SGLT2i and GLP-1 RA therapy in type 1 diabetes and reno-vascular outcomes: a real-world study

The incidence and prevalence of type 1 diabetes is increasing. It is estimated that 8.4 million individuals are living with the disease worldwide, and this is expected to increase [1]. Insulin has remained the cornerstone of treatment in type 1 diabetes since its discovery by Banting and Best in 1922 [2]. Despite advancements in ultra-fast and ultra-long-acting insulins, there has been a paucity of new non-insulin glucose-lowering therapies specifically licensed for type 1 diabetes.

Type 1 diabetes is unique amongst many of the chronic diseases where escalation of therapy does not routinely include the addition of alternative adjunctive pharmacotherapy. Intensification of glucose-lowering therapy in type 1 diabetes includes multiple daily insulin injections, continuous subcutaneous insulin infusion, alongside continuous glucose monitoring and increasingly closed-loop insulin delivery [3]. Pramlintide is an amylin analogue licensed for type 1 diabetes in the US but currently very few people are treated with it [4], and the risk of severe hypoglycaemia remains a barrier to its approval in the UK and Europe [5]. Early optimisation of glycaemic control is key to reducing the incidence of microvascular and macrovascular complications as demonstrated by the landmark DCCT [6]. However, the ‘ominous octet’ as described by DeFronzo, suggests that multiple drugs targeting different metabolic abnormalities in type 2 diabetes are required to reduce subsequent morbidity and mortality [7]. A similar rationale may also be applicable in type 1 diabetes.

Despite the advances in technology and insulin delivery over the past several decades, less than a third of adults with type 1 diabetes in England and Wales achieve a target HbA_1c of ≤58 mmol/mol [7.5%] [8], and only 20% of patients in the US T1D Exchange Registry achieved HbA_1c ≤53 mmol/mol [7.0%] [9]. Novel adjuvant therapies for type 1 diabetes have remained elusive.

The management of type 2 diabetes has included targeting multiple metabolic pathways to achieve optimal glycaemic control. Inhibitors of the sodium-glucose cotransporter 2 (SGLT2i) and glucagon-like peptide-1 receptor agonists (GLP-1 RA) have been studied and trialled in type 2 diabetes with an overwhelming body of evidence supporting their use in type 2 diabetes [10]. SGLT2is and GLP-1 RAs consistently demonstrate cardiovascular and nephroprotective effects [11]. SGLT2is, and GLP-1 RAs in particular promote weight loss, and therefore especially may be of benefit, given that overweight and obesity is increasingly prevalent (62%) in individuals with type 1 diabetes [12]. Furthermore, cardiovascular and renal disease are the predominant cause of premature mortality in type 1 diabetes [13]. Given that both SGLT2i and GLP-1 RA demonstrate robust and consistent cardio-renal benefits [14], both agents seem to be an attractive option to fulfil an unmet therapeutic need in type 1 diabetes. However, concerns regarding the safety of such agents, in particular the increased risk of diabetic ketoacidosis (DKA) with SGLT2is, have restricted their widespread adoption in type 1 diabetes [15]. Thus, all adjuvant glucose-lowering therapies in type 1 diabetes are used off-label and contrary to local and national guidance [16].

The administration of such agents could therefore be: (1) inappropriate and should not be prescribed off-label in type 1 diabetes; or (2) underutilised with a failure to capitalise on the significant glucose-lowering and cardio-renal benefits. This study has investigated the real-world impact of SGLT2i and GLP-1 RA therapy on individuals with type 1 diabetes in relation to efficacy, safety and cardio-renal outcomes.

Participants in the GLP-1 RA cohort were on average younger, more frequently female and had higher weight compared with those using an SGLT2i. At baseline, participants in the GLP-1 RA group had a preserved eGFR compared with those in the SGLT2i group and were less likely to have a diagnosis of hypertension, IHD and heart failure. The prevalence of diabetes-associated microvascular complications across both cohorts was low (3–4%).

Individuals with CKD were on average older, more likely to be male and have a higher burden of cardiovascular disease than those in the main cohort. Within the SGLT2i + CKD group, eGFR improved from 45.5 ml/min per 1.73 m² at baseline to 47.0 ml/min per 1.73 m² at 5 years. Within the GLP-1 RA + CKD group, eGFR reduced from 50.1 ml/min per 1.73 m² to 44.8 ml/min per 1.73 m² at 5 years. No individuals developed DKA.

We have conducted the largest real-world retrospective cohort study to date investigating the safety and outcomes of individuals with type 1 diabetes using SGLT2i and GLP-1 RA therapy, and the first to report cardio-renal benefits. We demonstrate that both therapies offer clinically significant reductions in HbA_1c with no difference in 5-year all-cause mortality. SGLT2i use is associated with an increased risk of DKA and UTI, but confers several favourable outcomes, notably preservation of renal function, reduced risk of developing heart failure and CKD, and of hospitalisation, despite a higher prevalence of heart failure, hypertension, IHD and CKD at baseline. We demonstrate that within a subgroup of individuals with established CKD treated with an SGLT2i, eGFR remains preserved over 5 years compared to those treated with a GLP-1 RA. Participants in the GLP-1 RA cohort had a greater weight and a more preserved renal function at baseline. These data suggest that physicians may be preferentially selecting patients with type 1 diabetes thought to benefit most from either cardio-renal benefits of SGLT2i therapy or the weight reduction associated with GLP-1 RA, established in type 2 diabetes [17]. We also report that 13% of patients with type 1 diabetes in our registry use adjuvant glucose-lowering therapy in addition to insulin, significantly higher than the 5.4% reported from the T1D Exchange Registry in the US [18].

