FormalPara Key Points

Sodium-glucose cotransporter type 2 inhibitors (SGLT2is) are the only glucose-lowering agents that specifically target the kidneys.

Exposure to SGLT2is moderately increases in patients with chronic kidney disease.

SGLT2i-induced urinary glucose excretion progressively decreases along with the decline of the estimated glomerular filtration rate (eGFR), yet sodium excretion and blood pressure reduction are maintained even with a reduced eGFR.

Reduction in glycosylated haemoglobin is lower in patients with moderate chronic kidney disease than in patients with normal renal function, but with similar reductions in body weight and blood pressure, and with a preserved safety profile.

Pharmacokinetic, pharmacodynamic and clinical results in randomised controlled trials in the presence of chronic kidney disease are similar in Asian and non-Asian patients with type 2 diabetes mellitus.

Cardiovascular and renal protection is observed in patients at high cardiovascular risk independently of glucose control and baseline eGFR.

New guidelines recommend the prescription of SGLT2is in patients with an eGFR between 30 and 90 mL/min/1.73 m2 and/or macroalbuminuria.

1 Introduction

Type 2 diabetes mellitus (T2DM) is a prevalent chronic disease worldwide, which is associated with numerous complications, especially atherosclerotic cardiovascular (CV) disease, heart failure (HF) and chronic kidney disease (CKD). In most Western countries, the leading cause of end-stage renal disease (ESRD) is diabetes mellitus [1, 2]. Chronic kidney disease is defined as a decrease in the estimated glomerular filtration rate (eGFR: stage 2 or mild CKD: 60 to < 90 mL/min/1.73 m2; stage 3 or moderate CKD: 30–59 mL/min/1.73 m2, divided into stage 3a (45–59 mL/min/1.73 m2) and stage 3b (30 to < 45 mL/min/1.73 m2); stage 4 or severe CKD: 15 to < 30 mL/min/1.73 m2; ESRD: < 15 mL/min/1.73 m2) [3]. Microalbuminuria is viewed as an early marker of renal damage, which may progress to overt proteinuria as the eGFR declines [3]. Chronic kidney disease is associated with a higher risk of CV disease and HF in patients with T2DM [2]. Thus, prevention of advanced CKD and ESRD is the main objective in the management of patients with T2DM [1, 4].

Sodium-glucose cotransporter type 2 inhibitors (SGLT2is) recently emerged as oral glucose-lowering agents that exert CV and renal protective effects [5,6,7,8,9,10,11]. These effects appear specific, independent of improvement of glucose control, and the renoprotective effects are better than those reported with any other glucose-lowering agent [12]. It is well known that the presence of CKD may contraindicate, limit the use or force dose reduction of several glucose-lowering agents [13,14,15]. This was also the case for SGLT2is, more for loss of efficacy than for obvious safety reasons [16, 17]. Indeed, SGLT2i glucose-lowering potency, which mainly results from increased glucosuria, is decreasing as the eGFR declines, a decrease explained by the reduction in urinary glucose excretion (UGE) [18]. As a consequence, SGLT2is, primarily developed as antidiabetic agents, have long been contraindicated in patients with stage 3 CKD, i.e. if the eGFR was below 60 mL/min/1.73 m2.

However, in recent years, there is an increasing interest to prescribe SGLT2is in patients with mild-to-moderate CKD to benefit from CV and renal protection [18, 19]. Indeed, a post-hoc meta-analysis of three landmark CV outcome trials with three different SGLT2is (EMPA-REG OUTCOME with empagliflozin, CANVAS with canagliflozin and DECLARE-TIMI-58 with dapagliflozin) showed a reduction in major cardiovascular events (MACEs), CV mortality, hospitalisation for HF and a composite renal outcome (including progression to ESRD), a prespecified secondary endpoint, in patients with T2DM at high CV risk. Of utmost interest, this cardiorenal protection was preserved even in patients with a baseline eGFR comprised between 30 and 59 mL/min/1.73 m2, despite a lower glucose-lowering efficacy [20].

