Unexpected Acute Hyperglycemia Following Pre-Surgical Discontinuation of SGLT2 Inhibitors in a Type 1 Diabetic Patient

Background

Sodium-glucose transporter 2 (SGLT2) inhibitors are a class of oral hypoglycemic agents that promote urinary glucose excretion by inhibiting SGLT2 in the proximal renal tubules. SGLT2 inhibitors are also available for use as an adjunct to insulin therapy in patients with type 1 diabetes due to their mechanism of action, which helps to reduce the required insulin dose and improve glycemic control [1–3].

However, SGLT2 inhibitors are typically associated with normoglycemic ketoacidosis, an adverse reaction unique to these agents [4,5]. Given that surgery is among the risk factors for the onset of diabetic ketoacidosis associated with the use of SGLT2 inhibitors, the American Diabetes Association and the European Association for the Study of Diabetes recommend that they should be discontinued in patients with type 2 diabetes undergoing surgery at least 72 h preoperatively [6]. The same precaution may be deemed necessary for type 1 diabetes, given that diabetic ketoacidosis is shown to have occurred in some patients with perioperative use of SGLT2 inhibitors [7].

We encountered a case of greater-than-expected acute hyperglycemia occurring in a patient with type 1 diabetes discontinuing SGLT2 inhibitor therapy 5 days before undergoing surgery, which made her glycemic control extremely difficult. As the patient had been followed up for her entire clinical course with intermittently scanned continuous glucose monitoring, we present a detailed case report describing the extent of hyperglycemia noted after discontinuation of SGLT2 inhibitor therapy and associated increases in the required insulin dose.

Case Report

Our patient was a 47-year-old woman with type 1 diabetes (height, 155 cm; body weight, 65.6 kg; body mass index, 27.3 kg/m2). She was diagnosed with type 1 diabetes at the age of 25 and was started on multiple daily insulin injection therapy. At the age of 43, when she was admitted for surgery for lipoma, her fasting serum C-peptide and 24-h urinary C-peptide value were <0.03 ng/mL and 0.3 μg/day (reference range, above 20 μg/day), respectively, and she was negative for anti-glutamic acid decarboxylase antibodies. From that time on, she was treated with insulin lispro 17 U/day (morning, 3 U; afternoon, 6 U; and evening, 8 U) and insulin degludec 11 U/day, while she found it difficult to adhere to the dietary intake prescribed, with her HbA1c level remaining in the 8–9% range.

Thus, at the age of 45, she was additionally prescribed dapagliflozin 5 mg; at the age of 46, the dose was increased to 10 mg based on improved glycemic control and body weight, with the insulin lispro dose reduced to 12 U/day (morning, 3 U; afternoon, 5 U; and evening, 4 U).

At the age of 44, the patient was found to have a goiter, which was later diagnosed with fine-needle aspiration as a follicular tumor, and she was closely monitored thereafter for enlargement in tumor size. At 46, however, surgery was indicated for her based on evidence of tumor growth.

Her preoperative glycemic control was unfavorable, with her HbA1c shown to be 8% because of inadequate adherence to her dietary regimen. Thus, she was admitted for glycemic control 1 week before the day of surgery. Before admission, her average sensor glucose level was 167 mg/dL and her%CV was 44%. With the daily total energy intake prescribed as 1600 kcal/day at admission, she received frequent insulin injections consisting of insulin lispro 12 U/day (morning, 3 U; afternoon, 5 U; and evening, 4 U) and insulin degludec 11 U/day (see Figures 1 and 2 for the clinical course described below for days 1–16 of admission). She had no extra food and drink during hospitalization.

Given that she had a mean sensor glucose level of 109.7 mg/dL with a time in range of 100% on day 1 of admission, she was treated for glycemic control, with her basal insulin dose unchanged but her bolus insulin dose adjusted (insulin lispro 12U/day to 11U/day). When SGLT2 inhibitor therapy was discontinued from day 2 of admission onwards, an acute increase was noted in her mean sensor glucose level, from 122.7 mg/dL on day 2 to 166.0 mg/dL on day 3. Again, she was found to have a mean glucose level of up to 273.9 mg/dL on day 6, despite up-titration of her insulin doses from day 5 onwards. Therefore, the patient had her dose of insulin lispro increased to 15 U/day (morning, 4 U; afternoon, 6 U; and evening, 5 U) and her dose of insulin degludec increased to 12 U/day.

Then, with a fasting plasma glucose level of 112 mg/dL achieved on day 7, she went on to have the left lobe of her thyroid gland removed. She received no steroid preoperatively or postoperatively. She received maintenance fluid (Na+ 35 mEq/L, K+ 20 mEq/L, Cl− 35 mEq/L, lactate− 20 mEq/L, glucose 43 g/L) at 60 mL per hour postoperatively until before the evening meal on day 8. On day 8, the day after the surgery, her hyperglycemia persisted, with a fasting plasma glucose level of about 200 mg/dL. Hence, her insulin regimen was changed to insulin degludec 11 U/day plus sliding scale therapy with insulin lispro. Her diet was resumed from the afternoon of that day onwards.

Given that she was expected to be unable to eat enough due to postoperative pain, her dose of insulin lispro on that day was set at 8 U/day (afternoon and evening, 4 U each), but she was found to be hyperglycemic, with sensor glucose levels >350 mg/dL after dinner. With her postoperative pain shown to be ameliorated on day 9, her insulin regimen was increased to a maximum dose of insulin lispro of 23 U/day (morning and afternoon, 5 U each; evening, 6 U; and correction bolus, 7 U) and insulin degludec 11 U/day, but her hyperglycemia persisted, with a mean sensor glucose level of 233.4 mg/dL.

