INTRODUCTION — Cerebral injury (cerebral edema) is an uncommon but potentially devastating consequence of diabetic ketoacidosis (DKA). This complication is far more common among children with DKA than among adults. Children presenting with more severe DKA (higher blood urea nitrogen levels and more severe acidosis and hypocapnia) are at greatest risk [1-4]. These children are often younger and newly diagnosed with diabetes due to the greater difficulty detecting symptoms in these children, which leads to delayed diagnoses and more severe presentations. Symptoms of cerebral injury may be present prior to initiation of therapy or emerge during treatment. In most cases, symptoms become apparent within 12 hours after the initiation of therapy and, rarely, later than 24 hours [2,5-7].
The pathophysiology, diagnosis, and treatment of cerebral injury in children with DKA will be discussed here. The diagnosis and treatment of DKA in children are discussed separately. (See "Diabetic ketoacidosis in children: Clinical features and diagnosis" and "Diabetic ketoacidosis in children: Treatment and complications".)
INCIDENCE — Clinically significant cerebral injury occurs in 0.3 to 0.9 percent of episodes of DKA in children and has a mortality rate of 20 to 25 percent [1-3,8,9]. Overall mortality rates for DKA in children and adolescents range from 0.15 to 0.51 percent in national population studies in Canada, the United Kingdom, and the United States [1,5,10-13]; 50 to 80 percent of diabetes-related deaths in children are caused by cerebral injury [2,14].
Subclinical cerebral edema, as detected by ventricular narrowing on imaging studies or alterations in brain water distribution on magnetic resonance imaging (MRI), has been reported in the majority of children with DKA, even in the absence of neurologic signs or symptoms (image 1) [6,15,16]. In a study of 41 children with DKA, the width of the lateral ventricles (measured by MRI) was significantly smaller in patients during treatment for DKA than after recovery (mean width 9.3±0.3 versus 10.2±0.3 mm, respectively) [16]. Fifty-six percent of the children had ventricular narrowing during treatment. In this study, ventricular narrowing was more common in children who had subtle mental status abnormalities during DKA (abnormal Glasgow Coma Scale scores). In addition, subclinical cerebral edema detected by diffusion-weighted MRI is more severe in children who present with risk factors for clinically apparent cerebral injury (greater hyperventilation, higher blood urea nitrogen levels) [17].
Multiple lines of evidence suggest that subtle cerebral injury occurs commonly during DKA, and it is possible that clinically apparent cerebral injury may represent the most severe presentation of an otherwise common phenomenon. (See 'Outcome' below.)
PATHOPHYSIOLOGY
Initial theories based on osmotic change — The cause of DKA-related cerebral injury is not fully understood. In the 1980s, retrospective studies documented an increased frequency of cerebral injury in children who received intravenous fluids at higher rates [18,19]. It was therefore proposed that cerebral injury results from osmotic changes and fluid shifts during DKA. According to this theory, the hyperosmolar state during DKA causes brain cells to accumulate intracellular osmolytes. During DKA treatment, a decline in intravascular osmolality results in osmotically mediated movement of water into brain cells, causing cerebral edema and increased intracranial pressure and resulting in cerebral injury [18,20,21]. This theory was widely adopted, leading many pediatric DKA guidelines to recommend conservative rehydration strategies with slow intravenous infusions of isotonic fluids.
Over time, however, multiple lines of evidence have challenged the theory that cerebral injury is caused primarily by osmotic changes and fluid shifts:
●PECARN FLUID trial (2018) – The most definitive investigation of the relationship of intravenous fluid treatment to cerebral injury in children with DKA comes from the "PECARN FLUID" (Pediatric Emergency Care Applied Research Network, Fluid Therapies Under Investigation in DKA) trial [22]. This trial examined neurologic outcomes of DKA in 1389 children randomized to one of four intravenous fluid treatment regimens to compare rapid versus slow rehydration and use of 0.45 versus 0.9% sodium chloride (NaCl) fluids, using a 2 × 2 factorial design. There were no significant differences between arms in mental status during DKA treatment (percentage of patients with a significant decline in Glasgow Coma Scale scores or alterations in tests of working memory), nor in rates of clinically apparent cerebral injury. Contrary to previous hypotheses, analyses of the subgroup with more severe DKA (pH or partial pressure of carbon dioxide [pCO2] in the lowest quartiles of the study population) suggested that more rapid fluid infusion was associated with more rapid improvement in scores on working memory tests during DKA treatment. Tests of memory and intelligence quotient (IQ) two to six months after recovery from DKA showed no differences between the four fluid treatment arms. These data underscore the lack of causal association between intravenous fluid therapies and DKA-related cerebral injury, at least within the range used in this trial.
