Review Article

Cerebral edema in diabetic

Daniel L. Levin, MD, FAAP, FCCM, FACC

Objective: To review the causes of in diabetic matter. Of the 194 references, there were 21 preclinical and 40 ketoacidosis (CEDKA), including pathophysiology, risk factors, clinical studies, 35 reviews, 15 editorials, 43 case reports, 29 letters, and proposed mechanisms, to review the diagnosis, treatment, three abstracts, six commentaries, and two book chapters. and prognosis of CEDKA and the treatment of diabetic ketoaci- Data Synthesis: The data are summarized in discussion. dosis as it pertains to prevention of cerebral edema. Conclusions: The causes and mechanisms of CEDKA are un- Data Source: A MEDLINE search using OVID was done through known. CEDKA may be due as much to individual biological 2006 using the search terms cerebral edema and diabetic keto- variance as to severity of underlying metabolic derangement of . the child’s state and/or treatment risk factors. Treatment recom- Results of Search: There were 191 citations identified, of which mendations for CEDKA and are made taking 150 were used. An additional 42 references listed in publications into consideration possible mechanisms and risk factors but are thus identified were also reviewed, and two book chapters were intended as general guidelines only in view of the absence of used. conclusive evidence. (Pediatr Crit Care Med 2008; 9:320–329) Study Selection: The citations were reviewed by the author. All KEY WORDS: diabetic ketoacidosis; brain edema; fluid therapy; citations identified were used except 25 in foreign languages and insulin; child; risk factors 16 that were duplicates or had inappropriate titles and/or subject

erebral edema in diabetic ke- many interesting but unsubstantiated the literature and almost 40 yrs of clinical toacidosis (CEDKA) has been and conflicting theories. Third, it is in- experience to discuss recommendations identified for 70 yrs and has herently difficult to study in vivo metab- for treatment of DKA to prevent symp- been a subject of much inves- olism and cellular integrity in the human tomatic CE and for treatment of symp- tigationC and debate during the 40 yrs brain. Recently, several technical devel- tomatic CEDKA once it occurs. since the inception of pediatric critical opments, especially magnetic resonance care medicine. Although many risk fac- imaging (MRI) technology, have im- Pathophysiology and Risk tors of both diabetic ketoacidosis (DKA) proved knowledge in this area. Fourth, Factors and its treatment have been identified the pathogenesis may be complex and Most evidence supporting different and have led to many proposed patho- multifactorial and may proceed in stages, risk factors is based on underpowered, physiologic mechanisms, there is no gen- making it impossible to understand by retrospective, noncontrolled, and/or non- eral agreement concerning the risk fac- unifocal analysis. tors, pathophysiology, and mechanisms randomized clinical studies. Several in- Symptomatic cerebral edema (CE) in teresting animal studies are also re- underlying CEDKA, for several reasons. pediatric patients with DKA is uncom- First, variable definitions of symptomatic viewed, and I will specify risk factors of mon and is defined by diagnostic criteria, the disease state itself vs. risk factors as- cerebral edema have been used. Second, including abnormal motor or verbal re- there is an absence of adequately pow- sociated with treatment. sponses to pain, decorticate posture, and Hypoxia and Ischemia. Dillon et al. ered, prospective, controlled, randomized abnormal neurogenic respiratory pat- clinical trials. Most reports are either (2) proposed a mechanism based on risk terns. Major, but not diagnostic criteria factors having to do with the disease small, retrospective, uncontrolled, and/or include altered mentation, sustained nonrandomized and have resulted in state: a reduced blood volume due to de- heart rate decelerations, and age-inap- hydration aggravated by a low PaCO2, at- propriate incontinence. Minor criteria in- tributable to acidosis and hyperventila- clude vomiting, headache, lethargy, dia- tion, all leading to vasoconstriction and From Dartmouth Medical School, Lebanon, NH. stolic blood pressure Ͼ90 mm Hg, and resulting in cerebral ischemia, hypoxia, Supported, in part, by the Susan J. Epply Quasi- age Ͻ5 yrs (1). Subclinical CEDKA is Endowment Fund, Pediatric Critical Care, Children’s and increased capillary permeability as Hospital at Dartmouth. usually defined by imaging studies and is the cause of CEDKA. Increased whole The author has not disclosed any potential con- probably common. Most of the deaths in blood viscosity (3) and arrest secondary flicts of interest. pediatric patients with type 1 diabetes mel- to abnormalities (4) have also For information regarding this article, E-mail: litus (DM) are due to symptomatic CE. [email protected] been proposed as causes of ischemia. Ce- Copyright © 2008 by the Society of Critical Care Here I review the risk factors for the rebral hypoxia and ischemia are sup- Medicine and the World Federation of Pediatric Inten- disease state and its treatment as well as ported by the work of Glaser and col- sive and Critical Care Societies the proposed pathophysiologic mecha- leagues (5–7), who found significantly DOI: 10.1097/PCC.0b013e31816c7082 nisms, using a combination of a review of higher serum urea nitrogen and lower

320 Pediatr Crit Care Med 2008 Vol. 9, No. 3 PaCO2 levels in CEDKA patients compared volume due to vasodilation and an in- regulation of release, sodium with matched controls. Using the same crease in brain tissue volume due to CE. bicarbonate administration, and intracel- database, Marcin et al. (8) found a poorer Glaser et al. (56) also pointed out that CT lular as well as extracellular fluid accu- outcome due to disease state risk factors scans cannot distinguish between vaso- mulation. in patients with greater neurologic de- genic edema due to blood-brain barrier The observation has been made (11) pression at the time of diagnosis of CE endothelial cell injury secondary to isch- that the role of fluid in causing symptom- and high initial serum urea nitrogen. emia and/or hypoxia and cytotoxic edema atic CEDKA is puzzling in that the inci- Lawrence et al. (9) found some support due to brain cell swelling secondary to dence has stayed the same even though for this theory with lower initial bicar- hyperosmolality. much more conservative fluid manage- bonate and higher initial blood urea ni- Recent MRI studies have more effec- ment is now practiced to prevent exces- trogen levels in patients with CE. Dunger tively distinguished these different sites sive free administration and lack of and Edge (10) and Edge (11) emphasized of injury and therefore mechanisms of serum sodium increase during therapy that evidence supporting this theory is edema and have indicated that edema (58, 59). Although that may be true, I find not completely convincing and there are may be more vasogenic (blood-brain bar- that many patients referred to our unit contradictory pieces of evidence as well. rier) than cytotoxic (cellular) in origin. are initially managed in small local clin- Hyperosmolar State and Fluid Admin- Figueroa et al. (50) performed MRI stud- ics and emergency rooms frequently by istration. Several authors emphasize the ies on four patients on admission, at 6–8 physicians trained in adult emergency importance of the risk of disease state due hrs, and 120 hrs after therapy began. medicine, and these patients still receive to the hyperosmolar state in long- These studies indicated that CE was generous amounts of fluid, hypotonic flu- standing hyperglycemia in these patients present in asymptomatic patients and be- ids, or both. (10–22) as well as in experimental mod- fore therapy was started. Edema was va- Insulin Dosage. In addition to initiat- els in rats (23–25), dogs (26–28), and sogenic in nature, indicating a perturba- ing fluid shifts due to relatively increased rabbits (29). The theory is due to the tion of the endothelial cells of the blood- intracellular central nervous system os- disease state; there is prolonged hyperos- brain barrier, which is a microcapillary molality after the administration of insu- molarity in the serum, and brain cells abnormality. Injury may be mediated by lin and a subsequent decrease in serum protect their volume status by producing an inflammatory process as well as met- osmolality (29, 60), insulin may have intracellular osmoles via production of abolic abnormalities and oxidative stress. other effects that cause CE. Using a rat metabolic products thought to be primar- Glaser et al. (56), also using MRI technol- model, Johanson and Murphy (61) stim- ily taurine (23, 24, 30) and myoinositol ogy, found vasogenic edema rather than ulated increased choroid cell concentra- (14, 30, 31). These osmoles dissipate from osmotic swelling, and Glaser (57) indi- tion of sodium with insulin, and this was the intracellular space slowly after serum cated that this may be due to hypoperfu- blocked by acetazolamide, a standard in- osmolality begins to decrease, taking sion and ketone bodies affecting the hibitor of cerebral spinal fluid produc- hours to days to return to normal due to blood-brain barrier. Those authors also tion. In a rat model in which DKA was the characteristics of the organic os- suggested that increases in metabolites, induced by streptozotocin, Tornheim molyte efflux channels and the down- such as taurine and myoinositol, found in (62) found an increase in brain edema regulation of cotransporters, which can some studies may be subsequent to cen- with aggressive treatment with fluid and take 16 hrs to begin to act after abrupt tral nervous system cellular injury rather insulin but not with fluid alone. This sup- reduction of extracellular osmolality in than the original cause of osmotically in- ported Arieff and Kleeman’s (29) findings rat cortical astrocytes (30). These condi- duced CE. in a hyperglycemic rabbit model, in tions favor movement of water from the When we debate 1) whether CE is which edema did not occur with reduc- serum into brain cells. Therefore, risks present in all patients with DKA; 2) tion of serum glucose by peritoneal dial- are also associated with therapy with in- whether CE (when present) is vasogenic ysis but edema did form when hypergly- sulin, which lowers blood glucose (18, in nature, indicating blood-brain barrier cemia was reduced by insulin. 32–34) and stimulates the sodium- endothelial cell injury due to ischemia, Blood Glucose Concentration. Investi- hydrogen exchange mechanism, increas- hypoxia, inflammation, metabolic de- gators have tried to correlate the pres- ing intracellular sodium concentration rangement, or oxidative stress; or 3) ence of CE with the rate of decrease in (35, 36) and fluid (especially hypotonic whether CE is cytotoxic in nature, indi- blood glucose (12, 63). As indicated in the fluid) (12, 37–46) given to repair dehy- cating the importance of the hyperosmo- hyperosmolar theory, a decrease in glu- dration. Both of these interventions can lar state and subsequent therapy with in- cose could be related to a decrease in cause a rapid decrease in serum osmola- sulin and fluids, it is important to realize serum osmolality, causing fluid shifts lity and favor movement of water into no one factor may be solely responsible into brain cells, and then insulin admin- brain cells, although the causal relation for the phenomenon and that several fac- istration would be a treatment risk factor. of fluid administration to CEDKA is de- tors may be important in a progressive Not all studies have found a tight corre- bated (5, 47, 48). sequence. Levitsky (44) pointed out that lation for this factor. Several authors doubt the hyperosmo- the initial response of the central nervous Sodium Bicarbonate Administration. lality and treatment explanation, point- system to the metabolic derangements of Several authors have implicated the ad- ing to the presence of CE before therapy the disease state may be vasogenic in ministration of sodium bicarbonate to pa- in some patients as a major contradiction nature, with blood-brain barrier pertur- tients as a treatment risk factor in the (5, 9, 49–55). Levitsky (44) reminded us bation occurring along with later devel- development of CE (5, 64, 65). The risk is that head ultrasonography and computed opment of symptomatic CE with cyto- presumably due to either decreased oxy- tomography (CT) scans cannot distin- toxic edema due to other secondary gen delivery to the brain secondary to a guish between an increase in brain blood insults, such as fluid administration, dys- shift in the oxygen dissociation curve (43)