Furthermore, CREDENCE (“Canagliflozin and Renal Events in Diabetes with Established Nephropathy Clinical Evaluation”), the first dedicated trial with renal outcome as a primary endpoint, which recruited patients with T2DM with mild-to-moderate CKD (eGFR 30–90 mL/min/1.73 m2) and macroalbuminuria (urinary albumin creatinine ratio (UACR) > 300 mg/g), fully confirmed both CV and renal protection in this specific population [21]. Thus, recent international guidelines now recommend the use of SGLT2is in patients with moderate CKD, a subgroup that was initially excluded, because of poorer glucose-lowering efficacy, to prevent both CV complications (including HF) and progression of renal impairment [22, 23].

The present updated comprehensive review is complementary to our previous paper published in 2015 [17]. It aims to describe the pharmacokinetic (PK)/pharmacodynamic (PD) characteristics of SGLT2is in patients with different stages of CKD and their clinical use in patients with T2DM and CKD, mainly stage 3 CKD with, if possible, a distinction between stage 3a and stage 3b. As the safety of SGLT2is in patients with stage 2–3 CKD was shown to be not different from the safety in patients with normal kidney function [24], this article mainly focuses on efficacy rather than on safety. Because of possible differences between Caucasian (non-Asian) and Asian patients [25], the data from the literature are analysed separately in the two ethnic groups.

2 Site and Mechanisms of Action of Sodium-Glucose Cotransporter Type 2 Inhibitors

Sodium-glucose cotransporter type 2 inhibitors inhibit the main glucose transporter SGLT2 on the luminal surface of the proximal tubule, thereby lowering the threshold for UGE in the kidney. This effect results in a desired glucose-lowering effect. In addition, enhanced glucosuria is associated with calorie loss, osmotic diuresis and natriuresis, which lead to some weight loss and blood pressure lowering [26]. While SGLT2is were initially developed as glucose-lowering medications, primarily for the management of T2DM, they have shown a range of actions involving multiple biochemical, metabolic, endocrine and hemodynamic pathways [9, 11, 27]. All these effects could contribute to the beneficial CV and renal effects consistently reported with all compounds belonging to this innovative pharmacological class. A detailed analysis of these numerous complex mechanisms is beyond the scope of this paper and may be found in several extensive reviews [7, 9, 11, 28].

Beneficial effects in the kidney result from many converging factors related not only to glucose but also to sodium. From a natriuretic perspective, it has been shown in models of diabetes-related hyperfiltration that increased distal natriuresis to the macula densa in response to SGLT2is activates a reflex called tubuloglomerular feedback [11]. This effect leads to afferent arteriolar vasoconstriction, likely via an adenosine-related mechanism [29]. Intriguingly, however, this effect has been challenged recently by results suggesting a post-glomerular vasodilation rather than a pre-glomerular vasoconstriction [30]. These hemodynamic effects may explain the characteristic acute initial fall in eGFR with SGLT2is, an effect that occurs quickly and is rapidly reversible after cessation of the medication. They also account for the substantial albuminuria-lowering effect of SGLT2is. Finally, a sustained reduction in intraglomerular pressure most probably contributes to the long-term renal protection. Aside from hemodynamic mechanisms, the SGLT2i-associated inhibition of glucose reabsorption at the proximal tubule suppresses pathways linked with inflammation and fibrosis, possibly by reducing energy demand and renal hypoxia [11]. Beyond simply understanding physiological mechanisms, the distinction between sodium vs glucose-related mechanisms involved in CV and renal protection is becoming of major interest because of the potential for future uses of SGLT2is in normoglycemic non-diabetic individuals (see Sect. 6).

3 Pharmacokinetic/Pharmacodynamic Considerations

Commercialised SGLT2is, both worldwide (canagliflozin, dapagliflozin, empagliflozin, ertugliflozin) and in Japan (ipragliflozin, luseogliflozin, tofogliflozin), share similar PK characteristics with a rapid oral absorption, a long elimination half-life allowing once-daily administration, an extensive hepatic metabolism mainly via glucuronidation to inactive metabolites and a low renal elimination as a parent drug [17, 31]. All these SGLT2is were evaluated after the oral administration of a single dose in different groups of patients with T2DM with normal kidney function (stage 1: eGFR > 90 mL/min/1.73 m2), mild CKD (stage 2: eGFR 60–90 mL/min/1.73 m2), moderate CKD (stage 3: eGFR 30–59 mL/min/1.73 m2) and severe CKD (stage 4: eGFR < 30 mL/min/1.73 m2). In all these studies, the PK characteristics may be summarised by the maximum observed plasma concentration (Cmax) and drug exposure assessed by the area under the plasma concentration–time curve from zero to infinity (AUC0–infinity). The PD response to SGLT2is was evaluated by the changes in UGE during the first 24 h following a single-drug administration.

