What can real-world studies really tell us about amputation risk for SGLT2 inhibitors?
Ever since the data from the CANVAS studies suggested an excess risk for amputations in patients with type 2 diabetes treated with sodium-glucose cotransporter (SGLT)2 inhibitors, there has been a flurry of activity to investigate whether there is a signal for amputation risk with other SGLT2 inhibitors. This includes an analysis of EMPA-REG OUTCOME (no increase) and more recently of DECLARE (no significant increase with dapagliflozin in 17,160 patients followed for over 4 years). There has also been much speculation about plausible biologic mechanisms to explain this, including the known increased haematocrit observed with SGLT2 inhibitors and “off target” effects such as the weak effects of canagliflozin on SGLT1, which might have effects on the vasculature. We also have to remember that the increase was only seen in the longer-term follow-up of the original CANVAS trial, and not in the shorter CANVAS-R cohort, and that the rate seen in the placebo group in CANVAS was lower than that seen in EMPA-REG.
Recent real-world analyses seem to have added to the confusion, with some studies suggesting no effect, and others an increase with all SGLT2 inhibitors. The numbers of amputations seen in these studies are small, and the event rate is about 10% of what was seen in high-risk populations in the clinical trials. Although the groups treated with SGLT2 inhibitors are matched using propensity scores, such matching cannot account for all possible confounding factors.
These studies are also of incredibly short duration, mostly involving exposures to the SGLT2 inhibitor or comparators of less than 6 months. In at least one of the studies, most of the amputations occurred in low-risk patients, which is the opposite to what was seen in CANVAS. Given the publicity surrounding SGLT2 inhibitors and amputations, it is possible that reporting bias also exists. Real-world evidence can be useful in confirming therapeutic effects such as glycated hemoglobin and weight loss in large datasets, and even important outcomes that are easily verifiable from claims data, such as hospitalisation for heart failure and mortality, as has been demonstrated in the CVD-REAL and EASEL studies. By contrast, the value of real-world studies in detecting rare adverse events is questionable, and has a history of generating spurious associations (eg, insulin glargine and cancer), which can then only be refuted by carefully conducted prospective trials.
So where does this leave us in relation to amputations and SGLT2 inhibitors? It remains possible that the observation in CANVAS was a chance finding, and if the CREDENCE study with canagliflozin, which will report next year, shows no signal for amputations that will be reassuring. So far, there is no convincing evidence suggesting that empagliflozin or dapagliflozin increase amputation risk, but until more evidence emerges, in particular from the ongoing VERTIS trial with ertugliflozin, and ongoing trials in renal impairment and heart failure with dapagliflozin and empagliflozin, it is wise for clinicians to avoid using SGLT2 inhibitors in patients with previous amputations or existing foot ulceration. All people with diabetes, no matter how they are treated, should be alert to the possibility of foot problems, and good foot care and regular screening, along with control of blood glucose and other risk factors, remains the best way to prevent amputation.