FormalPara Key Summary Points

Novel antidiabetic drugs showed renoprotective effects in patients with type 2 diabetes (T2D) based on hard renal composite endpoints in clinical trials

In terms of risk reduction of the hard renal composite endpoint, sodium-glucose co-transporter-2 (SGLT2) inhibitors showed the greatest benefit versus placebo in patients with T2D, followed by glucagon-like peptide-1 receptor agonists (GLP-1 RAs), while dipeptidyl peptidase-4 (DPP-4) inhibitors did not show a benefit

Unlike for hard renal endpoints, head-to-head comparisons of the efficacies of novel antidiabetic drugs on urinary albumin-to-creatine ratio (UACR) or albuminuria outcomes remain limited despite the recommended routine monitoring of these renal disease risk markers in T2D clinical practice

The qualitative synthesis of available evidence revealed that SGLT2 inhibitors consistently improved UACR and albuminuria outcomes in patients with T2D, while evidence on the benefit of GLP-1 RAs and DPP-4 inhibitors was relatively limited and showed potential drug-specific effects within the classes

Continuous management of albuminuria with specific novel antidiabetic drugs, particularly SGLT2 inhibitors, together with routine UACR monitoring, could delay the onset and progression of renal disease in patients with T2D

Introduction

Globally, approximately 537 million adults lived with diabetes in 2021, with most having type 2 diabetes (T2D) [1]. Chronic kidney disease (CKD) is a major complication of T2D, with a reported prevalence of 24–50% in this population [2,3,4,5]. CKD manifests clinically as persistent (> 3 months) albuminuria and/or impaired estimated glomerular filtration rate (eGFR) that eventually progresses to end-stage kidney disease (ESKD) [6] and is associated with increased risk of cardiovascular (CV) and all-cause mortality in patients with T2D [7, 8].

The rate of CKD progression in patients with T2D is highly variable, being characterized by different rate trajectories of renal function decline measured by the eGFR [9]. Albuminuria, indicated by elevated urinary albumin-to-creatine ratio (UACR), is a major risk factor for rapid renal function decline and progression to ESKD [10,11,12,13,14,15,16,17,18,19,20,21]. Albuminuria or increased UACR is also an important predictor of adverse outcomes, including CV events and mortality [10, 11, 16, 21]. The routine screening of UACR, together with eGFR, from the time of T2D diagnosis is recommended by clinical guidelines to detect CKD and monitor its progression [6, 22, 23].

The novel antidiabetic drugs, sodium-glucose co-transporter-2 (SGLT2) inhibitors and glucagon-like peptide-1 receptor agonists (GLP-1 RAs), have been shown to confer CV- and renoprotection in patients with T2D in CV and renal outcomes-focused trials and are recommended by clinical guidelines for the treatment of patients with T2D and risks for CV disease and/or CKD [23,24,25,26]. Dipeptidyl peptidase-4 (DPP-4) inhibitors showed a neutral effect in reducing the risks of renal and CV events [27, 28]. A recent network meta-analysis of 16 randomized, placebo-controlled trials showed that treatment with SGLT2 inhibitors resulted in the greatest risk reduction for CV and renal events, including the hard renal composite endpoint (doubling of serum creatinine, ≥ 40% decline in eGFR, development of ESKD and death due to renal causes), followed by GLP-1 RAs, while DPP-4 inhibitors did not improve these outcomes compared with placebo [29].

Although hard renal endpoints were frequently used in clinical trials to assess the renoprotective effects of novel antidiabetic drugs, limitations for their use include the need for a large study cohort and a long follow-up period [30]. UACR or albuminuria, being routinely determined in clinical practice, as well as an important predictor of renal function decline and clinical outcomes, is increasingly recognized as a potential short-term surrogate renal endpoint for the evaluation of novel antidiabetic drugs in T2D clinical trials [30,31,32]. To our knowledge, head-to-head comparison studies on the efficacy of novel antidiabetic drugs in improving UACR and albuminuria outcomes remain limited [33, 34]. We aimed to compare the efficacy of novel antidiabetic drug classes, namely SGLT2 inhibitors, GLP-1 RAs and DPP-4 inhibitors, on UACR and albuminuria outcomes in patients with T2D based on available clinical data.

