Euglycemic Ketoacidosis in a Patient without Diabetes Taking Sodium-Glucose Cotransporter 2 Inhibitors for Heart Failure

Introduction

Sodium-glucose cotransporter 2 (SGLT2) inhibitors increase urinary glucose excretion, reducing blood glucose levels and promoting weight loss [1]. SGLT2 inhibitors may also have a cardioprotective effect against heart failure owing to their diuretic and sympathetic modulatory effects, may improve myocardial energy metabolism, and may have a renoprotective effect [2]. Guidelines for heart failure treatment advocate the early use of SGLT2 inhibitors in patients with or without diabetes [3]. The use of these drugs is expected to increase in the future.

Although it is clear that SGLT2 inhibitors can cause ketoacidosis in patients with type 2 diabetes [4], it is unclear whether SGLT2 inhibitors are associated with a risk of ketoacidosis in people without diabetes. Currently, reports on SGLT2-related euglycemic ketoacidosis in patients without diabetes are limited [5–7]. Because physiologic stress-induced changes due to surgery in addition to fasting increase the risk of ketoacidosis through the release of catecholamines, cortisol production, and decreased insulin secretion and utilization [8], perioperative withdrawal of SGLT2 inhibitors is recommended in patients with or without diabetes before undergoing surgery [9].

To our knowledge, there are no reports in Japan of SGLT2 inhibitor-related ketoacidosis in people without diabetes, occurring independently of the perioperative period.

Here, we report a case of ketoacidosis in a Japanese woman without diabetes who was taking SGLT2 inhibitors for heart failure.

Case Report

Our patient was an 83-year-old Japanese woman with chronic heart failure (ejection fraction: approximately 30%) who was initiated on empagliflozin 10 mg daily about 8 months before admission. She had never been diagnosed with diabetes mellitus (HbA1c 5.9% before starting empagliflozin). About 6 months before hospitalization, her treatment was changed to dapagliflozin 5 mg. Her diuretics and other cardiac medications were not adjusted during that time. Two weeks before admission, she tripped and fell, sustaining vertebral and rib fractures. The patient became significantly debilitated from the pain from the fractures, which caused difficulty walking and anorexia. She could eat only 50–70% of her usual dietary intake. This got worse in the following week when she also developed nausea and vomiting and further reduced her food intake to only 10–20% of the usual amount. During this period, she continued to take dapagliflozin. She was admitted to the hospital.

On admission, she was conscious with a body temperature of 37.4°C, blood pressure of 124/68 mmHg, heart rate of 91 bpm, and oxygen saturation of 98% breathing room air. Blood tests revealed a blood glucose level of 124 mg/dL, pH of 7.3, HCO3– concentration of 14 mmol/L, base excess of −10 mEq/L, anion gap of 20 mEq/L, and β-hydroxybutyrate concentration of 5150 µmol/L (Table 1). These findings were consistent with the diagnosis of normoglycemic ketoacidosis. The patient had no history of alcohol consumption, smoking, or drug use. Other causes of high anion gap acidosis were ruled out. Her C-peptide level was consistent with the blood glucose levels on admission, indicating that she had adequate insulin secretion; therefore, insulin was not initiated. SGLT2 inhibitor use was discontinued, and the patient was initially treated with isotonic saline and a maintenance infusion of approximately 170 g glucose daily (Figure 1).

Despite not starting insulin, her pH (7.418), HCO3– concentration (20.7 mmol/L), and anion gap (13.3 mEq/L) improved on the day after admission. The Carb 60 diet was initiated, but intravenous glucose infusions were continued because the patient’s food intake was poor. The amount of glucose infusion administered was adjusted according to her food intake. On the fourth day of hospitalization, urinary ketones disappeared. The glucose infusion was terminated on the 12th day of hospitalization once the patient could eat full meals. Although the patient’s blood glucose level was within the normal range, the urinary glucose excretion was 3+ (>500 mg/dL) for 8 days after the last dose of dapagliflozin.

On admission, the patient was unable to walk due to pain from the fracture, general fatigue, and disuse syndrome, but after rehabilitation, she was able to walk independently. She was discharged on day 19 of hospitalization after undergoing rehabilitation. The SGLT2 inhibitors were not restarted.

