Summary

• Canagliflozin is not recommended in patients with type 1 diabetes mellitus (T1DM). It may increase the risk of diabetic ketoacidosis (DKA) in these patients.¹
• On December 4, 2015, the Food and Drug Administration (FDA) issued a Drug Safety Communication informing that sodium-glucose cotransporter (SGLT)2 inhibitor class of drugs had their Full Prescribing Information updated to include a warning regarding the risk of ketoacidosis.²
• Reports of ketoacidosis, a serious life-threatening condition requiring urgent hospitalization, have been identified in clinical trials and postmarketing surveillance in patients with T1DM and type 2 diabetes mellitus (T2DM) receiving SGLT2 inhibitors, including canagliflozin. The risk of ketoacidosis may be greater with higher doses. Fatal cases of ketoacidosis have been reported in patients taking canagliflozin.¹
• In the CREDENCE study, the overall incidence of confirmed DKA events occurred at a rate of 2.2 vs 0.2 events per 1000 PY in the INVOKANA and placebo groups, respectively (hazard ratio [HR], 10.8; 95% confidence interval [CI], 1.39-83.65).³
• In the CANVAS Program (N=10,142), comprised of 2 large INVOKANA cardiovascular (CV) outcome studies CANVAS and CANVAS-R, DKA was an adverse event (AE) of interest. Analyses included participants who received at least one dose of drug or placebo and had an event at any time during follow-up. A small number of DKA events were observed with INVOKANA and placebo (0.6 vs 0.3 per 1000 patient-years, respectively).⁴
• The incidence of serious AE cases reported as DKA, ketoacidosis, metabolic acidosis, or acidosis was <0.1% in a pooled analysis of safety data from 15 completed and ongoing randomized controlled trials of INVOKANAwhich enrolled more than 17,000 patients diagnosed with T2DM.⁵
• An observational study, a first of its kind, using a large US insurance claims database showed that there was no statistically significant difference for DKA risk between new users of SGLT2 inhibitors and non-SGLT2 inhibitors antihyperglycemic agents (AHAs) among patients with T2DM. The HR (95% CI) for DKA comparing new users of SGLT2 inhibitors to new users of non-SGLT2 inhibitors AHAs was 1.91 (0.94-4.11) (P=0.09) among the 30,196 exposure propensity scores (EPS)-matched pairs overall, and 1.13 (0.43-3.00) (P=0.81) among 27,515 EPS-matched pairs that excluded potential autoimmune diabetes.⁶
• A population-based cohort study by Fralick et al (2021) using data from a US commercial claims database (March 29, 2013-September 30, 2015) showed that canagliflozin was associated with DKA in two pairwise cohorts.⁷

BACKGROUND

In December 2015, the FDA issued a Drug Safety Communication² to report that the SGLT2 inhibitor class of drugs had their Full Prescribing Information updated to include a warning regarding the risk of ketoacidosis. More information is available at the FDA website.

Other international health authorities issued similar safety advisories on the risk of DKA in patients taking SGLT2 inhibitors.^8-18

A Dear Health Care Provider letter¹⁹ was issued by Janssen jointly with other pharmaceutical companies in July 2015 to ensure widespread dissemination of ketoacidosis safety information.

CLINICAL DATA - T2DM

CREDENCE (Canagliflozin and Renal Events in Diabetes with Established Nephropathy Clinical Evaluation), was a randomized, double-blind, placebo-controlled, parallel group multicenter, event driven clinical trial to assess the effects of INVOKANA (100 mg) compared to placebo on clinically important renal outcomes in people with T2DM and established chronic kidney disease (estimated glomerular filtration rate [eGFR] 30 to <90 mL/min/1.73 m²) and albuminuria (ratio of albumin to creatinine >300 to 5000 mg/g), who were receiving a stable, maximum tolerated or labelled dose (for >4 weeks prior to randomization) of an angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker.^3,20-24

• Patients with a history of DKA or T1DM were excluded from the study.
• Study drug administration was to be continued until completion of the study, commencement of dialysis, receipt of renal transplant, occurrence of DKA, pregnancy, or receipt of disallowed therapy.
  • DKA was designated as a key safety outcome, which was adjudicated by independent Endpoint Adjudication Committees blinded to treatment allocation, as well as a selected AE of interest.