Methods

Search Strategy

We conducted a systematic search of the MEDLINE database (via PubMed) to identify relevant randomized controlled trials (RCTs) published from inception to 5 December 2022. Only English-language articles were retrieved; we did not limit the search by text availability, article type or any other trial characteristics. The following search terms were used: (“sodium-glucose transporter 2 inhibitors” [Mesh] OR “dipeptidyl-peptidase IV inhibitors” [Mesh] OR (“glucagon-like peptide 1” [Mesh] OR “glucagon-like peptide-1 receptor” [Mesh])) AND (“diabetes mellitus, type 2” [Mesh]) AND (UACR OR (“Albuminuria” [Mesh]) AND ((randomized controlled) OR (randomized controlled trial)). To ensure completeness of the search, different entry terms for each aforementioned phrase or word, as well as the individual drugs in each class were included in the search string (Supplementary Table S1). We also manually screened the references of the systematic reviews, meta-analyses and pooled analyses shortlisted from the database search to identify additional relevant RCTs that were not initially captured by the search string.

Study Selection

The retrieved studies were included if they met the following criteria, based on the patient, intervention, comparison, and outcome (PICO) framework: (1) Phase 3 or 4 RCT; (2) enrolled adult patients with T2D regardless of concomitant CV disease, renal impairment and/or albuminuria; (3) evaluated SGLT2 inhibitors, GLP-1 RAs or DPP-4 inhibitors versus placebo; (4) reported UACR and/or albuminuria outcomes.

The articles retrieved from the database search were initially screened for eligibility based on the study type, patient population, intervention and comparator in their titles and abstracts. The shortlisted articles then underwent full text screening for all eligibility criteria specified, including study outcomes. For RCTs determined to be eligible, we additionally sourced for their primary, secondary, post hoc and/or exploratory analyses that were not initially captured by the database search and assessed the eligibility of each article for inclusion.

Data Extraction

The data extracted for each eligible study included patient characteristics, dosage of the intervention, study duration or median follow-up period (categorized as short term [≤ 1 year] and long term [≥ 2 years]) and outcomes of interest. Data extraction was conducted using a standardized Excel spreadsheet by one researcher and independently reviewed by two others.

Outcomes

Outcomes of interest were (1) changes in UACR from baseline over time (continuous) and (2) changes in albuminuria stages (categorical), namely normoalbuminuria (UACR < 30 mg/g), microalbuminuria (UACR 30–300 mg/g) and macroalbuminuria (UACR > 300 mg/g) [6, 35]. Albuminuria onset was defined as the shift to micro- or macroalbuminuria in patients with normoalbuminuria at baseline; albuminuria progression as the shift to macroalbuminuria in patients with microalbuminuria at baseline; and albuminuria regression as the shift to normo- or microalbuminuria in patients with macroalbuminuria at baseline or the shift to normoalbuminuria in those with microalbuminuria at baseline.

Due to heterogeneity of the trial characteristics (e.g., study duration) and the outcome definitions or measures, a meta-analysis was not considered meaningful. Instead, albuminuria outcomes were qualitatively synthesized based on study duration. Analyses of albuminuria outcomes in subgroups defined by baseline eGFR or albuminuria status were also extracted from the selected studies and summarized.

Risk of Bias Assessment

The risk of bias for included studies was assessed by two independent reviewers, separately for continuous and categorical albuminuria outcomes using the Cochrane Risk of Bias tool.

Compliance with Ethics Guidelines

This systematic review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [36]. This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Results

Study Selection

Among 211 records identified from the database search, 75 were shortlisted from title and abstract screening (Supplementary Fig. S1). An additional 25 records not identified from the database search were included for completeness. Hence, a total of 100 records underwent full text screening, from which 53 were deemed eligible. Twelve of the selected records were further excluded during data extraction. Finally, data were extracted from 41 articles that reported on 26 RCTs.

Among the 26 selected RCTs, 16 had enrolled patients with T2D regardless of baseline renal impairment and/or albuminuria status, while 10 enrolled patients with concomitant CKD. To ensure the homogeneity of the study population reported in this systematic review, herein, only the results in patients with T2D regardless of baseline CKD status (27 articles from 16 RCTs) were qualitatively synthesized. The risk of bias assessment of the included studies is presented in Supplementary Fig. S2.