Pediatr Crit Care Med 2008 Vol. 9, No. 3 321 resulting in hypoxia, paradoxic cerebral CEDKA. Cerebral vascular dysregulation uncontrolled DM in adults (98) and has spinal fluid acidosis (66–70), or both. may be mediated by an increase in pros- been postulated as causative in pediatric

Several authors have stated that sodium taglandin I2 produced from adipose tis- patients as well (99). This promotes fluid bicarbonate is dangerous and should not sue, as found in a DKA rat model (81). retention, but the osmotic diuresis in be used (5, 71, 72) with the possible ex- Several authors (82–85) indicate that DKA results in a balanced net effect. Once ception of severe myocardial depression the risk of CE seems to be related to therapy with insulin is begun and the (64) due to acidosis. It is controversial, severity of acidosis and that this is related glycosuria is stopped, vasopressin level is however, whether severe acidosis causes to 1) the severity of metabolic derange- still elevated and could cause water intox- myocardial depression (64, 73) in DKA ment causing CE; 2) treatment of DKA ication. Vasopressin is required for opti- patients. Others dispute this prohibition risk factors, such as receiving sodium bi- mal cell size in the kidney, but it is un- (11, 74) and do not find convincing evi- carbonate; or 3) another risk factor of known whether this occurs in the brain. dence of sodium bicarbonate as a treat- treatment of CE, such as acute alteration There is no known correlation between ment risk factor for CE. of pH and PaCO2 with intubation and ven- vasopressin levels and the degree of brain Intubation and Hyperventilation. tilation. edema. Marcin et al. (8) reported that intubation Although many disease state risk fac- Atrial natriuretic factor, which pro- and hyperventilation as treatment for tors, as well as treatment risk factors, duces natriuresis and excretion of volume symptomatic CE were an independent have been postulated to be associated via the kidney, is reported to be sup- risk factor for a worse outcome. This is with or causative in the development of pressed in the volume-depleted state in consistent with the theory that low PaCO2 CEDKA, it is not known whether any of adults with uncontrolled DM (11). The levels cause vasoconstriction and isch- them are responsible. Several authors in- initial low level in children is reported to emic injury to the blood-brain barrier, dicate that the risk of developing CEDKA increase during 24 hrs with therapy. It is resulting in vasogenic edema. Tasker et may be due not to treatment factors at all possible that this results in elevated atrial al. (75) argued that patients who are in- but rather to the severity of metabolic natriuretic factor levels, causing natri- tubated and ventilated for symptomatic derangement of the disease state (9–11, uresis and and favoring

CE need to be ventilated at PaCO2 levels 40, 74, 86–89) and that development of water movement into brain cells after ini- lower than the normal range, because CE is unpredictable (90–92). In that re- tiation of therapy for DKA. ventilation at normal PaCO2 levels repre- spect, earlier recognition and treatment Sodium/Proton Antiporter and Other sents a sudden increase in patients’ PaCO2 of new-onset disease and better manage- Membrane Cotransporters. Sodium- from their spontaneous hyperventilation ment and compliance in known DM pa- hydrogen exchange mechanisms at the state during DKA and has detrimental tients are probably the best ways to pre- cell membrane are important in regulat- effects. Fortune (76), Marcin et al. (77), vent CE (21, 93–95). ing intracellular ion concentration and and Ackerman (78) seem to agree with therefore cell volume in many cells, in- this, suggesting that patients intubated Proposed Mechanisms cluding the brain (30, 100). Sodium- and ventilated for CE should be ventilated of CEDKA bicarbonate cotransporters (101, 102) to PaCO2 levels representing their base- and sodium-potassium ion exchange line state at the onset of symptoms to Accumulation of Intracellular Os- channels (103, 104) are important in this avoid either too high or too low PaCO2 moles. As indicated previously, several role as well. These channels are affected levels. Ventilation at PaCO2 levels that lines of evidence in humans (10, 12–22, by many perturbations, including acido- may either cause further cerebral vaso- 58, 59), as well as models of hyperglyce- sis (100, 105), ischemia (105), insulin constriction and aggravate ischemia or mia or DKA in rat (23–25), dog (27, 28), (35, 76, 103), vasopressin (98), and even cause cerebral vasodilation and aggravate and rabbit (29), implicate generation of glucose (106) and ketosis (97). Incuba- increased (ICP) may intracellular osmoles (“idiogenic”) to tion of rat vascular smooth muscle cells be complicated by altered cerebral auto- protect brain cell volume when serum is in a high-glucose medium for 24 hrs in- regulation in DKA patients. Hoffman hyperosmolar, as in hyperglycemia, as creases sodium-hydrogen exchange chan- (79), using transcranial Doppler ultra- the cause of CEDKA. These intracellular nel activity, probably via phosphocreatine sound in five patients, found an increased osmoles are now thought to be taurine kinase activation (106). All of these pro- pulsatility index, suggesting an increased (23, 24, 30) and myoinositol (14, 30, 31). mote an increase in intracellular ionic ICP due to the existence of cerebral va- Acetoacetate (96) and ketones (97) may concentrations in brain cells, resulting in soparalysis based on low values of cere- be implicated in the intracellular hyper- glial swelling as well as alteration in en- bral vascular reactivity before treatment tonicity in these patients as well. These dothelial cells of the blood-brain barrier and 6 hrs after initiation of treatment. osmoles dissipate over 12 to 24 hrs after favoring vasogenic edema. Lam et al. There was a return of vascular tone at 24 initiation of fluid and insulin therapy to (105) were able to decrease CE formation hrs with a complete reversal by 48 hrs. correct and hyperglycemia in a streptozotocin-induced DKA rat Roberts et al. (80), also using a transcra- (30). Due to the slow dissipation of these model as determined by MRI scan when nial Doppler, determined that cerebral intracellular osmolytes, when serum os- giving the sodium-potassium-chloride autoregulation was impaired in five of six molarity decreases rapidly, there is a rel- cotransporter inhibitor bumetanide. In patients at 6 hrs and normalized by 36 ative hyperosmolarity in brain cells, fa- addition to the factors listed previously, hrs. There was no ischemia, with all pa- voring a shift of fluid into brain cells. use of sodium bicarbonate would buffer tients having normal to increased cere- This seems to require the presence of protons released from cells into the ex- bral blood flow and increased cerebral insulin (29, 33, 62). tracellular space, causing further move- oxygenation. Their findings favor a vaso- Role of Vasopressin and Atrial Natri- ment of protons out of brain cells and genic mechanism for the formation of uretic Factor. Vasopressin is elevated in therefore sodium transport into cells,