Results

• A total of 4401 participants were randomized in the intention-to-treat analysis set and 4397 participants in the on-treatment and on-study analysis sets.³
  • Analyses of DKA and other AEs of special interest were performed using the on-treatment analysis set.
• Baseline characteristics were similar between the INVOKANA and placebo groups. Across both treatment groups, the mean exposure to study drug was 115 weeks and median follow-up was 2.62 years.³
• In the CREDENCE study, the overall rate of confirmed DKA was 2.2 and 0.2 events per 1000 patient-years in the INVOKANA (n=11/2200) and placebo (n=1/2197) groups, respectively (HR, 10.80; 95% CI, 1.39-83.65).³
  • All potential ketone-related events were adjudicated for DKA by an independent adjudication committee based on clinical presentation and predefined biochemical parameters.²¹
• See Table: Baseline Characteristics of Patients with DKA Events Regardless of Treatment.

The CANVAS Program (N=10,142)^4,25 is comprised of 2 large INVOKANA CV outcome studies, including CANagliflozin cardioVascular Assessment Study (CANVAS)^26,27 and CANagliflozin cardioVascular Assessment Study-Renal (CANVAS-R)^28,29, designed to evaluate the potential for CV protection of INVOKANA in patients with T2DM who had either a prior history of CV disease or >2 CV risk factors.^4,25 The integrated analysis also evaluated the effects of INVOKANA on renal and safety outcomes.⁴ The mean follow-up in the CANVAS Program was 188.2 weeks, and the median follow-up was 126.1 weeks.

• A small number of DKA events (adjudicated) were observed with INVOKANA and placebo (0.6 vs 0.3 per 1000 patient-years; HR, 2.33; 95% CI, 0.76-7.17; P=0.14).

Erondu et al (2015)⁵ performed a pooled analysis of safety data from 15 randomized controlled trials (11 completed^30-44 and 4 ongoing^26,28,45,46 at the time of the analysis) to identify and assess serious AEs of DKA, ketoacidosis, metabolic acidosis, or acidosis that occurred among 17,596 patients in the INVOKANA T2DM Clinical Program. All 15 trials were designed to exclude patients with a history of T1DM and/or DKA.

• Through 11 May 2015, there were 12 patients with 13 unblinded serious AEs of DKA, ketoacidosis, metabolic acidosis, and acidosis, and 3 additional serious AEs (from the CANVAS trial²⁶) that remained blinded and were not included in the pooled analysis.⁵
• The overall incidence of serious AEs of DKA and related events was 0.07% (12 of 17,596).
  • The incidence by treatment group was 0.07% (4 of 5337), 0.11% (6 of 5350), and 0.03% (2 of 6909) with INVOKANA 100 and 300 mg and comparator (placebo or active-control), respectively.⁵
• Although history of T1DM was an exclusion criterion in all the studies, 6 of the 10 patients on INVOKANA who experienced DKA (3 on 100 mg, 3 on 300 mg) had evidence of autoimmune diabetes (T1DM, latent autoimmune diabetes of adulthood [LADA], or tested positive for GAD65 antibodies).
  • Excluding these 6 patients, the incidences of serious AEs of DKA and related events were 0.02% (1 of 5334), 0.06% (3 of 5347), and 0.03% (2 of 6909) with INVOKANA 100 and 300 mg and comparator, respectively.⁵
• Among the 10 INVOKANA-treated patients who had DKA, nine had blood glucose >300 mg/dL at presentation. One patient on INVOKANA had blood glucose levels ranging from 148 to 320 mg/dL, but dates and times of the measurements were not provided.⁵
• Among the 8 INVOKANApatients on insulin, 4 had questionable compliance with insulin therapy. One of the four had a second event post-cholecystectomy. The other four patients on insulin had concomitant diagnoses of pancreatic cancer, myocardial infarction, gastroenteritis, and viral infection.⁵
• Among the two INVOKANApatients not on insulin, one was found to have T1DM and one had a subcutaneous abscess and chronic pancreatitis.⁵
• More specific details about the 12 patients with serious AEs of DKA and related events are described in Table 2 and Table 3 of the published article.⁵

The Erondu et al (2015)⁵ publication has generated correspondence in the literature.^47-49

Qiu et al (2017)⁵⁰ reported on the longer-term safety of INVOKANA, including incidence of DKA, and assessed seven of the 15 randomized controlled trials evaluated in Erondu et al (2015)⁵.

