To the Editor:
Since their introduction more than a decade ago, sodium glucose transporter 2 inhibiting drugs have established themselves as important tools in the management of diabetes, chronic kidney disease, and heart failure.1 However, periprocedural euglycemic diabetic ketoacidosis has been associated with the drug class,2 causing the Food and Drug Administration (Silver Spring, Maryland) to recommend that the drugs be held for at least 3 days before procedures.3 This recommendation presents anesthesiologists and proceduralists with the difficult decision of whether to cancel procedures if patients have not held their sodium glucose transporter 2 inhibitor. Euglycemic diabetic ketoacidosis has been most associated with major physiologic stresses, including infection and emergency surgery,4 suggesting that proceeding with elective outpatient procedures may convey less risk than the consequences of canceling them.5 Here, we prospectively measured serum ketones in patients prescribed a sodium glucose transporter 2 inhibitor undergoing outpatient endoscopies to test whether patients undergoing these procedures are at risk for significant ketoacidosis if they do not hold their sodium glucose transporter 2 inhibitor for the recommended time.
This prospective cohort study was part of a perioperative quality improvement initiative and approved by the Massachusetts General Hospital (Boston, Massachusetts) Institutional Review Board (Protocol 2021P002931). The study included patients prescribed a sodium glucose transporter 2 inhibitor undergoing outpatient endoscopies at Massachusetts General Hospital between January 10, 2023, and March 10, 2023. A β-hydroxybutyrate was drawn with IV placement before the endoscopy. Assuming no events of ketoacidosis when sodium glucose transporter 2 inhibitors are appropriately held, we aimed to recruit 60 patients to power our study to detect a greater than 5% rate of clinically significant ketoacidosis if a sodium glucose transporter 2 inhibitor was continued at the α = 0.05 confidence level based on a binomial distribution.
Demographic information and baseline laboratory measurements were similar to large randomized controlled trials of sodium glucose transporter 2 inhibitors6,7 (table 1). Two patients who did not hold their sodium glucose transporter 2 inhibitors appropriately met our definition of clinically significant ketoacidosis (defined as greater than 3 mM, a suggested diagnostic criterion for diabetic ketoacidosis8 ; 4.8% incidence; 95% CI, 1.3 to 15.8%), while none who held the medication appropriately were found to have significant ketoacidosis (95% CI, 0 to 12.9%, P = 0.39 vs. continuation of sodium glucose transporter 2 inhibitor by Fisher exact test). In both patients with a β-hydroxybutyrate greater than 3 mM, the measurement resulted after the procedure began. After completion of the procedure, they were referred to the emergency department, treated with subcutaneous insulin, and discharged home once they were deemed to have adequate oral intake. Patients undergoing colonoscopies who continued their sodium glucose transporter 2 inhibitor presented with significantly elevated ketones compared with those who did hold the medication (fig. 1). They also demonstrated β-hydroxybutyrate concentrations greater than patients undergoing an esophagogastroduodenoscopy who similarly did not hold their sodium glucose transporter 2 inhibitor. Unlike patients undergoing colonoscopies, those undergoing an esophagogastroduodenoscopy did not appear to have significantly elevated ketone concentrations if they continued their sodium glucose transporter 2 inhibitor.
Here, we present evidence that patients taking sodium glucose transporter 2 inhibitors who undergo colonoscopies, but not necessarily esophagogastroduodenoscopies, are at increased risk for ketoacidosis. Patients prescribed sodium glucose transporter 2 inhibitors who continue the medication until a minor procedure continue to present clinicians with the dilemma of whether to cancel the procedure.9,10 We found no evidence of worsened ketosis in patients undergoing an esophagogastroduodenoscopy if they did not hold their medication, albeit in a relatively small sample size. This observation suggests that the risk of clinically significant ketosis is lower in these patients, and it may not be worth the consequences of delaying these minor procedures if an sodium glucose transporter 2 inhibitor is not held. However, if a patient does continue his or her sodium glucose transporter 2 inhibitor and undergoes the scheduled minor procedure, ketoacidosis must still be ruled out by either clinical or laboratory criteria before discharge because euglycemic diabetic ketoacidosis is life-threatening.2
In contrast to patients undergoing esophagogastroduodenoscopies, we observed significantly increased concentrations of ketoacids if patients continued their sodium glucose transporter 2 inhibitor until a colonoscopy. These increases likely stem from the physiologic stress caused by osmotic diarrhea during colon preparation and longer fasting period for colonoscopies than for esophagogastroduodenoscopies. Two patients with β-hydroxybutyrate concentrations greater than 3 mM would not have come to clinical attention had ketones not been measured. They would have been particularly vulnerable to metabolic complications had they not been able to tolerate oral intake after going home. We therefore suggest that patients who have not held their sodium glucose transporter 2 inhibitor and undergo colonoscopies or other procedures with significant physiologic stress be screened for euglycemic diabetic ketoacidosis with a serum ketone measurement, and extra effort be made to ensure that they are able to eat and can be monitored by a caretaker after discharge. We acknowledge the small sample size and observational nature of this study as limitations of the work. Although we observed no events of meaningful ketoacidosis if sodium glucose transporter 2 inhibitors were continued before an esophagogastroduodenoscopy, it cannot be inferred from these 13 patients that the incidence is low enough (95% CI, 0 to 23%) that it is truly safe to proceed with these minor procedures. We hope this work prompts larger studies of noncolonoscopy outpatient procedures to confirm if the rate of ketoacidosis in this population is low enough to support proceeding if a sodium glucose transporter 2 inhibitor is not appropriately held.
Acknowledgments
The authors thank the Massachusetts General Hospital (Boston, Massachusetts) endoscopy nursing group for assistance with drawing β-hydroxybutyrates, and Hang Lee, Ph.D., of the Massachusetts General Hospital Biostatistics Center for statistical advice.
Research Support
Support was provided solely from institutional and departmental resources.
Competing Interests
Dr. Wiener-Kronish has received support from the Fonds de la Recherche Scientifique (FNRS research federation; Brussels, Belgium) and honoraria from Washington University (St. Louis, Missouri) and University of Washington (Seattle, Washington). She is also an editor for the textbook Miller’s Anesthesia published by Elsevier. The other authors declare no competing interests.