Risk of lower extremity amputation in patients with type 2 diabetes mellitus and peripheral arterial disease receiving <scp>sodium‐glucose cotransporter‐2</scp> inhibitors versus other medications: A systematic review and meta‐analysis of observational cohort studies

Abstract

Type 2 diabetes mellitus (T2DM) is associated with significant microvascular and macrovascular complications, leading to an increased risk of morbidity and mortality. Advancement in the treatment of T2DM has led to the development of novel therapies, such as sodium-glucose cotransporter 2 (SGLT2) inhibitors, which have been shown to significantly reduce adverse health outcomes.1 A recent meta-analysis reported that SGLT2 inhibitors significantly reduced glycated haemoglobin and the risk of major adverse cardiac events.2 However, safety concerns were raised following the findings of the CANVAS (CANagliflozin cardioVascular Assessment Study) study. CANVAS was a randomized controlled trial that assessed the risk of cardiovascular outcomes in patients with T2DM and a high cardiovascular risk. The study reported an increased risk of lower extremity amputation (LEA) for patients who received canagliflozin (an SGLT2 inhibitor) compared to placebo (hazard ratio [HR] 1.97, 95% confidence interval [CI] 1.41, 2.75).3 This led to a product warning for canagliflozin from the European Medicines Agency and the US Food and Drug Administration, and increased uncertainty as to the harms and benefits of prescribing canagliflozin in high-risk T2DM patients such as those with peripheral artery disease (PAD), a history of amputation, neuropathy, or diabetic foot ulcers.4 Considering the severity of these complications, but also the benefits of SGLT2 inhibitors on cardiovascular and kidney outcomes, it is important to assess the risk of LEA in patients receiving SGLT2 inhibitors, particularly in high-risk patients such as those with PAD. There have been previous meta-analyses of trial-level data,1, 5 and one of observational studies in participants with T2DM,6 but as far as we are aware, the research presented here is the first to pool results from large observational cohort studies in patients with both T2DM and PAD. This is an important subgroup to consider as the randomized controlled trial evidence has already indicated that the amputation risk associated with SGLT2 inhibitors may differ in high-risk groups compared to the overall T2DM population. As such, the aim of this systematic review and meta-analysis was to use observational research studies to assess the LEA risk associated with SGLT2 inhibitors compared to other glucose-lowering therapies, in patients with T2DM and PAD. MEDLINE OVID, CINAHL, Embase, and the Central Cochrane Library were systematically searched from inception until 3 November 2023 according to a prespecified search strategy (Table S2). Two reviewers (A.H. and C.G.) performed the literature search, data extraction and quality assessments. These steps were checked independently by two investigators (L.H. and S.G.) and discrepancies were resolved by discussion. Studies eligible for inclusion were observational studies including all or a subgroup of participants with both T2DM and PAD, investigating LEA in those receiving SGLT2 inhibitors compared to other agents. This study was reported in accordance with the Meta-analyses Of Observational Studies in Epidemiology (MOOSE) checklist (Table S1) and registered in PROSPERO (CRD42022325598). Included studies were assessed using the National Institutes of Health study quality assessment tool for cohort studies.7 Hazard ratios and 95% CIs were extracted from relevant papers, and where only a subgroup of individuals with PAD were included in the study, results were extracted for this group alone. Results were pooled using random-effects meta-analysis to assess the risk of LEA in patients with T2DM and PAD receiving SGLT2 inhibitors compared to other agents. The analysis was split according to the comparator medication used in the study (glucagon-like peptide-1 receptor agonists [GLP-1RAs], dipeptidyl peptidase-4 [DPP-4] inhibitors, or sulphonylureas). The I2 statistic was calculated to assess between-study heterogeneity and publication bias was assessed using Egger's test. Statistical analyses were performed using STATA (StataCorp. 2021. Stata Statistical Software: Release 18. College Station, TX: StataCorp LLC). The systematic literature search retrieved 3146 records. A title and abstract screen identified 454 articles for full-text screening. Five cohort studies were identified for inclusion in the final meta-analysis (Figure S1), with a total of 303 619 participants, 56 526 of whom had PAD.8-12 Studies were carried out using routinely collected health data in Sweden, Denmark, Germany, Taiwan and the United States. The included studies are summarized in Table 1. Overall, the five studies were of good quality with low risk of bias (Table S3), although two studies were identified as having short average follow-up periods of less than 1 year.8, 11 Three of the included studies9-11 compared the risk of LEA in patients with T2DM and PAD receiving SGLT2 inhibitors compared to GLP-1RAs, in one study the comparator treatment was DPP-4 inhibitors,8 and in one study the comparator was sulphonylureas.12 Comparison groups were identified using propensity-matching methods8, 9, 11, 12 or inverse probability weighting10 to reduce bias. When the studies were pooled in a