●Additional evidence against osmolar change as a cause of DKA-related cerebral injury – Early studies comparing children with DKA-related cerebral injury with those with uncomplicated DKA were limited because they were not adjusted for factors related to DKA severity. By contrast, subsequent larger studies adjusting for DKA severity (including degree of acidosis and dehydration) did not find associations between fluid infusion rates and cerebral injury [2]. In addition, properly controlled studies of risk factors for DKA-related cerebral injury have not found associations between higher initial glucose levels or higher osmolality and risk for cerebral injury [2,3]. Similarly, declines in glucose and osmolality during treatment have not been associated with risk of cerebral injury [2]. Although retrospective studies found associations between declines in serum sodium levels during DKA treatment and risk of cerebral injury, a prospective study found no such association [23]. In this study, the frequencies of mental status changes and clinical diagnoses of cerebral injury were no different in children who had declines in sodium levels during treatment versus those who did not. These data suggest that associations between serum sodium decline and cerebral injury in previous retrospective studies may have reflected delayed diagnoses of cerebral injury, leading to complications involving abnormal sodium and fluid regulation (eg, syndrome of inappropriate secretion of antidiuretic hormone [SIADH] or cerebral salt-wasting syndrome). Finally, case reports and case series document the occurrence of severe and even fatal cerebral injury or cerebral edema occurring before treatment of DKA, including reports of deaths at home in children with diabetes found to have cerebral injury or cerebral edema at autopsy (image 1) [5,7,24]. Studies demonstrate that at least 5 to 19 percent of clinically apparent DKA-related cerebral injuries occur before initiation of DKA treatment in the emergency department [1,2], suggesting that intravenous fluid therapy cannot be solely responsible for causing cerebral injury. A study found that children with DKA who have cerebral injury at the time of presentation to the emergency department have the same risk factors (high blood urea nitrogen concentrations and severe acidosis) as those who develop cerebral injury during DKA treatment [25]. These findings again suggest that factors intrinsic to DKA, rather than treatment-related factors, are responsible for cerebral injury.
Studies documenting imaging findings in children with severe DKA-related brain injuries also provide important information regarding causation. One large case series examined radiographic findings in children with profound neurologic alterations during DKA who were diagnosed with "cerebral edema" [26]. In this series, 39 percent had no evidence of edema on radiographic studies done at the time of neurologic decline but developed edema hours to days later. These findings suggest that edema may be a consequence of injury rather than the main cause, similar to edema that results from cellular energy failure during hypoxic-ischemic brain injuries. Studies using magnetic resonance imaging (MRI) to evaluate changes in brain water distribution during DKA also provide evidence to inform theories on causation. Diffusion-weighted MRI studies demonstrate that subclinical cerebral edema occurring during DKA in children is vasogenic, with increased accumulation of water in the extracellular space [27-29]. This finding is counter to what would be expected from edema caused by osmotic fluid shifts, where increased intracellular water content (cell swelling) would be anticipated, and suggests that abnormalities in blood-brain barrier function may occur during DKA [30,31].
Current theories of cerebral hypoperfusion and neuroinflammation — Several lines of evidence suggest that cerebral hypoperfusion may occur during DKA and that hypoperfusion/reperfusion may play a role in causing DKA-related cerebral injury.
●Fluctuation in cerebral blood flow – In a rat model, studies using MR spectroscopy to measure cerebral metabolism during DKA demonstrate elevated levels of lactate in the brain, as well as decreased high-energy phosphate levels and low N-acetylaspartate:creatine ratio, a marker of neuronal integrity (figure 1) [30]. In humans who died from DKA-related cerebral injury, autopsy results also show findings similar to hypoxic-ischemic injury [32-34]. In addition, clinical studies demonstrate that factors affecting cerebral blood flow during DKA (dehydration and hypocapnia) are most strongly associated with risks of both life-threatening and more subtle cerebral injury in children [2,17]. In a rat DKA model, cerebral blood flow is reduced and gradually rises during treatment with insulin and intravenous fluids [35].
Although there are no data documenting cerebral blood flow during untreated DKA (ie, before treatment commences) in humans, several studies have assessed cerebral blood flow during DKA treatment. These studies show elevated cerebral blood flow, along with increased extracellular water content, consistent with hyperemia and associated vasogenic cerebral edema [27-29]. Studies using near-infrared spectroscopy during DKA treatment show elevated regional cerebral oxygen saturation, suggesting that cerebral blood flow is elevated in excess of metabolic demand [28]. This abnormal elevation in cerebral blood flow may persist for hours after DKA resolution. Studies using transcranial Doppler ultrasound also document abnormalities in cerebral autoregulation during DKA [36,37].