3.1 Non-Asian/Caucasian Patients

Maximum observed plasma concentration and drug exposure were increased in patients with T2DM with CKD compared with individuals with normal renal function whatever the SGLT2i considered, canagliflozin, dapagliflozin, empagliflozin, ertugliflozin and ipragliflozin [32,33,34,35,36] (Table 1). There were some differences when comparing the amplitudes of Cmax and AUC0–infinity changes across SGLT2is. However, in the absence of head-to-head trials, no definite conclusion could be drawn about possible clinically relevant differences between compounds. Of note, there was no clear-cut inverse relationship between Cmax or drug exposure and eGFR when renal impairment progressed from mild to severe CKD (Table 1). The maximum increase in AUC0–infinity was less than 100% in patients with moderate or severe CKD and the increase in Cmax was consistently lower than the increase of AUC0–infinity. These results explain why no dose adjustment is generally required in patients with mild-to-moderate CKD. However, for canagliflozin, the dose of 100 mg (dose used in the renal outcome trial CREDENCE: see below) instead of 300 mg is recommended in patients with mild-to-moderate CKD. Very few studies analysed the pharmacokinetics/pharmacodynamics of SGLT2is in patients with ESRD, one with canagliflozin [32] and one with empagliflozin [34] (Table 1). SGLT2is are not recommended in patients with severe CKD or ESRD, even those treated with dialysis [18].

Table 1 Maximum observed plasma concentration (Cmax), parent drug exposure (area under the plasma concentration–time curve (AUC0infinity) and increase from baseline in 24-h urinary glucose excretion (UGE) in non-Asian/Caucasian patients with type 2 diabetes mellitus with different degrees of chronic kidney disease (CKD)

In contrast to what was observed with the PK parameters, a clear-cut relationship was noticed between the PD response and the deterioration of renal function. Indeed, a progressive and marked reduction in UGE was consistently reported as the eGFR declined (Table 1). Urinary glucose excretion was reduced by 30–50% in patients with mild CKD, but the reduction may reach 60–75% in patients with moderate CKD and up to 80–90% in patients with severe CKD. Thus, this reduction in UGE may alter the glucose-lowering efficacy of SGLT2is, an observation that led to limiting the use of these medications in patients with T2DM with moderate-to-severe CKD when they were commercialised as antidiabetic agents [17, 37].

3.2 Asian/Japanese Patients

Similar results as those previously reported in non-Asian/Caucasian patients [32, 34, 36] were obtained in Japanese patients with T2DM and different degrees of CKD when PK characteristics of canagliflozin 100 mg or 200 mg [38], empagliflozin 25 mg [39] and ipragliflozin 50 mg or 100 mg [36, 40] were evaluated. Overall, the increases in Cmax were minimal while AUC0–infinity was ≤ 1.5 fold the reference value obtained in patients with normal kidney function, whatever the degree of CKD (Table 2). Pharmacokinetic results reported for luseogliflozin and tofogliflozin confirmed the absence of clinically relevant changes in Cmax and AUC0–infinity in patients with mild, moderate or severe CKD compared with patients with normal renal function [41, 42] (Table 2).

Table 2 Maximum observed plasma concentration (Cmax), parent drug exposure [area under the plasma concentration–time curve (AUC0–infinity)] and increase from baseline in 24-h urinary glucose excretion (UGE) in Asian patients with type 2 diabetes mellitus with different degrees of chronic kidney disease (CKD)

As in non-Asian patients, UGE steadily decreased with the decline in eGFR in Asian patients with T2DM, and this trend was consistently observed independently which SGLT2i was investigated [36, 38,39,40,41,42] (Table 2). By using a PK/PD model, a 1.17-fold increase in area under the plasma concentration–time curve from zero to 24 h of ipragliflozin and a 0.76-fold change in UGE over 24 h were estimated in patients with T2DM with moderate renal impairment compared with those with normal renal function [43].