random-effects meta-analysis the results did not indicate a statistically significant increased risk of LEA with use of SGLT2 inhibitors compared to GLP-1RAs in patients with T2DM and PAD; the pooled HR for LEA amputation was 1.29 (95% CI 0.96, 1.75; p = 0.095 [Figure 1]). A low risk of publication bias was indicated (Egger's test p = 0.166) and between-study heterogeneity was moderate (I2 = 34%). For the one study that compared SGLT2 inhibitors to DPP-4 inhibitors, SGLT2 inhibitors were associated with a significantly lower risk of LEA: HR 0.43 (95% CI 0.30, 0.62). Conversely SGLT2 inhibitors were associated with a significantly higher LEA risk when compared to sulphonylureas: HR 2.24 (95% CI 1.15, 4.36). In this research letter we report that using the evidence from large observational studies showed no statistically significant increase in risk of LEA for people with T2DM and PAD prescribed SGLT2 inhibitors compared to GLP-1RAs. Risk of LEA also appeared to be significantly lower in people with T2DM treated with SGLT2 inhibitors compared to DPP-4 inhibitors, and significantly higher in SGLT2 inhibitors compared to sulphonylureas. Given the potential beneficial effect of SGLT2 inhibitors on cardiovascular and diabetic outcomes, these are important findings with potential implications for clinical practice. The risk of LEA with SGLT2 inhibitors has been investigated in previous meta-analyses of randomized controlled trials. Huang et al. analysed cardiovascular outcome trials of SGLT2 inhibitors, to assess if they were associated with LEA events.5 They found there was no difference in associated risk of LEA with SGLT2 inhibitors in T2DM patients with PAD: risk ratio 1.22 (95% CI 0.87, 1.70; n = 4648). In a separate meta-analysis, Scheen et al.6 assessed the incidence of LEA in patients with T2DM receiving SGLT2 inhibitors compared to GLP-1RAs. They included six cohort studies, however, their analysis was not specific to patients with PAD. In a population with T2DM, they reported an increased risk of amputation with SGLT2 inhibitors compared to GLP-1RAs (HR 1.15, 95% CI 1.04, 1.24), but argued that the increased risk observed may be attributable to a protective effect of GLP-1RAs, rather than an increased risk due to SGLT2 inhibitors. The cohorts of the studies included in this review came from a wide range of geographical regions. However, the clinical features were quite similar and well characterized across the studies. As such, there is likely to be good generalizability of the data. Our review also has some limitations, including the low volume of research available, and the lack of data to enable comparisons to be made in other populations at high risk of LEA. Two studies included patients from health registries in Denmark, and it was unclear if these had an overlap in their participants.11, 12 Also, two studies had follow-up of less than 1 year, which may have limited their ability to compare incidence of the outcome between cohorts due to a low number of events. However, given the importance of this topic and the potential benefit of SGLT2 inhibitors, this review provides important information concerning the risks of SGLT2 inhibitors in this population and highlights the possibility that such medication could currently be being withheld unnecessarily in high-risk patients. Further high-quality studies in high-risk groups investigating risk of LEA in T2DM patients treated with a range of medications for controlling T2DM would be useful. Specifically, further analyses of the risks and benefits of SGLT2 inhibitors versus different comparators would be informative, as our study has shown there may be a difference in perceived risk of SGLT2 inhibitors depending on the comparator used. Randomized clinical trials are unlikely to be implemented to address risks and benefits by drug classes, so analyses of other large datasets may be a more practical approach to address these questions. This research project was developed by Anna Hodgson, Clare Gillies and Patrick Highton with support from Kamlesh Khunti. The literature search, data extraction, quality assessment and data analysis were undertaken by Anna Hodgson and Clare Gillies. Louise Haddon and Safoora Gharibzadeh performed second screening of the literature search. Clare Gillies undertook the analysis. Clare Gillies and Anna Hodgson wrote the paper, and all authors reviewed and edited the manuscript. We acknowledge the support from the MiFoot research team, and funding from National Institute for Health and Care Research (NIHR) Programme Grant for Applied Research (NIHR202021). This study was also supported by the NIHR Applied Research Collaboration East Midlands and Leicester NIHR Biomedical Research Centre. The views expressed are those of the author(s) and not necessarily those of the NIHR or the Department of Health and Social Care. Kamlesh Khunti has served as a consultant and participated in speaker bureaus for, or received research support from, Amgen, AstraZeneca, Berlin-Chemie AG/Menarini Group, BMS, Boehringer Ingelheim, Janssen, Lilly, MSD, Napp, Novartis, Novo Nordisk, Roche, Sanofi and Servier. All other authors declare no conflicts. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.15571. The data that support the findings of this study are available from the corresponding author upon reasonable request. DATA S1: Supporting Information. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Topics & Concepts