In this study, we investigated the incidence of postoperative metabolic acidosis after elective surgery in patients with diabetes mellitus; specifically, we compared the incidence in patients treated with SGLT2-is to the incidence in those who were not. We found that preoperative use of SGLT2-is was associated with an increased incidence of metabolic acidosis without hyperglycemia during the ICU stay after elective surgery. The results of the multivariable linear regression analysis suggested that SGLT2-is were the only type of medication that affected pH among the medications included in the model. Although we could only detect the incidence of metabolic acidosis with euglycemia, these results suggest that perioperative SGLT2-i use is associated with eDKA—one of the major causes of mortality among patients with diabetes [16].
To the best of our knowledge, this was the first study to evaluate the relationship between the preoperative use of SGLT2-is and the incidence of perioperative metabolic acidosis. A previous systematic review suggested that perioperative eDKA is a common condition in patients treated with SGLT2-is [9]. However, since that systematic review only included case reports or case series, the relationship between eDKA and SGLT2-is remained unclear. In the present study, we performed an adjustment for confounding factors using one-to-four matching for patients treated with an SGLT2-i and controls. We found an association between SGLT2-i use and an increased incidence of postoperative metabolic acidosis. A previous study showed that patients are exposed to food deprivation and dehydration during the perioperative period, and thus they may be vulnerable to developing eDKA [17].
In this study, both groups had a high incidence of metabolic acidosis compared to that reported in a previous study [10]. Two reasons may explain the high incidence of metabolic acidosis in this study. First, the study hospital is among the leading cardiovascular centers in Japan, and approximately half of the patients in this study underwent cardiovascular surgery, which is an invasive surgical procedure. Stress induced by the surgical procedure may increase the incidence of metabolic acidosis [18]. Second, in this study, patients stopped receiving SGLT2-is on the day of surgery to decrease the risk of perioperative eDKA [17]. The Japanese package inserts distributed with prescription drugs recommend that these drugs be stopped on the day of surgery. The patients in this study followed these recommendations. In contrast, the United States Food and Drug Administration (USFDA) recommends that this drug be stopped 3 or 4 days before surgery [19]. The patients in this study had a withdrawal period that was shorter than that of patients following the USFDA recommendation. The short withdrawal period may have increased the incidence of metabolic acidosis identified in this study.
These findings suggest that meticulous perioperative monitoring of arterial blood gas results may be needed in patients with diabetes mellitus who are treated with SGLT2-is. Patients who undergo highly invasive procedures (e.g., cardiovascular surgery) may be at a higher risk of developing eDKA. Withdrawal of medication on the day of surgery may be insufficient to prevent eDKA.
This study had several limitations. First, we did not examine the level of ketone bodies in the patients’ blood or urine. The ICU did not conduct routine urinalyses, and measurement of ketone bodies in the blood is difficult in Japanese hospitals because the required devices are not covered by the national insurance system. Therefore, we assessed the incidence of metabolic acidosis without accounting for hyperglycemia. However, ketoacidosis is among the common causes of a high anion gap in patients with metabolic acidosis [20]. These results provide preliminary evidence of an association between SGLT2-i use and the incidence of perioperative eDKA. Second, there were differences in the proportions of patients in the two groups who had undergone dialysis and those who presented with cardiovascular disease. SGLT2-i use may reduce the incidence of cardiovascular events [5, 21, 22] and help prevent decline in renal function [6, 22–24]. SGLT2-is may not be used to treat patients with anuria because the drug affects the proximal tubule of the kidney and promotes glucose excretion. The characteristics of this drug may account for the differences between the study groups. The electronic database did not include information on eGFR, dialysis, or history of cardiovascular disease. This limitation of the dataset may have introduced some selection bias to this study. However, a subgroup analysis of patients without chronic kidney disease showed larger between-group differences for the primary outcome, suggesting that the overall findings were robust. Third, the external validity of this study remains unclear. In December 2020, the Japan Diabetes Society recommended a preoperative withdrawal period of SGLT2-is to match that of the USFDA guideline [25]. However, in this study, the patients did not stop using SGLT2-is 3 or 4 days before surgery. This delay in withdrawal might have affected the incidence of metabolic acidosis. Further studies are required to estimate the incidence of ketoacidosis in contexts where an appropriate withdrawal period is observed. A multicenter observational study examining the incidence of postoperative ketoacidosis associated with SGLT2-i use is on-going in Japan [26].