Studies analyze amputation rates with sodium–glucose cotransporter-2 inhibitors

An ACP Journal Club commentary looked at studies on the effects of canagliflozin in patients with cardiovascular disease and dapagliflozin in patients with peripheral artery disease, concluding that amputation risks are small enough that they needn't play a major role in prescribing decisions.


A population-based study compared lower-limb amputation rates among more than 300,000 patients who started taking canagliflozin or a glucagon-like peptide-1 (GLP-1) receptor agonist. Patients were categorized by age younger than 65 years or age 65 years and older and by the presence or absence of cardiovascular disease (CVD). Although the hazard ratios favored an increase in amputations with canagliflozin in all four categories, the risk was only significant among patients who were age 65 years or older with CVD. The study by Fralick and colleagues was published Aug. 25, 2020, by The BMJ and summarized in the September 2020 ACP Diabetes Monthly.

The other study was a secondary analysis of the DECLARE-TIMI 58 trial, which randomized 17,160 patients with type 2 diabetes, 1,025 of whom had peripheral artery disease (PAD), to dapagliflozin or placebo. This analysis found that patients with PAD had increased risk of adverse outcomes, including cardiovascular death and limb ischemic events, compared to patients without PAD. However, there was no consistent pattern of risk for adverse limb events, including amputation, with dapagliflozin compared to placebo among patients with PAD. The study by Bonaca and colleagues was published Aug. 3, 2020, by Circulation.

The following commentary by Qiukui Hao, MD, and Gordon Guyatt, MD, MSc, discusses both the Fralick and Bonaca studies and was published in the ACP Journal Club section of the January Annals of Internal Medicine.

Sodium–glucose cotransporter-2 (SGLT2) inhibitors, including dapagliflozin, empagliflozin, and canagliflozin, show a consistent class benefit on CV [cardiovascular] and renal outcomes. However, whether these drugs increase risk for amputation remains a subject of debate. In the CANVAS study, canagliflozin was associated with a 2-fold higher risk for amputation compared with placebo (6 vs. 3 events per 1000 patient-y; hazard ratio, 1.97 [95% CI, 1.41 to 2.75]). None of the DECLARE-TIMI 58 trial of dapagliflozin, CREDENCE trial of canagliflozin, or recent systematic review and network meta-analysis of SGLT2 inhibitors (odds ratio, 1.14 [CI, 0.96 to 1.35]; moderate certainty) replicated this finding.

The large population-based cohort study by Fralick and colleagues again raised the possibility that canagliflozin increased the risk for amputation requiring surgery compared with GLP-1 agonists, particularly in patients aged ≥65 years with CVD. However, the secondary analysis of DECLARE-TIMI 58 by Bonaca and colleagues reported that the benefits of dapagliflozin in reducing risk for CV death or hospitalization for heart failure and progression of kidney disease did not differ in patients with and without PAD, and that dapagliflozin did not increase risk for amputation.

Sample size limitations may be responsible for the nonsignificant findings in the secondary analysis of the DECLARE-TIMI 58. There are, however, reasons for skepticism of the suggestion that canagliflozin increases amputations. The meta-analysis of all SGLT2 inhibitor trials did not show a persuasive increase in amputations, and all other outcomes of the SGLT2 suggest a class, rather than specific drug, effect. Fralick and colleagues' observational study suggesting an increase in amputation in a subgroup is limited by the residual confounding that limits all observational studies. Moreover, even if the effect is real—which remains uncertain—it is small: 7 more amputations per 1000 high-risk person-years with PAD as estimated by Bonaca and colleagues, and 4 more amputations per 1000 person-years in patients aged ≥65 years with previous CVD as estimated by Fralick and colleagues. This effect appears to be smaller than the unequivocally established reductions in other major CV and renal events in adults with type 2 diabetes and CVD (up to approximately 10 per 1000 patient-y).

Given the uncertainty of whether SGLT2 inhibitors, including canagliflozin, increase amputation at all, and that any potential effect is very small (perhaps smaller than the important benefits in CV and renal outcomes), the issue of amputation needn't play a major role in decisions regarding administration of SGLT2 inhibitors.