4 Trials in Patients with Type 2 Diabetes Mellitus with Chronic Kidney Disease

The clinical efficacy of SGLT2is was evaluated in dedicated randomised controlled trials (RCTs) in patients with T2DM with CKD, mainly in stage 3 (sometimes divided into 3a and 3b), with primary analysis at 24–26 weeks [42, 44,45,46,47,48,49,50,51,52,53] (Table 3). Some of these studies had an extension up to 52 weeks [44, 46, 50, 53, 54], with one also up to 104 weeks [44] (Table 4). The most representative criteria of evaluation were the diminution of glycosylated haemoglobin (HbA1c) and fasting plasma glucose (FPG), the reduction of body weight and the lowering of systolic blood pressure (SBP). Several of these RCTs also measured the changes in UACR with SGLT2is. They generally showed a trend for a greater reduction in UACR with SGLT2is than with placebo [48, 51, 52, 54], yet not in all studies [45, 47]. Besides the efficacy, drug safety in patients with T2DM with stage 3 (both 3a and 3b) was evaluated in each of these RCTs. No specific adverse events could be identified and the overall safety profile was similar to that previously reported in patients with T2DM with normal renal function [17, 24]. In particular, the risk of acute kidney injury was not increased, but rather reduced among patients with T2DM with moderate CKD treated with an SGLT2i [24] as it was also observed in the whole population with T2DM [55].

Table 3 Efficacy of sodium-glucose cotransporter type 2 inhibitors (SGLT2is) in patients with type 2 diabetes mellitus with mild (stage 2) or moderate (stage 3: 3a and 3b) chronic kidney disease (CKD) after 24–26 weeks (primary analysis)
Table 4 Efficacy of sodium-glucose cotransporter type 2 inhibitors (SGLT2is) in patients with type 2 diabetes mellitus with mild (stage 2) or moderate (stage 3: 3a and 3b) chronic kidney disease (CKD) after 52 weeks (extension study)

4.1 Non-Asian/Caucasian Patients

4.1.1 Canagliflozin

Two doses of canagliflozin (100 mg and 300 mg) were evaluated in patients with T2DM and stage 3 CKD (eGFR ≥ 30 to < 50 mL/min/1.73 m2) over a period of 26 weeks [48] (Table 3), with an extension up to 52 weeks (Table 4) [54]. Canagliflozin significantly reduced HbA1c, FPG, body weight and SBP both after 26 weeks and 52 weeks. Of note, albuminuria was reduced with the two doses of canagliflozin compared with placebo at both 26 weeks (Table 3) and 52 weeks (Table 4).

In a pooled analysis of integrated data from four phase III RCTs that enrolled patients with T2DM and stage 3 CKD, placebo-subtracted reductions in HbA1c (− 0.38 and − 0.47%; p < 0.001), body weight (− 1.6 and − 1.9%; p < 0.001) and SBP (− 2.8 and − 4.4 mmHg; p < 0.01) were observed with canagliflozin 100 and 300 mg, respectively [56]. When patients were separated according to baseline eGFR, the reduction in HbA1c was smaller in patients with stage 3b CKD than in patients with stage 3a CKD with both canagliflozin 100 mg (− 0.23% vs − 0.47%) and canagliflozin 300 mg (− 0.39% vs − 0.52%) [56].

4.1.2 Dapagliflozin

In patients with T2DM with mild-to-moderate CKD (most patients with stage 3a or 3b), the mean change in HbA1c after 24 weeks (primary endpoint) was not statistically different from placebo (− 0.41% and − 0.44% for 5- and 10-mg doses of dapagliflozin, respectively, vs − 0.32% for placebo) (Table 3) [44]. The placebo-subtracted mean reduction in HbA1c averaged 0.33–0.37% in patients in stage 3a CKD, whereas there was no change seen in patients in stage 3b CKD. In contrast, in both subgroups, there was a significant reduction in body weight of about 1.5–2.0 kg and in SBP of about 4–6 mmHg with the two doses of dapagliflozin compared with placebo; these changes were observed after 24 weeks (Table 3) and were sustained in patients with T2DM with moderate-to-severe CKD until 52 and 104 weeks (Table 4) [44]. These results were confirmed in DERIVE, a RCT that enrolled patients with T2DM with stage 3a CKD. Dapagliflozin 10 mg significantly decreased HbA1c, body weight, FPG and SBP at week 24, compared with placebo [47] (Table 3).