Inagaki et al (2015)⁵¹ conducted a randomized, open label, 52-week trial (N=1,299) in Japanese patients with T2DM to evaluate the safety and efficacy of INVOKANA 100 mg or 200 mg daily as monotherapy or in combination with one additional antidiabetic agent.

• One episode of hyperglycemic (≥600 mg/dL) severe DKA occurred 13 days after discontinuing study treatment in a patient receiving INVOKANA 100 mg in combination with a sulfonylurea. Laboratory data were generally consistent with fulminant T1DM, and the patient had recently developed infectious gastroenteritis.

An observational study by Wang et al (2017)⁶ using a large US insurance claims database showed that there was no statistically significant difference for DKA risk between new users of SGLT2 inhibitors and non-SGLT2 inhibitor AHAs among patients with T2DM. Pair-matched on exposure propensity scores (EPS) using Cox regression models showed incidence rates (95% CI) per 1000 patient-years for DKA of 1.69 (1.22-2.30) and 1.83 (1.58-2.10) among new users of SGLT2 inhibitors (n=34,442) and non-SGLT2 inhibitor AHAs (n=126,703). The HR (95% CI) for DKA comparing new users of SGLT2 inhibitors to new users of non-SGLT2 inhibitor AHAs was 1.91 (0.94-4.11) (P=0.09) among the 30,196 EPS-matched pairs overall, and 1.13 (0.43-3.00) (P=0.81) among 27,515 EPS-matched pairs that excluded potential autoimmune diabetes.

A population-based cohort study by Fralick et al (2021)⁷ using data from a US commercial claims database evaluated adult patients who were newly prescribed canagliflozin or an active comparator (dipeptidyl peptidase-4 inhibitor [DPP4] or glucagon-like-peptide 1 receptor agonist [GLP1]) between March 29, 2013-September 30, 2015. The analysis included two pairwise propensity score variable ratio match cohorts (44,733 canagliflozin versus 99,458 DPP4 initiators, and 55,974 canagliflozin versus 74,727 GLP1 initiators). When screened for inpatient and emergency room diagnoses, DKA was the only serious adverse event associated with canagliflozin initiation in both cohorts.

Additional citations related to DKA and SGLT2 inhibitors have been published.^52-56

CLINICAL DATA - T1DM

Henry et al (2015)⁵⁷ conducted a phase 2, randomized, parallel-group, double-blind, 18-week study to assess the efficacy and safety of INVOKANA 100 mg and 300 mg, compared with placebo, as add-on to insulin for the treatment of T1DM (N=351).

• Patients were excluded if they had a DKA episode within 6 months prior to study start.
• Safety and tolerability were assessed based on AE reports, safety laboratory tests, vital sign measurements, and physical examination.
• Analysis of ketone-related AEs was performed using a pre-specified list of preferred
• Safety and tolerability were similar to that seen in T2DM studies, except for an increased incidence of ketone-related AEs, including DKA.
  • Ketone-related AEs occurred in 5.1% (6 of 117) on INVOKANA 100 mg, 9.4% (11 of 117) on INVOKANA 300 mg, and no patients on placebo.
• Among the 17 patients with ketone-related AEs, 12 were female, four had prior history of DKA, and 12 were serious (requiring hospitalization). Blood glucose levels at time of hospitalization ranged from 170 to >800 mg/dL with five patients having blood glucose level <250 mg/dL.
• All serious AEs of DKA had precipitating factors (eg, flu, pneumonia, insulin pump failure, inappropriate insulin use).

Peters et al (2016)⁵⁸ conducted a post hoc analysis of the DKA events from the phase 2 study reported by Henry et al (2015)⁵⁷.

• No apparent differences were observed in baseline characteristics in patients with compared to those without a ketone-related AE that would predict greater or lesser risk.