Study Characteristics

The characteristics of the studies included in this qualitative synthesis are summarized in Table 1. Patients had concomitant CV disease or CV risk factors or were overweight/obese. Eleven studies involved long-term follow-up (median ≥ 2 years): three on SGLT2 inhibitors (CANVAS PROGRAM [37,38,39,40,41], DECLARE-TIMI-58 [42] and EMPA-REG-OUTCOME [43,44,45,46]), five on GLP-1 RAs (AMPLITUDE-O [47], ELIXA [48, 49], LEADER [50, 51], REWIND [52] and SUSTAIN-6 [53, 54]), and three on DPP-4 inhibitors (CARMELINA [55,56,57], SAVOR-TIMI-53 [58] and TECOS [59]), while five studies involved short-term follow-up (≤ 1 year): two on SGLT2 inhibitors (DECREASE [60] and EMPEROR-Pooled [61]), four on GLP-1 RAs (DECREASE [60], EXSCEL [62], SAFEGUARD [63] and SCALE [64]), and one on a DPP-4 inhibitor (SAFEGUARD [63]); the DECREASE and SAFEGUARD studies evaluated two different drug classes each in separate arms [60, 63]. Nine studies, all of which were of long-term duration, investigated albuminuria outcomes in subgroups defined by baseline albuminuria and/or eGFR status (Table 1).

Table 1 Characteristics of studies in the overall T2D population included in the systematic review
Full size table

Effects of Novel Antidiabetic Drugs on Changes in UACR

Overall T2D Population

The effects of novel antidiabetic drugs on changes in UACR were mainly reported in studies with long-term follow-up (Table 2). SGLT2 inhibitors (canagliflozin and empagliflozin) decreased UACR by 19–22% relative to placebo (P < 0.05 for all studies) [37, 40, 44]. GLP-1 RAs decreased UACR by 17–33% (P < 0.05 for all studies) [47, 50, 52, 54]. Lixisenatide also showed a significant benefit in slowing the increase in UACR versus placebo (P = 0.004) [49]. The effect of different DPP-4 inhibitors on UACR varied greatly. Saxagliptin showed the greatest magnitude of UACR reduction (− 34.3 mg/g; P < 0.004) relative to placebo, followed by sitagliptin at − 0.18 mg/g (P = 0.031) [58, 59]. Conversely, no meaningful difference in UACR (− 0.05 mg/g) was seen with linagliptin versus placebo [55].

Table 2 Changes in UACR in patients with T2D treated with novel antidiabetic drugs
Full size table

Data on the short-term effects of novel antidiabetic drugs on UACR outcomes were limited. At 52 weeks, empagliflozin decreased UACR by 19% compared with placebo [61]. At 56 weeks, liraglutide (3.0 mg once daily [qd]) decreased UACR by 20% versus placebo, although a statistically significant difference was not observed at the 1.8 mg qd dose [64]. Nevertheless, in a post hoc analysis of the LEADER study at 1 year, UACR was decreased by 15% (95% confidence interval [CI] 13, 18) in patients treated with liraglutide (0.6–1.8 mg qd) and increased by 10% (95% CI 7, 14) in those receiving placebo [51]. At a shorter follow-up duration of 16 weeks, the DECREASE study showed that dapagliflozin and exenatide resulted in higher, but statistically nonsignificant, percentage decreases in UACR compared with placebo [60]. In the 12-week SAFEGUARD study, UACR decreased by 24% and 32% with liraglutide and sitagliptin, respectively, versus placebo, although the differences were not statistically significant [63].

Subgroup Analysis by Baseline Albuminuria Status

Two SGLT2 inhibitor studies (CANVAS PROGRAM and EMPA-REG OUTCOME) and two GLP-1 RA studies (ELIXA and LEADER) reported UACR changes by baseline albuminuria status during long-term follow-up; no relevant DPP-4 inhibitor study was identified (Supplementary Table S2). SGLT2 inhibitors lowered UACR compared with placebo regardless of baseline albuminuria status (P < 0.05 for all albuminuria subgroups in both CANVAS PROGRAM and EMPA-REG OUTCOME), although the magnitude of the treatment effect was observed to be greater in patients with microalbuminuria (− 34 to − 30%) or macroalbuminuria (− 36 to − 32%) versus those without (− 12 to − 9%) [40, 43]. Conversely, GLP-1 RAs showed varying effects on UACR reduction by baseline albuminuria status. Liraglutide decreased UACR relative to placebo by 14%, 24% and 13% in patients with normo-, micro- and macroalbuminuria, respectively, although the between-group difference only reached statistical significance in patients with normo- and microalbuminuria [50]. Compared with placebo, lixisenatide significantly decreased UACR by 39.18% (P = 0.0070) in the macroalbuminuria subgroup; the reduction in UACR was smaller and nonsignificant in the microalbuminuria subgroup (− 21.10%; P = 0.0502) and marginal in the normoalbuminuria subgroup (− 1.69%; P = 0.7398) [48].