322 Pediatr Crit Care Med 2008 Vol. 9, No. 3 thus promoting more water movement perturbation of the endothelial cells of Table 1. Bedside evaluation of neurological state into cells. the blood-brain barrier are also consis- of children with diabetic ketoacidosis Hypoxia and Ischemia. In addition to tent with an increase in proinflammatory Diagnostic criteriaa the clinical risk factors already cited (2, cytokines (14, 31, 44, 50, 56, 57). A gen- Abnormal motor or verbal response to pain 5), activation of excitatory amino acid eralized inflammatory response to meta- Decorticate or decerebrate posture receptors (N-methyl-D-aspartate) during bolic factors, such as hyperglycemia and Cranial nerve palsy (especially III, IV, and VI) postischemic, peri-infarct depolarizations acidosis, is also consistent with the clin- Abnormal neurogenic respiratory pattern is implicated in the mechanism of CE by ical findings of pulmonary edema con- (e.g. grunting, tachypnea, Cheyne-Stokes which the ischemic penumbra is re- comitantly found in some patients with respiration, apneusis) Major criteria cruited into the core of cerebrovascular CE (114–119). Altered mentation/fluctuating level of stroke. Hypoxia increases glutamate, Aquaphorin Channels. Aquaphorin consciousness which causes an ion flux and cell swelling channels in glial cells transport water Sustained heart rate deceleration (decline following activation of N-methyl-D- from the extracellular to the intracellular more than 20 beats per minute) not aspartate receptors. This could occur in space and demonstrate a compensatory attributable to improved intravascular hypoxia associated with DKA as well (11). increase in expression in a streptozotocin volume or sleep state Age-inappropriate incontinence Ketones and Acidosis: Initiation of the DKA rat model (81). In an aquaphorin-4 Minor criteria Proinflammatory Cytokine Cascade. As knockout or depletion model in mice, Vomiting mentioned previously, several authors there is reduced CE formation in both Headache have tried to relate the severity of acidosis water intoxication and ischemic stroke Lethargy or being not easily aroused from in DKA to the risk of CE and/or the de- (120). Abnormalities in the number of sleep Diastolic blood pressure Ͼ90 mm Hg gree of altered mental status (82–85). It functioning aquaphorin channels in Age Ͻ5 years is interesting to note, however, that one brain glial cells on either a genetic or an patient who was not acidotic was reported acquired basis as the result of metabolic aSigns that occur before treatment should to have developed CE (107). Acetoacetate stresses could account for the develop- not be considered in the diagnosis of cerebral and ␤-hydroxybutyrate, in addition to ment of CE in certain individuals. edema. having an osmotic effect (96) and stimu- Reproduced with permission from Muir AB et al (1). lating the sodium/proton exchanger (74), Diagnosis are proinflammatory agents that affect the endothelial cells of the blood-brain Presentation. Although the phenome- barrier (50, 55, 108). Acetoacetate in- non of CEDKA was first identified in 1936 steroids, less developed mechanisms of creases intracellular calcium concentra- in adults (2), since being reported by brain cell volume regulation, differences tion in endothelial cells, causing vaso- Fitzgerald et al. (121) and Young and in taurine or other osmolyte metabolism, constriction (108). ␤-hydroxybutyrate Bradley (122) in children, CEDKA has differences in blood-brain barrier effi- increases vascular permeability factor, generally been regarded as a disease of ciency, and more rapid metabolism and also leading to an increase in endothelial young children (1, 11, 123–125). Muir et water turnover in children (11, 84, 91). cell intracellular calcium concentration al. (1) proposed a useful system of major However, the exact reason for the appar- (108). Hyperglycemia itself promotes in- and minor diagnostic criteria for the clin- ently more frequent occurrence in chil- flammation via an increased glucose me- ical diagnosis of CEDKA (Table 1). The dren than adults is unknown. It is also tabolite (109) and reactive oxygen species minor criteria are frequently present in possible that subclinical CE is common (110). There is increasingly convincing patients who do not develop symptomatic in both children and adults but that evidence of the proinflammatory state in CE. The authors indicate that signs that symptomatic CE is more common in patients with DKA at the time of admis- occur before treatment should not be children. It is also possible that the true sion and an increase in levels at the ex- considered in the diagnosis of CE, al- incidence of CE in adults in DKA is un- pected time of onset of symptomatic CE. though symptomatic CE may occur be- derappreciated. C-reactive protein, which induces adhe- fore treatment. Some authors indicate Subclinical vs. Symptomatic Cerebral sion molecule expression in endothelial that the first signs of CE may be respira- Edema. In 1985, Krane et al. (146) re- cells and stimulates macrophage produc- tory arrest and fixed dilated pupils (11), ported a series of CT scans in six children tion of cytokines at sites of inflammation, although this is disputed by those who 11–14 yrs of age with DKA who were all both of which lead to an increase in en- believe that careful monitoring will re- asymptomatic for CE; the authors com- dothelial cell permeability (vasogenic veal signs and symptoms before cata- pared scans taken early in the course edema), is increased in humans with DKA strophic collapse (1, 54). after treatment was started with scans without signs of infection (111). Interleu- Even though symptomatic CE does repeated at discharge, and all showed ev- kin-6, interleukin-1␤, and tumor necro- occur in adults (2, 16, 126–145), it is idence of CE. The authors concluded that sis factor-␣ levels were also elevated be- generally considered a disease of young subclinical swelling was common in chil- fore, during, and following treatment of children. Ninety-five percent of patients dren with DKA after treatment began. DKA. Interleukin-10 is elevated as well, are Ͻ20 yrs of age, and 33% are Ͻ5 yrs of This conclusion had been reached previ- increasing at the time of symptomatic CE age. Sixty-seven percent are patients with ously in a study that used head ultra- (112). Pathologic examination of human new-onset DKA (11). The reasons for this sonography to examine 18 children and brain tissue also has provided evidence of finding are unknown. Potential explana- adults in DKA who were asymptomatic activation of the complement system (113). tions include a relatively greater brain for CE (132). MRI studies that report findings con- volume for overall body size, more rapid In a study of eight patients ages 4–15 sistent with vasogenic edema due to a changes in plasma osmolality, lack of sex yrs in DKA (one with new-onset DKA)

