Sodium–glucose cotransporter 2 (SGLT2) inhibitors have been linked with diabetic ketoacidosis (DKA) (14). However, characteristics of “at-risk” patients are less well described.

Clinical cases were collected from South Australia between December 2015 and March 2017. The audit was approved by the Royal Adelaide Hospital Human Research Ethics Committee (HREC reference no. HREC/17/RAH/185).

The Therapeutic Goods Administration (TGA) of the Australian Government Department of Health was contacted, and a search of the Database of Adverse Event Notifications (medicines) for all reports of SGLT2 inhibitor–associated DKA was requested up until April 2017 using the search terms ketoacidosis, diabetic ketoacidosis, and generic and brand drug names. Full case line listings for each notification were provided by the TGA.

Autoantibodies against glutamic acid decarboxylase (GAD) and insulinoma-associated antigen 2 (tyrosine phosphatase IA-2) were quantitated using commercial ELISA test kits (Euroimmun AG, Lubeck, Germany). These assays are calibrated in international units (IU) using the first World Health Organization reference reagent for islet cell antibodies (1999, reagent 97/550; National Institute for Biological Standards and Control, Hertfordshire, U.K.), which contains 100 IU of anti-GAD and 100 IU of anti–IA-2 per ampoule by definition.

We identified thirteen cases of SGLT2 inhibitor–associated DKA occurring in South Australia between December 2015 and March 2017 (Table 1). Eight case subjects had type 2 diabetes (T2D) and five had type 1 diabetes (T1D), although two patients were identified as having T1D or latent autoimmune diabetes of adults (LADA) retrospectively after DKA presentation based on positive anti-GAD and anti–IA-2 antibodies. Insulin was coprescribed in half of the patients with T2D (n = 4), was not prescribed in three patients, and was unknown in one patient, raising the possibility of misdiagnosis of T1D or LADA despite antibodies not being detected or indicating advanced T2D. Nine patients required intensive care or high dependency care, and all patients required i.v. insulin and dextrose. Dapagliflozin was implicated in nine and empagliflozin in four cases of DKA. One patient died due to Takotsubo cardiomyopathy. Treating doctors initially overlooked the diagnosis of DKA in two patients and were unaware of the association of SGLT2 inhibitors and DKA in six patients.

Precipitating events were identified in most cases. These included missed insulin (n = 5), undiagnosed T1D (n = 2), infection (n = 5), and surgery (n = 3). Five case subjects had reduced carbohydrate intake precipitating DKA (fasting for surgery, low carbohydrate diet, or anorexia).

Up until April 2017, 82 unique notifications of SGLT2 inhibitor–associated DKA have been reported to the Australian TGA. Reports include twenty-four patients with T2D and nine patients with T1D (two diagnosed retrospectively). Two reports of DKA occurred with off-label drug use for weight loss or insulin resistance, both precipitated by a gastrointestinal illness. Diabetes status was not reported in the majority (n = 47). SGLT2 inhibitors were mostly discontinued; however, they were recommenced in three reports (at lower doses in two). Most reports described treatment with i.v. insulin and dextrose. Eighteen reports of DKA were managed in intensive care, and sixteen were classified as life-threatening.

Precipitants were surgery or perioperative (n = 13), reduction or cessation of insulin (n = 2), low-carbohydrate diet (n = 1), acute coronary events (n = 3), infections (n = 3), and liver and kidney impairment (n = 1). Average duration of SGLT2 inhibitor use prior to DKA presentation was 11.6 weeks (range 1 day to 76 weeks). DKA was generally serious, with mean pH 7.06, bicarbonate 7.35 mmol/L, ketones 6.2 mmol/L, and anion gap 23.7 mmol/L. The mean glucose was 14.1 mmol/L (range 4.8 to 35 mmol/L), and four patients had glucose ≤8.0 mmol/L.

We suggest temporary cessation of these drugs during acute illness and surgery. The SGLT2 inhibitor–associated DKA illness prodrome was similar to that of DKA in T1D; however, DKA often occurs in the absence of marked hyperglycemia. Thus, early detection of the ketotic state while symptomatic with malaise, nausea, and/or vomiting, particularly in the context of precipitants, could prompt temporary cessation of SGLT2 inhibitor, hydration, frequent carbohydrate consumption, and administration of full-dose insulin to prevent progression to DKA (2). Consider excluding a diagnosis of T1D before prescribing these medications or reevaluating diabetes diagnosis upon DKA presentation.

The TGA reports are unlikely to have captured all cases of SGLT2 inhibitor–associated DKA as adverse drug events are generally underreported (5). Further limitations include variable and incomplete case information, with reporting bias subject to media influence.

Within South Australia between December 2015 and March 2017, 20,548 empagliflozin and 65,303 dapagliflozin scripts were dispensed. This equates to 85,851 SGLT2 inhibitor scripts dispensed over the 16-month duration in which our case series was conducted. This equates to 7,154.25 patient-years of SGLT2 inhibitor treatment in which 13 SGLT2 inhibitor–associated DKA cases occurred, resulting in an adverse event rate of 1.8 cases of SGLT2 inhibitor–associated DKA per 1,000 patient-years. While this estimate is higher than all but one previous, it must be treated with caution (14). Limitations include uncertainty around case detection and medication compliance. SGLT2 inhibitor–associated DKA incidence requires further validation.

Key issues identified were that 1) most patients did not recognize DKA, 2) treating physicians often did not initially recognize DKA due to relative euglycemia, and 3) effective treatment was delayed. Most cases were severe, with one associated death, the second within the literature (4). Identifiable precipitants were often present, suggesting the potential for risk mitigation.

Acknowledgments. The authors thank Kylies Moore (pharmacist, South Australia Pharmacy Medicines Information Service) for providing South Australian SGLT2 inhibitor prescription data, Sonya J. Conrad (Women’s Information Service, South Australia) for analysis of TGA reports, and the TGA for providing nationally reported notifications of adverse events of SGLT2 inhibitor–associated DKA.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. E.J.M. researched data, performed the audit, and wrote the manuscript. G.G. contributed to the discussion and reviewed the manuscript. D.J. contributed to the audit and reviewed the manuscript. E.J.M. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the Australian Diabetes Society and Australian Diabetes Educators Association Annual Scientific Meeting, Perth, Australia, 30 August to 1 September 2017.

1.
Fralick
M
,
Schneeweiss
S
,
Patorno
E
.
Risk of diabetic ketoacidosis after initiation of an SGLT2 inhibitor
.
N Engl J Med
2017
;
376
:
2300
2302
2.
Rosenstock
J
,
Ferrannini
E
.
Euglycemic diabetic ketoacidosis: a predictable, detectable, and preventable safety concern with SGLT2 inhibitors
.
Diabetes Care
2015
;
38
:
1638
1642
3.
Handelsman
Y
,
Henry
RR
,
Bloomgarden
ZT
, et al
.
American Association of Clinical Endocrinologists and American College of Endocrinology position statement on the association of SGLT-2 inhibitors and diabetic ketoacidosis
.
Endocr Pract
2016
;
22
:
753
762
4.
Isaacs
M
,
Tonks
KT
,
Greenfield
JR
.
Euglycaemic diabetic ketoacidosis in patients using sodium-glucose co-transporter 2 inhibitors
.
Intern Med J
2017
;
47
:
701
704
5.
Hazell
L
,
Shakir
SA
.
Under-reporting of adverse drug reactions: a systematic review
.
Drug Saf
2006
;
29
:
385
396
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. More information is available at http://www.diabetesjournals.org/content/license.