Subgroup Analysis by Baseline eGFR Status

Five studies investigated the effects of the three novel antidiabetic drug classes on changes in UACR according to baseline eGFR (< 60 and ≥ 60 ml/min/1.73 m2) over long-term follow-up (Supplementary Table S3). The SGLT2 inhibitor, canagliflozin, led to UACR reduction in both baseline eGFR subgroups. UACR was reduced by 13 to 26% in patients with eGFR > 30 to < 60 ml/min/1.73 m2 and by 17% in those with eGFR ≥ 60 ml/min/1.73 m2 [41]. The effect of a GLP-1 RA, liraglutide, on UACR changes differed by baseline eGFR status as reported in the LEADER study. Liraglutide decreased UACR by 10–24% versus placebo across all eGFR subgroups, although statistical significance was only observed among patients with eGFR ≥ 60 ml/min/1.73 m2 [50]. Subgroup analyses by baseline eGFR status revealed varying results with DPP-4 inhibitors. The magnitudes of mean UACR change from baseline with saxagliptin were greater in the eGFR < 30 and 30–50 ml/min/1.73 m2 subgroups at − 245.2 mg/g and − 105.0 mg/g, respectively, although in the former subgroup it did not reach statistical significance; a smaller but significant improvement in UACR (− 19.3 mg/g; P = 0.033) was observed in the eGFR > 50 ml/min/1.73 m2 subgroup [58]. In contrast, linagliptin and sitagliptin did not lead to statistically significant improvements in UACR versus placebo across any of the eGFR subgroups [55, 59]. The effect on UACR changes in the eGFR < 30 ml/min/1.73 m2 subgroup was reported only for GLP-1 RAs and DPP-4 inhibitors, although no statistically significant benefit was observed [50, 55, 58].

Effects of Novel Antidiabetic Drugs on Changes in Albuminuria Categories

Overall T2D Population (by Baseline Albuminuria Status)

Among the selected T2D studies that evaluated the effect of novel antidiabetic drugs on changes in albuminuria status, all except one involved long-term follow-up (Tables 3, 4, 5). Albuminuria onset was reported for SGLT2 inhibitors and GLP-1 RAs, but not DPP-4 inhibitors (Table 3). Compared with placebo, SGLT2 inhibitors significantly decreased the risk for albuminuria onset by 16–20%, that for microalbuminuria onset by 20–21% and that for macroalbuminuria onset by 21–42% (P < 0.05 for all studies) [40, 42, 43]. GLP-1 RA studies mainly reported effects on macroalbuminuria onset. Among GLP-1 RAs, dulaglutide, liraglutide and semaglutide significantly decreased the risk for macroalbuminuria onset by 22–46% compared with placebo (P < 0.05 for all studies) [50, 52, 53]. The decrease in risk for macroalbuminuria onset with lixisenatide was, however, not statistically significant [48]. EXSCEL was the only short-term study identified where exenatide significantly reduced the proportion of patients with albuminuria onset versus placebo (4.0% versus 3.0%; P = 0.03) at 6 months; this was driven by a decrease in the incidence of microalbuminuria onset (2.9% versus 2.0%; P = 0.02) rather than macroalbuminuria onset (1.1% versus 1.0%; P = 0.79) [62].

Table 3 Albuminuria onset (from baseline normoalbuminuria) in patients with T2D treated with novel antidiabetic drugs
Full size table
Table 4 Albuminuria progression (from baseline normo- and/or microalbuminuria) in patients with T2D treated with novel antidiabetic drugs over long-term follow-up
Full size table
Table 5 Albuminuria regression (from baseline micro- and/or macoalbuminuria) in patients with T2D treated with novel antidiabetic drugs over long-term follow-up
Full size table