Discussion

SGLT2 inhibitor use in patients with type 2 diabetes is known to be a risk for diabetic ketoacidosis [4], whereas people without diabetes maintain adequate insulin secretion and are usually considered unlikely to experience ketoacidosis [10]. Ketoacidosis in patients without diabetes has not been reported in large placebo-controlled trials of SGLT2 inhibitors used by patients in heart failure [11–14]. However, one case of ketoacidosis was reported in a non-diabetic patient treated with empagliflozin in the EMPA-KIDNEY study, although case details were not provided [15]. Recently, a few cases of SGLT2 inhibitor-associated ketoacidosis in the absence of diabetes have also been reported outside of clinical trials [5–7], suggesting that SGLT2 inhibitor use may occasionally cause ketoacidosis in non-diabetic patients.

SGLT2 inhibitors are thought to increase urinary glucose excretion, decreasing blood glucose levels. This results in a relative decrease in insulin secretion and a decrease in the insulin-to-glucagon ratio, thereby increasing lipolysis. Free fatty acids are metabolized in the liver to produce ketone bodies [16]. Notably, there are reports that kidney ketone excretion may also be decreased [17]. SGLT2 inhibitors have further been shown to decrease the insulin-to-glucagon ratio and promote lipolysis and ketogenesis in patients with diabetes and those without diabetes [18]. These findings suggest that ketoacidosis may occur even in patients without diabetes mellitus with preserved insulin secretion. Patients with heart failure often take diuretics in addition to SGLT2 inhibitors, and their simultaneous use may contribute to dehydration. In turn, dehydration promotes glucagon secretion and is a potential risk factor for ketoacidosis [7,19]. The mechanism by which ketoacidosis occurs in patients without diabetes mellitus remains unclear; however, the action of SGLT2 inhibitors may contribute to the risk of ketoacidosis in these patients.

In patients without diabetes, ketosis can be induced by starvation or alcohol. Starvation-induced ketosis can be caused by ketogenic diets or intentional or unintentional malnutrition, but generally, the bicarbonate ion (HCO3–) remains above 18 mEq/L [19]. In addition, in healthy individuals, elevated ketone bodies appear after 2–4 days of fasting, and it takes approximately 2 weeks to exceed 5000 µmol/L [20].

In the present case, there was no history of chronic alcohol consumption. Although the patient had hardly eaten for a few days, the marked elevation of ketones was not consistent with the level of reduction in food intake. Although it is not easy to clearly distinguish SGLT2 inhibitor-related ketoacidosis from starvation-induced ketosis in patients without diabetes because both are triggered by low carbohydrate intake, SGLT2 inhibitors aggravate this condition.

A few reports on SGLT2 inhibitor-associated ketoacidosis in patients without diabetes have shown that a significantly lower dose of insulin is generally required for treatment compared with the dose required in diabetic ketoacidosis [5,6]. Seki et al [7] reported treatment of SGLT2 inhibitor-related ketoacidosis with glucose without insulin. This may be because insufficient glucose primarily causes SGLT2-related euglycemic ketoacidosis with preserved insulin secretion. In the present case, the patient developed ketoacidosis due to insufficient carbohydrate intake, combined with the use of SGLT2 inhibitors and diuretics for heart failure. She was treated with glucose supplementation and discontinuation of the SGLT2 inhibitors. As a result, the patient had a temporary period of blood glucose levels exceeding 200 mg/dL. This may have been caused by a rapid glucose infusion rate under high-stress conditions, decreased stimulation of insulin secretion by SGLT2 inhibitors, and relative insulin deficiency. Although we considered insulin infusion, we did not use insulin infusion in this case because insulin secretion was maintained in response to blood glucose on admission.

In addition, despite blood glucose levels being in the normal range for a long period, the patient’s glucosuria persisted. This suggests that the effects of dapagliflozin persisted. Persistent urinary glucose excretion has been reported even after the discontinuation of SGLT2 inhibitors [21]. Considering that the estimated half-life of dapagliflozin is 11.2–16.6 hours [22], the presence of glucosuria in this case until 8 days after the last use of dapagliflozin suggests that glucose excretion occurred as a sustained pharmacological response even after the blood concentration of dapagliflozin decreased. This suggests that caloric balance should be monitored for some time after the discontinuation of SGLT2 inhibitors, regardless of whether the patient has diabetes, and education and reminders regarding their use should be provided even if the patient has not undergone surgery.

Conclusions

The continued use of SGLT2 inhibitors combined with an inadequate dietary intake of glucose due to various causes, such as bone fractures or depression, may lead to euglycemic ketoacidosis, as illustrated by the present case. Thus, SGLT2 inhibitors should be used with caution during periods when dietary glucose may be inadequate, even in patients who have not undergone surgery. All patients should be cautioned about the risk of ketoacidosis when SGLT2 inhibitors are prescribed, regardless of whether they have diabetes.