Insulin and glucose infusion could prevent euglycemic diabetic ketoacidosis associated with sodium-glucose cotransporter 2 inhibitors

Abstract

Perioperative euglycemic diabetic ketoacidosis (euDKA) is a serious adverse effect of sodium-glucose cotransporter 2 inhibitor (SGLT2i) treatment. We observed perioperative euDKA immediately after discontinuing insulin infusion that was started during surgery in a patient with type 2 diabetes mellitus (T2DM) for whom empagliflozin could not be withdrawn before emergency off-pump coronary artery bypass grafting (OPCAB). Insulin infusion that was started during surgery unexpectedly prevented euDKA until its discontinuation. Therefore, we hypothesized that insulin and glucose infusion initiated at the start of emergency surgery in patients receiving SGLT2is prevents perioperative euDKA. We implemented this strategy during emergency OPCAB in another patient with T2DM who received empagliflozin 2 days before surgery and observed that the patient did not develop perioperative euDKA. With the increasing use of SGLT2is, surgeons may encounter more SGLT2i users who require emergency surgeries. The administration of insulin and glucose infusion in advance emergency surgery can prevent perioperative euDKA.

Introduction

Sodium-glucose cotransporter 2 inhibitors (SGLT2is) are the newest class of oral antidiabetic agents available for treating type 2 diabetes mellitus (T2DM). Some SGLT2is have also been approved for treating type 1 DM. These SGLT2is have several pleiotropic effects and are, therefore, expected to be prescribed more frequently even to patients without DM in the future. Perioperative euglycemic diabetic ketoacidosis (euDKA) is a life-threatening adverse effect of these agents, and it is well-known that SGLT2is should be discontinued several days before a scheduled surgery to prevent euDKA [1]. However, in several cases, it is not possible to discontinue SGLT2is in time owing to emergency situations; hence, there is a need to identify strategies to prevent perioperative euDKA in such cases.

Case Reports

Case 1

Case 1 involved a 55-year-old man with a 20-year history of T2DM who started empagliflozin 18 months earlier because of intolerance to insulin therapy. His medications included empagliflozin 10 mg, metformin 500 mg, glimepiride 2 mg, vildagliptin 100 mg, and acarbose 300 mg. He was referred to our department for unstable angina. His preoperative parameters were as follows: creatinine, 0.65 mg/dL; estimated glomerular filtration rate (eGFR), 98 mL/min; glucose, 168 mg/dL; and HbA1c, 8.0%. Preoperative urinalysis revealed no ketonuria. As the patient suffered from unstable angina pectoris, we decided to perform emergency off-pump coronary artery bypass grafting (OPCAB), and we could withdraw all antidiabetic agents only on the day of surgery. During the surgery, continuous insulin infusion (CII) was started because of hyperglycemia. The patient was extubated 3 h after the surgery, and CII was discontinued because his blood glucose level had reduced to 71 mg/dL. Thereafter, the patient immediately developed high anion gap metabolic acidosis. A blood test showed no hyperglycemia or elevation in lactate level. Figures 1 and 2 show his postoperative transition of pH, anion gap, and lactate and glucose levels. Eight hours after extubation, the patient’s parameters were as follows: pH, 7.25; anion gap, 17.4; glucose level, 149 mg/dL; and lactate level, 0.6 mmol/L. We performed fluid replacement and electrolyte correction; however, metabolic acidosis lasted 26 h after extubation. Further laboratory investigation revealed hyperketonemia (serum total ketone body, 6849 μmol/L; 3-hydroxybutyrate, 4538 μmol/L) and ketonuria; thus, we diagnosed the patient with euDKA. We commenced insulin and glucose infusion, and as a result, ketoacidosis improved. Empagliflozin was not restarted, and we initiated a regimen of multiple daily insulin injections on postoperative day (POD) 4 after consulting a diabetologist. His postoperative blood glucose level ranged from 110 to 220 mg/dL. The patient was discharged without any complications on POD 15.

Fig. 1.

Transition of pH and anion gap in the arterial blood gas of case 1 patient

Fig. 2.