Albuminuria progression was defined differently across studies, for example, some studies considered deterioration in albuminuria categories among patients with normo- or microalbuminuria, i.e., combined albuminuria onset and progression (Table 4). Among patients with baseline microalbuminuria, SGLT2 inhibitors significantly lowered the risk for albuminuria progression compared with placebo [38, 42, 46]. The greatest risk reduction in albuminuria progression of approximately 50% (P < 0.0001) was observed with dapagliflozin, followed by empagliflozin and canagliflozin, which brought about risk reductions of 38% (P < 0.001) and 27% (P < 0.05), respectively [38, 42, 46]. Furthermore, the risk of combined albuminuria onset (in patients with baseline normoalbuminuria) and progression (in patients with baseline microalbuminuria) was significantly reduced with SGLT2 inhibitors (dapagliflozin: hazard ratio [HR] 0.82 [95% CI 0.77, 0.88]; empagliflozin: odds ratio 0.67 [95% CI 0.59, 0.77]) versus placebo [42, 43]. The risk of progression to macroalbuminuria in patients with baseline normoalbuminuria and microalbuminuria (i.e., combined macroalbuminuria onset and albuminuria progression) was also significantly decreased by 55% (95% CI 44, 63) with empagliflozin versus placebo [43, 45]. The GLP-1 RA, dulaglutide, significantly reduced the risk for albuminuria progression by 24% (95% CI 11, 35) compared with placebo [52]. While efpeglenatide led to lower rates of combined macroalbuminuria onset and albuminuria progression (32% risk reduction) [47], lixisenatide did not result in a statistically significant difference on this outcome (P = 0.0908) [48]. A DPP-4 inhibitor study revealed that linagliptin significantly reduced the risk of combined albuminuria onset and progression by 14% (P = 0.003) compared with placebo [56].

The effect of novel antidiabetic drugs on albuminuria regression, including microalbuminuria and macroalbuminuria regression, is shown in Table 5. Compared with placebo, SGLT2 inhibitors were significantly associated with albuminuria regression (HR 1.70 [95% CI 1.51, 1.91] for albuminuria regression with canagliflozin; HR 1.46 [95% CI 1.31, 1.62] and 1.82 [95% CI 1.51, 2.20] for microalbuminuria and macroalbuminuria regression with dapagliflozin, respectively; HR 1.43 [95% CI 1.22, 1.67] and 1.82 [95% CI 1.40, 2.37] for microalbuminuria and macroalbuminuria regression with empagliflozin, respectively) [38, 42, 43]. In contrast, available studies did not show statistically significant macroalbuminuria regression with the GLP-1 RA, lixisenatide (P = 0.4086), and the DPP-4 inhibitor, linagliptin (P = 0.31), compared with placebo [48, 57].

Subgroup Analysis by Baseline eGFR Status

Subgroup analyses of changes in albuminuria categories by baseline eGFR status were available for a GLP-1 RA and a DPP-4 inhibitor in two studies with long-term follow-up (Supplementary Table S4). Compared with placebo, the GLP-1 RA, dulaglutide, significantly decreased the risk of combined albuminuria onset and progression (HR 0.70 [95% CI 0.59, 0.81]) in the eGFR ≥ 60 ml/min/1.73 m2 group, while a smaller nonsignificant decrease in risk (HR 0.91 [95% CI 0.73, 1.13]) was observed among patients with eGFR < 60 ml/min/1.73 m2 [52]. Linagliptin did not show a statistically significant association with albuminuria regression versus placebo across all eGFR subgroups [55].

Discussion

To our knowledge, this is, to date, the most comprehensive qualitative evidence synthesis of albuminuria outcomes (both continuous and categorical) in patients with T2D treated with novel antidiabetic drugs. A previous meta-analysis of 26 studies (N = 14,929) compared the effect of novel antidiabetic drugs on percentage changes in albuminuria as a continuous outcome (measured by either urinary albumin excretion or UACR) versus placebo or other conventional glucose-lowering drugs [34]. Another meta-analysis of 19 studies (N = 140,851) compared the effect of the novel antidiabetic drugs on albuminuria progression and regression as categorical outcomes [33]. Results from the present qualitative synthesis and prior meta-analyses provide evidence on the comparative efficacies of novel antidiabetic drugs in improving UACR and albuminuria outcomes, beyond the risk reduction for hard renal endpoints, which will further inform the use of these interventions for the prevention and management of CKD in patients with T2D.

Effects of Novel Antidiabetic Drugs on Changes in UACR

Changes in UACR over time with novel antidiabetic drug treatment were reported using different outcome measures (as absolute mean, absolute geometric mean and percentage) across the selected studies, thus precluding a meaningful meta-analysis. SGLT2 inhibitors and GLP-1 RAs showed benefit in improving UACR outcomes in patients with T2D during long-term follow-up, while varying effects were observed for DPP-4 inhibitors [37, 40, 44, 47, 49, 50, 52, 54, 55, 58, 59]. The short-term effects of novel antidiabetic drugs on changes in UACR have not been adequately studied, being reported in only five trials [51, 60, 61, 63, 64]. At around 1 year, empagliflozin and liraglutide showed benefit in reducing UACR versus placebo [51, 61, 64]. The nonsignificant benefit of novel antidiabetic drugs in improving UACR outcomes at 12–16 weeks of the DECREASE and SAFEGUARD trials may be due to their relatively small sample sizes [60, 63]. Hence, the benefit of novel antidiabetic drugs in improving UACR outcomes in the shorter term (≤ 1 year) warrants further investigation.