●Inflammatory responses – Although alterations in cerebral blood flow appear to play a role in DKA-related cerebral injury, the cerebral blood flow reduction caused by dehydration and hypocapnia alone is unlikely to be sufficient to cause substantial ischemic injury. Studies suggest that inflammatory factors present during DKA may act synergistically with hypoperfusion/reperfusion to cause injury or may increase the vulnerability of the brain to ischemic injury. DKA elicits a marked systemic inflammatory response, and several inflammatory factors may play a role in causing neuroinflammation or affecting the function of the blood-brain barrier [38-41]:
•As an example, children with DKA have altered matrix metalloproteinases (MMP) levels and these are correlated with levels of various inflammatory mediators (CXCL8, IL1-beta), some of which also increase leukocyte adhesion to blood-brain barrier endothelia [42,43]. MMPs are endopeptidases that degrade tight junction proteins and endothelial basement membranes, allowing fluid and blood-borne proinflammatory/neurotoxic proteins to invade the central nervous system. MMPs have been implicated as mediators of blood-brain barrier dysfunction in many disease states characterized by systemic inflammation with associated cerebral injury. The observed patterns of change in MMPs during DKA are similar to those documented during cerebral ischemia/reperfusion in studies of stroke [42].
•Furthermore, studies using immunohistochemistry in a rat DKA model demonstrate a neuroinflammatory response involving reactive astrogliosis in the hippocampus and activation of microglia in most brain regions (picture 1) [44]. Microglia are inflammatory cells that respond to alterations in levels of cytokines; they are activated during various types of cerebral injury, including the secondary injury that occurs during reperfusion after ischemia [45,46]. These studies therefore suggest that various inflammatory pathways may participate in DKA-related cerebral injury.
•Finally, studies in a rat DKA model have shown that DKA increases levels of inflammatory mediators in brain tissue lysates. The most marked elevations were noted among chemokines involved in recruitment and activation of microglia (CCL3 and CCL5). Interestingly, this study also found evidence of persistent neuroinflammation long after recovery from DKA, suggesting that DKA might trigger a chronic neuroinflammatory response that could contribute to neurodegeneration over time in patients with diabetes [47].
The precise roles of neuroinflammation and altered cerebral perfusion in causing DKA-related cerebral injury are the subject of ongoing investigation.
CLINICAL PRESENTATION AND EVALUATION
Risk factors — Studies documenting risk factors for clinically apparent cerebral injury in children with DKA are useful in identifying children requiring more intensive monitoring [1-4]. Important risk factors for cerebral injury are:
●Severe acidosis at presentation of DKA [1,3,20].
●Increased blood urea nitrogen at presentation (which may represent a greater degree of hypovolemia) [1,2].
●A lower initial partial pressure of carbon dioxide (pCO2) [2,4].
●Young age (<5 years) and new onset of diabetes – These are not independent risk factors, but they are markers for more severe DKA because they are associated with delayed diagnosis of DKA.
Use of bicarbonate therapy for correction of acidosis in DKA is also associated with increased risk of cerebral injury [2,48] and is no longer recommended for pediatric DKA treatment, except under specific rare circumstances. (See "Diabetic ketoacidosis in children: Treatment and complications".)
Monitoring protocol
●Initial evaluation – All children presenting with DKA should have a rapid assessment of level of consciousness using the Glasgow Coma Scale (table 1B) or a similar neurologic evaluation, with particular attention to findings associated with cerebral injury (see 'Clinical criteria' below). Assessment of risk factors for cerebral injury is also helpful in determining the level of monitoring that is necessary. (See 'Risk factors' above.)
●Care setting – Patients with risk factors for cerebral injury (see 'Risk factors' above), abnormalities in mental status, or other findings concerning for cerebral injury (see 'Clinical criteria' below) should be managed in a specialized unit in which careful clinical and biochemical monitoring can occur (most commonly in a pediatric intensive care unit) because abrupt deterioration in neurologic status can occur and should be treated immediately.
●Monitoring – Neurologic monitoring using the Glasgow Coma Scale or other similar assessments should be repeated at least hourly during the first 12 to 24 hours of treatment (table 1B). More frequent assessments may be necessary for children with altered mental status.
DIAGNOSIS — The presence of any of the symptoms listed below should raise suspicion for the possibility of cerebral injury [26]. Altered mental status in children with DKA can be caused by a variety of factors other than cerebral injury, including acidosis, other metabolic derangements, and sleep deprivation. Nonetheless, clinicians should maintain a high level of suspicion for evidence of cerebral injury and intervene promptly if the diagnosis is suspected.