POSTMARKETING CASE REPORTS IN THE PUBLIC DOMAIN

The FDA review of the FDA Adverse Event Reporting System (FAERS) database from March 2013 through May 2015 identified 73 postmarketing cases (INVOKANA [n=48]; dapagliflozin [n=21]; empagliflozin [n=4]) of ketoacidosis reported with SGLT2 inhibitors.²

• Forty-four cases were reported in patients with T2DM, 15 cases were reported in patients with T1DM, and 13 cases did not report a diabetes type. One case was reported in a patient with latent autoimmune diabetes.
• In all cases, the patients were hospitalized or treated in the emergency department.
• Forty-four of the 73 total cases reported at least 1 diagnostic laboratory criteria suggestive of ketoacidosis.
• The median time from initiation or increase in dose of the SGLT2 inhibitor to the onset of ketoacidosis reported was 43 days (range: 1 day to 1 year).
• In 57 of the 73 total cases, the SGLT2 inhibitor was discontinued.
• A trend to describe the relationship between the SGLT2 inhibitor dose and risk of ketoacidosis could not be identified.
• Potential predisposing factors for ketoacidosis were identified: reduction of insulin dose, acute febrile illness, reduced caloric intake due to illness or surgery, pancreatic disorders suggesting insulin deficiency (eg, T1DM, history of pancreatitis, or pancreatic surgery), and alcohol abuse.
• In many of the postmarketing reports, the presence of ketoacidosis was not immediately recognized, and institution of treatment was delayed because presenting blood glucose levels were below those typically expected for DKA (often less than 250 mg/dL).
• Signs and symptoms at presentation were consistent with dehydration and severe metabolic acidosis and included nausea, vomiting, abdominal pain, generalized malaise, and shortness of breath.
• Postmarketing Requirements for Ketoacidosis: The FDA is requiring the company to conduct an analysis of spontaneous postmarketing reports of ketoacidosis in patients treated with INVOKANA. This required enhanced pharmacovigilance study would continue for 5 years.²

He et al (2023)⁵⁹ conducted disproportionality analyses to evaluate the association between DKA or euglycemic DKA and SGLT2 inhibitors using cases from the FAERS database. Between January 1, 2013, and December 31, 2021, a total of 10,195 reports of AEs associated with SGLT2 inhibitors were received in FAERS, of which 1680 were of euglycemic DKA (peaking in 2021 [n=589]) and 8151 were of DKA (peaking in 2017 [n=1770]).

• Among the euglycemic DKA cases (INVOKANA, n=46; dapagliflozin, n=153; empagliflozin, n=332, ertugliflozin, n=4), the reporting odds ratio (ROR) and information components (IC), respectively, were 1.95 (95% CI, 1.45-2.61) and 0.92 (95% CI, 0.68-1.23) for INVOKANA, 14.43 (95% CI, 12.16-17.12) and 3.59 (95% CI, 3.03-4.26) for dapagliflozin, 25.51 (95% CI, 22.44-29.0) and 4.16 (95% CI, 3.66-4.73) for empagliflozin, and 7.44 (95% CI, 2.77-19.98) and 2.11 (95% CI, 0.79-5.66) for ertugliflozin.
• Among the DKA cases (INVOKANA, n=1459; dapagliflozin, n=733; empagliflozin, n=1222, ertugliflozin, n=10), the ROR and IC, respectively, were 11.39 (95% CI, 10.74-12.08) and 3.13 (95% CI, 2.95-3.32) for INVOKANA, 10.96 (95% CI, 10.14-11.85) and 3.19 (95% CI, 2.95-3.44) for dapagliflozin, 12.68 (95% CI, 11.90-13.51) and 3.30 (95% CI, 3.10-3.52) for empagliflozin, and 2.71 (95% CI, 1.45-5.09) and 1.30 (95% CI, 0.69-2.44) for ertugliflozin.
• Overall, with SGLT2 inhibitors, the median time to onset was 21 days for euglycemic DKA and 34 days for DKA. INVOKANA had the longest median time to onset of euglycemic DKA (96.5 days) and DKA (75 days) compared with dapagliflozin (euglycemic DKA, 13 days [P=0.013]; DKA, 27.5 days [P<0.0001]) and empagliflozin (euglycemic DKA, 21 days [P=0.052]; DKA, 15 days [P<0.0001]).