Pediatr Crit Care Med 2008 Vol. 9, No. 3 323 who were admitted in a coma, Smedman In the United States, Glaser et al. (5) in Table 1. In the report by Muir et al. (1), et al. (147) performed CT scans 10 hrs conducted a retrospective study of 6,977 at least two warning signs of neurologic after therapy began and repeated the scan episodes of DKA from 1982 to 1997 and compromise or increased ICP were doc- after full recovery (all patients). The au- found an incidence of 0.9% (61 of 6,977). umented in the records of 10 of 12 pa- thors found evidence of CE in only two of In Canada, Lawrence et al. (9) conducted tients before a catastrophic event. Gen- eight patients and concluded that asymp- a prospective study of cases of CEDKA eral overall recommendations for tomatic CE does not occur commonly. from 1999 to 2001 and found 13 cases of treatment are prescribed by several au- Glaser et al. (148), using MRI brain scans CE from a total of 1,960 cases of DKA, for thors, suggesting good or better results in 41 children during treatment and after an incidence of 0.51%. Some of the diffi- with early aggressive intervention (123, recovery, found the lateral ventricles to culty in determining the incidence of 163–165, 167, 168). Specific recommen- be significantly smaller during DKA subclinical CEDKA by clinical means dations for use of mannitol, hypertonic treatment than after recovery, and those alone has to do, in part, with the different saline, intubation and hyperventilation, with smaller ventricles during treatment definitions used for diagnosis (1, 5, 6, steroids, and ICP monitoring follow. were most likely to have mental status 125, 148, 149). Mannitol. Roberts et al. (169) de- changes, which were seen in half the pa- Differential Diagnosis. Certainly not scribed 11 patients with severe DKA with tients. These data suggest that there is a all patients with DKA and altered mental cerebral complications, nine of whom continuum of clinically relevant CE from status have symptomatic CE. As indicated were treated early with mannitol and had modest to severe (149). However, Muir et earlier, patients with minor criteria alone complete recovery. Mannitol may im- al. (1), as well as others, emphasized that may not be diagnosable by clinical prove patients’ outcomes by osmotic di- many symptomatic patients with CE have means. Other intracranial pathology may uresis and by decreasing blood viscosity, a normal early CT scan that subsequently exist in as many as 10% to 20% of symp- thereby improving cerebral blood flow becomes abnormal later in the course tomatic patients (11, 54, 123, 137, 152– and oxygen delivery. Rosenbloom (54) and that CEDKA is not a CT scan but a 159). Processes to consider include hypo- also indicated a role for early treatment clinical diagnosis. glycemia, nonketotic hyperosmolality, with mannitol. Of 23 patients treated be- Early imaging studies as well as re- drug ingestions, infection (e.g., meningi- fore respiratory arrest, 13 survived with ports of other risk factors, such as fluid, tis, encephalitis), hemorrhage (spontane- independent function and three were se- insulin, and sodium bicarbonate admin- ous or traumatic), thrombosis, emboli, verely disabled. Of 46 patients not treated istration, emphasized the occurrence of stroke, infarction, extrapontine myelino- before respiratory arrest, only three sur- both subclinical and symptomatic CE af- sis, obstructive , and vived and were normal. Two patients ter initiation of treatment for DKA, im- trauma. Some clinicians remind us that were not treated and did not experience plicating the treatments themselves as after initiation of treatment for mental respiratory arrest. Shabbir et al. (170) risk factors in its development (32). How- status changes, central nervous system and Franklin et al. (171) also reported ever, Hoffman et al. (53) and Steinhart imaging is advisable in all patients, espe- good results with mannitol. and Hoffman (55), as well as others (5, 9, cially those who do not clearly improve, Hypertonic Saline. Some authors 49–52, 54), documented occurrences of because of these other causes of altered have reported using hypertonic saline in- subclinical and symptomatic CEDKA, in- mental status (11, 26, 160, 161). stead of mannitol to treat CEDKA with cluding even a death (51), before onset of Although less frequent than central some success (172, 173). The primary ad- therapy. Hoffman et al. (53) studied nine nervous system pathology and CE, other vantages proposed for hypertonic saline consecutive patients with DKA aged 6–17 causes of morbidity are found in pediatric are that it prevents or mitigates hypona- yrs who were asymptomatic for CE. The patients with DKA. These include hyper- tremia and prevents hypovolemia associ- investigators performed CT scans before kalemia and with arrhyth- ated with osmotic diuresis produced by treatment began, 6–8 hrs after starting mia, sepsis, pneumonia, pulmonary mannitol, thus aggravating cerebral isch- treatment, and 7 days later. All patients edema, acute respiratory distress syn- emia. Hypertonic saline also provides had evidence of CE, and it was present drome, alveolar rupture with free air, and similar rheologic effects as mannitol, de- before treatment was initiated. rhabdomyolysis (123). creasing blood viscosity and improving Incidence. The incidence of symptom- cerebral blood flow. atic CEDKA is thought to be approxi- Therapy of Cerebral Edema in Intubation and Hyperventilation. In- mately Ͻ1% and is remarkably similar in Diabetic Ketoacidosis tubation and ventilation are indicated to several studies, both prospective and ret- protect the airway in comatose patients rospective, and in different countries (5, Here I review the currently proposed and provide ventilation in apneic individ- 9, 10, 150, 151). Edge et al. (151), con- recommendations for treating symptom- uals (54, 166). Although some authors ducting a population-based study from atic CEDKA. These recommendations are suggest using hyperventilation to treat 1995 to 1998 in the United Kingdom, based in large part on individual case CE, this has been identified as a risk fac- found an incidence of 7.1 in 1,000 from a reports and opinions. tor for a worse outcome (8,163). Con- total of 2,941 episodes of DKA in children Early Signs of Cerebral Edema. Sev- versely, when patients are spontaneously

Ͻ16 yrs of age. The authors noted that eral authors (1, 54, 162) emphasize that hyperventilating to lower PaCO2 levels, CEDKA was more common in new-onset careful neurologic and cardiorespiratory ventilating them at normal levels may be disease, with an incidence of 11.9 in monitoring of patients often can lead to detrimental as well (75). Patients with 1,000 in patients with new-onset DKA but recognition of early warning signs and symptomatic CE who are intubated only 4.3 in 1,000 in those with known symptoms of symptomatic CE and there- should probably be ventilated to PaCO2 disease. This may relate to later detection fore early treatment and better outcome. levels existing at the time intubation is in new-onset patients. These signs and symptoms are presented performed (75–80).

324 Pediatr Crit Care Med 2008 Vol. 9, No. 3 Steroids. Only one author (129, 174) 13 patients with symptomatic CE; three hypotension, cold extremities, and/or an- has attributed a successful outcome to (23%) died and two (15%) recovered with uria. The degree of dehydration in these use of steroids in a patient with CE. Ste- neurologic impairment. Long-term mor- patients is frequently overestimated (167, roids certainly will complicate the man- bidity (178) and complete or partial hy- 194). If a bolus is given for hemodynamic agement of hyperglycemia, and most au- popituitarism (153, 179, 180) have been instability, do not continue to use boluses thors do not recommend their use (123, reported in other survivors. once circulatory stability is demon- 163). strated. Use isotonic (0.9%) sodium chlo- Intracranial Pressure Monitoring. Therapy of DKA: Prevention ride solution for boluses and to correct Two reports (175, 176) advocate the use of CE deficits. Maintenance fluids can be given of ICP monitoring to manage a patient with 0.45% sodium chloride solutions. with CEDKA. Others report use of ICP Certainly the best way to prevent Deficits should be estimated at 5% to 7% monitors (151). CEDKA is to prevent DKA (93–95). Once of weight unless shock is present, in DKA occurs, however, clinicians can refer which case assume 10% to 15% loss of Prognosis of Symptomatic to several excellent guidelines for man- body weight. Replace fluid deficits at a Cerebral Edema in Diabetic agement (59, 123, 124, 161, 167, 181– rate of 1/48 per hour, evenly over 48 hrs. Ketoacidosis 186) as well as excellent general guide- After any initial bolus, fluids can be ad- lines for fluid management in pediatric ministered at a rate rarely in excess of In considering the relationship of patients (186, 187). These guidelines 1.5–2.0 times the usual daily requirement treatment to outcome in symptomatic show a change over time from recom- (Table 2). Urinary losses should not be CE, it is interesting to note that some mendations for bolus doses to continuous added to the calculation of replacement reports indicate a good outcome without insulin infusions, more cautious fluid ad- fluids. This recommendation has only specific treatment (54, 151, 177). In ex- ministration, and avoidance of sodium class E support, and many clinicians do amining 31 patients with symptomatic bicarbonate, although some recent rec- recommend replacing urinary losses. Do CE, Edge et al. (151) reported that four of ommendations vary (188), and variation not decrease serum osmolality rapidly 34 patients received no specific treat- in practice has been documented to be (191). Many clinicians decrease osmolal- ment; eight of 34 died, and 26 of 34 considerable (189, 190). I next discuss ity at a rate Յ1.5–2.0 mOsm/hr. survived. Seventeen of 26 survivors were recommendations specifically addressing Insulin Administration. Bolus insulin normal, nine of 26 had persisent morbid- fluid, insulin, and sodium bicarbonate ad- administration is no longer used in pedi- ity. Morbidities in the nine included mo- ministration and correction of electrolyte atric patients (123). Administer insulin at tor deficits in eight visual impairment in abnormalities as they relate to prevention a rate of 0.1 unit/kg/hr (123, 184). This is six, memory loss in six, and seizures in of CE. These recommendations are made a class A evidence recommendation of the two. As indicated previously, Rosenbloom in the absence of definitive evidence of European and American Endocrinology (54) reported on 69 patients with a mor- the association of specific disease or Societies based on prospective, con- tality rate of 64%; most of the patients treatment risk factors for CE or an un- trolled, clinical trials. Many authors and who died were not treated until experi- derstanding of specific underlying mech- clinicians (especially intensivists) recom- encing respiratory arrest. Mahoney et al. anisms. mend 0.05 units/hr. If blood glucose de- (83) reported on nine patients with symp- Fluid Administration. It seems pru- creases at a rate of Ͻ50 mg/dL/hr or if the tomatic CE, five of whom died. Two of dent to avoid excessive amounts of fluid, patient fails to start to correct in 2–4 hrs, four survivors were intact and two were fluid given too rapidly, and use of hypo- increase the insulin to 0.15 units/kg/hr. If impaired. Glaser et al. (5) reported on 61 tonic fluid (33, 90, 91, 123, 184, 191– the blood glucose decreases to Ͻ350 patients with symptomatic CE. Thirty- 193). Most authors do not recommend mg/dL or Ͼ100 mg/dL/hr, add glucose in five of the 61 recovered without sequelae; that a bolus of fluid be given unless there a ratio of 4–5 g/unit of insulin. As a 13 had permanent neurologic sequelae is evidence of cardiovascular compromise guideline, many clinicians do not let the and 13 died. Lawrence et al. (9) studied as demonstrated by extreme tachycardia, glucose decrease Ͼ50–100 mg/hr. Due to

Table 2. Water and salt replacement in diabetic ketoacidosisa

● Water and salt deficits must be replaced. IV or oral fluids that may have been given before the child presents for treatment and prior to assessment should be factored into calculation of deficit and repair (A). ● Initial IV fluid administration and, if needed, volume expansion should begin immediately with an isotonic solution (0.9% saline or balanced salt solutions such as Ringer’s lactate). The volume and rate of administration depend on circulatory status, and where it is clinically indicated, the volume is typically 10 to 20 mL/kg over 1 to 2 hours, repeated if necessary (E). ● Use crystalloid (C). ● Subsequent fluid management should be with a solution with a tonicity Ն0.45% saline (C): This can be achieved by administering 0.9% saline or balanced salt solution (Ringer’s lactate or 0.45% saline with added potassium) (E). Rate of IV fluid should be calculated to rehydrate evenly over at least 48 hours (E). ● In addition to clinical assessment of dehydration, calculation of effective osmolality may be valuable to guide fluid and electrolyte therapy (E). ● The severity of dehydration may be difficult to determine and can be overestimated, infuse fluid each day at a rate rarely in excess of 1.5 to 2 times the usual daily requirement based on age, weight, or body surface area. Urinary losses should not be added to the calculation of replacement fluids (E).

aLetters in parenthesis are classes of evidence A through E; IV, intravenous. See reference 123, appendix. Reproduced with permission from Dunger et al (123).