Clinical criteria — The following criteria have been proposed for clinical diagnosis of DKA-related cerebral injury (table 1A) (modified from [26]):
●Minor criteria (moderately suspicious findings):
•Headache – Although headache is frequently present at diagnosis of DKA, worsening or recurrence of headache during treatment is concerning
•Vomiting – Vomiting is suspicious if it develops or recurs during treatment
•Irritability, lethargy, or not easily aroused from sleep – These features are particularly suspicious if they occur or worsen after initiation of therapy
•Elevated blood pressure – Eg, diastolic blood pressure >90 mmHg
●Major criteria (very suspicious findings):
•Abnormal or deteriorating mental status after initiation of therapy, agitated behavior, or fluctuating level of consciousness
•Inappropriate slowing of heart rate – Eg, decline more than 20 beats per minute that is not attributable to improved intravascular volume or sleep state
•Incontinence inappropriate for age
●Diagnostic criteria (signs of significant brain injury, increased intracranial pressure, or brain herniation):
•Abnormal motor or verbal response to pain
•Decorticate or decerebrate posture
•Abnormal pupillary response or other cranial nerve (CN) palsy – Key steps are to evaluate extraocular movements (CN III, IV, and VI) and pupillary dilation and reactivity (CN II and III) (see "Detailed neurologic assessment of infants and children", section on 'Cranial nerves')
•Abnormal neurogenic respiratory pattern – Eg, grunting, abnormal tachypnea, Cheyne-Stokes respiration, apnea
In some cases, DKA-related cerebral injury may have other manifestations. Seizures have been described in some patients, while some other patients develop diabetes insipidus, manifested as a rapidly rising serum sodium level.
Provisional diagnosis and indications for treatment — We suggest immediate treatment for patients with any of the following, using the criteria above [26]:
●One diagnostic criterion
●Two major criteria
●One major and two minor criteria
●One major and one minor criterion (for children under five years of age)
(See 'Treatment' below.)
TREATMENT — Treatment for cerebral injury should be initiated as soon as the disorder is diagnosed, based on the clinical criteria outlined above. It is appropriate to treat promptly based on clinical suspicion because untreated cerebral injury has a high rate of mortality and morbidity, and treatment with a hyperosmolar agent (mannitol or hypertonic saline) may be beneficial. Evidence supporting the use of hyperosmolar treatments for cerebral injury consists of case reports and case series documenting rapid clinical improvement after administration of these agents in DKA-associated cerebral injury [49-52] and successful use in patients with cerebral edema due to other causes.
Although cerebral edema may be detectable on imaging studies in some children, edema is not uniformly present and may develop hours to days after neurologic decline [26]. For this reason, initial treatment decisions should not rely on findings from imaging studies [53]. Imaging studies should be considered after initiating treatment, depending on the clinical response, to detect cerebral herniation or impending herniation, cerebral infarction, hemorrhage, or thrombosis. All patients with suspected cerebral injury should be managed in an intensive care unit.
Treatment for cerebral injury consists of the following [54]:
●Immediately give mannitol 0.5 to 1 g/kg intravenously over 10 to 15 minutes. Mannitol administration can be repeated after 30 minutes. Effects of mannitol generally are rapidly apparent (within 15 minutes).
●Adjust fluid administration as indicated to maintain normal blood pressure and optimize cerebral perfusion. Avoid hypotension that might compromise cerebral perfusion pressure.
●Administer oxygen as needed to maintain normal oxygen saturation.
●Intubation may be necessary for patients with apnea or abnormal respirations or to protect the airway. Aggressive hyperventilation (ie, that brings the partial pressure of carbon dioxide [pCO2] below the level appropriate to the patient's level of acidosis) has been associated with poor outcomes of DKA-related cerebral injury and should be avoided unless absolutely necessary to treat elevated intracranial pressure [8].
●Monitoring of intracranial pressure may be necessary in selected patients.
The mechanism of action of mannitol in DKA-related cerebral injury is unclear. Improvements in mental status are often more rapidly evident than would be anticipated based on reductions in edema. These effects may result from improvements in blood viscosity and red blood cell deformability that have been described with mannitol treatment, with beneficial effects on cerebral blood flow [55,56].