A pharmacovigilance assessment of the FAERS database for SGLT2 inhibitor-associated DKA was conducted by Fadini et al (2017).⁶⁰

• A direct mine of the FAERS files between first quarter 2014 to third quarter 2016 found 2397 DKA reports with an SGLT2 inhibitor. These included 680 DKA reports among 5694 individuals with reports for dapagliflozin (11.9%), 1362 DKA reports among 14,117 for INVOKANA (9.6%), and 355 DKA reports among 2719 for empagliflozin (13.1%). In addition, 1.34% of reports for other drugs included DKA.
• Overall, 47.7% of reports included T2DM as a drug indication, 18.5% other diabetes-related diagnoses, 11.4% T1DM, and >20% of reports had no diabetes-related indication.
• The proportional reporting ratio of DKA in reports with vs those without an SGLT2 inhibitor was 7.9. This only reflects disproportionality within the FAERS and does not inform on the risk in the real world.
• Most events resulted in hospitalization or were deemed to be life-threatening or serious. Thirty-seven cases (1.54%) were fatal: 17 with dapagliflozin, 14 with INVOKANA, and 6 with empagliflozin. There were no demographic differences between fatal and nonfatal cases.
• There were no consistent trends in DKA outcome over time or in relation to type of SGLT2 inhibitor.

Moumouni et al (2018)⁶¹ conducted a case-control study of the World Health Organization’s adverse drug reactions database, VigiBase, from January 2013 to March 2016, to investigate reports of ketoacidosis related to T2DM medications. The study identified 842 cases of ketoacidosis, of which 484 (57.7%) were associated with SGLT2 inhibitor use.

• Reports of ketoacidosis occurred more than 15 times greater with SGLT2 inhibitor use than with other glucose-lowering drugs.
• Of the ketoacidosis cases identified, 261 cases were associated with INVOKANA, 147 cases with dapagliflozin, 97 cases with empagliflozin. The results regarding ipragliflozin (available in Japan only) and ertugliflozin were based on very few cases (19 and 0 reported cases, respectively; ertugliflozin was sparingly used).
• The adjusted reporting odds ratios (ROR) for ketoacidosis reports were 15.47 (95% CI, 12.82-18.65) for any SGLT2 inhibitor, 14.33 (95% CI, 11.53-17.80) for INVOKANA, 19.48 (95% CI, 14.91-25.44) for dapagliflozin, and 35.91 (95% CI, 23.65-54.52) for empagliflozin.
  • The ROR was adjusted for age, gender, reporting region, reporting year, notifier type, and other glucose-lowering drugs
• The disproportionality signal of an increased reporting of ketoacidosis with SGLT2 inhibitors was found to be present and statistically significant prior to the first FDA warning in May 2015; however, with a lower ROR.

The EMA reported a search of Janssen’s global safety database (through June 15, 2015) for the terms DKA, diabetic ketoacidotic hyperglycemic coma, ketoacidosis, ketonuria, ketosis, urine ketone body present, acidosis, metabolic acidosis, blood ketone body present, and blood ketone body increased and identified 214 spontaneous cases in patients treated with INVOKANA.¹²

• Among the 214 cases (93 T2DM, 63 T1DM, 20 DM, 38 unspecified indication), 208 were serious. Of 85 cases that had a reported outcome, 78 were resolved or resolving, and none were fatal. Time to onset was recorded for 64 cases, of which 33 occurred within a month, and the remainder occurred within 2 months to 1 year. In the 66 cases with recorded blood glucose results, 42 had levels ≤250 mg/dL or were described as euglycemic or otherwise atypically low. There were an additional 4 cases in T2DM patients on combination INVOKANA/metformin therapy.¹²
• The EMA reported that searches of other manufacturers’ global safety databases also identified 271 spontaneous cases in patients treated with dapagliflozin (through October 4, 2015) and 91 reports in  patients taking empagliflozin (through December 10, 2015).¹²

Health Canada reported receiving 5 Canadian case reports of DKA associated with the use of INVOKANA (4 T2DM and 1 T1DM). Each of these events was reported as serious and some patients were hospitalized. Blood glucose was generally observed to be normal or mildly elevated.¹⁵

In Japan, where 6 SGLT2 inhibitors are on the market, 33 postmarketing cases (4 involving INVOKANA) of ketoacidosis had been reported by drug manufacturers as of September 15, 2015.¹⁷ Blood glucose levels on presentation were not reported for any of the 4 INVOKANA cases, but possible contributing factors were identified as cessation of insulin in 2 cases, influenza in 1 case, and heavy alcohol use in 1 case.⁶²

Peters et al (2015)⁶³ published a case series describing 9 patients (7 T1DM and 2 T2DM) from various practice sites within the United States who experienced euglycemic DKA or ketosis associated with the use of INVOKANA.

