Effects of ertugliflozin on renal function over 104 weeks of treatment: a post hoc analysis of two randomised controlled trials

Beyond lowering blood glucose levels and inducing weight loss in individuals with type 2 diabetes mellitus, sodium–glucose cotransporter 2 (SGLT2) inhibitors are associated with protective renal and cardiovascular effects. These agents reduce BP by approximately 4.0 mmHg systolic and 2.0 mmHg diastolic, possibly through reducing endothelial dysfunction and arterial stiffness [1], and by inducing a modest degree of circulating volume contraction [2]. Regardless of the responsible mechanism, some SGLT2 inhibitors have been demonstrated to reduce cardiovascular events compared with placebo in renal and cardiovascular outcome trials [3‐6]. However, mechanisms are still being elucidated [7‐10].

At the kidney level, SGLT2 inhibitors attenuate intraglomerular hypertension and single nephron hyperfiltration by activating tubuloglomerular feedback [11, 12], a mechanism likely linked with albuminuria lowering in the setting of diabetes [13, 14]. Importantly, anti-albuminuric effects of SGLT2 inhibitors have been reported to be largely independent of other factors that influence urine albumin excretion (i.e. declines in body weight, HbA_1c and BP) [14]. Mechanisms related to changes in haemodynamic function, renal hypoxia and inflammation may also contribute to albuminuria-lowering effects and renal protection with SGLT2 inhibitors [7, 15, 16].

The hypothesis that SGLT2 inhibitors confer renal protection was tested in the Canagliflozin and Renal Events in Diabetes with Established Nephropathy Clinical Evaluation (CREDENCE) trial, which reported a 30% reduction in the primary composite endpoint (end-stage kidney disease, doubling of the serum creatinine level or death from renal or cardiovascular causes) [3]. The Dapagliflozin and Prevention of Adverse Outcomes in Chronic Kidney Disease (DAPA-CKD; ClinicalTrials.​gov NCT03036150) and EMPA-KIDNEY studies (NCT03594110) will further assess the renal protective effects of SGLT2 inhibitors in patients with chronic kidney disease (CKD) with or without diabetes.

Ertugliflozin is a selective SGLT2 inhibitor approved for use in adults with type 2 diabetes mellitus as a glucose-lowering agent [17, 18], and is being evaluated in the ongoing cardiovascular outcome trial eValuation of ERTugliflozin effIcacy and Safety (VERTIS) CV [19]. Ertugliflozin exerts metabolic and BP effects that are comparable to other SGLT2 inhibitors [20]. Some of the longer-term renal effects of ertugliflozin, including changes in urine albumin/creatinine ratio (UACR) and renal safety, have not been reported.

Accordingly, in this post hoc, exploratory analysis, we used data from two RCTs (VERTIS MET [ClinicalTrials.​gov registration no. NCT02033889] and VERTIS SU [NCT01999218]) that evaluated ertugliflozin vs glimepiride or placebo/glimepiride (non-ertugliflozin) in patients with type 2 diabetes mellitus [21, 22], as described in the electronic supplementary material (ESM) Table 1 and ESM Figs 1 and 2. These studies were combined for the purposes of this analysis as they included patients with similar background glucose-lowering medication, had the same study duration of 104 weeks and used glimepiride as the active comparator. The inclusion of studies with an active comparator was important in order to attenuate the impact of glucose lowering on exploratory outcomes. The two studies were pooled to evaluate the effects of ertugliflozin on eGFR (using the Modification of Diet in Renal Disease study equation) and albuminuria (UACR, geometric mean of change from baseline) in the overall population and by baseline UACR category over 104 weeks. The renal safety profile of ertugliflozin was assessed in the overall population.

Data were pooled from two Phase III studies from the VERTIS programme in patients with type 2 diabetes mellitus: VERTIS MET (protocol MK-8835-007; ClinicalTrials.​gov NCT02033889) and VERTIS SU (protocol MK-8835-002; ClinicalTrials.​gov NCT01999218).

