The New England Journal of Medicine -- January 25, 2001 -- Vol. 344, No. 4

Cerebral edema is a devastating complication of diabetic ketoacidosis and remains the leading cause of serious illness and death in children with diabetes mellitus. (1) The causes of cerebral edema are unknown, largely because of the lack of large-scale population-based studies. However, several hypotheses relating to possible antecedent risk factors and the effects of various treatment regimens have been proposed. (1)

In this issue of the Journal, Glaser et al. (2) report the results of a retrospective analysis of cases of children with diabetic ketoacidosis at 10 centers. Among 6977 such children, the authors identified 61 with cerebral edema. The incidence of cerebral edema was 0.9 percent, a rate remarkably similar to that reported historically in the United States (3) and recently in the United Kingdom. (4) As in previous studies, the risk was highest among younger children with newly diagnosed diabetes. (1) The 21 percent mortality rate associated with cerebral edema was similar to that in the United Kingdom, (4) as was the morbidity rate, with a substantial 27 percent of the survivors having neurologic sequelae.

The development of cerebral edema may be the result of the treatment that children receive for diabetic ketoacidosis; treatments such as high doses of insulin and the administration of bicarbonate or large volumes of hypotonic fluid would be the major culprits. (1) However, it is also possible that the condition is an idiosyncratic response to diabetic ketoacidosis. Thus far, no adequate proof for either hypothesis has been advanced.

The study by Glaser et al. confirms previous reports that cerebral edema can become evident even before treatment for diabetic ketoacidosis is initiated, but usually it develops 4 to 12 hours later. (1,2) The only treatment association detected by the authors was with the administration of bicarbonate; other treatments that previously have been implicated, (1) such as a high rate of fluid administration or the administration of hypotonic fluid, were not associated with an increased risk of cerebral edema. Nevertheless, a smaller increase in the serum sodium concentration during therapy was associated with cerebral edema, as suggested in previous reports. (1,5) However, as the authors state, the failure of serum sodium concentrations to rise substantially could be a consequence, rather than a cause, of cerebral edema. (2)

Glaser et al. are to be commended for their systematic study, but does it improve our understanding of the pathophysiology of cerebral edema or provide new guidelines for therapy? The authors note that cerebral edema was associated with higher serum urea nitrogen concentrations and lower partial pressures of arterial carbon dioxide at the time of diagnosis; they argue that this finding supports the hypothesis, first proposed by Dillon et al., (6) that cerebral edema is the result of reduced blood volume, aggravated by a reduced partial pressure of arterial carbon dioxide and leading to cerebral vasoconstriction, cerebral ischemia, and hypoxia. The reduced partial pressure of arterial carbon dioxide at presentation in children with cerebral edema may imply an increased respiratory drive in response to acidosis. However, caution must be used in interpreting blood pH values, partial pressures of arterial carbon dioxide, and bicarbonate concentrations when corrections have been made to account for the collection of both arterial and venous blood samples. Furthermore, whereas rapid reductions in the partial pressure of arterial carbon dioxide may lead to acute vasoconstriction, this process is often reversed when low partial pressures are chronic, as in cases of adaptation to metabolic acidosis. However, the hypoxia theory has received support from those who have suggested that bicarbonate therapy might reduce the partial pressure of oxygen in the cerebrospinal fluid, leading to vasoconstriction and further brain hypoxia. (7)

The higher initial serum urea nitrogen concentrations noted by Glaser et al. in children in whom cerebral edema developed (and, by inference, the greater degrees of renal insufficiency in these children) have been linked to pure anion-gap acidosis, rather than hyperchloremic acidosis. (8) The type of acidosis may therefore be important in predicting cerebral edema, since regulation of cell volume by the sodium-hydrogen ion exchanger may be compromised by severe acidosis. (9) The higher serum urea nitrogen concentrations in these children may reflect the chronic nature of diabetic ketoacidosis, because a long duration of symptoms has previously been associated with development of cerebral edema. (3)

The data presented by Glaser et al. confirm previous reports that the administration of bicarbonate may be a risk factor for cerebral edema in children with diabetic ketoacidosis; however, over the past 10 years, during which time the use of bicarbonate treatment has been greatly reduced, there has been no corresponding reduction in the incidence of cerebral edema. Bicarbonate is now recommended only for children with severe circulatory failure and a high risk of cardiac decompensation due to profound acidosis. Glaser et al. do not provide details about the indications for bicarbonate administration in their study; its administration may have reflected the severity of diabetic ketoacidosis and circulatory collapse, factors that may be related to the lower partial pressures of arterial carbon dioxide and higher sodium urea nitrogen concentrations in children with cerebral edema. Given these data, it would be difficult to justify the prohibition of bicarbonate therapy, irrespective of the severity of diabetic ketoacidosis.

We must be cautious about interpreting the data of Glaser et al. as revealing anything more than an association between the severity of diabetic ketoacidosis and dehydration and the risk of cerebral edema. The increased risk in children with newly diagnosed diabetes and the occurrence of cerebral edema before treatment support the importance of other antecedent factors, such as the accumulation of idiogenic, osmotically active substances that regulate cell volume. (1)

The current study, with its rigorous design and careful analysis, is important because it counters the common supposition that cerebral edema is an iatrogenic disease. Rather, the data suggest that the risk of cerebral edema is related to the duration and severity of diabetic ketoacidosis. Furthermore, because cerebral accumulation of water occurs to some extent in the majority of children with diabetic ketoacidosis, (10) the risk may also relate to individual biologic responses. Although a smaller change in the serum sodium concentration during therapy and the use of bicarbonate administration may be associated with cerebral edema, we must consider whether the cause of this condition is idiosyncratic rather than iatrogenic.

David B. Dunger, M.D. University of Cambridge, Cambridge CB2 2QQ, United Kingdom

Julie A. Edge, M.D. John Radcliffe Hospital, Oxford OX3 9DU, United Kingdom