• Presenting plasma glucose levels ranged between 96 and 233 mg/dL, anion gap ranged from 16 to 35 mmol/L in the 8 patients for which it was reported, and urine ketones were strongly positive in all cases. Three T1DM patients had recurrence of ketosis upon re-administration of INVOKANA.⁶³
• Upper respiratory tract infection, alcohol consumption, exercise, decreased food intake, and decreased insulin dosage were listed as possible contributing factors for development of DKA and ketosis in 6 T1DM patients. No contributing factors were identified for one T1DM patient.⁶³
• Both T2DM patients developed DKA postoperatively (ie, 12 hours after bilateral cervical foraminotomy and a week after sigmoid colectomy).⁶³
• The authors suggest that the absence of severe hyperglycemia may have delayed recognition of DKA by patients and healthcare providers.  In turn, some patients responded to symptoms by decreasing (instead of increasing) their insulin dosage, and some healthcare providers ordered unnecessary testing and treatment upon arrival of the patient for acute medical care.⁶³
• Once DKA was identified, all patients readily responded to fluids and insulin.⁶³

In a retrospective review of medical records from an adult diabetes practice, Argento and Nakamura (2016)⁶⁴ reported 2 cases of DKA among 27 patients with longstanding T1DM who were on Dexcom G4Platinum personal continuous glucose monitors >1 year (mean, 4.6 years) and were prescribed INVOKANA 100 mg daily. The authors reported the precipitating cause of DKA was pump site failure in 1 case and running out of basal insulin in the other case.

Goldenberg et al (2016)⁶⁵ identified 46 patients with SGLT2 inhibitor-associated DKA in a literature review of case reports (27 of the 46 patients were reported to be taking INVOKANA).

• Twenty-four T2DM patients, 15 T1DM patients, 5 LADA patients, and 2 pancreatic diabetes patients
• Euglycemic DKA occurred in 70% (32/46) of patients.
• Common precipitants were inappropriate insulin reduction or omission among those who were insulin-deficient, bariatric or other surgery, excessive alcohol intake, physical exertion/exercise, and dietary restriction (low-carbohydrate or reduced intake).

Burke et al (2017)⁶⁶ reviewed a limited number of SGLT2 inhibitor-associated DKA case reports, including several INVOKANA-related cases.

A case report describing the development of INVOKANA-associated severe prolonged euglycemic DKA in a patient with T2DM has been identified in the literature but not summarized in this response.⁶⁷

Additional postmarketing case reports of DKA associated with the use of INVOKANA have been described in detail in the literature, though are not presented here.

Additional case reports describing onset of metabolic acidosis and DKA in patients treated with INVOKANA monotherapy or combination therapy have been published.^68,69

POSTULATED MECHANISM(S)

The exact mechanism(s) for the development of DKA in patients receiving SGLT2 inhibitors, including INVOKANA, is unknown.^63,70 Several mechanisms have been postulated in the literature:

• Plasma glucagon levels are increased through uncertain mechanisms in association with urinary loss of glucose by SGLT2 inhibition. Elevated glucagon could enhance lipolysis, and free fatty acids released from adipose tissues are converted into ketone bodies, which might result in ketoacidosis.^{5,58,62,63,71-76}
• When an SGLT2 inhibitor is added to insulin therapy, a reduction in insulin dose may be required to minimize the risk of hypoglycemia. A lower dose of insulin may be insufficient to suppress adipose tissue lipolysis and hepatic ketogenesis.^58,73,76 Insulin nonadherence⁷¹, insulin pump failure⁷³, and difficulty with pump attachment site⁷⁷ may also contribute to insulin deficiency. Insulin-independent urinary glucose excretion by SGLT2 inhibition might maintain glycemic control concurrent with an absolute or relative deficiency in insulin. Absence of hyperglycemia through SGLT2 inhibition may contribute to delay in the recognition of euglycemic DKA.^62,63,71,77
• Dehydration may accelerate the development of DKA.⁷⁸ SGLT2 inhibitors may predispose to negative fluid balance, thereby compounding the hypovolemic state of DKA.  Hypovolemia elevates glucagon, cortisol, and epinephrine, which further increase insulin resistance, lipolysis, and ketogenesis.⁶³
• It has also been proposed that SGLT inhibitors decrease renal clearance of ketone bodies.⁶⁸
• Additional biologically plausible mechanisms are hypothesized by authors.^{5,12,57,58,62,63,73,74,79-90}