The studies were conducted in accordance with principles of Good Clinical Practice and were approved by the appropriate institutional review boards and regulatory agencies. Informed consent was obtained from individuals in each study. The designs for the two studies have been previously published [21‐24] and are summarised in ESM Table 1. Both studies had two treatment phases and a treatment period of 104 weeks. The VERTIS MET study comprised a double-blind, placebo-controlled, 26 week treatment period (Phase A) and a double-blind, 78 week treatment extension period (Phase B; ESM Fig. 1). In Phase A, patients with an HbA_1c of 53–91 mmol/mol (7.0–10.5%) who were receiving metformin monotherapy were randomised to placebo, ertugliflozin 5 mg or ertugliflozin 15 mg administered once daily. In Phase B, participants who had a fasting capillary glucose ≥6.1 mmol/l and were not rescued during Phase A had the addition of blinded glimepiride (for participants randomised to placebo) or glimepiride placebo (for participants randomised to ertugliflozin). Overall, 621 participants were randomised at the start of the study (209, 207 and 205 patients in the placebo, ertugliflozin 5 mg and ertugliflozin 15 mg groups, respectively), and 581 participants entered Phase B (190, 201 and 190 participants in the glimepiride, ertugliflozin 5 mg and ertugliflozin 15 mg groups, respectively) and received ≥1 dose of study medication in Phase B [21]. The VERTIS SU study comprised a double-blind, active-controlled, 52 week treatment period (Phase A), with a double-blind, active-controlled 52 week treatment extension period (Phase B; ESM Fig. 2). In Phase A, patients with an HbA_1c of 53–75 mmol/mol (7.0–9.0%) who were receiving metformin monotherapy were randomised to glimepiride, ertugliflozin 5 mg or ertugliflozin 15 mg administered once daily. These treatments were continued during the Phase B extension period. Overall, 1315 participants were randomised and had ≥1 dose of study medication in the study (435, 445 and 435 patients in the glimepiride, ertugliflozin 5 mg and ertugliflozin 15 mg groups, respectively). Of these, 1037 participants entered Phase B (349, 337 and 351 patients in the glimepiride, ertugliflozin 5 mg and ertugliflozin 15 mg groups, respectively). Participants receiving glimepiride in VERTIS SU, and those receiving placebo (26 weeks) and glimepiride (78 weeks) in VERTIS MET, comprised the non-ertugliflozin control group for the purpose of all analyses reported here.

The analyses were performed on the safety population (randomised patients who took ≥1 dose of study medication) and included data after the initiation of glycaemic rescue therapy. Mean changes from baseline in eGFR and UACR over 104 weeks were evaluated using the longitudinal data analysis model. The model contained fixed effects for treatment, time, trial, treatment by time interaction and baseline covariates (HbA_1c, systolic BP and eGFR for the eGFR analysis, and HbA_1c, systolic BP and UACR for the UACR analysis). In this model, time was treated as a categorical variable so that no restriction was imposed on the trajectory of the means over time. The treatment difference in terms of mean change from baseline to a given time point was estimated and tested. An unstructured covariance matrix was used to model the correlation among repeated measurements. The analysis was performed on data from the overall population and patients with baseline UACR ≥3.39 mg/mmol. Due to the non-normal distribution of UACR, log transformation of UACR data was applied before analysis. Adjusted least squares means (LSMs) with 95% CIs and differences between treatments were back-transformed to the original scale. Adjusted percentage changes (derived from the exponentiation of adjusted LSMs) in UACR from baseline are presented.

The slopes for changes in eGFR per week were analysed by random coefficient models. The model included the eGFR value as a response variable with treatment group, time (in weeks), treatment by time interaction and baseline eGFR as linear covariates. The model allowed individual participant slopes to vary by random effects of intercept and time.

Adverse events (AEs) occurring up to 14 days after the final dose of study medication were included. Key endpoints included AEs related to decreased eGFR (defined by a Custom Medical Dictionary for Regulatory Activities [MedDRA; version 19.0] Query, which comprised the following preferred terms: blood creatinine increased, GFR decreased, creatinine renal clearance decreased and hypercreatininaemia); renal-related AEs (defined by the standardised MedDRA query of acute renal failure, narrow); and renal events adjudicated for causality by a blinded, external, independent committee.

Changes from baseline in metabolic, haemodynamic and volume-related biochemical variables (HbA_1c, fasting plasma glucose, weight, systolic and diastolic BP, pulse rate, haematocrit, haemoglobin, albumin, sodium, potassium, calcium, bicarbonate, magnesium, phosphate levels and uric acid) at week 104 were assessed by treatment group in the overall population. The analysis was based on a mixed model for repeated measures with treatment, time, trial, the interaction of time by treatment and baseline covariates (HbA_1c, eGFR and systolic BP).