Transition of the lactate and glucose levels in the arterial blood gas of case 1 patient

Case 2

Case 2 involved an 80-year-old female with a 30-year history of T2DM and had been receiving daily antidiabetic medications (metformin, 500 mg; glimepiride, 1 mg; alogliptin, 25 mg) and insulin therapy. She was diagnosed with unstable angina and referred to our department for emergency OPCAB. Her preoperative parameters were as follows: creatinine, 0.80 mg/dL; eGFR, 52 mL/min; blood glucose, 310 mg/dL; and HbA1c, 9.3%. Preoperative urinalysis revealed no ketonuria. Treatment with 10 mg empagliflozin daily was initiated 2 days earlier, by a physician, to control DM. We could not withhold empagliflozin before emergency surgery because of a lack of time. Therefore, we decided to initiate insulin and glucose infusion (Humulin R 7 units + 50% dextrose 40 mL at the fixed flow rate of 10 mL/H) at the start of surgery to prevent euDKA. OPCAB was performed uneventfully, and the patient was extubated 4.5 h after surgery. We continued this insulin and glucose infusion at the fixed flow rate even after surgery and added another CII for perioperative blood glucose management when needed (Fig. 3). On postoperative day 1, we terminated this insulin and glucose infusion after confirming that the patient could eat well. The patient did not develop euDKA throughout the perioperative course.

Fig. 3.

Transition of pH and glucose level in case 2 patient with clinical events and our perioperative management

Discussion

Diabetic ketoacidosis (DKA) is a serious complication in DM and is accompanied with high anion gap metabolic acidosis, hyperglycemia, and increased ketone bodies in the blood and urine. Generally, DKA is also accompanied with considerably elevated glucose level, but a rare type of DKA (euDKA) is accompanied with only a mild increase in blood glucose. Munro et al. described euDKA for the first time in 1973 [2]. Currently, euDKA is defined with the blood glucose level < 250 mg/dL. euDKA is rapidly becoming a serious threat. euDKA can occur in patients on SGLT2is, and delayed diagnosis and suboptimal treatment of euDKA may result in severe clinical complications [3].

SGLT2is reduce blood glucose level by preventing SGLT2 on the proximal tubular epithelial cells from reabsorbing glucose from the primary urine. Their effects are insulin-independent, leading to the reduction in insulin secretion by pancreatic beta cells. Consequently, SGLT2i therapy results in insulin insufficiency. The lowering of blood insulin level accelerates lipolysis and increases the production of free fatty acids converted to ketone bodies in the liver. SGLT2is also increase the production of glucagon. The lowering of the insulin-to-glucagon ratio further stimulates lipolysis and increases free fatty acid and lipid oxygenation [4], resulting in ketoacidosis. Because the blood glucose level in patients on SGLT2is tends to be low, they are predisposed to be euglycemic.

The perioperative period also contributes to the body’s ketogenic state as fasting decreases oral carbohydrate intake, dehydration, and surgical stress [5]. Recently, Perry et al. reported that the following two factors are involved in euDKA development: dehydration and deficiency of insulin [6]. Under the combination of dehydration and deficiency of insulin perioperatively, surgical stress may become a trigger for euDKA development. Perioperative metabolic changes in cardiac surgery are induced by tissue injury, cardiopulmonary bypass (CPB), perioperative hypothermia, drugs, and blood products [7]. Although CPB was considered a major factor inducing metabolic changes in cardiac surgery, several studies have reported that CPB does not induce significant metabolic changes in coronary artery bypass grafting surgery [8]. In the case series reported by Lau et al., one-third of patients who developed euDKA underwent OPCAB [3], and our case 1 patient also underwent OPCAB. Irrespective of whether or not CPB is performed, any kind of cardiac surgery could lead to euDKA perioperatively.

Only a mild increase in blood glucose level in patients with euDKA may delay its detection, as seen in case 1. However, timely and adequate diagnosis is important as euDKA is a medical emergency. Once diagnosed, the management of euDKA involves the supplementation of insulin and glucose despite euglycemia, fluid replacement, and electrolyte correction. These should be continued until anion gap and bicarbonate level are corrected.

The US Food and Drug Administration in 2020 approved the changes to the prescribing guidelines for SGLT2is, and it recommends halting treatment approximately 3 or 4 days before a scheduled surgery [1]. However, it is possible that the number of emergency surgeries in patients who cannot discontinue SGLT2is in time will increase in the near future. The optimal time for withdrawing SGLT2is before an emergency surgery to prevent euDKA development remains unclear. Currently, the management of patients receiving SGLT2is undergoing emergency surgeries is performed on a case-by-case basis.