When analyzed by baseline albuminuria categories, SGLT2 inhibitors were the only class of novel antidiabetic drugs that led to significantly decreased UACR versus placebo across all subgroups [39, 40, 43]. The treatment effect of SGLT2 inhibitors on UACR was observed to be greater in patients with albuminuria than in those without [39, 40, 43]. These findings were corroborated by studies in patients with T2D and concomitant albuminuria (CREDENCE, DAPA-CKD and Study MB102029), where UACR was consistently shown to be decreased with SGLT2 inhibitors versus placebo [65,66,67]. In contrast, GLP-1 RAs showed varying effects on UACR in different baseline albuminuria subgroups, although this may be due to the small subgroup sample sizes and warrants further investigation [48, 50]. No subgroup analysis of UACR changes with DPP-4 inhibitors by baseline albuminuria status has been identified to date.

When analyzed by baseline eGFR, canagliflozin showed statistically significant UACR improvement across all subgroups [41]. Consistently, findings from CREDENCE, DAPA-CKD and Study MB102029 showed that SGLT2 inhibitors improved UACR versus placebo among enrolled patients with T2D and concomitant renal impairment (eGFR > 25 to < 90 ml/min/1.73 m2) [65, 67,68,69]. Furthermore, similar to the findings in CANVAS PROGRAM, a greater magnitude of UACR reduction was observed in the subgroup with eGFR < 60 versus ≥ 60 ml/min/1.73 m2 in CREDENCE [41, 68]. Despite showing a benefit in the overall study cohort, the GLP-1 RA, liraglutide, and the DPP-4 inhibitor, saxagliptin, did not lead to statistically significant UACR reductions in subgroups with eGFR < 60 and < 30 ml/min/1.73 m2, respectively [50, 58]. Small subgroup sample sizes, which may not be sufficiently powered to detect treatment effects on UACR cannot be ruled out and necessitates further investigation. Similarly, the subtle treatment benefit observed with sitagliptin in the overall study cohort lost statistical significance in the subgroup analysis, likely because of the reduced sample sizes [59].

Our observations on the comparative efficacy of novel antidiabetic drugs on UACR changes in the overall T2D population corroborated the conclusions of an earlier meta-analysis by Luo et al. [34]. They determined that both SGLT2 inhibitors and GLP-1 RAs significantly reduced UACR compared with other conventional therapies or placebo and that SGLT2 inhibitors showed a greater magnitude of UACR reduction than GLP-1 RAs (weighted mean difference [WMD] − 26.23% versus − 13.85%) [34]. While all SGLT2 inhibitors studied led to UACR reduction, a statistically significant effect was not observed with lixisenatide among GLP-1 RAs [34]. UACR reduction with DPP-4 inhibitors (WMD − 6.19%) compared with other conventional therapies or placebo was not statistically significant [34]. Furthermore, subgroup analysis from Luo et al. revealed that SGLT2 inhibitors led to UACR reduction regardless of baseline albuminuria status, with a greater effect in patients with albuminuria versus those without, consistent with the observations from the present evidence synthesis [34]. GLP-1 RAs also decreased UACR across all baseline albuminuria subgroups with the greatest benefit being observed among patients with macroalbuminuria [34]. SGLT2 inhibitors decreased UACR in patients with eGFR ≥ 60 ml/min/1.73 m2, with a greater reduction in patients with eGFR > 60 to 90 ml/min/1.73 m2 than in those with eGFR > 90 ml/min/1.73 m2 (− 33.36% versus − 9.40%, P < 0.05 for both). However, no statistically significant treatment benefit was seen in patients with eGFR 30–60 ml/min/1.73 m2, although a wide 95% CI was noted (–35.59% [95% CI –83.11, 11.94]) [34]. GLP-1 RAs significantly decreased UACR only in patients with eGFR > 60 to 90 ml/min/1.73 m2, with numerical improvements observed for those with eGFR 30–60 and > 90 ml/min/1.73 m2 [34]. Notably, although UACR outcomes with novel antidiabetic drug treatment reported in Luo et al. were generally consistent with those observed in the present systematic review, the meta-analysis by Luo et al. evaluated a different set of trials [34]. First, recent CV and renal outcomes-focused trials were included in the present systematic review while several glycemic control trials were included in the prior meta-analysis by Luo et al. Second, the present systematic review focused on studies in the overall T2D population and excluded those that enrolled only patients with T2D and concomitant CKD for a more generalizable and uniform study cohort, unlike in the meta-analysis by Luo et al. Last, in addition to placebo-controlled RCTs, those evaluating novel antidiabetic drugs against conventional therapies were included in the meta-analysis by Luo et al.

