Prolonged glucosuria and relapse of diabetic ketoacidosis related to SGLT2‐inhibitor therapy

Abstract SGLT2 inhibitors (SGLT2i) are glucose‐lowering medications which increase the renal threshold for glucose reabsorption and promote glucosuria. Treatment with these agents raises serum ketone levels, and cases of diabetic ketoacidosis (DKA) during therapy have been reported. The duration of glucosuria and inpatient course of SGLT2i‐related DKA, however, is not well‐characterized. We report 11 inpatient cases of SGLT2i‐related DKA, including a subset of patients who experienced prolonged glucosuria and relapse of DKA during their hospitalization.

To the Editor Sodium-glucose cotransporter-2 inhibitors (SGLT2i) have become increasingly popular glucose-lowering agents, and interest in this class has broadened due to demonstration of cardioprotection in patients with established cardiovascular disease and slowing the progression of renal dysfunction. 1 There is also interest in extending this therapy to patients without diabetes, as it appears to lower the risk of heart failure or cardiovascular death, even in patients with heart failure in the absence of diabetes. 2 SGLT2i increase serum ketones, and a recognized adverse effect of SGLT2i therapy is diabetic ketoacidosis (DKA). While the incidence and risk factors for developing DKA on SGLT2i have been studied, 3 and recognition and prevention of DKA have become part of society and consensus guidelines, 4,5 it has been assumed that DKA in this setting mimics classical DKA typically seen in type 1 diabetes. However, there are no detailed studies of the inpatient course of patients who develop this adverse medication effect. Therefore, we performed a systematic medical records search at a large academic medical centre in Boston, MA to identify common patterns and inpatient course for cases with DKA associated with SGLT2i. Venous pH is reported, as arterial pH is not routinely obtained at this institution to diagnose DKA. Serum ketones were not available for most cases. The duration of intravenous insulin infusion, persistence of glucosuria and time to close anion gap (5), defined as time elapsed from the first available serum chemistry to the chemistry measuring an AG ≤ 16, were extracted from medical records.
A total of 273 unique patient records were identified and reviewed for inclusion. Reasons for exclusion included: no inpatient or emergency room admission for DKA (249 cases) and not on SGLT2i prior to admission (13 cases). The remaining 11 cases were analysed for this study.
Data for the 11 cases are summarized in Table 1. Patients with type 2 diabetes tended to be older and have more comorbidities than those with type 1. Among all patients, the average time to resolve DKA as defined by closure of the AG was 37.8 hours, ICU stay averaged 4.5 days, and total length of stay was 9.5 days. Our investigation reveals a subset of six patients who had prolonged ICU courses of four or more days, were on an IV insulin infusion for over 40 hours, had glucosuria for at least 3 days following admission, and in three cases, had relapse of DKA while admitted. Glucosuria persisted for up to 10 days after discontinuing the SGLT2i despite adequate glycemic control in most cases, suggesting that glucosuria is evidence of continued drug action.
For one case (#5), urine glucose concentration was quantified during hospitalization (see Figure 1) and demonstrated significant elevation of urine glucose for several days after discontinuation of the drug despite normal plasma glucose levels. Of note, two cases (#7 and #8) had recently initiated a ketogenic diet.
Our study suggests not only that the time to close the anion gap may be longer in patients with SGLT2i-related DKA, but also that even once the gap has closed, inadequate insulin or carbohydrate delivery can result in recurrent ketosis and risk of DKA relapse due to the ongoing glucosuric effects of the medication. Treatment of DKA relies principally on volume resuscitation and adequate insulin and carbohydrates to suppress ketosis. In classical DKA in patients with type 1 diabetes, this is often accomplished relatively quickly; multiple prior studies have indicated that median time to resolution of classical DKA is approximately 11 hours and total hospital length of stay is 3 days, though patients with classical DKA tend to be younger and with fewer comorbidities compared with those on SGLT2i. 6,7 This study is the first detailed analysis of the inpatient course of all SGLT2i-related DKA cases within a single centre and including an evaluation of glucosuria. Prior investigations have focused primarily

ACK N OWLED G EM ENTS
We thank Karla Pollick, MHA and George S. Silva, BA of the InSIGHT Core in the Center for Healthcare Delivery Science at BIDMC for their assistance searching the electronic medical record for data collection, for which they were not compensated.

CO N FLI C T O F I NTE R E S T
The authors have no relevant conflicts of interest to disclose.