according to a new large database analysis.
The findings, which include data on the use of three different SGLT2 inhibitors in Canada and the United Kingdom and suggest a class effect, were published online July 27 in Annals of Internal Medicine by Antonios Douros, MD, PhD, of McGill University and the Centre for Clinical Epidemiology, Lady Davis Institute, Montreal, and colleagues.
“Our results provide robust evidence that SGLT2 inhibitors are associated with an increased risk for DKA. Of note, increased risks were observed in all molecule-specific analyses, with canagliflozin [Invokana, Janssen] showing the highest effect estimate,” they noted.
And because the beneficial effects of SGLT2 inhibitors in the prevention of cardiovascular and renal disease will probably increase their uptake in the coming years, “Physicians should be aware of DKA as a potential adverse effect,” Dr. Douros and colleagues wrote.
Analysis “generally confirms what has already been published”
Asked for comment, Simeon I. Taylor, MD, PhD, professor of medicine at the University of Maryland, Baltimore, said that the study “generally confirms what has already been published” on the topic. He noted that overall “the risk of SGLT2 inhibitor–induced ketoacidosis is quite low in type 2 diabetes, perhaps on the order of 1 episode per 1000 patient-years.”
However, Dr. Taylor cautioned: “Published evidence suggests that the risk of DKA is increased if patients are unable to eat,” such as when hospitalized patients are not permitted to eat.
“In that setting, it is probably prudent to discontinue an SGLT2 inhibitor. Also, it may be prudent not to prescribe SGLT2 inhibitors to patients with a history of DKA,” he added.
Dr. Taylor also advised: “Although not necessarily supported by this publication, I think that caution should be exercised in prescribing SGLT2 inhibitors to insulin-dependent type 2 diabetes patients. … Some late-stage type 2 diabetes patients may have severe insulin deficiency, and their physiology may resemble that of a type 1 diabetes patient.”
Dr. Taylor has previously advised against using SGLT2 inhibitors altogether in patients with type 1 diabetes.
Increased DKA risk seen across all SGLT2 inhibitors
The study involved electronic health care databases from seven Canadian provinces and the United Kingdom, from which 208,757 new users of SGLT2 inhibitors were propensity-matched 1:1 to new dipeptidyl peptidase-4 (DPP-4) inhibitor users.
Over a mean 0.9-year follow-up, 521 patients were hospitalized with DKA, for an overall incidence rate of 1.41 per 1,000 person-years.
The rate with SGLT2 inhibitors, 2.03 per 1,000 person-years, was nearly three times that seen with DPP-4 inhibitors, at 0.75 per 1,000 person-years, a significant difference (hazard ratio, 2.85).
By individual SGLT2 inhibitor, the hazard ratios compared with DPP-4 inhibitors were 1.86 for dapagliflozin, 2.52 for empagliflozin, and 3.58 for canagliflozin, all statistically significant. Stratification by age, sex, and incident versus prevalent user did not change the association between SGLT2 inhibitors and DKA.
Asked about the higher rate for canagliflozin, Dr. Taylor commented: “It is hard to know whether there are real and reproducible differences in the risks of DKA among the various SGLT2 inhibitors. The differences are not huge and the populations are not well matched.”
But, he noted, “If canagliflozin triggers more glucosuria, it is not surprising that it would also induce more ketosis and possibly ketoacidosis.”
He also noted that the threefold relative increase in DKA with canagliflozin versus comparators is consistent with Janssen’s data, published in 2015.
“It is, of course, reassuring that both [randomized clinical trials] and epidemiology produce similar estimates of the risk of drug-induced adverse events. Interestingly, the incidence of DKA is approximately threefold higher in the Canadian [data] as compared to Janssen’s clinical trials.”
Dr. Taylor also pointed out that, in the Janssen studies, the risk of canagliflozin-induced DKA appeared to be higher among patients with anti-islet antibodies, which suggests that some may have actually had autoimmune (type 1) diabetes. “So the overall risk of SGLT2 inhibitor-induced DKA may depend at least in part on the mix of patients.”
In the current study, individuals who never used insulin had a greater relative increase in risk of DKA with SGLT2 inhibitors, compared with DPP-4 inhibitors, than did those who did use insulin (hazard ratios, 3.96 vs. 2.24, both compared with DPP-4 inhibitors). However, just among those taking SGLT2 inhibitors, the absolute risk for DKA was higher for those with prior insulin use (3.52 vs. 1.43 per 1,000 person-years).
The results of sensitivity analyses were consistent with those of the primary analysis.
The study was funded by the Canadian Institutes of Health Research and supported by ICES. Dr. Douros has reported receiving a salary support award from Fonds de recherche du Quebec – sante. Dr. Taylor was previously employed at Bristol-Myers Squibb. He is currently a consultant for Ionis Pharmaceuticals and has reported receiving research support provided to the University of Maryland School of Medicine by Regeneron.
A version of this article originally appeared on.