To prevent euDKA in emergency surgeries, determining the accurate perioperative insulin requirement for patients is difficult as the blood glucose level can be misleading. For case 1 patient, the anesthesiologist started CII for intraoperative hyperglycemia, and at the time of extubation, we discontinued CII to prevent hypoglycemia. Immediately, the patient developed euDKA. Goldenberg et al. reported that the reduction in insulin dose is one of the predisposing factors for ketoacidosis [9]. However, this implies “a reduction in insulin without proper advice in usual insulin users.” We emphasize that the postoperative decrease in, or discontinuation of insulin infusion, temporarily started during surgery can also be a trigger for euDKA development, based on the case 1 experience. Lau et al. reported euDKA development in a current insulin user who underwent coronary artery bypass grafting after the discontinuation of insulin infusion started intraoperatively [3]. We observed euDKA in a non-current insulin user in the same situation. We expect that this report will serve as a cautionary case when handling similar situations.

Considering the course of case 1, we hypothesized that if we start insulin infusion before surgical stress, there will be no deficiency of insulin in patients; therefore, euDKA may not develop. Lau et al. also discussed the possibility of this strategy to prevent euDKA in their report [3]. We implemented this strategy in case 2, and the patient did not develop perioperative euDKA. In fact, we used insulin and glucose infusion for this purpose. Glucose infusion can prevent insulin-induced hypoglycemia. The ratio of insulin and glucose infusion was referred to as glucose-insulin (GI) therapy for hyperkalemia. Humulin R 7 units + 50% dextrose 40 mL (glucose 20 g) was used. This solution was administered as continuous intravenous infusion at 10 mL/h. The recommended perioperative target glucose level for most intensive care unit and cardiac surgery patients with hyperglycemia and diabetes is between 140 and 180 mg/dL; therefore, we adopted this target for perioperative blood glucose management. We assumed that this insulin and glucose infusion at a fixed flow rate plays a role to cover the deficiency of insulin associated with SGLT2i therapy. Hence, when the blood glucose level exceeded the upper limit, we added CII at another transfusion line. When the blood glucose dropped below the lower limit, we did not decrease its flow rate but adjusted another CII (if another CII was started). We do not assume that this insulin and glucose infusion leads to hypoglycemia. As shown in Fig. 3, we did not observe hypoglycemia in case 2 despite the use of another CII. To alleviate persistent hypoglycemia, glucose should be supplemented via another transfusion line.

Because this strategy, similar to GI therapy for hyperkalemia, may induce hypokalemia and hypoglycemia and as euDKA can occur during the course of treatment, electrolyte, blood glucose, and urine ketone body levels, pH and the anion gap should be closely monitored. A previous clinical review of GI therapy for hyperkalemia recommends hourly monitoring of blood glucose level to prevent hypoglycemia after initiating insulin infusion [10]. Accordingly, we monitored them hourly and observed no clinical manifestations associated with either hypokalemia or hypoglycemia throughout the course. Insulin and glucose infusion was discontinued when the patient was feeling better and could eat and drink to maintain hydration after surgery, which corresponded with the time of SGLT2i re-initiation reported by Goldenberg et al. [9]. Nevertheless, optimizing the ratio of insulin/glucose is necessary to establish the usefulness and safety of this strategy and this aspect remains to be explored. We cannot conclude that insulin and glucose infusion definitively prevent euDKA based only on the experience with this case. In the future, we will keep track of more such cases.

There is a serious risk of euDKA during cardiac surgery. Further investigations and collaborations are warranted to establish an adequate preventive measure in cases where SGLT2is cannot be discontinued in time before surgery. For patients’ safety, we recommend the withdrawal of SGLT2is at the optimum time in scheduled surgeries; currently, we expect that insulin and glucose infusion in advance will be beneficial in several emergency cases.

Conclusions

We prevented perioperative euDKA in a patient who could not discontinue SGLT2i before an emergency surgery by initiating insulin and glucose infusion in advance before surgical stress. Although the electrolyte and blood glucose levels should be monitored closely, we believe that this is a valid method to prevent perioperative euDKA during emergency surgery in patients receiving SGLT2i.

Funding

None.

Data availability

Not applicable.

Code availability

Not applicable.

Declarations

Ethics approval

This case report was approved by the ethics committee of Tokyo General Hospital and SHIN-YURIGAOKA General Hospital.

Informed consent

In accordance with the ethical standards of Tokyo General Hospital and SHIN-YURIGAOKA General Hospital, informed consent was obtained from the patients before the submission of this manuscript.

Consent for publication

In accordance with the ethical standards of Tokyo General Hospital and SHIN-YURIGAOKA General Hospital, informed consent for publication was obtained from the patients before the submission of this manuscript.

Human and animal rights

This study has been reported in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. This paper also does not describe any experiment involving animals performed by any of the authors.

Conflict of interest

The authors declare no competing interests.