SGLT2 inhibitors and amputation risk: Real-world data from a diabetes foot wound clinic

Sodium-glucose cotransporter 2 inhibitors (SGLT2i) are effective agents in the management of type 2 diabetes mellitus (T2DM), with significant improvements in glycaemic control and favourable cardiovascular and renal profiles [1], [2]. In 2017, however, the CANVAS and CANVAS-R trials reported a two-fold increase in the occurrence of lower limb amputations (LLAs) in participants receiving canagliflozin [2]. These were predominantly toe or metatarsal amputations and they occurred in individuals without established peripheral vascular disease prior to study commencement. These findings prompted the United States Food and Drug Administration (FDA) to issue a caution on the use of canagliflozin in individuals at risk of amputation [3]. Other SGLT2i have not been associated with an increased risk of LLA. In the EMPA-REG OUTCOME study, rates of LLAs were not significantly higher in those randomised to empagliflozin [4]. Similarly, a pooled analysis of 30 phase II and III trials did not find a significant association between dapagliflozin and LLAs [5]. However, real-world data are lacking, in particular regarding the use of SGLT2i amongst people at high baseline risk of amputation. Whether SGLT2i increase the risk of poor wound healing in subjects with established diabetic ulcers and severe peripheral vascular disease remains an open question. On this basis, the European Medicines Agency has adopted a cautious approach, advising against the use of any SGLT2i in patients at risk of amputation until further data are available [6].

We conducted a retrospective case-controlled study of people with T2DM attending a foot-wound clinic in a tertiary hospital in Sydney, Australia over a 30-month period (April 2015-September 2017). Incidences of LLAs, including minor and major amputations, were compared in participants with active diabetic foot wounds who were receiving SGLT2i or not.

Twenty-seven people with open foot wounds who were receiving SGLT2i therapy were identified. Among them, 16 were on dapagliflozin (59%), 9 were on empagliflozin (33%), 1 person took dapagliflozin which was changed to empagliflozin, and another person took canagliflozin which was changed to dapagliflozin. They were matched by age, duration of diabetes, HbA1c, and smoking status, in a 1 to 3 ratio, with control subjects with diabetic foot wounds who were not receiving SGLT2i for the study duration. Renal function was not statistically different between the groups and on average, control subjects did not have a degree of renal dysfunction that would contraindicate SGLT2i use. In the SGLT2i group, only LLAs that occurred more than one month following the commencement of therapy were included. The mean duration of SGLT2i use was 12.9 ± 5.3 months, and the control group was observed for a similar duration. Baseline characteristics and amputation rates in each group are listed in Table 1. Ten out of the 27 (37.0%) people in the SGLT2i group had at least 1 episode of LLA, compared to 37 of the 81 (45.7%) people in the control group. The odds ratio of having an LLA in association with SGLT2i use was not significant (OR 0.70 [95% CI 0.29, 1.71]; p = 0.43). Similarly, there were a total number of 11 LLAs in the SGLT2i group (0.41 per patient) while there were 49 LLAs in the control group (0.60 per patient) and the difference between the groups in the number of LLAs was not statistically significant (difference −0.19 [95% CI −0.52, 0.13]; p = 0.23). Therefore, the number of people having LLA and the number of LLAs in each of the groups were not significantly higher in the SGLT2i group.

Table 1.

Baseline characteristics and amputation rates of the patients who did and did not receive SGLT2i*.

	SGLT2i, n = 27	Control, n = 81		P value
Age (years)	59.1 ± 9.0	61.1 ± 9.9	–	0.192
Female sex – n (%)	7 (25.9%)	23 (28.4%)	–	0.801
Duration of T2DM (years)	15.8 ± 4.6	16.2 ± 7.6	–	0.803
Glycated haemoglobin (%)	9.0 ± 1.7	8.9 ± 1.8	–	0.739
Glycated haemoglobin (mmol/mol)	75 ± 19	74 ± 20	–	0.739
Estimated GFR (ml/min)	82.1 ± 27.1	68.0 ± 32.1	–	0.05
Current Smokers, n (%)	7 (25.9%)	19 (23.5%)	–	0.798
Patients who had amputation(s) – n (%)	10 (37.0%)	37 (45.7%)	OR 0.70 (0.29–1.71)	0.434
Number of amputations	11 (0.41 per patient, 37.9 per 100 patient-years)	49 (0.60 per patient, 60.5 per 100 patient-years)	Difference −0.20 (−0.52 to 0.13)	0.232
Types of Amputations, n (%) Minor amputations Major amputations	10 (90.9%) 1 (9.1%)	42 (85.7%) 7 (14.3%)

Data are presented as the mean ± SD. Minor amputation is any resection through or distal of the articulation of the ankle, whilst major amputation is any resection proximal of the ankle. SGLT2i, sodium-glucose cotransporter 2 inhibitor; T2DM, type 2 diabetes; OR, odd ratio.

To date, safety data exonerating empagliflozin and dapagliflozin did not specifically stratify subjects on the basis of their baseline amputation risk or on the presence of established foot wounds. By studying such a high-risk population, our study seeks to clarify whether SGLT2i pose additional risk to people at a high baseline risk of amputation. Despite matching the groups for age, duration of diabetes, HbA1c, and smoking status, other confounding variables may not have been accounted for in this retrospective study. Similarly, the SGLT2i users had a trend for better renal function (p = 0.05), but this would not mask a higher incidence of LLA relating to SGLT2i. Study power may have been inadequate to exclude a significant difference between the groups. A non-inferiority power calculation indicated that a sample size of 214 would be necessary to exclude a difference of greater than 20% in LLAs between SGLT2i users and control subjects in our study population (for a HR < 1.5, baseline LLA rate 45.7% in controls, alpha 0.05) [7]. Our study included one subject on canagliflozin, which was the SGLT2i that initially raised this safety signal. During our study period, canagliflozin was removed from the government pharmaceutical subsidy scheme and as such, this agent is now infrequently used in Australia. The strength of this study lies in the assessment of a group of people with a relatively high baseline risk of amputation, as established lower extremity ulcers are known risk factors for LLA [8]. In fact, the incidence of LLAs in our study group was 80-fold higher than that seen in EMPA-REG study (548.4 versus 6.5 per 1000 patient-years) [6]. This study is therefore the first to provide real-world data on the use of SGLT2i in patients with open foot wounds and offers some reassurance on the use of empagliflozin and dapagliflozin in people at high risk of amputation.