In the recent DELIGHT study, a multinational RCT that recruited patients with T2DM and CKD (eGFR of 25–75 mL/min/1.73 m2 and UACR 30–3500 mg/g), positive effects on UACR were also reported with dapagliflozin compared with placebo (− 21.0%, 95% confidence interval (CI) − 34.1, − 5.2; p = 0.011), despite only a non-significant trend for a reduction in HbA1c, FPG, body weight and SBP. In this study, the reduction in UACR was enhanced with the dapagliflozin–saxagliptin combination (− 38.0%, 95% CI − 48.2, − 25.8; p < 0.0001) [45].

In a pooled analysis of 11 phase III RCTs in patients with T2DM and an eGFR between 12 and < 45 mL/min/1.73 m2, placebo-subtracted changes in HbA1c with dapagliflozin 5 mg and 10 mg were negligible (only 0.03% with both doses) during the overall 102-week period. These disappointing results contrasted with significant reductions in body weight (− 1.7 kg and − 2.2 kg), SBP (− 1.4 mmHg and − 3.8 mmHg) and albuminuria (− 47.1% and − 57.6%), with 5 mg and 10 mg of dapagliflozin, respectively [49].

4.1.3 Empagliflozin

The efficacy and safety of empagliflozin as an add-on treatment were evaluated in patients with T2DM and stage 2 or 3a CKD for 52 weeks [46]. In patients with stage 2 CKD, mean placebo-corrected changes from HbA1c baseline at week 24 were − 0.52% (95% CI − 0.72, − 0.32) for empagliflozin 10 mg and − 0.68% (95% CI − 0.88, − 0.49) for empagliflozin 25 mg (both p < 0.0001). In patients with stage 3a CKD, the corresponding change in HbA1c at week 24 was lower but still significant (− 0.42%, 95% CI − 0.56, − 0.28) for empagliflozin 25 mg (p < 0.0001) (Table 3). The reductions in HbA1c were sustained at 52 weeks in all groups. Empagliflozin showed significant reductions in body weight and SBP at week 24 (Table 3) and week 52 (Table 4) compared with placebo in patients with T2DM with stage 2 and stage 3a CKD [46]. In patients with stage 3a CKD, fewer patients taking empagliflozin 25 mg than placebo shifted from no albuminuria at baseline to microalbuminuria or from microalbuminuria to macroalbuminuria, whereas more patients improved from macroalbuminuria to microalbuminuria or from microalbuminuria to no albuminuria [46].

4.1.4 Ertugliflozin

The efficacy of ertugliflozin was assessed over 52 weeks, with a primary analysis after 26 weeks in patients with stage 3 CKD. At week 26, reductions from baseline in HbA1c were observed across groups: − 0.3% (95% CI − 0.4, − 0.1), − 0.3% (95% CI − 0.4, − 0.1) and − 0.4% (95% CI − 0.6, − 0.3) for placebo and ertugliflozin 5 mg and 15 mg, respectively (Table 3). Greater reductions from baseline in HbA1c, FPG, body weight and SBP were observed with ertugliflozin vs placebo at week 26 in the subgroup with stage 3a CKD compared with the subgroup with stage 3b CKD (Table 3). Overall, changes after 26 weeks were maintained after 52 weeks (Table 4) [50].

4.2 Asian/Japanese Patients

Results obtained with three SGLT2is commercialised in Japan (ipragliflozin, luseogliflozin and tofogliflozin) were almost similar to those previously reported in non-Asian/Caucasian patients with other SGLT2is. These observations argue for a class effect and tend to confirm the absence of disparity in the clinical efficacy and safety of SGLT2is between Asian and non-Asian patients with T2DM [57]. In a meta-analysis that included a total of 17 trials with Asian patients and 39 trials with non-Asian patients, comparison of the placebo-subtracted reductions in HbA1c showed that there was a minimal non-significant difference of 0.05% between ethnic groups. Similarly, comparisons of the body weight and SBP placebo-corrected reductions did not show a significant difference between Asian and non-Asian patients [58].

