triggering brain insulin resistance.^,,

Similarly, to the mechanism that takes place in peripheral insulin resistance, brain insulin resistance occurs when the brain cells fail to respond to insulin. The following paragraphs will focus on the main metabolic pathways involved in brain glucose homeostasis which may be altered by insulin resistance and, consequently, fail to provide an adequate energy supply during neuronal activation (Fig 1).

It has been demonstrated that insulin receptors and the related insulin signaling cascade are pivotal factors in brain metabolism, both in neurons and astrocytes. The insulin receptor (IR) in mammals occurs in two isoforms, IR-A and IR-B, which are expressed in different relative proportions in various organs and tissues. Moreover, it has been observed that their expression varies during development, aging and disease states. On binding to IR-A, insulin triggers the classical mitogenic signaling cascade (non-metabolic effects), whilst if it binds to IR-B it activates the metabolic phenotype pathway.^,,

The IR‐B/IR‐A mRNA ratio is predominant in human tissues like the liver, adipose tissue, skeletal muscle and kidney and is associated with the metabolic effects of insulin. Conversely, insulin acts as a mitogenic agent in fetal and cancer tissues where, the IR‐A/IR‐B mRNA ratio prevails.^,, Although the expression of both IR-A and IR-B in human astrocytes had been previously described, Spencer et al. used an innovative, investigative method (in situ RT-PCR/ FISH assay) and was the first to demonstrate that both IR-A and IR-B are expressed in the neurons of the adult human frontal cortex brain tissue. These IR-A and IR-B receptors have distinct activation and regulation mechanisms. There is a downregulation of IR-B in chronically high levels of insulin, without it affecting the brain IR-A. Indeed, animal studies have demonstrated that an increased IR-A/IR-B ratio is related to insulin resistance and glucose intolerance in mice. Moreover, IR-A is favored in diabetic or pre-diabetic subjects in a state of peripheral hyperinsulinism, whilst IR-B metabolic processes are reduced.

Alterations in insulin receptor signaling have been associated with dementia and, interestingly, it has been reported that an IR-B analogue, discovered in C. elegans, is involved in learning and memory. Therefore, as postulated by Spencer et al., IR-A and IR-B may have distinct functions in neurons.

It has been suggested that brain mitochondrial dysfunction takes place in association with brain insulin resistance and recent evidence in support of a mitochondrial dysfunction in migraine has also been reported. Neuroimaging studies have demonstrated a lower adenosine triphosphate (ATP) and ‘mitochondrial phosphorylation potential’ in the brain of migraineurs without aura during the interictal period, than in controls. Interestingly, the lowest ATP concentrations were observed in the most severely affected migraine patients. Several studies have described that the lowest ATP concentrations in the migraineurs’ brains, evaluated by phosphorus magnetic resonance spectroscopy and compared to controls, are associated with a reduced glucose metabolism in the parietal, temporal and frontal lobes, as reported in Table 2.^,,,,, However, further research is required to elucidate whether brain insulin resistance could be implicated in brain mitochondrial dysfunction.

Specific areas of the brain, like the Brodmann areas 10 and 47, seem to suffer from glucose hypometabolism, both in subjects with insulin resistance and those with chronic migraine (Table 2). This evidence strengthens our hypothesis that brain insulin resistance, stemming from peripheral insulin resistance extending to the brain and impairing a correct astrocytes and/or neurons glucidic metabolism, may well trigger the neuronal cell stress implicated in migraine chronification. This hypothesis is supported by other experimental evidence. Firstly, GLUT4 is mainly expressed by the cerebral areas that regulate memory, learning, emotional and cognitive functions, ie, the hippocampus, the amygdala and the cerebral cortex.^,, This suggests that the insulin signaling pathway may well play a key role in the utilization of glucose in these areas. Noteworthy is the fact that all these areas are affected both in subjects with insulin resistance and those with migraine (Table 2). Moreover, it has been observed in rats that insulin activation of GLUT4 improves glucose flux into neurons during periods of high metabolic demand, like during learning or other cognitive tasks.^,,

Therefore, we hypothesize that, if this increased glucose demand is not satisfied – in subjects with episodic migraine partly due to postprandial hypoglycemia, and in subjects with chronic migraine partly due to brain insulin resistance – and if the brain is not able to use ketone bodies efficiently, as should happen during fasting or carbohydrate restriction (conditions unlikely in Western countries with a carbohydrate-laden diet), then this would lead to an energy deficit, which would, in turn, trigger a migraine attack.