The increased risk of DKA in the SGLT2i group, and the associated hospitalisation, was not great enough to offset the increased rate of hospitalisation of any cause in the GLP-1 RA group. These findings suggest that SGLT2i therapy may overall have more benefit than GLP-1 RA therapy in type 1 diabetes.

The benefit of SGLT2i therapy beyond improved glycaemic control was first reported in the EMPA-REG study which reported a reduced incidence of major adverse cardiovascular events in patients with type 2 diabetes [19]. Subsequently, multiple large outcome trials of SGLT2is have demonstrated a reduction in all-cause mortality and hospitalisation for heart failure in patients with HFrEF [20] and HFpEF [21], irrespective of diabetes status. A meta-analysis of 13,275 patients further supports that SGLT2i therapy significantly reduces the risk of hospitalisation and all-cause mortality [22]. The benefits of SGLT2i therapy also extends to CKD where there is a significantly reduced risk of decline in eGFR and progression to end stage renal failure, regardless of presence or absence of type 2 diabetes [23]. Given that atherosclerotic cardiovascular disease and CKD are the leading causes of morbidity and mortality in type 1 diabetes [24], it is reassuring that we show that SGLT2i therapy also confers similar benefits in type 1 diabetes. A previous real-world study on SGLT2i therapy showed a promising reduction in HbA_1c, weight and insulin requirements in type 1 diabetes but did not present long-term safety or reno-cardiovascular outcome data [25].

There is consistent evidence from RCTs and real-world studies that SGLT2i use in type 1 diabetes is associated with an increased risk of DKA [14]. Despite legitimate safety concerns and the lack of FDA (US Food and Drug Administration) approval for their use in type 1 diabetes, SGLT2i are used as adjunctive therapy in individuals with type 1 diabetes [18]. The UK’s National Institute for Health and Care Excellence and European Medicines Agency approved dapagliflozin as an adjunctive treatment of type 1 diabetes in 2019 and 2018, respectively [26, 27]. To mitigate the well-established risk of DKA, patients were required to have a total daily insulin requirement of ≥0.5 U/kg, undergo frequent home ketone monitoring and receive structured education [28]. However, in November 2021, approval was withdrawn after the marketing authorisation holder withdrew the indication for type 1 diabetes across Europe and the UK. This was notably not due to any new safety concerns and may potentially be driven by the manufacturer’s reluctance to introduce a ‘black triangle’ to ensure clinician and patient awareness of the increased monitoring requirements in type 1 diabetes [29].

Various guidelines have been proposed to mitigate the risk of adverse events, which includes avoiding initiation in those at extremes of age, recent history of DKA and frequent history of severe fungal infection [30]. SGLT2i therapy below the lowest approved dose may further reduce the risk of DKA [31]. There are currently an equal number of studies in SGLT2i and dual SGLT2i/SGLT1i therapy reporting both an increased [32–34] and decreased risk of DKA at maximal permitted doses [34–36] when compared with minimally licensed doses. The dose dependent relationship of SGLT2i and incidence of DKA remains unclear and further investigation is warranted. Concerns have been raised that the rate of DKA in real-world practice would surpass rates seen in closely supervised RCTs [37]. Whilst RCTs have reported DKA ranging from 0.8–4.3% [34], real-world rates of DKA have ranged from 0.0 [38]–3.5% [25] with one outlier of 12.5% [14]. We report a DKA rate at 5 years of 2.1%. Thus, the current literature does not support the concern of an increased risk of DKA in real-life practice, beyond levels seen in clinical trials. An initial analysis of the FDA adverse event reporting system (FAERS) data demonstrated no consistent trends in DKA with SGLT2i use over time [39], however an alternative cohort study of 50,220 individuals with type 2 diabetes showed the highest risk of DKA shortly after SGLT2i initiation, which subsequently reduced with time [40]. We demonstrate a positive association between length of time on therapy and DKA; further investigation is required to establish if this time-based association persists and remains significantly elevated above that of the baseline type 1 diabetes population. There is much debate surrounding the ‘tolerability’ of an increased risk of DKA, with comparisons being made with troglitazone and its subsequent removal from the market where the risk of fatal liver failure was greater than the current risk of DKA with SGLT2i use [31]. However, the long-term cardiovascular and renal benefits need to be carefully evaluated against the risk of DKA.