4.2.1 Ipragliflozin

After 24 weeks, ipragliflozin 50 mg improved glycaemic control and reduced body weight and SBP in Japanese patients with T2DM with mild-to-moderate CKD recruited in the LANTERN study. The reduction in HbA1c was positively correlated with renal function: a statistical 0.35% (p < 0.001) reduction in HbA1c was observed in patients with mild renal impairment, whereas only a modest non-significant trend (− 0.17%, p = 0.215) was noticed in patients with moderate renal impairment. In contrast, reductions in body weight and SBP were not smaller in patients with stage 3a CKD than in patients with stage 2 CKD (Table 3) [52].

4.2.2 Luseogliflozin

Glycosylated haemoglobin, FPG and body weight significantly decreased at week 24 in patients with T2DM with moderate renal impairment treated with luseogliflozin 2.5 mg/day vs placebo (Table 3). At 52 weeks, these reductions and lowering of SBP were significant with luseogliflozin (2.5 mg, up-titrated to 5 mg/day, if necessary, after 24 weeks) compared with placebo (Table 4) [53].

In a pooled analysis of four 52-week, phase III studies with luseogliflozin, patients with T2DM were stratified into three groups by baseline eGFR: ≥ 90 mL/min/1.73 m2, ≥ 60 to < 90 mL/min/1.73 m2 and ≥ 30 to < 60 mL/min/1.73 m2. At week 52, HbA1c, FPG and body weight were significantly decreased from baseline in all groups. However, the reductions in HbA1c and body weight were significantly smaller in patients with moderate impairment than in those with normal function; of note, the HbA1c-lowering efficacy was more reduced in patients with stage 3b CKD than in patients with stage 3a CKD, whereas a much lower difference concerned changes in body weight (Table 4) [53].

4.2.3 Tofogliflozin

A 24-week open-label study compared the effects of tofogliflozin 40 mg in patients with T2DM with normal renal function and moderate renal impairment [42]. Glycosylated haemoglobin, FPG and body weight changes from baseline to week 24 were − 0.65%, − 1.66 mmol/L and − 2.2 kg, respectively, in patients with eGFR ≥ 90 mL/min/1.73 m2 and − 0.23%, − 0.93 mmol/L and − 1.75 kg, respectively, in patients with eGFR ≥ 30 to < 60 mL/min/1.73 m2. As in other studies, whereas improvement of glycaemic parameters by tofogliflozin was attenuated as renal function declined, body weight reduction was not markedly affected by the renal status [42].

5 Cardiorenal Protection in Patients with Type 2 Diabetes Mellitus with Chronic Kidney Disease

5.1 Subgroup Analyses from Cardiovascular Outcome Trials

Meta-analyses [20, 59] of three CV outcome trials (EMPA-REG OUTCOME [60, 61], CANVAS [62,63,64] and DECLARE-TIMI 58 [65, 66]) demonstrated that both the reduction in MACEs (primary endpoint) and the reduction in composite hard renal outcomes (pre-specified secondary endpoint) were observed independently of the stage of CKD, across an eGFR ≥ 90 mL/min/1.73 m2 down to 30 mL/min/1.73 m2, without significant heterogeneity between subgroups separated by baseline eGFR levels (Table 5).

Table 5 Effects of sodium-glucose cotransporter type 2 inhibitors (SGLT2is) on composite cardiovascular and renal outcomes in placebo-controlled outcome trials according to the degree of severity of chronic kidney disease (CKD)

A post-hoc analysis in patients with prevalent CKD (defined as an eGFR < 60 mL/min/1.73 m2 and/or UACR > 300 mg/g) at baseline showed that empagliflozin, compared with placebo, significantly reduced the risk of CV death by 29%, the risk of all-cause mortality by 24%, the risk of hospitalisation for HF by 39% and the risk of all-cause hospitalisation by 19%. Of note, positive effects of empagliflozin on these outcomes were consistent across categories of eGFR and UACR at baseline [67].