Effects of Novel Antidiabetic Drugs on Changes in Albuminuria Categories

Albuminuria onset, progression and regression were defined differently across available studies; standardized definitions of these outcomes in future studies will allow comparisons among novel antidiabetic drug classes via a meta-analysis. Changes in albuminuria categories with novel antidiabetic drugs were mainly reported in studies with long-term follow-up. EXSCEL was the only study shortlisted from the systematic search that had a time point of < 1 year, showing reduced microalbuminuria onset risk with exenatide versus placebo; effects on albuminuria progression and regression over short-term follow-up were not reported [62].

Compared with placebo, SGLT2 inhibitors showed significant benefit in reducing the risks for albuminuria onset as well as albuminuria progression [38, 40, 42, 43, 45, 46]. In contrast to SGLT2 inhibitor studies, GLP-1 RA and DPP4 inhibitor studies used varying outcome definitions besides those defined in this systematic review, which included microalbuminuria onset, macroalbuminuria onset, deterioration in albuminuria categories among patients with normo- and microalbuminuria (i.e., combined albuminuria onset and progression) and incident macroalbuminuria in patients with normo- and microalbuminuria (i.e., combined macroalbuminuria onset and albuminuria progression). Among GLP-1 RAs, lixisenatide failed to show a statistically significant benefit in reducing the risk for deterioration in UACR categories [48], which may indicate a drug-specific effect within the class. Liraglutide, dulaglutide and semaglutide were shown to reduce the risks for micro- or macroalbuminuria onset [50, 52, 53]. The risk for albuminuria progression was reduced with dulaglutide [52] and that for combined macroalbuminuria onset and albuminuria progression was reduced with efpeglenatide [47]. Available data for DPP-4 inhibitors remained limited with only one relevant study (CARMELINA) identified that showed the reduced risk for combined albuminuria onset and progression with linagliptin [56].

SGLT2 inhibitors were associated with albuminuria regression (including that for micro- and macroalbuminuria) [38, 42, 43, 45]. Consistently, the DAPA-CKD study, which recruited patients with T2D and albuminuria (UACR between 200 and 5000 mg/g), showed that dapagliflozin was associated with reduced albuminuria progression (HR 0.39 [95% CI 0.29, 0.51]) and increased macroalbuminuria regression (HR 2.06 [95% CI 1.78, 2.39]) [67]. In contrast, the limited evidence available did not reveal a significant benefit of a GLP-1 RA (lixisenatide) and a DPP-4 inhibitor (linagliptin) in promoting macroalbuminuria regression; there were no reports on the effect of these two drug classes on albuminuria regression or microalbuminuria regression [48, 57].

Reports on the effect of novel antidiabetic drugs on categorical changes in albuminuria by baseline eGFR status were scarce. The reduction in the risk for combined albuminuria onset and progression with dulaglutide was statistically significant only in patients with eGFR ≥ 60 ml/min/1.73 m2, which was consistent with findings from the eGFR subgroup analysis of UACR changes with another GLP-1 RA, liraglutide [50, 52]. A DPP-4 inhibitor, linagliptin, was not associated with albuminuria regression across all baseline eGFR subgroups, which corroborated its lack of effect on UACR and albuminuria changes in the overall study cohort [55, 57]. Although no eGFR subgroup analyses of albuminuria regression with SGLT2 inhibitors were identified, the DAPA-CKD trial showed that dapagliflozin induced macroalbuminuria regression (HR 2.06 [95% CI 1.78, 2.39]) in enrolled patients with T2D and impaired renal function (eGFR >25 to <75 mL/min/1.73m2) [67].