GLP-1 RAs promote weight loss [41], hence our findings of an increase in weight with GLP-1 RA therapy at 5 years was interesting. On further time-based analysis, we demonstrated a modest reduction in weight with GLP-1 RA use within the first 3 years (−2.7% of baseline body weight) before rebounding to above baseline at 5 years. Several factors could compound the disappointing degree of sustained weight loss seen in our cohort. Lower doses of GLP-1 RA are typically used for glycaemic control than in weight loss. Efficacy of GLP-1 RA at reducing weight is proportional to the higher baseline weight, i.e. greater degree of weight loss is seen when initiating from a higher starting weight [42]. A significant proportion of our cohort used dulaglutide (24%) and exenatide (13%), which are less efficacious for weight loss compared with liraglutide and semaglutide [43]. A proportion of participants also discontinued GLP-1 RA therapy after 3 years and it is well described that individuals can re-gain two-thirds of their initial weight loss upon cessation of GLP-1 RA [44]. We did, however, observe a −5.4 mmol/mol (−0.5%) reduction in HbA_1c in the GLP-1 RA cohort, similar to reports from randomised controlled trials [45, 46]. GLP-1 RA use in type 1 diabetes has been associated with an increased risk of symptomatic and asymptomatic hypoglycaemia [45]. Whilst day-to-day hypoglycaemia was not coded in our network, we employed hypoglycaemia requiring hospital admission as a surrogate outcome of severe hypoglycaemia and there was no difference in the incidence in our cohort compared to the general population with type 1 diabetes. We have not observed any difference in the rate of acute pancreatitis or gastrointestinal disturbance associated with GLP-1 RA compared to SGLT2i therapy and patients with type 1 diabetes. Only one previous study has investigated the effect of GLP-1 RA on lipid status in type 1 diabetes and found no difference compared with placebo [47]. We report a significant reduction in total cholesterol in individuals with type 1 diabetes treated with a GLP-1 RA. Given the relatively positive safety profile of GLP-1 RA with a reduction in weight and HbA_1c, without an increased risk of DKA, they are an appealing adjunctive therapy for people with type 1 diabetes.

This study does, however, have limitations typical of real-world data and we recognise the early exploratory nature of our work would benefit from further dedicated longitudinal studies with more robust design methodology in order to confirm our findings. We relied on accurate coding of medical data by healthcare providers, and we are unaware of the precise method of diagnosis of type 1 diabetes. However, given our data are predominantly North American, we would expect the diagnosis to be based on clinical presentation and biochemistry/antibody testing. Additionally, the overwhelming majority of our data are from the US, which limits generalisability to the rest of the world. Due to a lack of coding in the dataset, we were unable to report more nuanced and specific metrics such as duration of diabetes, total daily dose of insulin, percentage time in range and reason for starting adjunctive therapy, all of which are clinically relevant. However, given the size of the network, we were able to identify and include 2814 individuals for analysis, the largest real-world observational study of GLP-1 RA and SGLT2i use in type 1 diabetes to date. We were unable to record the average length of time each drug was prescribed for, but more than half of each cohort were still on SGLT2i and GLP-1 RA therapy 3 years post initiation. We did not differentiate between different types and doses of drugs in the two groups. As this was a real-world study, we intended to capture the greatest amount of data possible and report findings as a total aggregate by drug class for simplicity. Minor adverse reactions such as UTIs and candidiasis may have been underreported as patients may present to health services such as primary care physicians or community pharmacies, which were not part of the network. We did not include individuals treated with insulin alone as an active comparator as it would introduce a significant amount of bias with the lack of a codable index event (i.e. initiation of a drug or a new diagnosis) in an ‘insulin-only’ arm, but this is an area of future study. Introducing a scale of disease severity such as the New York Heart Association (NYHA) classification for heart failure and degree of albuminuria in CKD may have further strengthened our propensity matching.

Studies to clarify the risk–benefit ratio of DKA and improved cardio-renal outcomes with SGLT2i use in type 1 diabetes are required to help inform future guidelines and standardise clinical practice. Stratifying by the presence or absence of endogenous C-peptide as a potential protective factor against DKA should also be investigated. Multiple studies have demonstrated beneficial reductions in weight and HbA_1c with short acting GLP-1 RA in individuals with overweight or obesity and type 1 diabetes [45–48]. However, there has been no study to date investigating longer acting GLP-1 RA in individuals without overweight or obesity [49]. The improvement in HbA_1c following GLP-1 RA may primarily be driven by weight loss [50], but the mechanism may be more nuanced in individuals with type 1 diabetes and normal body weight. The glucagonostatic characteristics of GLP-1 RA could prove to be more relevant to individuals with beta cell deficiency [51] which contrasts with type 2 diabetes where beta cell dysfunction and impaired incretin effect may dominate [52]. To what extent this relationship translates into alterations in HbA_1c, time in range and total daily dose of insulin in patients with ideal body weight and type 1 diabetes is yet to be determined. The efficacy and safety of combination treatment with both GLP-1 RA and SGLT2i has not been studied to date in type 1 diabetes. Pivotal clinical trials in continuous ketone monitoring, akin to currently available continuous glucose monitoring, are taking place in 2023 [53]. If successful, future work investigating the combination of an SGLT2i with a dual continuous glucose and ketone sensor may allow individuals with type 1 diabetes to access SGLT2i therapy.