After ensuring that her dietary intake remained stable, SGLT2 inhibitor therapy was restarted on day 11 (4 days after surgery), which led to a declining tendency in her sensor glucose level on the following day (day 12). Ultimately, with her mean sensor glucose level decreased to 160.4 mg/dL, her dose of insulin lispro decreased to 12 U (morning, afternoon, and evening, 4 U each), and her dose of insulin degludec decreased to 10 U/day, the patient was discharged on day 17 and she had no postoperative complications such as infection or recurrent nerve palsy.

Discussion

The use of SGLT2 inhibitors has been associated with an increased risk of diabetic ketoacidosis in patients with type 1 diabetes [8], and perioperative use of SGLT2 inhibitors may pose a risk of diabetic ketoacidosis [9]; hence, the rationale for discontinuing SGLT2 inhibitor therapy in our patient 5 days before surgery. The patient’s glycemic control while on SGLT2 inhibitor therapy was improved, but she had an acute rise in her glucose level the day after discontinuation of SGLT2 inhibitor therapy, which made glycemic control difficult to achieve. Intermittently scanned continuous glucose monitoring readings showed a tendency towards increased sensor glucose levels from the first day after discontinuation of SGLT2 inhibitor therapy onwards, eventually leading to her bolus insulin dose being increased to close to double that required before discontinuation of SGLT2 inhibitor therapy. Generally, a patient with HbA1c 8% is shown to have an average blood glucose level of around 180 mg/dL [10]. With her HbA1c level shown to be about 8%, it was thought conceivable that her average blood glucose level could be elevated to around 180 mg/dL; ultimately, however, it rose to as high as 273 mg/dL on day 6. The 5-day preoperative discontinuation of SGLT2 inhibitor therapy led to a greater-than-expected rise in her glucose level, which proved difficult to control, even with delayed up-titration of the insulin dose. Subsequently, her glucose level decreased from the first day after re-initiation of SGLT2 inhibitor therapy, with her insulin dose also decreased to that required before discontinuation of SGLT2 inhibitor therapy. This clinical course may be explained by the recovery of insulin resistance associated with invasive surgery, but it may also have reflected the effect of restarting SGLT2 inhibitor.

According to Araki et al, dapagliflozin 10 mg produces a reduction in mean HbA1c value of 0.36%, as well as a reduction in insulin dose of about 13% [3]. However, to date, no study has investigated how much blood glucose or required insulin dose can be increased in patients with type 1 diabetes discontinuing dapagliflozin or SGLT2 inhibitor therapy. Indeed, while 1 study reported that discontinuing SGLT2 inhibitor therapy did not lead to increased glucose levels in patients with type 2 diabetes undergoing surgery [11], it remains unclear how discontinuing SGLT2 inhibitor therapy affects glycemic control in patients with type 1 diabetes before undergoing surgery. Our patient’s glucose level tended to increase from day 1 after discontinuation of SGLT2 inhibitor therapy onwards, with her mean sensor glucose level ranging between 109.7 and 273 mg/dL and their standard deviations deteriorating from 25.3 to 55.2 mg/dL. Given that dapagliflozin is assumed to have a half-life of about 8–12 h [12], its discontinuation is thought likely to lead to an increase in glucose level that continues for the following few days in patients who have regularly received the drug – a clinical course also confirmed in our patient as well. Indeed, before re-initiation of SGLT2 inhibitor therapy, our patient had to have her bolus insulin dose increased from her usual dose of 12 U to 23 U for glycemic control, which may be accounted for not only by the discontinuation of SGLT2 inhibitor therapy but also by increased insulin resistance associated with invasive surgery. Although our patient was not evaluated for ketones, it appears advisable to watch for higher-than-expected glycemic elevations, given the risk of diabetic ketoacidosis associated with perioperative SGLT2 inhibitor discontinuation.

To avoid these conditions, therefore, it was suggested that when a patient with type 1 diabetes discontinues the SGLT2 inhibitor before surgery, as in this case, the insulin dose should be increased as soon as the glucose level increases. Also of note, as poor glycemic control was noted in our patient even after re-initiation of SGLT2 inhibitor therapy, her insulin dose could not be decreased on the same day that SGLT2 inhibitor therapy was re-initiated; it was only gradually decreased over time, with the monitoring of changes in her glucose level. Furthermore, as dapagliflozin is assumed to reach its maximum concentration within 1 h after administration [12], it appears that insulin doses need to be reduced as rapidly as possible in likely responders to SGLT2 inhibitors, such as our patient, upon re-initiation of SGLT2 inhibitor therapy and thereafter.

Conclusions

We encountered a case of acute hyperglycemia occurring in a patient with type 1 diabetes discontinuing SGLT2 inhibitor before undergoing surgery, which led to an unexpected increase in mean sensor glucose level to 273 mg/dL within 5 days after drug discontinuation and required increasing her bolus insulin dose. The present case is important because similar cases are likely to be encountered with preoperative discontinuation of SGLT2 inhibitor therapy in patients with type 1 diabetes before undergoing surgery. It was deemed advisable that, whenever possible, attention be focused on increasing insulin doses as rapidly as possible, under continuous glucose monitoring, in patients with type 1 diabetes discontinuing SGLT2 inhibitor therapy before undergoing surgery.