5d. Peripheral vascular disease and amputation risk

5d.1 Background and Evidence Review

An interim safety analysis of the CANVAS trial (37) raised concern over an increased amputation signal with canagliflozin which led to protocol amendments to the contemporaneously recruiting CREDENCE trial (7) to exclude those with recent amputation history and to interrupt therapy in the event of foot disease onset. However, whilst the CANVAS trial reported an almost two-fold increased risk of amputation with canagliflozin (HR=1.97, 95%CI 1.41-2.75), no significantly increased amputation risk was detected in CREDENCE (HR=1.11, 95%CI 0.79-1.56). Amputation events in the CANVAS trial were significantly increased for both major (ankle and above) and minor amputation, but the majority (71%) were minor (predominantly toe amputations).

Relative risks of amputation in CANVAS were also similar across a range of subgroups, including history of peripheral vascular disease (PVD), prior amputation, and eGFR <60 vs ≥60 mL/min.1.73m². Secondary analyses of CANVAS and CREDENCE have failed to identify any participant or trial factors to explain the difference in reported amputation risk, with the aforementioned late protocol amendment (implemented about three quarters of the way through recruitment) not thought to have contributed to the absence of amputation signal in CREDENCE (53, 54).

The RSG and SMART-C meta-analysis (3) investigated the risk of lower limb amputation overall and by diabetes status. After excluding the hypothesis-generating CANVAS trial, there was no significant increase in risk of amputation identified in the other 12 trials (relative risk 1.06, 95%CI 0.93-1.21) with evidence for statistical heterogeneity between CANVAS vs these other 12 trials (p=0.0007). There was also no evidence that the risk of lower limb amputation was modified by trial-level eGFR or by diabetes status. When assessing absolute risks of SGLT 2 treatment using the conservative 15% increased risk (representing a pessimistic estimate of risk from analysis of all 13 trials including CANVAS), SGLT 2 inhibition would cause an estimated 1.1 amputation events per 1000 person years of treatment among people with CKD and diabetes, while an estimated 11 kidney disease progression events, 4 AKI events, and 11 hospitalisations for heart failure or cardiovascular events would be avoided (Figure 5.1). Due to the considerably lower absolute risk of amputation among people without diabetes, the absolute number of amputations caused per 1000 person years is estimated to be small compared to the absolute benefits in the studied populations where there are indications for use of SGLT 2 inhibitors.

Our recommendation is to avoid initiation of SGLT 2 inhibitors in individuals with active foot disease and withhold SGLT 2 inhibitors should this complication arise. We also stress the importance of shared decision-making in initiating SGLT 2 inhibitors, and reinstating use following resolution of foot complications, recognising that individuals at high risk of amputation may also stand to gain significant cardiorenal benefit from these agents as outlined above. Given the absence of amputation signal seen in CREDENCE, there is insufficient evidence to disadvantage one of the SGLT 2 inhibitors over the other for individuals at high risk of amputation events. However, attention to routine preventative foot care should be advised for all people with DM initiated on SGLT 2 inhibitors.

5d.2 Recommendations for Implementation

1. We suggest avoiding initiation of SGLT 2 inhibitors in the presence of active foot disease (infection, ulceration and ischaemia) and withholding treatment in those who develop foot complications whilst taking an SGLT 2 inhibitor (Grade 2B).
2. We suggest a shared decision-making approach, with appropriate counselling on risks and benefits of treatment and the importance of routine preventative foot care measures for:

• Individuals at high risk of amputation (previous amputations, existing PVD, peripheral neuropathy)
• Re-initiation of SGLT 2 inhibitors after treatment and satisfactory resolution of a foot complication that occurred whist taking SGLT 2 inhibitors (Grade 2B).

Rationale: A significant finding from a single large trial using the SGLT 2 inhibitor canagliflozin alerted clinicians to the possibility that SGLT 2 inhibitors could increase the risk of lower limb amputations. This finding has not been confirmed in other large trials and furthermore, it is important to appreciate that people with PVD are a group of individuals who have more to gain from the initiation of SGLT 2 inhibitors in relation to protection against risk of cardiovascular death, myocardial infarction, heart failure complications and progression of CKD. It is therefore important not to exclude these individuals from the potential benefits of SGLT 2 inhibitors, but to ensure that these medicines are used appropriately and safely in people at risk, or with evidence of PVD.