A total of 1936 randomised, treated patients were included in the analyses. Of these, 644 patients received non-ertugliflozin, 652 ertugliflozin 5 mg and 640 ertugliflozin 15 mg. Baseline characteristics in the treatment groups were balanced (Table 1). Overall, mean (SD) baseline eGFR was 88.2 (18.8) ml min⁻¹ (1.73 m)⁻² and geometric mean (95% CI) of baseline UACR was 1.31 mg/mmol (1.23, 1.38). In the overall cohort, 22.5%, 23.1% and 20.1% of patients had UACR ≥3.39 mg/mmol at baseline in the non-ertugliflozin, ertugliflozin 5 mg and ertugliflozin 15 mg groups, respectively. The proportions of patients taking antihypertensive, lipid-lowering and glucose-lowering agents are shown in Table 1.

In this exploratory analysis, ertugliflozin was associated with an initial, early dip in eGFR, followed by a return in eGFR back to or above baseline levels by 104 weeks. By contrast, eGFR decreased over time in the non-ertugliflozin group. These findings were observed even though changes in HbA_1c were comparable in ertugliflozin vs non-ertugliflozin treatment groups. In addition, among patients with albuminuria at baseline, the ertugliflozin groups had greater reductions in UACR at all measured time points up to week 104 even though changes in HbA_1c were similar vs non-ertugliflozin therapy. The longer-term renal safety profile of ertugliflozin was generally consistent with observations made with other SGLT2 inhibitors [4, 13].

Experimental models of diabetes and mechanistic studies in humans have shown that SGLT2 inhibition reduces renal hyperfiltration through activation of tubuloglomerular feedback, leading to afferent arteriolar vasoconstriction and a decline in intraglomerular pressure [10, 25, 26]. Importantly, hyperfiltration is associated with CKD initiation and progression, and a variety of physiological factors that promote CKD progression [27‐29]. Reduced glomerular pressure by SGLT2 inhibition is also mediated by increases in tubular pressure that reduce net filtration pressure and glomerular hypertension [12, 30]. In patients with type 2 diabetes mellitus, albuminuria decreases by 7–41% after 12 weeks in response to SGLT2 inhibition with empagliflozin, and stays low even after 3 years of treatment in clinical studies [13, 14]. This is similar to experimental models and is a consequence of changes in renal physiology [31].

Similar to observations in experimental studies demonstrating SGLT2 inhibition effects on intraglomerular hypertension and hyperfiltration, these agents reduce inulin-based measures of hyperfiltration and hyperperfusion in human mechanistic studies [11]. In larger clinical studies, SGLT2 inhibitors acutely induce a dip in eGFR [32, 33], which happens even after a single dose of drug [34] and tends to return towards baseline over time. Importantly, after a 2–4 week washout period, eGFR tends to return to baseline even after prolonged periods of therapy [35, 36]. This pattern of renal function change is thought to reflect known haemodynamic effects of SGLT2 inhibitors in the renal microcirculation. In a similar fashion to these previous observations with other SGLT2 inhibitor agents, ertugliflozin induced an eGFR dip during weeks 0–6 in the current exploratory analysis. Based on known mechanisms, changes in renal function are mainly attributed to sodium-related pathways [37], rather than renal glucose handling or glycaemic lowering [7, 8, 15, 38]. Similar to effects on BP lowering [39], the characteristic early dip in eGFR occurs in patients with normal kidney function and in those with CKD stages 2–4 [35, 40].

Interestingly, in the current analysis, after the initial dip, eGFR tended to increase back towards or above baseline over time. These findings are consistent with the pharmacodynamic effects of ertugliflozin. While the mechanisms responsible for the eGFR dip have relatively accepted explanations [16, 41, 42], the mechanism for the rise in eGFR back to and above baseline is incompletely understood but may be related to compensatory upregulation of sodium–glucose cotransporter 1 [43]. This results in increased sodium–glucose cotransport and afferent arteriolar redilatation, resulting in increases in renal perfusion and GFR. Aside from hypothetical pathways leading to the rise in eGFR over time, similar increases in eGFR towards baseline levels have been reported in the Canagliflozin Cardiovascular Assessment Study (CANVAS) and EMPA-REG OUTCOME studies [13, 44]. The clinical relevance of this pattern of renal function change over time is not yet understood but appears in part to reflect preservation of renal function. Beyond these possible haemodynamic effects in the kidney, ertugliflozin was associated with increases from baseline in serum albumin, haemoglobin and haematocrit—changes that have been most closely associated with haemoconcentration in the systemic circulation on the basis of natriuresis and osmotic diuresis. While these volume-related effects have been linked with cardiovascular protection, especially against heart failure hospitalisation [10, 45], their clinical relevance is not currently known. Interestingly, despite the natriuresis and volume-related effects attributed to SGLT2 inhibitors, as with previous work, ertugliflozin did not impact plasma electrolyte levels in a meaningful way, aside from small changes in magnesium and phosphate that were unlikely to be clinically significant. As expected based on previous work, uric acid decreased—an effect that has been attributed to glucosuria-related uricosuria, and associated with cardiovascular benefits in exploratory mediation analyses from the EMPA-REG OUTCOME study [45, 46].