Pediatr Crit Care Med 2008 Vol. 9, No. 3 325 the frequent large decrease in blood glu- is understood (23, 92). It is best to be 17. Anonymous: Cerebral oedema in diabetes. cose seen with initial fluid therapy, some extremely vigilant in monitoring pa- Lancet 1971; 2:694–695 clinicians do not start insulin until after tients; most will have been ill for many 18. Hale PM, Rezvani I, Braunstein AW, et al: initial fluid boluses are given when nec- days by the time they present, and there- Factors predicting cerebral edema in young essary. fore careful, slow correction of their met- children with diabetic ketoacidosis and new onset type 1 diabetes. Acta Paediatr 1997; Sodium Bicarbonate Administration. abolic derangement is prudent. 86:626–631 It is difficult to justify a complete prohi- 19. Jayashree M, Singhi S: Diabetic ketoacido- bition on use of sodium bicarbonate (10, sis: Predictors of outcome in a pediatric 11, 123). One may give 0.5–1.0 mEq/kg REFERENCES intensive care unit of a developing country. over 30–60 mins in children with severe Pediatr Crit Care Med 2004; 5:427–433 circulatory compromise and high risk of 1. Muir AB, Quisling RG, Yang MCK, et al: 20. Lee JH, Arcinue E, Ross B: Brief report: decompensation due to profound acido- Cerebral edema in childhood diabetic keto- Organic osmolytes in the brain of an infant sis. There is debate as to whether this is acidosis. Diabetes Care 2004; 27:1541–1546 with . N Engl J Med 1994; necessary (64, 73), since acidosis may not 2. Dillon ES, Riggs HE, Dyer WW: Cerebral 331:439–442 cause myocardial depression in these pa- lesions in uncomplicated fatal diabetic aci- 21. Segar WE: Clinical quiz. Pediatr Nephrol 1991; 5:99–101 tients. Various definitions of severe aci- dosis. Am J Sci 1936; 192:360–365 22. Lebovitz H: Diabetic ketoacidosis. Lancet dosis are used, including pH Ͻ7.1, Ͻ7.0, 3. Ditzel J: Raised C.S.F. pressure during treatment of diabetic ketosis. Lancet 1971; 1995; 345:767–772 and Ͻ6.9. Although it is not completely 2:925–926 23. Harris G, Lohr J, Fiordalisi I, et al: Brain convincing that sodium bicarbonate is 4. Beer S: Fatal ketoacidosis. Lancet 1986; osmoregulation during extreme and mod- dangerous, there is no evidence that it is 1:563–564 erate dehydration in a rat model of severe beneficial in these patients. 5. Glaser N, Barnett P, McCaslin I, et al: Risk DKA. Life Sci 1993; 53:185–191 Correction of Electrolyte Abnormali- factors for cerebral edema in children with 24. Rose SJ, Bushi M, Nagra I, et al: Taurine ties. It is not known whether failure of diabetic ketoacidosis. N Engl J Med 2001; fluxes in insulin dependent diabetes melli- sodium to increase during treatment or a 344:264–269 tus and rehydration in streptozotocin decrease in serum sodium causes CE or is 6. Glaser N: Cerebral edema in children with treated rats. Adv Exp Med Biol 2000; 483: a consequence of already existing cere- diabetic ketoacidosis. Curr Diabetes Rep 497–501 25. Silver M, Clark C, Schroeder M, et al: Patho- bral injury (7, 58). In either case, use 2001; 1:41–46 7. Glaser N: New perspectives on the patho- genesis of cerebral edema after treatment of isotonic (0.9%) sodium chloride solution genesis of cerebral edema complicating di- diabetic ketoacidosis. Kidney Int 1997; for bolus and initial therapy to correct abetic ketoacidosis in children. Pediatr En- 5:1237–1244 fluid deficits until serum sodium is in docrinol Rev 2006; 3:379–386 26. Clements R, Prockop L, Winegrad A: A normal range. 8. Marcin J, Glaser N, Barnett P, et al: Factors mechanism to explain acute cerebral edema Patients are usually potassium and associated with adverse outcomes in chil- during the treatment of diabetic acidosis. phosphate depleted (123, 184). Potassium dren with diabetic ketoacidosis-related ce- Diabetes 1968; 17:299 depletion is in the range of 3–6 mmol/kg, rebral edema. J Pediatr 2002; 141:793–797 27. Clements RS Jr, Prockop LD, Winegrad AI: and insulin administration will accentu- 9. Lawrence S, Cummings E, Gaboury I, et al: Acute cerebral oedema during treatment of ate this by drawing potassium into cells. Population-based study of incidence and hyperglycaemia. Lancet 1968; 2:384–386 28. Prockop LD: Hyperglycemia, polyol accu- Proper potassium replacement may help risk factors for cerebral edema in pediatric diabetic ketoacidosis. J Pediatr 2005; 146: mulation, and increased intracranial pres- prevent arrhythmias. Phosphate deficits 688–692 sure. Arch Neurol 1971; 25:126–140 are in the range of 0.5–2.5 mmol/kg and 10. Dunger DB, Edge JA: Predicting cerebral 29. Arieff AI, Kleeman CR, Keushkerian A, et al: need to be addressed as well (184). Ade- edema during diabetic ketoacidosis. N Engl Effects of hyperglycemia and rapid lowering quate phosphate levels may help preserve J Med 2001; 344:302–303 of plasma glucose in normal rabbits. J Clin energy availability. 11. Edge JA: Cerebral oedema during treatment Invest 1973; 52:571–583 of diabetic ketoacidosis: Are we any nearer 30. McManus M, Churchwell K, Strange K: Regulation of cell volume in health and CONCLUSIONS finding a cause? Diabetes Metab Res Rev 2000; 16:316–324 disease. N Engl J Med 1995; 333:1260–1266 The causes and mechanisms of 12. Bello FA, Sotos JF: Cerebral oedema in di- 31. Kreis R, Ross D: Cerebral metabolic distur- bances in patients with subacute and CEDKA are unknown. CE may have as abetic ketoacidosis in children. Lancet 1990; 336:64 chronic diabetes mellitus: Detection with much to do with individual biological 13. Bullock DG: Management of diabetic keto- proton MR spectroscopy. Radiology 1992; variance as with the severity of the un- acidosis and cerebral edema. Am Fam Phy- 184:123–130 derlying metabolic derangement at the sician 1994; 50:1468 32. Warren P, Villaluz ES, Rosenberg H: Dia- time of presentation or any one or com- 14. Cameron FJ, Keant MJ, Wellard RM, et al: betic ketoacidosis with fatal cerebral edema. bination of risk factors associated with Insights into the acute metabolic changes Pediatrics 1969; 43:620–622 treatment of DKA. Although I have pre- associated with childhood diabetes. Diabet 33. Guisado R, Arieff A: neurologic manifesta- sented treatment recommendations, at- Med 2005; 22:648–653 tions of diabetic comas: Correlation with tempting to consider proposed risk fac- 15. Carlotti APCP, Bohn D, Halperin ML: Im- biochemical alterations in the brain. Metab- tors and mechanisms of CE, I cannot be portance of timing of risk factors for cere- olism 1975; 24:665–679 34. Drash A: The treatment of diabetic ketoac- dogmatic about these recommendations bral oedema during therapy for diabetic ke- toacidosis. Arch Dis Child 2003; 88: idosis. J Pediatr 1977; 91:858–860 in the absence of better evidence. CE still 170–173 35. Moore R: Stimulation of NA:H exchange by occurs when these treatment guidelines 16. Clements RS Jr, Blumenthal SA, Morrison insulin. Biophys J 1981; 33:203–210 are followed, and well-documented cases AD, et al: Increased cerebrospinal-fluid 36. Matz R: Cerebral oedema in diabetic keto- representing the unpredictability of this pressure during treatment of diabetic keto- acidosis. Lancet 1987; 2:689 phenomenon remind us of how poorly it sis. Lancet 1971; 2:671–675 37. Duck S, Weldon V, Pagliara A, et al: Cere-

