SGLT-2 Inhibitors and Lower-limb Amputations in Diabetes: A Misunderstood Relationship? Insights from a Retrospective Study

Editorial

Diabetic foot ulcers (DFUs) represent one of the most severe and prevalent complications of diabetes mellitus (DM), increasing amputations and mortality.^1,2 The complex pathophysiology of DFUs involves multiple risk factors, including poor glycaemic control, peripheral arterial disease (PAD), distal symmetrical polyneuropathy, impaired immune function, infection and trauma. ³

Sodium-glucose co-transporter-2 inhibitors (SGLT-2is) are relatively newer oral antidiabetic agents reducing serum glucose via its increased renal excretion. ⁴ In addition, they exert multiple beneficial effects on mortality, major cardiovascular events, heart failure, and renal outcomes in people with type 2 DM (T2DM).^5–8

However, for almost 10 years, data on the risk of lower-limb amputations in individuals treated with SGLT-2is have remained controversial. The first and only randomised controlled trial (RCT) to demonstrate an increased risk of lower-limb amputation associated with an SGLT-2i was the Canagliflozin Cardiovascular Assessment Study (CANVAS) Program. ⁵ In CANVAS, canagliflozin was linked with increased risk of amputation (6.3 vs 3.4 participants per 1000 patient-years, hazard ratio [HR]: 1.97, 95% confidence interval [CI]: 1.41-2.75). ⁵ Among amputations, 71% were minor (at the level of the toe or metatarsal) and 29% major. ⁵ A later analysis of CANVAS identified prior history of amputation, male sex, race, history of PAD, history of neuropathy, albuminuria, higher baseline glycated haemoglobin (HbA_1c) and canagliflozin treatment as independent predictors of amputation. ⁹ Nevertheless, the exact underlying mechanisms of canagliflozin-related risk of amputation are unknown. ¹⁰

When CANVAS was published, CREDENCE (Canagliflozin and Renal Outcomes in Type 2 Diabetes and Nephropathy) ¹¹ was still an ongoing trial. As a result, the CREDENCE protocol was modified, requiring foot examinations at every study visit, and temporary interruption of treatment in participants with active foot conditions that could increase amputation risk. Nevertheless, no significant increase in lower-limb amputations versus placebo was observed. ¹¹ Similar data emerged from the EMPA-REG ⁶ (Empagliflozin Cardiovascular Outcome Event Trial in Type 2 Diabetes Mellitus Patients-Removing Excess Glucose) and DECLARE-TIMI 58 ⁷ (Dapagliflozin Effect on Cardiovascular Events-Thrombolysis in Myocardial Infarction 58) studies that did not demonstrate an increased risk of amputation with two other SGLT-2is, namely empagliflozin and dapagliflozin, respectively.

Moreover, there is data from meta-analyses^12–16 and cohort studies.^17,18 On the one hand, the risk of amputation was slightly increased in people with T2DM treated with SGLT-2is (odd ratio [OR]: 1.23, 95% CI 1.08-1.40, p = .002) in one of these meta-analyses, ¹² as well as in one population-based cohort study. ¹⁸ The latter included >25 000 participants and showed that SGLT-2is initiation was associated with an almost 2-fold higher risk of below-knee amputation (0.17 vs 0.09 events per 100 person-years, HR: 1.99, 95% CI: 1.12-3.51). ¹⁸ Conversely, there are several meta-analyses suggesting that there is no such increased risk.^13–16 In this context, a large real-world study ¹⁷ with over 118 000 new users of SGLT-2is strongly demonstrated that the hazard of lower-limb amputations was not increased with canagliflozin (HR: 0.98, 95% CI 0.68-1.41, p = .92).

Accordingly, the recently published retrospective, single-centre study by Dumortier et al ¹⁹ is of great interest. The aim of this study was to determine whether the use of an SGLT-2i could influence lower-limb amputation and wound healing in patients with DFUs. The primary end point was amputation rate at 1 year from inclusion in the study. ¹⁹ Secondary end points were healing rate at 6 months, healing time, minor amputation rate at 1 year, major amputation rate at 1 year and mortality rate at 1 year. The study included all consecutive adults with T2DM referred for a new DFU. Neuropathic, ischaemic, and neuro-ischaemic ulcers, infected or non-infected, superficial or penetrating to tendon or capsule or to the bone or joint, were included. ¹⁹ All patients received similar regular follow-up in accordance with current standards of care, as defined by the International Working Group on the Diabetic Foot (IWGDF). ²⁰