RECOMMENDATIONS and MANAGEMENT

The FDA announcement² and updated canagliflozin Prescribing Information⁹¹ provide recommendations to healthcare professionals, including the following:

• Precipitating conditions for DKA or other ketoacidosis include under-insulinization due to insulin dose reduction or missed insulin doses, acute febrile illness, reduced caloric intake, ketogenic diet, surgery, volume depletion, and alcohol abuse. Consider ketone monitoring in patients at risk for ketoacidosis if indicated by the clinical situation. Assess for ketoacidosis regardless of presenting blood glucose levels in patients who present with signs and symptoms consistent with severe metabolic acidosis. If ketoacidosis is suspected, discontinue INVOKANA, promptly evaluate, and treat ketoacidosis, if confirmed. Monitor patients for resolution of ketoacidosis before restarting INVOKANA.
• Signs and symptoms are consistent with dehydration and severe metabolic acidosis and include nausea, vomiting, abdominal pain, generalized malaise, and shortness of breath. Blood glucose levels at presentation may be below those typically expected for DKA (eg, <250 mg/dL). If ketoacidosis is suspected, discontinue INVOKANA, promptly evaluate, and treat ketoacidosis, if confirmed. Treatment of ketoacidosis may require insulin, fluid, and carbohydrate replacement.
• Inform patients that INVOKANA can cause potentially fatal ketoacidosis and that T2DM and pancreatic disorders (eg, history of pancreatitis or pancreatic surgery) are risk factors. Educate all patients on precipitating factors (such as insulin dose reduction or missed doses, infection, reduced caloric intake, ketogenic diet, surgery, dehydration, and alcohol abuse) and symptoms of ketoacidosis (including nausea, vomiting, abdominal pain, tiredness, and labored breathing). Inform patients that blood glucose may be normal even in the presence of ketoacidosis. Advise patients that they may be asked to monitor ketones. If symptoms of ketoacidosis occur, instruct patients to discontinue INVOKANA and seek medical attention immediately.

The American Association of Clinical Endocrinologists (AACE) and the American College of Endocrinology (ACE) held a Consensus Conference entitled “AACE/ACE Scientific and Clinical Review: Association of SGLT2 Inhibitors and DKA”, in which experts convened in October of 2015 to conduct an intensive review of the available scientific and clinical data on the possible relationship of SGLT2 inhibitors and DKA. Upon review of the available evidence, together with a thorough discussion of the impact of SGLT2 inhibition on human metabolism, the experts concluded:^83,84,92

• The prevalence of DKA is infrequent.
• The incidence of DKA in T2DM does not appear to exceed the low levels occurring in the general diabetes population.
• The risk-benefit ratio overwhelmingly favors continued use of SGLT2 inhibitors.

To minimize the risk of DKA associated with SGLT2 inhibitors, the AACE recommends the following:⁸⁴

• Consider stopping the SGLT2 inhibitor at least 24 hours prior to elective surgery, planned invasive procedures, or anticipated severe stressful physical activity such as running a marathon.
• For patients with T2DM who receive exogenous insulin therapy, the insulin dose should not routinely be reduced when SGLT2 inhibitors are begun. Instead, consider adjusting the insulin dose based on the individual response to the SGLT2 inhibitor, and carbohydrate intake should be maintained. Avoid stopping insulin or decreasing the dose excessively.
• Routine measurement of urine ketones is not recommended during the use of SGLT2 inhibitors in T2DM. Urine ketone measurement can be misleading because ketonuria may be an insensitive biomarker for hyperketonemia. However, measurement of blood ketones is preferred for diagnosis and monitoring of DKA in symptomatic patients.
• Stop the drug immediately for emergency surgery or any extreme stress event.
• Avoid excess alcohol intake and very-low-carbohydrate/ketogenic diets.