326 Pediatr Crit Care Med 2008 Vol. 9, No. 3 bral edema complicating therapy for dia- imizing the risk of during sound assessment of intracranial hemody- betic ketoacidosis. Diabetes 1976; 25: treatment of diabetic keto academia: A ret- namics in children with diabetic ketoacido- 111–115 rospective and prospective study. J Pediatr sis. J Clin Ultrasound 1995; 23:517–523 38. Duck SC, Kohler E: Cerebral edema in dia- 1990; 117:22–31 80. Roberts J, Vavilala M, Schenkman K, et al: betic ketoacidosis. J Pediatr 1981; 98: 59. Merkley K: Treating diabetic ketoacidosis in Cerebral hyperemia and impaired cerebral 674–676 children while preventing cerebral edema: autoregulation associated with diabetic ke- 39. Vlcek BW: Risk factors for cerebral edema One hospital’s protocol. J Emerg Nurs 2004; toacidosis in critically ill children. Crit Care associated with diabetic ketoacidosis. Ann 30:571 Med 2006; 34:2217–2223 Neurol 1986; 20:407 60. Garre M, Boles JM, Garo B, et al: Cerebral 81. Ouyyumi AA, Iaffaldano R, Guerrero JL, et 40. Krane EJ: Cerebral edema in diabetic keto- oedema in diabetic ketoacidosis: Do we use al: Prostacyclin and pathogenesis of hemo- acidosis. J Pediatr 1989; 114:166–168 too much insulin? Lancet 1986; 1:220 dynamic abnormalities of diabetic ketoaci- 41. Kaufman FR: Diabetes in children and ad- 61. Johanson CE, Murphy VA: Acetazolamide dosis in rats. Diabetes 1989; 38:1585–1594 olescents: Areas of controversy. Med Clin and insulin after choroid plexus epithelial 82. Edge JA, Roy Y, Bergomi A, et al: Conscious North Am 1998; 82:721–738 cell [Naϩ], pH, and volume. Am J Physiol level in children with diabetic ketoacidosis 42. Woodward GA, Levy RJ, Weinzimer SA: Di- 1990; 258:F1538–F1546 is related to severity of acidosis and not to abetes and transport: A potentially bitter- 62. Tornheim PA: Regional localization of cere- blood glucose concentration. Pediatr Dia- sweet combination. Pediatr Emerg Care bral edema following fluid and insulin ther- betes 2006; 7:11–15 1998; 14:71–76 apy in streptozotocin-diabetic rats. Diabetes 83. Mahoney CP, Vlcek BW, DelAguila M: Risk 43. Inward CD, Chambers TL, Edge J: Fluid 1981; 30:762–766 factors for developing brain herniation dur- management in diabetic ketoacidosis. Arch 63. Durr J, Hoffman W, Sklar A, et al: Corre- ing diabetic ketoacidosis. Pediatr Neurol Dis Child 2002; 86:443–444 lates of brain edema in uncontrolled IDDM. 1999; 21:721–727 44. Levitsky LL: Symptomatic cerebral edema Diabetes 1992; 41:627–632 84. Van Der Meulen JA, Klip A, Grinstein S: in diabetic ketoacidosis: The mechanism is 64. Maury E, Vassal T, Offenstadt G: Cardiac Possible mechanism for cerebral oedema in clarified but still far from clear. J Pediatr contractility during severe ketoacidosis. diabetic ketoacidosis. Lancet 1987; 2: 2004; 145:149–150 N Engl J Med 1999; 341:1938 306–308 45. Petrie JC: Diabetes mellitus—Problems of 65. Taubin H, Matz R: Cerebral edema, diabetes 85. Van Der Meulen J: Cerebral oedema in dia- keto-acidosis. BMJ 1972; 3:409–412 insipidus, and sudden death during the betic ketoacidosis. BMJ 1992; 305:1366 46. Gebara B: Risk factors for cerebral edema in treatment of diabetic ketoacidosis. Diabetes 86. Mel JM, Werther GA: Incidence and out- children with diabetic ketoacidosis. N Engl 1968; 17:108–109 come of diabetic cerebral oedema in child- J Med 2001; 344:1556 66. Bureau MA, Begin R, Berthiaume Y, et al: hood: Are there predictors? J Pediatr Child 47. Krane E: Cerebral edema in diabetic keto- Cerebral hypoxia from bicarbonate infusion Health 1995; 31:17–20 acidosis. J Pediatr 1989; 114:166–168 in diabetic acidosis. J Pediatr 1980; 96: 87. Nejsum LN, Kwon T, Marples D, et al: Com- 48. Brown AFT: Aetiology of cerebral oedema in 968–973 pensatory increase in AQP2, p-AQP2, and diabetic ketoacidosis. Emerg Med J 2004; 67. McGarry JD, Foster D: Regulation of keto- AQP3 expression in rats with diabetes mel- 21:754–756 genesis and clinical aspects of the ketotic litus. Am J Physiol Renal Physiol 2001; 49. Couch R, Acott P, Wong G: Early onset fatal state. Metabolism 1972; 21:471–489 280:F715–F726 cerebral edema in diabetic ketoacidosis. Di- 68. Winegrad A, Clements R: Diabetic ketoaci- 88. Rosenbloom AL, Riley WJ, Weber FT, et al: abetes Care 1991; 14:78–79 dosis. Med Clin North Am 1971; 55: Cerebral edema complicating diabetic keto- 50. Figueroa R, Hoffman W, Momin Z, et al: 899–911 acidosis in childhood. J Pediatr 1980; 96: Study of subclinical cerebral edema in dia- 69. Baruh S, Sherman L, Markowitz S: Diabetic 357–361 betic ketoacidosis by magnetic resonance ketoacidosis and coma. Med Clin North Am 89. Winegrad A, Kern EF, Simmons DA: Cere- imaging T2 relaxometry and apparent diffu- 1981; 65:117–132 bral edema in diabetic ketoacidosis. N Engl sion coefficient maps. Endocr Res 2005; 31: 70. Sperling M: Diabetic ketoacidosis. Pediatr J Med 1985; 312:1184–1185 345–355 Clin North Am 1984; 31:591–610 90. Finberg L: Why do patients with diabetic 51. Glasgow A: Devastating cerebral edema in 71. Kaye R: Diabetic ketoacidosis-the bicarbon- ketoacidosis have cerebral swelling, and diabetic ketoacidosis before therapy. Diabe- ate controversy. J Pediatr 1975; 87:156–159 why does treatment sometimes make it tes Care 1991; 14:77–78 72. Moore J: Cerebral oedema in diabetic keto- worse? Arch Pediatr Adolesc Med 1996; 150: 52. Greene SA, Jefferson IG, Baum JD: Cerebral acidosis. BMJ 1975; 3:540 785–786 oedema complicating diabetic ketoacidosis. 73. Edge J: Management of diabetic ketoacido- 91. Finberg L: Appropriate therapy can prevent Dev Med Child Neurol 1990; 32:633–644 sis in childhood. Br J Hosp Med 1996; 55: cerebral swelling in diabetic ketoacidosis. 53. Hoffman W, Steinhart C, Gammal T, et al: 508–512 J Clin Endocrinol Metab 2000; 85:508–522 Cranial CT in children and adolescents with 74. Brown TB: Cerebral oedema in childhood 92. Rosenbloom AL: Cerebral edema in diabetic diabetic ketoacidosis. Am J Neuroradiol diabetic ketoacidosis: Is treatment a factor? ketoacidosis. J Clin Endocrinol Metab 2000; 1988; 9:733–739 Emerg Med J 2004; 21:141–144 85:507J 54. Rosenbloom AL: Intracerebral crises during 75. Tasker RC, Lutman D, Peters MJ: Hyperven- 93. Rosenbloom AL, Schatz D, Krischer J, et al: treatment of diabetic ketoacidosis. Diabetes tilation in severe diabetic ketoacidosis. Pe- Prevention and treatment of diabetes in Care 1990; 13:22–33 diatr Crit Care Med 2005; 6:405–411 children. J Clin Endocrinol Metab 2000; 55. Steinhart CM, Hoffman WH: Cerebral 76. Fortune P: Diabetic emergencies in chil- 85:494–506 edema in diabetic ketoacidosis. J Pediatr dren. Hosp Med 2004; 65(4):234–237 94. Sperling M: No gold at the end of the gly- 1989; 114:905–906 77. Marcin JD, Glaser N, Kupperman N: Venti- cemic rainbow. J Pediatr 2002; 141: 56. Glaser N, Wootton-Gorges S, Marcin JP, et lation in pediatric ketoacidosis—Not too 754–756 al: Mechanism of cerebral edema in chil- much but not too little. Pediatr Crit Care 95. Sperling M: Diabetic ketoacidosis in chil- dren with diabetic ketoacidosis. J Pediatr Med 2005; 6:489–450 dren: The problems continue. Pediatr Dia- 2004; 145:164–171 78. Ackerman A: Cerebral edema in pediatric betes 2005; 6:67–68 57. Glaser NS: Cerebral metabolic alterations in diabetic ketoacidosis: Can six patients make 96. Puliyel J: Osmotonicity of acetoacetate: Pos- children with diabetic ketoacidosis. Diabet a difference? Crit Care Med 2006; 34: sible implications for cerebral edema in di- Med 2005; 22:515–516 2258–2259 abetic ketoacidosis. Med Sci Monit 2003; 58. Harris G, Fiordalisi I, Harris W, et al: Min- 79. Hoffman WH: Transcranial Doppler ultra- 9:170–173