Hypertonic (3%) saline (2.5 to 5 mL/kg over 10 to 15 minutes) can be used if initial treatment with mannitol does not result in improved mental status. For most patients, we suggest using hypertonic saline as a secondary intervention in patients with progressive symptoms and no response to mannitol. Some experts use hypertonic saline as a first-line treatment, especially for children who present with signs and symptoms of cerebral injury in the early phases of treatment, while they are still severely dehydrated and acidotic. One retrospective study found higher mortality rates in patients treated with hypertonic saline alone than in those treated with mannitol; however, these results may not reflect a causal association and should be interpreted with caution [12]. (See "Elevated intracranial pressure (ICP) in children: Management", section on 'Hyperosmolar therapy'.)
OUTCOME — The mortality rate among children with DKA who develop definite cerebral injury is approximately 20 to 25 percent. Among survivors, approximately 21 to 26 percent have permanent neurologic sequelae [8,14,57].
Risk factors for death or survival in a vegetative state were identified in a retrospective multicenter study of 61 children [8]:
●Elevated blood urea nitrogen at the time of initial presentation
●Intubation with aggressive hyperventilation (targeting a partial pressure of carbon dioxide [pCO2] of less than 22 mmHg)
●Severe neurologic compromise at diagnosis of cerebral injury (all patients who either died or survived in a persistent vegetative state had Glasgow Coma Scale scores ≤7 when cerebral injury was diagnosed)
In addition to clinically apparent episodes of cerebral injury, multiple lines of evidence suggest that subtle cerebral injury occurs commonly during DKA. It is possible that clinically apparent cerebral injury may represent the most severe presentation of an otherwise common phenomenon. As an example, studies demonstrate subtle alterations in cognitive functioning, including deficits in memory, attention, and verbal intelligence quotient (IQ), in children with type 1 diabetes who have had one or more DKA episodes compared with children with diabetes but without a history DKA [58-60]. In a large study involving 1134 children with diabetes, a single episode of DKA was associated with subtle deficits in memory shortly after diagnosis of diabetes [61]. Repeated DKA episodes over time were associated with sizable declines in IQ. In addition, studies using magnetic resonance imaging (MRI) document cerebral microstructural alterations in children after recovery from DKA [59,62], while other studies have found elevated levels of neuron-specific enolase, a marker of neuronal injury, in children with uncomplicated DKA, as well as cerebral metabolic alterations suggestive of hypoxia/ischemia [63-66].
SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Diabetes mellitus in children" and "Society guideline links: Hyperglycemic emergencies".)
SUMMARY AND RECOMMENDATIONS
●Incidence and outcomes – Cerebral injury (cerebral edema) occurs in 0.3 to 0.9 percent of cases of diabetic ketoacidosis (DKA) in children and is responsible for most DKA-related deaths. Among children with DKA who develop cerebral injury, the mortality rate is 20 to 25 percent and 21 to 26 percent of survivors have permanent neurologic sequelae. More subtle forms of cerebral injury occur commonly during DKA. A single episode of DKA may be associated with subtle deficits in memory, and multiple episodes of DKA are associated with cognitive declines. (See 'Incidence' above and 'Outcome' above.)
●Risk factors – Children who present with elevated blood urea nitrogen or more profound acidosis and hypocapnia are at greatest risk for cerebral injury. Younger children and those with new onset of diabetes are more likely to present with these risk factors because recognition of DKA is often delayed in these patients. (See 'Risk factors' above.)
●Clinical monitoring – Children undergoing DKA treatment should be monitored for signs and symptoms of cerebral injury (table 1A-B). (See 'Monitoring protocol' above and 'Clinical criteria' above.)
●Decision to treat – Changes detectable on cerebral imaging studies may occur late in the development of cerebral injury. Therefore, the decision to treat should be based on clinical evaluation (table 1A). Imaging may be useful to exclude other causes of neurologic deterioration but should not delay treatment. (See 'Treatment' above.)
●How to treat – Cerebral injury is associated with high mortality and morbidity and must be treated promptly with osmotic therapy when strongly suspected based on clinical criteria, even before imaging is obtained (table 1A) (see 'Treatment' above):
•We suggest treatment with mannitol (0.5 to 1 g/kg) rather than hypertonic saline (Grade 2C). Very limited evidence suggests that hypertonic saline may be associated with a higher mortality rate than mannitol.
If there is no clinical improvement in mental status, the same dose should be repeated after 30 minutes. It is also reasonable to use hypertonic saline (2.5 to 5 mL/kg over 10 to 15 minutes) if initial treatment with mannitol does not result in improved mental status.
•Intubation and mechanical ventilation may be required; however, aggressive hyperventilation (ie, that brings the partial pressure of carbon dioxide [pCO2] below the level appropriate to the patient's level of acidosis) should be avoided.
ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges George Jeha, MD, and Morey W Haymond, MD, who contributed to earlier versions of this topic review.