Subgroup analyses according to ethnicity showed a global CV and renal protection almost comparable in non-Asian (Caucasian) and Asian patients with T2DM and high CV risk. Indeed, no significant heterogeneity (p for interaction > 0.05) was reported between patients with T2DM separated according to their race in EMPA-REG OUTCOME [60, 68], CANVAS [62,63,64] and DECLARE-TIMI 58 [65, 66], and this finding was also recently confirmed in DAPA-HF [69]. In EMPA-REG OUTCOME (1517 among 7020 treated patients, i.e. 26.1%, were Asian subjects), empagliflozin improved composite kidney outcomes (doubling of serum creatinine, initiation of renal replacement therapy or renal death: hazard ratio 0.48, 95% CI 0.25, 0.92) vs placebo in Asian patients with T2DM, consistent with the overall trial population findings (hazard ratio 0.54, 95% CI 0.40, 0.75) [68]. A similar improvement was also observed regarding the slowing of eGFR decline and the lowering of albuminuria in the two ethnic groups.

5.2 CREDENCE, A Dedicated Renal Outcome Trial

CREDENCE is the first RCT specifically dedicated to assess the efficacy of an SGLT2i to retard the progression of CKD. All patients with T2DM had an eGFR of 30 to < 90 mL/min/1.73 m2 and albuminuria (UACR > 300 to 5000 mg/g) at baseline and were already treated with blockers of the renin-angiotensin-aldosterone system. They were assigned to receive canagliflozin 100 mg or placebo and the study was interrupted after a median follow-up of 2.62 years. The primary outcome (a composite of ESRD, a doubling of the serum creatinine level, or death from renal or CV causes) occurred less frequently in the canagliflozin group than in the placebo group (hazard ratio 0.70, 95% CI 0.59, 0.82; p = 0.00001). The relative risk of the renal-specific composite (thus excluding CV death) was lower by 34% (p < 0.001), and the progression to ESRD was reduced by 32% (p = 0.002) with canagliflozin compared with placebo (Table 5). In addition, the canagliflozin group had a lower risk of MACE-3 points (− 20%; p = 0.01) and hospitalisation for HF (− 39 %; p < 0.001) [21]. No significant heterogeneity was detected when comparing the positive effects of canagliflozin on MACEs or composite renal outcomes according to baseline renal status, especially when comparing patients with T2DM with an eGFR of 30–59 mL/min/1.73 m2 and patients with an eGFR of 60–90 mL/min/1.73 m2 (Table 5). Subgroup analysis of the primary composite endpoint according to patient race, including 877 Asian patients vs 2931 white patients, did not show any significant interaction (p = 0.91) [21]. Subsequent analysis showed that the risks of the primary composite renal outcome and the composite of CV death or hospitalisation for HF were consistently reduced in both groups in primary and secondary CV prevention [70].

Another post-hoc exploratory analysis showed that canagliflozin reduces the risk of both CV and renal events in patients with T2DM and CKD without a significant interaction across the spectrum of baseline HbA1c levels [71]. As CREDENCE included patients with baseline HbA1c between 6.5% (48 mmol/mol) and 7.0% (53 mmol/mol), these results suggest that treatment with an SGLT2i of patients with mild-to-moderate CKD and albuminuria is warranted, even if blood glucose is well controlled, in agreement with previous observations that focused on MACEs in CV outcome trials [72].

5.3 Impact on Recent Guidelines

The clinical efficacy and safety as well as the positive effects on both CV and renal outcomes of SGLT2is in patients with T2DM and CKD are now well recognised [19, 37]. These positive findings markedly impacted guidelines for the management of T2DM at risk of CV disease or progressive CKD in 2018 [73, 74]. The recent results of new large prospective RCTs that demonstrated a positive effect on CV outcomes and renal outcomes in patients with T2DM at high CV risk and mild-to-moderate CKD resulted in a further major change in the last American Diabetes Association-European Association for the Study of Diabetes consensus report [23]. Indeed, the use of SGT2is is now recommended in patients with T2DM with CKD (eGFR 30 to ≤ 60 mL/min/1.73 m2 or UACR > 30 mg/g, particularly > 300 mg/g) to prevent the progression of CKD, hospitalisation for HF, MACE and CV death [23]. Of note, the prescription is recommended whatever the baseline level of HbA1c. A preferred place of SGLT2is in patients with T2DM at risk of progression of CKD is also mentioned in guidelines published by scientific societies of cardiology [22, 75] and nephrology [76]. Even if there is an increasing evidence supporting the prescription of an SGLT2i in patients with T2DM with CKD, independently of baseline HbA1c, a closed monitoring is recommended in these patients to avoid and early detect possible adverse events [24, 55, 77].