Pediatr Crit Care Med 2008 Vol. 9, No. 3 327 97. Puliyel JM, Bhambhani V: Ketoacid levels al: Complement and activation in diabetic 133. Foster DW, McGarry JD: The metabolic de- may alter osmotonicity in diabetic ketoaci- ketoacidosis brains. Exp Mol Pathol 2006; rangements and treatment of diabetic keto- dosis and precipitate cerebral edema. Arch 80:283–288 acidosis. N Engl J Med 1983; 309:159–169 Dis Child 2003; 88:366 114. Matz R, Carroll P: Cerebral edema and dia- 134. Gordon D, MacCuish AC: Cerebral oedema 98. Matz R, Carroll P: Cerebral edema in dia- betic ketoacidosis. Ann Intern Med 1982; in diabetic ketoacidosis. Scot Med J 1988; betic ketoacidosis. Ann Intern Med 1982; 97:141–142 33:212–213 97:141–142 115. Brun-Buisson CJL, Bonnet F, Beregeret S, 135. Matz R: IDDM consensus guidelines. Diabet 99. Duck S, Wyatt D: Factors associated with et al: Recurrent high-permeability pulmo- Med 1994; 11:601–602 brain herniation in the treatment of dia- nary edema associated with diabetic ketoac- 136. Chan NN, Manchanda S, Feter MD: Fatal betic ketoacidosis. J Pediatr 1988; 113: idosis. Crit Care Med 1985; 13:55–56 cerebral edema associated with hyponatre- 10–14 116. Griffith D, Yudkin J: (Disturbed homeosta- mia. Practical Diabetes International 1997; 100. Kempski O, Staub F, Von Rosen F, et al: sis) diabetic ketoacidosis. Br J Hospital Med 15:209–211 Molecular mechanisms of glial swelling in 1986; 35:82–88 137. Ertl-Wagner B, Jansen O, Schwab S, et al: vitro. Neurochem Pathol 1988; 9:109–125 117. Keller U: Diabetic ketoacidosis: Current Bilateral basal ganglion haemorrhage in di- 101. Bevensee MO, Schmitt BM, Choi I, et al: An views on pathogenesis and treatment. Dia- abetic ketoacidotic coma: Case report. Neu- electrogenic Naϩ-HCO3 cotransporter betologia 1986; 29:71–77 roradiology 1999; 41:670–673 (NBC) with a novel COOH-terminus, cloned 118. Young MC: Simultaneous acute cerebral 138. Chan N, Darko D, O’Shea D: Meningeal syn- from rat brain. Am J Physiol Cell Physiol and pulmonary edema complicating dia- drome in diabetic ketoacidosis. Diabetes 2000; 278:C1200–C1211 betic ketoacidosis. Diabetes Care 1995; 18: Care 1999; 22:370–371 102. Schmitt BM, Berger UV, Douglas RM, et al: 1288–1290 139. McIntyre EA, Abraha HD, Perros P, et al: Na/HCO3 cotransporters in rat brain: Ex- 119. Dixon AN, Jude EB, Banerjee AK, et al: Serum S-100␤ protein is a potential bio- pression in glia, neurons, and choroids Simultaneous pulmonary and cerebral oe- chemical marker for oedema complicating plexus. J Neurosci 2000; 20:6839–6848 dema, and multiple CNS infarctions as severe diabetic ketoacidosis. Diabet Med 103. Amoroso S, Di Renzo G, Taglialatela M, et complications of diabetic ketoacidosis: A 2000; 17:807–809 al: Cytoplasmic alkalinization induced by case report. Diabet Med 2006; 23:571–573 140. Chan NN, Hurel SJ: Biochemical marker for insulin through an activation of Naϩ-Hϩ 120. Manley GT, Fujimura M, Tonghui M, et al: cerebral oedema complicating severe dia- antiporter inhibits tyrosine hydroxylase ac- Aquaporin-4 deletion in mice reduces brain betic ketoacidosis. Diabet Med 2001; 18: tivity in striatal synaptosomes. Biochem edema after acute water intoxication and 1007–1010 Pharmacol 1991; 41:1279–1282 ischemic stroke. Nat Med 2000; 6:159–163 141. Azzopardi J, Gatt A, Zammit A, et al: Lack of 104. Dryer SE: Naϩ-activated Kϩ channels: A 121. Fitzgerald MG, O’Sullivan DJ, Malins JM: evidence of cerebral oedema in adults new family of large-conductance ion chan- Fatal diabetic ketosis. BMJ 1961; 1:247–249 treated for diabetic ketoacidosis with fluids nels. Trends Neurosc 1994; 17:155–161 122. Young E, Bradley RF: Cerebral edema with of different tonicity. Diabetes Res Clin Pract 105. Lam T, Anderson SE, Glaser N, et al: Bu- irreversible coma in severe diabetic ketoac- 2002; 57:87–92 metanide reduces cerebral edema formation idosis. N Engl J Med 1967; 276:665–669 142. Leary P, Dunbar K: A very close call. in rats with diabetic ketoacidosis. Diabetes 123. Dunger DB, Sperling MA, Acerini CL, et al: Am J Med 2005; 118:968–971 2005; 54:510–516 European Society for Paediatric Endocri- 143. Troy P, Clark R, Kakarala S, et al: Cerebral 106. Soleimani M, Singh G: Physiologic and mo- nology/Lawson Wilkins Pediatric Endocrine edema during treatment of diabetic ketoac- lecular aspects of the Naϩ/Hϩ exchangers Society consensus statement on diabetic ke- idosis. Neurocrit Care 2005; 2:55–58 in health and disease processes. J Investigat toacidosis in children and adolescents. Pe- 144. Hiller K, Wolf S: Cerebral edema in an Med 1995; 43:419–430 diatrics 2004; 113:e133–e140 adult patient with diabetic ketoacidosis. 107. Shastry R, Bhatia V, Sahu R, et al: Cerebral 124. Glaser N: Pediatric diabetic ketoacidosis Am J Emerg Med 2005; 23:399–400 edema without ketoacidosis or hyperosmo- and hyperglycemic hyperosmolar state. Pe- lar coma in a 16-year-old boy. Diabetes diatr Clin North Am 2005; 52:1611–1635 145. Speirs AL: Diabetic ketoacidosis. BMJ 1970; Care 2005; 28:753–754 125. Rosenbloom AL: Cerebral edema in diabetic 3:746–747 108. Isales CM, Min L, Hoffman WH: Acetoace- ketoacidosis and other acute devastating 146. Krane EJ, Rockoff MA, Wallaman JK, et al: tate and ␤-hydroxybutyrate differentially complications: Recent observations. Ped Di- Subclinical brain swelling in children dur- regulate endothelin-1 and vascular endo- abetes 2005; 6:41–49 ing treatment of diabetic ketoacidosis. thelial growth factor in mouse brain micro- 126. Hayes TM, Woods CJ: Unexpected death N Engl J Med 1985; 312:1147–51 vascular endothelial cells. J Diabetes Com- during treatment of uncomplicated diabetic 147. Smedman L, Escobar R, Hesser U, et al: plicat 1999; 13:91–97 ketoacidosis. BMJ 1968; 4:32–33 Sub-clinical cerebral oedema does not oc- 109. Knudsen GM, Jakobsen J: Blood-brain bar- 127. Felig P: Diabetic ketoacidosis. N Engl J Med cur regularly during treatment for diabetic rier permeability to sodium modification by 1974; 290:1360–1363 ketoacidosis. Acta Paediatr 1997; 86: glucose or insulin. J Neurochem 1989; 52: 128. Keller U, Berger W, Ritz R, et al: Course and 1172–1176 174–178 prognosis of 86 episodes of diabetic coma: A 148. Glaser NS, Wootton-Georges SL, Buono- 110. Stentz FB, Umpierrez GE, Cuervo R, et al: five year experience with a uniform sched- core MH, et al: Frequency of sub-clinical Proinflammatory cytokines, markers of car- ule of treatment. Diabetologia 1975; 11:93 cerebral edema in children with diabetic diovascular risks, oxidative stress and lipid 129. Frier BM, McConnell JB: Cerebral oedema ketoacidosis. Pediatr Diabetes 2006; peroxidation in patients with hyperglycemic in diabetic ketoacidosis. BMJ 1975; 3:208 7:75–80 crises. Diabetes 2004; 53:2079–2086 130. Zimmerman B: Acute diabetic complica- 149. Sperling M: Cerebral edema in diabetic ke- 111. Dalton R, Hoffman WH, Passmore G, et al: tions. Minn Med 1976; 757–765 toacidosis: An underestimated complica- Plasma C-reactive protein levels in severe 131. Skillman T: Diabetic ketoacidosis. Heart tion? Pediatr Diabetes 2006; 7:73–74 diabetic ketoacidosis. Ann Clin Lab Sci Lung 1978; 7:594–602 150. Edge JA, Hawkins MM, Winter DL, et al: 2003; 33:435–442 132. Fein A, Rackow E, Sprung CL, et al: Rela- Incidence, presentation, management and 112. Hoffman WH, Burek CL, Waller JL, et al: tion of colloid osmotic pressure to arterial outcome of cerebral edema associated with Cytokine response to diabetic ketoacidosis hypoxemia and cerebral edema during crys- diabetic ketoacidosis (DKA) in Great Brit- and its treatment. Clin Immunol 2003; 108: talloid volume loading of patients with dia- ain. Arch Dis Child 1999; 80 (Suppl 1):A11 175–181 betic ketoacidosis. Ann Intern Med 1982; 151. Edge JA, Hawkins MM, Winter DL, et al: The 113. Hoffman WH, Cudrici CD, Zafranskaia E, et 96:570–575 risk and outcome of cerebral oedema devel-