AACE/ACE noted that the diagnosis of DKA is often missed or delayed due to atypical patient presentation involving lower-than-anticipated glucose levels or other misleading laboratory values. Therefore, AACE/ACE encourages greater healthcare professional education that focuses on proper diagnosis and treatment of DKA with SGLT2 inhibitors.⁸⁴

A cross-Canada expert panel and writing group offered the following recommendations for prevention of SGLT2 inhibitor-associated DKA:⁶⁵

Hold SGLT2 inhibitors:

• At onset of an acute illness (eg, infection, gastroenteritis, myocardial infarction/stroke) and restart when the patient is feeling well and able to eat and drink
• While on a preoperative low-carbohydrate diet for bariatric surgery and re-evaluate postoperatively
• Three days before major surgical procedures and restart when feeling well and able to eat and drink
• If there is a risk of dehydration (eg, extensive exercise, preparing for colonoscopy) until able to maintain hydration
• If on a low-carbohydrate diet and resume with normal diet

Stop SGLT2 inhibitors immediately and reassess at a later date:

• If excessive alcohol intake

The American Diabetes Association (ADA) published a report regarding international consensus on risk management of DKA in patients with T1DM treated with SGLT inhibitors. The Consensus Report reviews current data regarding SGLT inhibitor use and provides recommendations to minimize the risk of SGLT inhibitors in patients with T1DM. The following are key points noted in the Consensus Report:⁷⁶

• Some patients experience DKA with only slightly elevated glucose levels (<250 mg/dL), a condition known as euglycemic DKA. The usual hyperglycemia clinical alert is absent, which can lead to delays in recognition, diagnosis, and treatment.
• To minimize the risk of DKA associated with SGLT inhibitors, the ADA recommends the following:
  • SGLT inhibitor therapy should not be used in pregnant patients (T1DM) and in those on low-carbohydrate or ketogenic diets. Patients who skip meals, consume excessive alcohol, use an insulin pump, miss insulin doses (T1DM), have recurrent DKA episodes, experience prolonged significant hyperglycemia, or are not compliant with their diabetes regimen are also at high risk of DKA when on SGLT inhibitor therapy.
  • Upon initiation of SGLT inhibitor therapy in T1DM patients, insulin must be reduced cautiously to prevent ketosis and DKA.
  • SGLT inhibitors should be initiated at the lowest available dose and blood ketone levels should be tested prior to initiating SGLT inhibitor therapy. Ketone testing is necessary as development of euglycemic DKA cannot be detected by glucose monitoring. Although there is no evidence to support specific testing regimens, the consensus is that frequency of ketone testing should be based on the patient’s lifestyle and/or risk factors.
  • SGLT inhibitor therapy should be immediately discontinued following nausea, vomiting, or abdominal discomfort. Therapy should be withheld if the patient is hospitalized, acutely ill, or unable to eat/drink, prior to any medical procedure (especially if unable to eat/drink), and during switching of insulin therapy/adjustment of dose.
  • If ketones are elevated, SGLT inhibitor therapy should be discontinued. Treatment, including insulin, carbohydrates, and hydration, should be initiated immediately to avoid DKA and potential hospitalization. If symptoms and/or ketones worsen, patients should seek immediate medical assistance.
  • The ADA Consensus Report encourages patient and clinician education regarding DKA risk factors and symptoms, euglycemic DKA, ketone monitoring, and treatment protocols.    

Additional recommendations and management of DKA associated with SGLT2 inhibitors are summarized by authors.^{12,57,58,63,79-81}

BLOOD and URINE KETONES

A subject randomized to INVOKANA 300 mg in a 52-week, phase 3, active-control study⁴¹ developed ketonemia. A brief synopsis of this case is reported in the public domain.⁹³ In clinical studies in Japanese patients with T2DM, an increase in blood ketone bodies was reported with INVOKANA treatment.^51,94-98 One 24-week Japanese study (N=272) reported urine ketone body present in 1 patient on INVOKANA 100 mg daily, 3 patients on INVOKANA 200 mg daily, and no patients randomized to placebo.⁹⁶ An observational study that assessed real-world efficacy and safety of INVOKANA added to glucagon-like peptide-1 agonist therapy in 75 patients with T2DM found trace or 1+ urine ketones detected in 9 of the 54 patients in whom a urine analysis was performed.⁹⁹

Ghosh et al (2016)¹⁰⁰ reported 3 cases of ketonemia and/or ketonuria associated with the use of SGLT2 inhibitors (2 cases with INVOKANA) in New Delhi, India. All patients had diabetes of long duration (>15 years) and had recently started taking an SGLT2 inhibitor prior to presentation with ketonemia and/or ketonuria (3 to 14 days).

LITERATURE SEARCH

A literature search of MEDLINE^®, Embase^®, BIOSIS Previews^®, and Derwent Drug File (and/or other resources, including internal/external databases) pertaining to this topic was conducted on 21 January 2025.