(571) Clinical and Pharmacologic Risk Factors for Cerebral Edema in Diabetic Ketoacidosis

Introduction: Diabetic ketoacidosis-associated cerebral edema (DKA-CE) is the leading cause of death and disability from DKA. DKA management avoids rapid glucose and sodium drops to prevent osmotic edema. A recent clinical trial did not show increased risk of brain injury in those who received higher volumes or lower concentrations of fluids, suggesting other mechanisms. Sulfonylurea (SU) drugs, used in diabetes treatment, mitigate cerebral edema due to stroke. This study investigates clinical and pharmacologic DKA-CE risk factors, hypothesizing lower DKA-CE risk with SU drugs.

Methods: Data was retrospectively collected from a deidentified databank with electronic health records linked to a biorepository for genotype-phenotype association studies. An automated search identified records with DKA (control) using ICD-9/10 codes. DKA-CE (case) was defined as presence of DKA plus altered mental status and/or osmolar therapy. Records with traumatic brain injury, stroke, or brain tumor were excluded. Cases were manually reviewed. Demographic, clinical, laboratory and pharmacologic data are presented as median[IQR] and compared using the Wilxocon rank-sum test.

Results: Records were obtained from 4053 individuals with DKA (without CE), and 283 with DKA-CE. DKA-CE patients were younger (14[10-22] vs 16[11-36] years, p< 0.001), and more likely to have type 1 diabetes (84% vs 76%, p=0.002). They presented with higher heart rate (115[101-132] vs 100[83-119], p < 0.001), lower pH (7.10[6.92-7.25] vs 7.25[7.15-7.32], p< 0.001), higher blood urea nitrogen (BUN) (20[14-31] vs 15[11-21]mg/dL, p< 0.001) and glucose (428[258-622] vs 348[219-518]mg/dL, p< 0.001). Glucose-corrected sodium did not differ. Of cases, 104 (37%) received 3% saline, 28 (10%) received mannitol, and 44 (16%) required intubation. Neuroimaging showed cerebral edema in 10 (3.5%), and 6 (2%) died during an episode of DKA-CE. The databank was queried for prior SU use; 55 (1.4%) DKA controls and 0 cases received glyburide (p < 0.05).

Conclusions: Automated search of a deidentified databank allows feasible DKA-CE risk identification. DKA-CE is associated with higher BUN, glucose, and heart rate, and lower pH. SU drugs are associated with lower risk of DKA-CE. Further study is needed to explore potential genetic associations with DKA-CE using this databank.