328 Pediatr Crit Care Med 2008 Vol. 9, No. 3 oping during diabetic ketoacidosis. Arch Dis 167. Halperin M, Maccari C, Kamel K: Strategies 181. Padilla AJ, Loeb JN: “Low-dose” vs. “high- Child 2001; 85:16–22 to diminish the danger of cerebral edema in dose” insulin regimens in the management 152. Scibilia J, Finegold D, Dorman J, et al: Why pediatric diabetes. Pediatr Diabetes 2006; of uncontrolled diabetes: A survey. do children with diabetes die? Acta Endo- 7:191–195 Am J Med 1977; 63:843–848 crinol 1986; 279:326–333 168. Shastry RM, Bhatia V: Cerebral edema in 182. Kandel G, Aberman A: Selected develop- 153. Keller RJ, Wolfsdorf JI: Isolated growth hor- diabetic ketoacidosis. Indian Pediatr 2006; ments in the understanding of diabetic ke- mone deficiency after cerebral edema com- 43:701–708 toacidosis. Can Med Assoc J 1983; 128: plicating diabetic ketoacidosis. N Engl 169. Roberts MD, Slover RH, Chase HP: Diabetic 392–396 J Med 1987; 316:857–859 ketoacidosis with intracerebral complica- 183. Klekamp J, Churchwell KB: Diabetic keto- 154. Krane EJ: Arterial thrombosis causing cere- tions. Pediatr Diabetes 2001; 2:109–114 acidosis in children: Initial clinical assess- bral edema in association with diabetic ke- 170. Shabbir N, Oberfield SE, Corrales R, et al: ment and treatment. Pediatr Ann 1996; 25: toacidosis. Crit Care Med 1988; 16:100 Recovery from symptomatic brain swelling 387–393 155. Eskander E, Weller SJ, Frim DM: Hydro- in diabetic ketoacidosis. Clin Pediatr 1992; 184. Marks J, Gonzales J: Diabetic ketoacidosis. cephalous requiring urgent external ven- 31:570–573 In: Essentials of Pediatric Intensive Care. tricular drainage in a patient with diabetic 171. Franklin B, Lui J, Ginsberg-Fellner F: Ce- Second Edition. Levin DL, Morriss FC ketoacidosis and cerebral edema: case re- rebral edema and ophthalmoplegia reversed (Eds). St. Louis, QMP, 1997, pp 565–570 port. Neurosurgery 1997; 40:836–839 by mannitol in a new case of insulin- 156. Keane S, Gallagher A, Ackroyd S, et al: dependent diabetes mellitus. Pediatrics 185. Wolfsdorf J, Glaser N, Sperling M: Diabetic Cerebral venous thrombosis during diabetic 1982; 69:87–90 ketoacidosis in infants, children, and ado- ketoacidosis. Arch Dis Child 2002; 86: 172. Curtis JR, Bohn D, Daneman D: Use of hy- lescents: A consensus statement from the 204–205 pertonic saline in the treatment of cerebral American Diabetes Association. Diabetes 157. Bonkowsky J: Extrapontine myelinolysis in edema in diabetic ketoacidosis (DKA). Pedi- Care 2006; 29:1150–1159 a pediatric case of diabetic ketoacidosis and atr Diabetes 2001; 2:191–194 186. Perkin RM, Levin DL: Common fluid and cerebral edema. J Child Neurol 2003; 18: 173. Kamat P, Vats A, Gross M, et al: Use of electrolyte problems in the pediatric inten- 144–147 hypertonic saline for the treatment of al- sive care unit. Pediatr Clin North Am 1980; 158. Hatun S, Cizmecioglu F, Toprak D: Cere- tered mental status associated with diabetic 27:567–586 bral complications in diabetic ketoacidosis. ketoacidosis. Pediatr Crit Care Med 2003; 187. Marks JF: Abnormalities of fluids, electro- Turk J Pediatr 2005; 47:170–172 4:239–242 lytes, and calcium. In: Essentials of Pediat- 159. Ho J, Pacaud D, Hill M, et al: Diabetic ke- 174. Frier BM: Cerebral edema in diabetic keto- ric Intensive Care. Second Edition. Levin toacidosis and pediatric stroke. CMAJ 2005; acidosis. BMJ 1975; 3:704 DL, Morriss FC (Eds). St. Louis, QMP, 1997, 172:327–328 175. Wood E, Go-Wingkun J, Luisiri A: Symp- pp 543–547 160. White N: Management of diabetic ketoaci- tomatic cerebral swelling complicating dia- 188. Patel N: Diabetic ketoacidosis in the pediat- dosis. Rev Endocrinol Metab Disord 2003; betic ketoacidosis documented by intraven- ric patient. Indian J Pediatr 2002; 69:75–77 4:342–353 tricular pressure monitoring: Survival 189. Edge JA, Dunger DB: Variations in the man- 161. Glaser N, Kuppermann N: The evaluation without neurologic sequela. Pediatr Emerg agement of diabetic ketoacidosis in chil- and management of children with diabetic Care 1990; 6:285–288 dren. Diabet Med 1994; 11:984–986 ketoacidosis. Pediatr Emerg Care 2004; 20: 176. Shrier D, Shibata D, Wang H, et al: Central 190. Glaser N, Kupperman N, Yee C, et al: Vari- 477–481 brain herniation secondary to juvenile dia- ation in the management of pediatric dia- 162. Warne GL: Fatal cerebral oedema with in- betic ketoacidosis. Am J Neuroradiol 1999; betic ketoacidosis by specialty training. sulitis in diabetic ketoacidosis: Still puz- 20:1885–1888 Arch Pediatr Adolesc Med 1997; 151: zling and often fatal. Med J Austr 1973; 177. Metzger AL, Rubenstein AH: Reversible ce- 1125–1132 2:934–937 rebral oedema complicating diabetic keto- 191. Bohn D, Daneman D: Diabetic ketoacidosis 163. Hammond P, Wallis S: Cerebral oedema in acidosis. BMJ 1970; 3:746–747 and cerebral edema. Pediatrics 2002; 14: diabetic ketoacidosis. BMJ 1992; 305: 178. Rogers B, Sills I, Cohen M, et al: Diabetic 287–291 203–204 ketoacidosis: Neurologic collapse during 164. Kalis NN, Van Der Merwe PL, Schoeman JF, treatment followed by severe developmen- 192. Felner E, White PC: Improving manage- et al: Cerebral edema with coning in dia- tal morbidity. Clin Pediatr 1990; 29: ment of diabetic ketoacidosis in children. betic keto-acidosis. S Afr Med J 1991; 79: 451– 456 Pediatrics 2001; 108:735–740 727–731 179. Dunlop KA, Woodman D, Carson DJ: Hy- 193. Harris G, Fiordalisi I: Physiologic manage- 165. McAloon J, Carson D, Crean P: Cerebral popituitarism following cerebral oedema ment of diabetic ketoacidemia: A 5-year oedema complicating diabetic ketoacidosis. with diabetic ketoacidosis. Arch Dis Child prospective pediatric experience in 231 ep- Acta Paediatr Scand 1990; 79:115–117 2002; 87:337–338 isodes. Arch Pediatr Adolesc Med 1994; 148: 166. Strachan MWJ, Nimmo GR, Noyes K, et al: 180. Lufkin E, Reagan T, Doan D, et al: Acute 1046–1052 Management of cerebral oedema in diabe- cerebral dysfunction in diabetic ketoacido- 194. Koves IH, Neutze J, Donath S, et al: Improv- tes. Diabetes Metab Res Rev 2003; 19: sis: Survival followed by panhypopituita- ing our estimate of dehydration in DKA. 241–247 rism. Metabolism 1977; 26:363–369 Diabetes Care 2004; 27:2485–2487

Pediatr Crit Care Med 2008 Vol. 9, No. 3 329