Both the efficacy and the safety of SGLT2is were shown to be comparable in Asian populations and in non-Asian/Caucasian people [57, 58]. These findings confirmed the place of SGLT2is in patients with T2DM with an Asian phenotype [78], including in patients with moderate CKD to prevent CV events and the progression of the renal disease [79].

6 Perspectives in Non-Diabetic Patients with Chronic Kidney Disease

Because the CV and renal protection observed in patients with T2DM appears independent of baseline HbA1c and reductions in HbA1c under treatment in EMPA-REG OUTCOME [80] and CREDENCE [71], it has been postulated that the positive effects both on hospitalisation for HF and progression of CKD might also be present in non-diabetic patients.

A first demonstration was given by the results of DAPA-HF, a RCT that recruited both diabetic and non-diabetic patients with HF and reduced left ventricular ejection [69]. In patients treated with dapagliflozin compared with those treated with placebo, a significant reduction in hospitalisation for HF has been reported in non-diabetic individuals similar to that observed in patients with T2DM. Of note, about 40% of patients enrolled in DAPA-HF had eGFR < 60 mL/min/1.73 m2 [69]. Other RCTs are currently investigating the efficacy of empagliflozin in non-diabetic patients with HF: a rather small trial (EMPA-TROPISM) [81] and a much larger trial (EMPEROR), which recruited patients with and without T2DM (about half in each group) and reduced [82] or preserved [83] left ventricular ejection fraction.

According to preliminary results in animal models, studies examining the impact of SGLT2is on markers of kidney disease in patients with non-diabetic CKD are needed [84]. Two large ongoing RCTs are investigating the effect of SGLT2is in non-diabetic patients with CKD: dapagliflozin in DAPA-CKD [85] and empagliflozin in EMPA-KIDNEY [86]. If the renoprotective effects of SGLT2is are confirmed in patients with CKD, irrespective of whether they have diabetes, these findings would bring a new perspective for this pharmacological class and lead to possible new indications in an increasing number of patients with mild-to-moderate CKD [87].

7 Conclusions

Drug exposure of all SGLT2is is moderately increased in patients with CKD, yet the increase is not linear with the reduction in eGFR; the increase in Cmax and AUC0–infinity is less than two-fold even in patients with severe CKD, thus no drug reduction is generally recommended at least in patients with mild-to-moderate CKD (except for canagliflozin). No PK differences were noticed between Asian and non-Asian patients with T2DM.

In both ethnic groups, the efficacy as assessed by the reduction in the changes of UGE, FPG and HbA1c declines with increasing severity of renal impairment, especially if eGFR is reduced below 45 mL/min/1.73 m2. Of note, the positive effects on body weight and SBP are still preserved. Nevertheless, the glucose-lowering efficacy of SGLT2is is almost comparable in patients with mild CKD (stage 2) and only moderately reduced in patients with stage 3 CKD (more in stage 3b than in stage 3a) as compared with patients with normal kidney function. In patients with severe CKD (stages 4–5), the use of SGLT2is is presently contra-indicated. Because of the decline of the glucose-lowering efficacy, the initiation of SGLT2i therapy was generally not allowed if eGFR is < 60 mL/min/1.73 m2 and it was classically considered that the treatment should be interrupted if eGFR falls < 45 mL/min/1.73 m2 according to existing prescribing information.

However, a recent major breakthrough was the demonstration that both CV protection, including a reduction in CV death, all-cause mortality and hospitalisation for HF, and renal protection, including the progression to ESRD and renal death, are also observed in patients with T2DM with moderate CKD. The amplitude and the statistical significance of these effects were comparable in patients with stage 3 CKD (with an eGFR between 30 and 59 mL/min/1.73 m2) to patients with mild renal impairment or patients with normal renal function. These results from large prospective CV and renal outcome RCTs had a major impact on the most recent guidelines and position statements. Obviously, the use of SGLT2is in patients with CKD will increase in the next few years and this trend might be further triggered after the publication of results from two ongoing RCTs specifically dedicated to patients with CKD, DAPA-CKD and EMPA-Kidney.