Ertugliflozin is a highly selective SGLT2 inhibitor approved for use as a glucose-lowering therapy in patients with type 2 diabetes mellitus. Beyond changes in eGFR in the current analysis, ertugliflozin also reduced UACR in patients with elevated albuminuria at baseline. In previous work, empagliflozin reduced UACR in patients with type 2 diabetes mellitus with either baseline micro- or macroalbuminuria [13]. These changes were largely independent of changes in body weight, BP or HbA_1c. In other work, Heerspink et al. [47, 48] demonstrated that dapagliflozin and canagliflozin similarly reduced UACR—effects that were also largely independent of changes in these variables. The current analysis in a large study cohort using an active glucose-lowering comparator minimised differences in HbA_1c over time, supporting observations from placebo-controlled trials and suggesting that favourable effects on eGFR and albuminuria are not due to differences in glycaemic control, which is a more distinct feature of this analysis. Moreover, in this analysis we examined both acute and chronic eGFR changes in this cohort. One caveat is that, compared with non-ertugliflozin-treated patients, preservation of eGFR was only apparent in ertugliflozin-treated patients in the >6 (>0.5%) and >3 to ≤6 mmol/mol (>0.3 to ≤0.5%) HbA_1c reduction categories, but not in the treatment groups with ≤3 mmol/mol (≤0.3%) HbA_1c reduction, at week 104. Non-ertugliflozin-adjusted reductions from baseline in UACR in ertugliflozin-treated patients were also observed in the >3 to ≤6 mmol/mol (>0.3 to ≤0.5%) HbA_1c reduction category compared with groups in the other HbA_1c-lowering categories. Therefore, an interaction between the degree of HbA_1c lowering and kidney protection cannot be ruled out. Based on existing data, it may be speculated that, rather than being based on alterations in body weight, BP or HbA_1c, reductions in albuminuria and eGFR preservation are based primarily on reduced glomerular hypertension or improvements in glomerular membrane barrier function [49]—a hypothesis that is being tested in dedicated kidney outcome studies [50].

In the context of previous work demonstrating that changes in UACR are largely independent of other clinical factors that reduce albuminuria [14], our analysis emphasises that, even when compared with a sulfonylurea with comparable glucose-lowering efficacy, ertugliflozin reduced UACR. It is therefore important to define possible albuminuria-lowering and renal protective effects of SGLT2 inhibitors in patients with non-diabetic kidney disease, where hyperglycaemia is not implicated in kidney disease progression. This critical question is being examined in mechanistic studies, such as the Effects of Dapagliflozin in Non-diabetic Patients With Proteinuria (DIAMOND) study (ClinicalTrials.​gov registration no. NCT03190694), and in long-term renal outcome studies (DAPA-CKD [NCT03036150] and EMPA-KIDNEY [NCT03594110]), which are including patients with non-diabetic kidney disease. These last two trials will test the hypothesis that renal protective effects of SGLT2 inhibition are independent of changes in ambient glucose levels. Similarly, for ertugliflozin specifically, the VERTIS CV trial (ClinicalTrials.​gov registration no. NCT01986881), which is being performed in over 8000 patients with established cardiovascular disease to assess cardiovascular safety, may elucidate possible renal protective effects with this agent since kidney outcomes are important secondary endpoints [19].

In conclusion, transient and modest reductions in eGFR observed in the ertugliflozin groups at week 6 gradually returned to baseline by week 104. By contrast, eGFR declined from baseline through week 104 in the non-ertugliflozin group. The results are consistent with the pharmacodynamic effect seen with SGLT2 inhibitors and are suggestive of preservation of renal function with ertugliflozin. In patients with baseline albuminuria, ertugliflozin demonstrated a reduction in albuminuria over 104 weeks.