The efficacy of novel antidiabetic drugs on albuminuria progression and regression was also compared in a meta-analysis by Chewcharat et al., where all three novel antidiabetic drug classes significantly decreased the risk for albuminuria progression compared with placebo (by 30% for SGLT2 inhibitors, 24% for GLP-1 RAs and 11% for DPP-4 inhibitors) [33]. Furthermore, higher HRs for albuminuria regression were observed with SGLT2 inhibitors (HR 1.73 [95% CI 1.54, 1.94]; P < 0.001) and DPP-4 inhibitors (HR 1.20 [95% CI 1.02, 1.40; P = 0.027) compared with placebo [33]. Comparatively, the present qualitative review revealed limited evidence and benefit of GLP-1 RAs and DPP-4 inhibitors in preventing albuminuria progression or promoting albuminuria regression. This difference can be attributed to the inclusion of a different set of studies; the present review compared studies with standard defined outcomes in the overall T2D population over long-term follow-up durations (≥ 2 years). Furthermore, a meta-analysis captures the statistical combination of results across the different studies, which may not reflect outcomes of the individual studies as reported in the present qualitative review. GLP-1 RA studies analyzed in Chewcharat et al. that were excluded for the evaluation of albuminuria progression in the present review were SAFEGUARD, LEADER and EXSCEL; DPP-4 inhibitor studies that were excluded were TECOS, SAVOR-TIMI-53, SAFEGUARD and Cooper et al. [33]. Reasons for the exclusion of these studies were differences in study type, population, follow-up duration and outcome definition [50, 51, 58, 59, 62, 63, 70, 71]. The effect of SGLT2 inhibitors in inducing albuminuria regression was based on the CANVAS PROGRAM and Study MB102029 in the meta-analysis by Chewcharat et al. [33]. We identified additional studies, namely DECLARE-TIMI-58 and EMPA-REG-OUTCOME, which supported the same conclusion [42, 43, 45]. The benefit of DPP-4 inhibitors in promoting albuminuria regression, as reported by Chewcharat et al., was based on MARLINA-T2D and SAVOR-TIMI-53 [33]. However, albuminuria regression results from these DPP-4 inhibitor studies were not extracted for comparison in this review because MARLINA-T2D specifically enrolled patients with T2D and concomitant CKD and had a short-term (24 weeks) follow-up duration, and SAVOR-TIMI-53 only presented the number of patients in each UACR category at baseline and end of treatment, without the HRs and 95% CIs for changes in UACR categories [58, 72].

Limitations

A limitation lies in the heterogeneous definitions and measures of the UACR and albuminuria outcomes in the selected RCTs, which did not allow for a meaningful meta-analysis. The differences in the magnitudes of treatment effect among the novel antidiabetic drug classes cannot be quantitatively compared. Furthermore, changes in UACR and albuminuria were post hoc and/or exploratory outcomes in some of the included studies. Post hoc outcomes may be biased because of the possibility of data selection after unblinding, as indicated in the risk of bias analyses; exploratory outcomes warrant confirmatory studies; and those outcomes reported without accompanying P values or 95% CIs cannot be evaluated in terms of statistical significance.

Conclusions

SGLT2 inhibitors decreased UACR, reduced the risk for albuminuria onset and progression and promoted albuminuria regression in patients with T2D during long-term follow-up. UACR and albuminuria outcomes with GLP-1 RAs and DPP-4 inhibitors warrant further investigation, as potential drug-specific effects were observed within each class. Available evidence did not show a benefit for GLP-1 RAs and DPP-4 inhibitors in inducing albuminuria regression. Furthermore, SGLT2 inhibitors, but not GLP-1 RAs and DPP-4 inhibitors, potentially improved UACR and albuminuria outcomes in patients with T2D regardless of baseline eGFR and albuminuria status. Short-term (≤ 1 year) effects of novel antidiabetic drugs on changes in UACR and albuminuria remained limited, necessitating further studies. Standardized definitions of UACR and albuminuria outcomes across future studies will allow for the quantitative comparison of treatment effect among novel antidiabetic drug classes in a meta-analysis.

Continuous management of albuminuria with novel antidiabetic drugs may delay CKD onset and slow its progression in patients with T2D. The routine monitoring of UACR or albuminuria changes in clinical practice can inform outcomes with novel antidiabetic drug treatment. The present evidence review suggests that among the novel antidiabetic drugs, SGLT2 inhibitors consistently improved UACR and albuminuria outcomes in patients with T2D, with evidence that continuous treatment could bring about benefit over the long term.