The treatment group consisted of patients receiving an SGLT-2i from study inclusion until occurrence of an event, such as scarring, amputation, or death. ¹⁹ In contrast, patients in the control group with DFUs did not receive SGLT2is. Initiation, continuation, or discontinuation of SGLT-2i treatment in patients presenting with a new DFU was left to the discretion of the individual physician. Therefore, patients in the control group did not receive SGLT-2i, either because there was no clinical indication for their use or because treatment had been discontinued or not initiated due to the presence of a DFU. Among the 94 patients treated with an SGLT-2i, 79 received dapagliflozin and 15 received empagliflozin. ¹⁹ In total, the study population comprised 452 patients, with 94 in the SGLT-2is group and 358 in the non-SGLT2i group. ¹⁹ Regarding baseline characteristics, the two groups were well balanced across most variables. However, differences were observed in the presence of ischaemic heart disease (p = .039 for treated vs not treated with SGLT-2is) and haemodialysis (p = .016, respectively), as well as in the concomitant use of glucagon-like peptide-1 receptor agonists (GLP-1 RAs) (p = .002, for treated vs not treated with SGLT-2is) and rivaroxaban or apixaban (p = .025, respectively). ¹⁹ At inclusion, mean age of the study population was 70.2 ± 11.6 years, and 77% of individuals were men. Mean HbA_1c was 8.1%±1.8%. ¹⁹ Median ulcer duration was 30 days. Infection was present in 46.7% of participants. Wound characteristics were not significantly different between the two groups. ¹⁹

There was no significant difference in 1-year amputation rate between patients treated with an SGLT-2i and those not receiving SGLT-2i therapy (20.9% vs 22.9%, respectively, p = .687). ¹⁹ The estimated average treatment effect of SGLT-2is on amputation risk was not significant (−0.03, 95% CI: −0.12 to 0.05, p = .473). Similarly, the 6-month wound healing rate was insignificantly higher in the SGLT-2i group (54.4% vs 44.5%, respectively, p = .091). ¹⁹

Impressively, healing time was significantly shorter among patients treated with SGLT-2is, with a mean difference of approximately 44 days (136.5 ± 97.8 days vs 181.2 ± 159.8 days, p = .04). Notably, a significant reduction in 1-year mortality was also observed in the SGLT-2i group vs controls (1.1% vs 9.2%, p = .009). ¹⁹

The results of this study ¹⁹ are encouraging: they suggest that use of SGLT-2is is not associated with increased rate of amputation at 1 year and does not negatively affect the healing rate of DFU at 6 months. Notably, SGLT-2i treatment significantly shortened healing time and lowered 1-year mortality rate in patients with DFU. ¹⁹ These findings are also supported by previous large RCTs, which have consistently demonstrated the cardiovascular and renal benefits of SGLT-2is, along with reductions in overall mortality.^5–7

While this study demonstrates very promising data rejecting an association between SGLT-2i treatment with and increased rate of lower-limb amputation in patients with a first DFU, it has certain limitations. ¹⁹ First, it is single-centre and retrospective, which reduces the level of evidence. Secondly, there were some differences in baseline characteristics between the two groups. Thirdly, no patient received canagliflozin, which is the agent fist giving rise to the discussion on amputations.^21,22 Additionally, there were differences between the two groups in terms of concomitant GLP-1 RA use (45.7% vs 28.6%, p = .002 for treated vs not treated with SGLT-2is). Even though there was no significant association between GLP-1 RA use and the amputation rate at 1 year or the 6-month healing rate in the univariate analysis, we should bear in mind that in a recent RCT ²³ the GLP-1 RA semaglutide has been demonstrated to improve maximal walking distance and pain-free walking distance vs placebo, independently of weight loss, suggesting a direct vascular benefit in PAD. Accordingly, differences in GLP-1 RA use between the two groups might have, at least partly, influenced outcomes. Moreover, although no significant difference was observed between the two groups regarding severe PAD, the absolute number of participants with severe PAD in the SGLT-2i group was very small (n = 3). This limits generalisability of findings in this specific subgroup. Furthermore, SGLT-2is use was also not assessed in patients with end-stage kidney disease. Finally, the long diabetes duration in both groups (approximately 19 years) and the relatively high HbA_1c, indicating suboptimal glycaemic control, limit the interpretation and generalisability of findings to individuals with shorter diabetes duration and better glycaemic control.

In conclusion, the findings by Dumortier et al ¹⁹ are particularly important, given that the IWGDF has recommended avoiding initiation of SGLT-2is in drug-naïve individuals with active diabetic foot disease and considering temporary discontinuation in those already receiving these agents until complete wound healing is achieved. ²⁰ Nevertheless, cardiovascular disease and infection remain the leading causes of death in this population.^1,2 Consequently, SGLT-2is may represent a profoundly valuable therapeutic option for these high-risk individuals by virtue of their well-established cardiovascular and renal benefits.^5–8 Thus, the new study ¹⁹ contributes to reducing the underuse of SGLT-2is in patients with DFUs. SGLT-2is may, then, be assumed to offer benefits in patients with DFUs, in addition to the improvement in PAD shown with the GLP-1 RA semaglutide^23,24 and to the accelerated wound healing recently shown with another class of antidiabetic agents, the dipetidyl peptidase-4 inhibitors.^25,26 Perhaps, the progress with all these new agents may be combined with the benefits of continuous glucose monitoring towards improved glycaemic control.^27,28 Obviously, further research is needed to delineate this, perhaps overstated and still controversial, relationship between SGLT-2is and lower-limb amputations, which has previously attracted interest in our journal. ²¹