Patients with type 2 diabetes (T2D) on sodium–glucose cotransporter 2 inhibitors (SGLT2i) may be at increased risk of pericolonoscopy ketoacidosis, attributed to bowel preparation, fasting, and inappropriate SGLT2i management. We recently reviewed the SGLT2i-linked perioperative euglycemic diabetic ketoacidosis (EDKA) risk (1) and reported pericolonoscopy EDKA in this context (2). Subsequently, we educated our institution about the importance of ketone monitoring in mitigating this risk and implemented pre- and postprocedure capillary ketone assessment for T2D patients undergoing colonoscopy. Using retrospective analysis, we aimed to determine the incidence of pericolonoscopy hyperketonemia in patients with T2D prescribed SGLT2i therapy, comparing its magnitude with non-SGLT2i therapy. We reviewed electronic records of patients undergoing colonoscopy at The Queen Elizabeth Hospital between July 2019 and August 2020 (Central Adelaide Local Health Network Human Research Ethics Committee, no. 14166). The FreeStyle Optium Neo H point-of-care kit (Abbott Laboratories, Abbott Park, U.K.) capable of measuring capillary blood glucose and β-hydroxybutyrate (BHB) was used in the measurement. The highest BHB value either before or after colonoscopy was deemed the periprocedure estimate. Hyperketonemia was defined as BHB >1.0 mmol/L.
Over 14 months, 3,609 patients underwent colonoscopy; 16.4% (592 of 3,609) had T2D. Of these, 94 (15.9%) were on SGLT2i, 384 (64.9%) were prescribed non-SGLT2i medication, 103 (17.4%) were on diet control, and medication details were unavailable for 11. Monotherapy with an oral agent was noted in approximately two-thirds of the patients in both groups. Insulin was prescribed for 33% (31 of 94) and 27.3% (105 of 384), respectively, in SGLT2i and non-SGLT2i groups.
Periprocedure BHB was available for 77 of 94 (82%) on SGLT2i and for 313 of 384 (81.5%) prescribed a pharmacological glucose-lowering regimen that did not include an SGLT2i. The incidence of hyperketonemia was 18% (14 of 77) in the SGLT2i group and 15.7% (49 of 313) in the non-SGLT2i group (P = 0.589). In the SGLT2i group, 10 patients had BHB values ranging 1.0–1.9 mmol/L, 3 patients 2.0–2.9 mmol/L, and 1 patient >3.0 mmol/L. Data on the preprocedure instructions/documentations on withholding SGLT2i were available for 73 of 94 patients. Instructions as per the Australian Diabetes Society (3) and U.S. Food and Drug Administration (4) recommendations (withholding for 72 h) were given in only 11 cases (15%): 26 patients (37%) were instructed to cease at 24 h and 31 (42%) to cease at 48 h. No patients took SGLT2i medication on the day of the procedure. Of the 14 cases of SGLT2i-associated hyperketonemia, one patient was prescribed insulin and instructions on preprocedure SGLT2i withholding were documented in 11 cases. Five patients were instructed to withhold for 24 h, and three each were instructed to withhold for 48 and 72 h, respectively (Fig. 1). There was a nonsignificant weak negative correlation observed between periprocedure BHB levels and preprocedure duration of withholding of SGLT2i (Pearson correlation coefficient r = −0.125; P = 0.731).
There were four cases where EDKA was diagnosed by the attending anesthesiologists without a predefined criterion. Although the capillary glucose levels were within the euglycemic range (111–158 mg/dL [6.2–8.8 mmol/L]) in these cases, upon review by the study authors, only one case fulfilled the American Diabetes Association criteria for DKA, with BHB 2.0 mmol/L, pH 7.29, and anion gap 23 mmol/L. Insulin and dextrose were administered in three cases at the discretion of the attending anesthesiologists.
Of the 49 cases of hyperketonemia in the non-SGLT2i group, 9 patients were prescribed insulin and 40 had BHB values in the range of 1.0–1.9 mmol/L, 7 had values within 2.0–2.9 mmol/L, and 2 had values >3.0 mmol/L. Two cases of BHB values >2.0 mmol/L were managed with insulin and dextrose at the judgement of the attending anesthesiologists. Even though nine patients had BHB values >2.0 mmol/L within the non-SGLT2i group, acid-base balance estimates were not documented for this group and there were no clinical diagnoses of EDKA made. The hyperketonemia status did not result in colonoscopy deferral in either of the groups.
In a recent study investigators reported a 24% incidence of pericolonoscopy hyperketonemia across 37 SGLT2i-treated patients with T2D where SGLT2i were inconsistently held (range 14–77 h) (3). Findings revealed 5% (5 of 102) and 9% (13 of 151) rates of hyperketonemia, respectively, for patients with T2D on non-SGLT2i and patients without diabetes (3). The study included only assessment of preprocedure BHB values, and the lower rate observed in the non-SGLT2i cohort is difficult to explain (3).
In summary, colonoscopy and its preparation may predispose to low-grade ketosis in patients with T2D, regardless of treatment. The incidence of hyperketonemia was 18% and 15.7% for the SGLT2i and non-SGLT2i groups, respectively. Although SGLT2i withholding was inconsistent, we did not encounter severe cases of ketoacidosis and no cases were deferred. In situations where SGLT2i agents are stopped later than recommended, colonoscopy may still proceed with vigilance toward identifying ketoacidosis until adequate oral intake is established.
V.T. and G.S.-T. share co-first authorship.
Article Information
Acknowledgments. The authors thank Jim Wang, senior scientist, Department of Endocrinology and Metabolism, The Queen Elizabeth Hospital, for assistance with the statistical analysis.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. V.T. was the main author of article and was involved in study conception and analysis and drafting the manuscript. G.S.-T. was involved in data extraction and analysis and manuscript drafting. A.J. was involved in data extraction and analysis and drafting the manuscript. E.M. was involved in data analysis and in writing and review of the manuscript. B.L. was involved in data extraction and analysis. D.J. was involved in critical review and editing of the manuscript and provided an expert opinion. N.N. was involved in planning the study, data analysis, writing the manuscript, and critical review of the manuscript. R.V.W. was involved in critical review and editing of the manuscript and provided an expert opinion. V.T. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Publication. Parts of this study were presented in abstract form at the Australasian Diabetes Conference 2021, 12–10 August 2021.